chiropractic care and risk for acute lumbar disc ...©... · chapter 1 general introduction and...
TRANSCRIPT
Chiropractic Care and Risk For Acute Lumbar Disc Herniation: A Mixed Methods Approach
by
Cesar Alberto Hincapié
A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy
Dalla Lana School of Public Health Division of Epidemiology
University of Toronto
© Copyright by Cesar Alberto Hincapié, 2015
ii
Chiropractic Care and Risk For Acute Lumbar Disc Herniation:
A Mixed Methods Approach
Cesar Alberto Hincapié
Doctor of Philosophy
Dalla Lana School of Public Health Division of Epidemiology
University of Toronto
2015
Abstract
Objective: To advance knowledge of the risk for acute lumbar disc herniation (LDH)
following chiropractic care. The objectives were to: (1) synthesize current evidence on
the incidence and determinants of LDH with radiculopathy in adults; (2) determine
clinicians’ beliefs regarding the risk for acute LDH associated with chiropractic spinal
manipulation treatment (SMT); and, (3) investigate associations between both
chiropractic and primary care physician (PCP) care, and acute LDH with incident early
surgery.
Methods: (1) Systematic review of the literature. (2) Belief elicitation interviews with 47
chiropractors, family physicians and spine surgeons. Probability distributions for the
relative risk (RR) for acute LDH associated with chiropractic SMT were derived. (3) Self-
controlled case series study using Ontario administrative health data. Incidence ratios
for acute LDH with early surgery in exposed periods after chiropractic visits relative to
unexposed periods were estimated within individuals, and compared with incidence
ratios following PCP visits.
iii
Results: (1) The annual incidence of LDH with radiculopathy ranges between 0.1% and
10%, and varies with the case definition. Risk factors included individual, behavioural
and work-related characteristics. (2) In the belief elicitation study, chiropractors
expressed the most optimistic belief (median RR, 0.56; IQR, 0.39-1.03); family
physicians, a neutral belief (median RR, 0.97; IQR, 0.64-1.21); and spine surgeons, a
more pessimistic belief (median RR, 1.07; IQR, 0.95-1.29). (3) The self-controlled case
series study found positive associations between both chiropractic and PCP visits, and
surgically managed acute LDH. There was no more risk associated with visits to
chiropractors than with visits to PCPs.
Conclusions: (1) The best evidence synthesis shows the varying quality and
heterogeneity of the epidemiologic literature on LDH with radiculopathy. There is a need
to develop standardized case definitions that validly classify the clinical spectrum. (2)
Clinicians’ beliefs about the risk for acute LDH associated with chiropractic SMT varied
systematically across professions. The derived probability distributions can serve as
prior probabilities in Bayesian analyses of this exposure-outcome association. (3)
Patients with prodromal back pain related to a developing disc herniation may seek
healthcare from both chiropractors and PCPs before full clinical expression of acute
LDH that is subsequently surgically managed.
iv
Acknowledgments
First, I would like to acknowledge my parents, Juan Manuel and Lilia Hincapié; my
brother, Juan Carlos Hincapié, and sister, Maria Cristina Choy (née Hincapié); and the
rest of my family, for all their love, support and encouragement. Má y Pá, soy quien
soy, en gran parte por tu ejemplo único de amor, sacrificio y esfuerzo. Gracias con toda
alma, vida y corazón!
I would like to express my sincere gratitude and appreciation to my supervisor, J. David
Cassidy, for giving me the opportunity to undertake this PhD under your great
mentorship and for providing many learning and research opportunities throughout the
process. Your dedication, excellence and leadership have burned a bright path for so
many others to follow, and I am truly honoured and blessed to count you as a mentor,
colleague and friend.
A heart-felt thanks to all of my advisory committee members: Pierre Côté, Alejandro
Jadad, George Tomlinson, Raj Rampersaud and Eleanor Boyle, for their enthusiasm,
support and guidance with this thesis. This work and I have been enriched by each of
your brilliant minds and many contributions along the way. A special thanks goes to
George, for your incredible generosity with both your time and sharp analytical mind,
and for your friendship. I will remember with great fondness the many hours of
“R-awesome” coding and analysis that we shared.
I thank my doctoral program director, Nancy Kreiger, and the support staff in the
Division of Epidemiology, at the Dalla Lana School of Public Health, for their helpful
administrative support and guidance throughout the program.
I would like to acknowledge the Canadian Institutes of Health Research and the
Canadian Chiropractic Research Foundation, for their interest and funding of this work.
A special mention goes to Allan Gotlib, for your tireless effort and enthusiasm for the
development of Canadian chiropractic clinician-scientists; thank you, Allan.
Many thanks to my chiropractic and epi research friends: Maja Stupar, Carol
Cancelliere, James Donovan, Heather Shearer, Craig Jacobs, Mana Rezai, Darren
v
Brenner, Paul Arora, Beth Rachlis, Nancy Carnide and Silje Mæland, for the memorable
times shared and many insightful clinical, epidemiologic and personal discussions along
the way. Silje, a special thanks to you and Ketil (and Sunniva, Gustav and Wilhelm), for
your friendship and encouragement, and the great times our families shared.
Thank you to my beautiful, supportive and understanding children, Maia Liliana, Julia
Marie, Daniel Joseph and Michael Juan Hincapié. You are the most precious gift and
blessing in my life, and I give God thanks for allowing me to share in your lives.
Finally, thank you to my wonderful and amazing wife, Barb Hincapié (née Kral). Amor,
there are no words to express my love and gratitude for you and all that you do for our
family. You are a shining beacon of love, dedication and giving. Thank you for your
support, patience, kindness and encouragement. I look forward to continuing on life’s
journey together!
vi
Table of Contents
Acknowledgements iv
Table of Contents vi
List of Tables vii
List of Figures viii
List of Appendices ix
List of Abbreviations x
Thesis Overview xii
Chapter 1 General introduction and outline 1
Chapter 2 Incidence and determinants of lumbar disc herniation with
radiculopathy in adults: a systematic review
In preparation for submission, 2015
22
Chapter 3 Chiropractic spinal manipulation treatment and the risk for acute
lumbar disc herniation: a belief elicitation study
In preparation for submission, 2015
65
Chapter 4 Chiropractic care and risk for acute lumbar disc herniation: a
population-based self-controlled case series study
In preparation for submission, 2015
87
Chapter 5 Summary and synthesis 115
Chapter 6 Appendices 131
vii
List of Tables
Table 1.1. Low risk of bias RCTs on spinal manipulation for lumbar disc herniation
with radiculopathy
Table 2.1. Characteristics of eligible studies examining the incidence of and/or risk
factors for lumbar disc herniation with radiculopathy in adults
Table 2.2. Summary of eligible studies examining the incidence of and/or risk factors
for lumbar disc herniation with radiculopathy in adults
Table 3.1. Characteristics of belief elicitation study participants
Table 3.2. Clinicians’ beliefs about the risk for acute lumbar disc herniation and
surgically managed acute lumbar disc herniation associated with
chiropractic spinal manipulation treatment
Table 3.3. Characteristics of clinicians with optimistic versus pessimistic beliefs about
the risk for acute lumbar disc herniation and surgically managed acute
lumbar disc herniation associated with chiropractic spinal manipulation
treatment
Table 4.1. Characteristics of study participants at the time of the recording of an
acute lumbar disc herniation with incident early surgery
Table 4.2. Distribution of chiropractor and primary care physician visits before the
event index date for all cases and stratified by event fiscal year
Table 4.3. Incidence rate ratios for an acute lumbar disc herniation with first early
surgery event in exposed periods following chiropractic and primary care
physician visits
Table 4.4. Results of sensitivity analyses compared to primary analysis of an acute
lumbar disc herniation with first early surgery event in risk periods 0 to 7
days after chiropractor and primary care physician visits
viii
List of Figures
Figure 1.1. The case-crossover and self-controlled case series designs
Figure 2.1. Flow diagram of information through phases of the systematic review
Figure 3.1. Beliefs regarding the risk for acute lumbar disc herniation associated with
chiropractic spinal manipulation treatment: a. among all clinicians, b.
among clinician groups, c. among optimistic versus pessimistic clinicians
Figure 3.2. Beliefs regarding the risk for acute lumbar disc herniation associated with
chiropractic spinal manipulation treatment: a. by gender, b. by age group,
c. by clinical experience
Figure 4.1. Representation of the self-controlled case series design
Figure 4.2. Case definition flow diagram
Figure 4.3. Bootstrap incidence rate ratios for primary analysis of any visit exposure
with a risk period of 0-7 days following chiropractic versus primary care
physician care
Figure 4.4. Kernel density estimate of the ratio of bootstrap incidence rate ratios for
chiropractic versus primary care physician care
Figure 5.1. Preliminary Bayesian triplots of risk for acute lumbar disc herniation with
early surgery associated with chiropractic care
ix
List of Appendices
Appendix A. Risk of bias assessment summary (Chapter 1)
Appendix B. Comparison of the case-crossover (CCO) and self-controlled case series
(SCCS) designs (Chapter 1)
Appendix C. Systematic review search strategies (Chapter 2)
Appendix D. Detailed evidence tables for systematic review (Chapter 2)
Appendix E. Belief elicitation questionnaire (Chapter 3)
Appendix F. Belief elicitation script (Chapter 3)
Appendix G. Belief elicitation computer protocol (Chapter 3)
Appendix H. Case and exposure definition tables for self-controlled case series study
(Chapter 4)
x
List of Abbreviations
CCO Case-crossover
CI Confidence interval
CT Computerized tomography
DAD Discharge Abstract Database
DC Doctor of chiropractic
ED Emergency department
HEA Home exercise and advice
HNP Herniated nucleus pulposus
HRR Hazard rate ratio
ICD International classification of diseases
IQR Interquartile range
IRR Incidence rate ratio
LBP Low back pain
LDH Lumbar disc herniation
MRI Magnetic resonance imaging
MSK Musculoskeletal
N Number
NACRS National Ambulatory Care Reporting System
ND No data
NR Not reported
NRS Numerical rating scale
NSAID Nonsteroidal anti-inflammatory drug
OHIP Ontario Health Insurance Plan
OR Odds ratio
PCP Primary care physician
PRISMA Preferred Reporting Items for Systematic Reviews and Meta-Analyses
PROSPERO International prospective register of systematic reviews
QTFSD Quebec Task Force on Spinal Disorders
R R project for statistical computing
RCT Randomized controlled trial
xi
REB Research ethics board
ROB Risk of bias
RR Relative risk or risk ratio
SCCS Self-controlled case series
SF-36 Short form-36 health survey
SHELF Sheffield elicitation framework
SIGN Scottish Intercollegiate Guidelines Network
SMR Standardized morbidity ratio
SMT Spinal manipulation treatment
SNP Single nucleotide polymorphism
US United States of America
UK United Kingdom
VAS Visual analog scale
VNTR Variable number of tandem repeats
xii
Thesis Overview
The objective of this thesis is to advance knowledge of the risk for acute lumbar disc
herniation (LDH) following chiropractic care using a mixed methods approach.
Chapter 1 includes a brief introduction to the public health and epidemiologic
challenges of low back pain and LDH with radiculopathy; a description of the
epidemiology of LDH with radiculopathy, chiropractic care and the clinical management
of back pain; and a summary of the uncertainty surrounding the link between spinal
manipulation and LDH. The challenges in studying rare serious adverse events
following spinal manipulation are described. Finally, the rationale for the study designs
of this thesis is provided and the thesis aim and objectives are stated.
Chapter 2 presents a systematic review of the incidence and determinants of LDH with
radiculopathy in adults. This helps clarify the epidemiology of this important condition
and advances understanding of the risk factors for LDH with radiculopathy.
Chapter 3 describes and quantifies clinicians’ beliefs about the risk for acute LDH
associated with chiropractic spinal manipulation treatment. This involves a Bayesian-
inspired approach that can determine the magnitude of a potential risk expected by
clinician experts and describe the presence of uncertainty regarding this exposure-
outcome association.
Chapter 4 investigates the association between chiropractic care and acute LDH with
early surgery, and compares this to the association between primary care physician
care and acute LDH with early surgery. We provide a valid epidemiologic assessment
of the link between both chiropractic and primary medical care and acute LDH with early
surgical intervention.
Chapter 5 is the concluding chapter. The main findings of our studies are summarized,
synthesized and placed in broader clinical, research and public health context. We
elaborate on the implications and limitations of our work and provide guidance for future
research.
1
Chapter 1
General Introduction and Outline 1
Chiropractic care is popular for low back pain, but little is known about its potential link
to acute lumbar disc herniation (LDH) with radiculopathy. Several systematic reviews
suggest that chiropractic treatment or spinal manipulation is a viable therapeutic option
for low back pain, but the summarized studies are of varying quality and too small to
examine rare serious adverse events.1-4 Moreover, randomized clinical trials have
shown benefit of chiropractic spinal manipulation for the management of LDH with
radiculopathy,5-9 yet little is known about serious adverse events possibly related to this
treatment. This void of knowledge represents a significant problem because an
effective treatment is one that cost-effectively improves health outcomes while first
doing no harm.
Today, no valid estimate of the risk for acute LDH with radiculopathy following
chiropractic treatment is available in the scientific literature. The current literature
contains case reports and small case series linking lumbar spine manipulation to acute
LDH with radiculopathy and cauda equina syndrome.10-12 However, case reports and
case series provide the lowest level of evidence with respect to the assessment of risk
and should not be used to make inferences about the lack of safety of a treatment.
They have, however, raised the hypothesis of potential harm.
This chapter includes a brief introduction to the public health and epidemiologic
challenges of low back pain and LDH with radiculopathy; a description of the
epidemiology of LDH with radiculopathy, chiropractic care and the clinical management
of back pain; followed by a summary of the uncertainty surrounding the link between
spinal manipulation and LDH. The challenges in studying rare serious adverse events
following spinal manipulation are described. Finally, the rationale for the study designs
of this thesis is provided and the thesis aim and objectives are stated.
2
1.1 Low Back Pain and Lumbar Disc Herniation With Radiculopathy – Public Health and Epidemiologic Challenges
Around the world, low back pain is recognized as an important public health concern,
mainly because it is a leading cause of disability and work absence, and associated with
a huge economic burden on individuals, industry and society.13 It affects about 70% of
all people in their lifetime, and between 15% to 30% on any given day with varying
types of clinical presentations.14-16 The Global Burden of Disease 2010 project provides
the clearest evidence to date of the huge and increasing burden on global health from
musculoskeletal conditions—it showed that low back pain is the leading cause of
disability worldwide.17
LDH with radiculopathy, defined as the localized displacement of disc material beyond
the normal margins of the intervertebral disc space with pain, weakness or numbness in
a myotomal or dermatomal pattern,18 can be one of the most recognizable presentations
of low back pain. The diagnosis is typically based on a combination of clinical features
suggesting lumbar spinal nerve root compression or irritation, such as: lumbosacral
radiculopathy (i.e., radicular leg pain or sciatica), nerve root tension signs, neurologic
deficits (i.e., muscle weakness and reflex changes), and advanced imaging (i.e., MRI or
CT) findings that correlate with the clinical syndrome.19-21 However, many patients
present with a less definitive clinical picture, involving low back pain in the early
(prodromal) phase that then progresses to radicular leg pain with or without neurologic
signs.22,23 In addition, diagnostic imaging may only be indicated in the prodromal phase
when there is suspicion of underlying pathology, such as infection or malignancy.24
Finally, the predictive values of common physical examination tests for LDH with
radiculopathy are low.20,25,26
All these factors contribute in making the diagnosis during the early phases of the
syndrome especially difficult. Moreover, this multifaceted clinical presentation, resultant
variable case definitions,19 and difficulties with clinical diagnosis and classification,20,25,26
are notable reasons why the descriptive epidemiology of this condition is not precisely
known.
3
1.2 Descriptive Epidemiology of Lumbar Disc Herniation With Radiculopathy
Exact data on the prevalence and incidence of LDH with radiculopathy are lacking.
General population studies involving clinical assessment have estimated the point
prevalence of lumbar disc syndrome at about 5%,27,28 varying by sex and age. It is
most common among persons 30-50 years of age, with men more frequently affected
than women.29 Among acute low back pain patients seeking primary health care, the
prevalence of lumbar radiculopathy suggesting disc herniation is, not surprisingly,
substantially higher. Prospective clinical cohort studies have found as many as 20% to
25% of acute low back pain patients reporting prevalent radiating leg pain below the
knee at the time of initial presentation.30,31 One recent systematic review of the
prevalence of sciatica found considerable variation in estimates, ranging from 1.6% in
the general population to 43% in a population of nursing aides.32
Few epidemiologic studies have estimated the incidence of LDH with radiculopathy.
One study calculated the incidence of hospitalized lumbar disc herniation or sciatica
among the general population at 1.5 per 1,000 person-years,33 while another estimated
7.8 cases of hospitalized lumbar disc disorder with radiculopathy per 1,000 person-
years among young Finnish male military conscripts.34
With respect to risk factors, the literature has suggested several focusing on activities
involving sudden or sustained rotation and flexion of the low back, such as heavy lifting,
twisting, bending, and driving.33,35-41 However, other factors such as smoking,
pregnancy, diabetes, body mass index, hypertension, high cholesterol, and family
history have also been identified as possible determinants.42-45 Overall, little is known
about the incidence of symptomatic LDH, and consequently, risk factors for this
condition are not well understood.
1.3 Chiropractic Care and the Clinical Management of Back Pain
In Canada, medical doctors, chiropractors and physical therapists are the main health
care providers that manage spinal pain.46 Approximately 12% of American and
Canadian adults seek chiropractic care annually, and over 9 out of 10 visits involve
treatment with spinal manipulation.47,48 Canadian chiropractic patients, compared to
4
those seeking medical care for back pain, tend to be younger and have higher
socioeconomic status and fewer health problems.46,49 Patients have reported very good
or excellent satisfaction or perceived helpfulness with chiropractic care that has
included spinal manipulation.50
No widely adopted definition of serious adverse event exists in the back pain literature.
For the purpose of this thesis, we define a serious adverse event as an untoward
occurrence that results in death, is life threatening, requires hospitalization, or results in
significant or permanent disability.12,51 Examples of serious adverse events resulting
from lumbosacral spinal manipulation include disc herniation and cauda equina
syndrome.
The medical and chiropractic management of back pain rests on the delivery of different
therapeutic modalities, with varying risk-benefit profiles. Medical doctors mainly
prescribe medications, such as nonsteroidal anti-inflammatory drugs (NSAIDs), while
chiropractors primarily use spinal manipulation. Both types of interventions can benefit
low back pain,4,52 and are potentially related to adverse events. Compared with nonuse,
the use of NSAIDs has been associated with up to a 3-fold increased risk for
dyspepsia,53 about a 5-fold greater risk for serious upper gastrointestinal bleeding,54,55
and up to a 2-fold increase in the risk of death56. Even the newer and less gastrotoxic
cyclooxegenase-2 selective inhibitors are associated with an increased risk of serious
vascular events, such as stroke and myocardial infarction.57,58
With respect to spinal manipulation, prospective studies of harm have mainly reported
minor and transient adverse events, such as local pain and muscle soreness, but have
not observed the occurrence of serious adverse events. LeBoeuf-Yde et al48 described
a Swedish prospective case series including 625 chiropractic patients who had 1858
treatment visits, of which 99% involved spinal manipulation. Adverse events were
reported to be common, benign and self-limiting (usually within 1 day). The most
commonly reported adverse event was local discomfort in the area of treatment,
accounting for roughly two thirds of all adverse events. Similarly, in their prospective
case series of 4712 treatment visits of 1058 new patients by 102 Norwegian
chiropractors, Senstad et al59 found that 55% of patients reported at least one adverse
event, and 25% of all treatments (75% of which involved the lumbar spine) resulted in at
5
least one event. The onset of most events was on the day or the day after the
treatment visit. Adverse events consisted primarily of short-term local discomfort,
headache, tiredness or radiating discomfort (defined as discomfort in an area away from
the one treated), and were characterized as mild or moderate by 85% of patients.
Neither study, however, reported a serious adverse event associated with spinal
manipulation.
Nonetheless, the reporting of harms is generally poor in the primary literature,4 and
case reports of serious adverse events, such as disc herniation and cauda equina
syndrome, have been described, but their incidence have yet to be validly estimated.60
Hebert et al12 undertook a systematic review of cases of serious adverse events
following lumbopelvic spinal manipulation. They found that over two thirds of all
reported serious adverse events consisted of cauda equina syndrome (29 cases, 38%
of total) or LDH (23 cases, 30% of total). However, important case details were often
not captured in the case reporting, including pre-manipulation presentation of the
patient, specific details of the serious adverse event, time from spinal manipulation to
onset of the event, and clinical outcome. Improved knowledge of the risk for serious
adverse events associated with spinal manipulation could inform clinical decision-
making and is very much needed.
1.4 Spinal Manipulation and Lumbar Disc Herniation – Potential Benefit Or Potential Harm?
Spinal manipulation treatment is generally considered to be safe, although some
concern has been raised about its potential link with LDH.61,62 Several systematic
reviews have found that spinal manipulation can benefit low back pain with little
evidence of serious harm.4,63 Randomized clinical trial evidence also supports the use
of spinal manipulation for the treatment of LDH with radiculopathy (Table 1.1, see
Appendix A for risk of bias assessment summary).7-9
6
Table 1.1. Low risk of bias RCTs on spinal manipulation for lumbar disc herniation with radiculopathy
Authors, year
Subjects, setting, case
definition and study size
Description of Interventions
(n patients)
Outcomes measured and
follow-up Key findings
Santilli et al, 20067 Patients (18-65yo) seen in 2
rehabilitation centers in
Rome, Italy
Case definition: acute LBP
(<10d and pain-free
during prior 3mo) of
moderate to severe
intensity (VAS, ≥5/10),
moderate to severe
radiating pain to one leg
(VAS, ≥5/10) and +ve MRI
n=102
SMT by DC: up to 20
treatments over 30d (53)
Sham SMT by DC: up to 20
treatments over 30d (49)
Validated: VAS1 (local low
back pain), VAS2
(radiating pain), SF-36
(quality of life)
Non-validated: Disc
morphology, Kellner score
(psychological profile)
Follow-up: 15, 30, 45, 90
and 180d
A significantly greater
number of patients treated
with spinal manipulation
had no back, buttock or
leg pain at 180 days. VAS
1: 28% vs. 6 %. VAS 2:
55% vs. 20%.
No significant difference in
the SF-36, psychological
testing and disc
morphology between the
groups.
No adverse events.
McMorland et al, 20108 Patients (>18yo) screened
by 3 spine neurosurgeons
for symptoms of unilateral
lumbar radiculopathy
secondary to LDH at L3-4,
L4-5, L5-S1 disc levels, in
Calgary, Canada
Case definition: leg-
dominant symptoms with
SMT by DC: 2-3 visits/wk
for the first 4 wk reducing
to 1-2 visits/wk for the
next 3-4 wk; at the 8-wk
mark, follow-up visits
were scheduled based on
the patient's symptoms;
average of 21 treatments
(20)
Validated: McGill Pain
Questionnaire, Aberdeen
Back Pain Scale, Roland-
Morris Disability Index,
SF-36 (quality of life)
Follow-up: 3, 6, 12, 24 and
52wk
Significant improvement in
both treatment groups
compared to baseline
scores over time was
observed in all outcome
measures.
Of patients with lumbar
radiculopathy due to LDH
that failed 3mo of medical
7
Authors, year
Subjects, setting, case
definition and study size
Description of Interventions
(n patients)
Outcomes measured and
follow-up Key findings
objective signs of nerve
root tethering ± neurologic
deficit correlated with +ve
MRI
Notable inclusion criterion:
patients had failed ≥3mo
of nonoperative care,
including analgesics,
lifestyle modification,
physiotherapy, massage
therapy and/or
acupuncture
n=40
Surgical microdiscectomy
by neurosurgeon (20)
Crossover to the alternate
treatment was allowed
after 3mo, if patient felt
they were not responding
to primary intervention
management, 60%
benefited from SMT to the
same degree as if they
underwent surgical
intervention.
For the 40% that did not
respond to SMT, surgery
provided an excellent
outcome.
No new neurologic deficits
from either of the
interventions and no
significant adverse
events.
Bronfort et al, 20149 Patients (≥21yo) with
subacute or chronic back-
related leg pain at 2
research centers in
Minnesota and Iowa, US
Case definition: back-
related leg pain based on
QTFSD classifications 2,
3, 4 or 6 (radiating pain
into the proximal or distal
SMT plus HEA: SMT by DC
up to 20 visits and HEA
by DC, exercise therapist
and personal trainer 4
sessions (96)
HEA: HEA by DC, exercise
therapist and personal
trainer 4 sessions (96)
12wk course of care for
both groups
Validated: 11-point NRS,
Roland-Morris Diability
Questionniare, SF-36
(quality of life)
Non-validated: self-reported
LBP
Follow-up: 3, 12, 26 and
52wk
For leg pain, SMT plus HEA
had a clinically important
advantage over HEA
(difference, 10 percentage
points [95%CI, 2 to 19]; P
= 0.008) at 12wk but not
at 52wk (difference, 7
percentage points [CI, -2
to 15]; P = 0.146).
Nearly all secondary
8
Authors, year
Subjects, setting, case
definition and study size
Description of Interventions
(n patients)
Outcomes measured and
follow-up Key findings
part of the lower
extremity, with or without
neurologic signs); severity
of ≥3 on 0-10 scale;
current episode ≥4 weeks;
and stable prescription
medication plan in past
mo
Notable exclusion criterion:
QTFSD classifications 1,
5, 7, 8, 9, 10 and 11 (pain
without radiation into the
lower extremities
N=192
outcomes improved more
with SMT plus HEA at
12wk, but only global
improvement, satisfaction,
and medication use had
sustained improvements
at 52wk.
No serious treatment-
related adverse events or
deaths occurred.
Abbreviations: +ve, positive; CI, confidence interval; d, days; DC, doctor of chiropractic; HEA, home exercise and advice; LBP, low back pain; LDH,
lumbar disc herniation; mo, months; MRI, magnetic resonance imaging; NRS, numerical rating scale; QTFSD, Quebec Task Force on Spinal
Disorders; SF-36, short form-36 health survey; SMT, spinal manipulation treatment; VAS, visual analog scale; wk, weeks; yo, years old
9
On the other hand, some believe, on the basis of case reports and small case series,
that spinal manipulation ought to be contraindicated for the treatment of disc herniation
and have raised the hypothesis of causal harm.10,64-69 The variability in belief was
captured in a survey of the opinions and practices of 453 UK osteopaths, in which over
half of the respondents reported that they use manipulation for disc herniation, while
most of the others described the practice as “dangerous.”70 In sum, there seems to be
a disconnect between the scientific evidence and beliefs in this area, in that the best
available evidence indicates a beneficial effect of chiropractic spinal manipulation on
LDH, yet beliefs and opinion vary and may overly focus on potential harm.
1.5 Challenges In Studying Rare Serious Adverse Events Following Spinal Manipulation Treatment
Attempts to investigate serious adverse events related to chiropractic spinal
manipulation face several challenges. First, the valid study of uncommon adverse
events is limited simply by the rarity of the events. Clinically important but unexpected
adverse effects of spinal manipulation are too rare to be detected in randomized trials.
In addition, well-designed prospective cohort studies are infeasible due to the inability to
recruit and follow a large enough sample of patients that will have a sufficient number of
events.
Another challenge is the issue of confounding. Confounding refers to a situation in
which a noncausal association between a given exposure and an outcome is observed
as a result of the influence of some third variable (or group of variables), usually
designated as a confounder.71 The phenomenon can be described as follows: the
confounding variable is (1) a determinant of the outcome and (2) associated with the
exposure, but (3) is not an intermediate variable in the causal pathway between
exposure and outcome. The essential notion of confounding is that the association
between an exposure and a given outcome can be induced, strengthened, weakened,
or eliminated by a third variable or group of variables. Persons who seek health care
may differ from those who do not in ways that can be difficult to measure and control
for. Some of these differences may also be associated with the future risk for LDH,
10
which makes a conventional observational (case-control or cohort) design a less valid
source of information on this association.
Finally, this is a sensitive topic open to controversy and litigation,72,73 with an extremely
limited evidence base, and thus, susceptible to opinion and anecdotal information.
1.6 The Bayesian Paradigm – A Way Forward and Rationale
A Bayesian approach to health services research provides an opportunity to advance
knowledge in spite of the above challenges.74,75 Bayesian methods start with existing
“prior” beliefs, formally quantified as probability distributions, and update these using
new data to arrive at “posterior” beliefs, which may be used as the basis for inferential
decisions.76 Quantifying currently held beliefs can determine the magnitude of a
potential risk expected by experts and describe the presence of uncertainty or clinical
equipoise about an exposure-outcome association of interest. Experts in a field can
have knowledge of the risk of using a treatment through years of clinical experience.
When quantified, the knowledge gained from their clinical experience can be included in
models estimating risk, which may help to bridge the gap between evidence and beliefs.
A notable strength of this approach is that probability distributions obtained through the
elicitation of beliefs can be used to augment scarce data,74,75,77,78 and be formally
incorporated as Bayesian priors into analyses of association. Although infrequently
used in epidemiologic research, a Bayesian approach may have important utility in the
study of treatment-related rare serious adverse events, particularly in the absence of
definitive scientific evidence.74,75,79,80
1.7 A Primer On Case-Only Designs and Rationale for the Self-Controlled Case Series Design
Epidemiologic research studies, especially those that use secondary data analysis, may
test the hypothesis of interest through the selection of one of several potential study
designs. Conventional designs, including cohort and nested case-control studies, may
yield estimates that are biased by residual confounding because obtaining complete
information on all relevant confounders may be difficult or infeasible.81,82 Recently,
increasing attention has been paid to case-only study designs, including self-controlled
11
case series and case-crossover designs. Case-only designs are considered to offer
improved control over confounding arising from variables that are constant within a
person by focusing on comparisons between different periods of time within each
person’s follow-up.83 The self-controlled case series and case-crossover designs use
restricted samples, only including data for participants who experienced the outcome of
interest.
The self-controlled case series design adopts a prospective cohort perspective by
comparing the rate of the outcome of interest between exposed and unexposed time
periods for each participant (Figure 1.1).84,85 It starts with identifying the exposure of
interest (e.g., a chiropractic visit) and defining the “effect period of exposure”, that is, the
window of time when there is a hypothesized susceptibility to an adverse effect due to
the exposure. The likelihood of the outcome of interest occurring in a time period
following exposure (i.e., “exposed” period) is compared with the likelihood of the
outcome occurring in “unexposed” periods, with the comparison being in the same
individual. Control for risk factors that are constant within an individual is achieved
through fitting Poisson regression models, conditional on the individual, to estimate the
incidence rate ratio (IRR),85 as in a matched cohort study.
The case-crossover design adopts a case-control perspective by comparing exposure
status between a time period just before occurrence of the outcome (i.e., case period)
and one or more referent time periods (i.e., control periods) prior to the case period
when the outcome did not occur (Figure 1.1).83,86 The case-crossover design starts
with identifying the outcome of interest and assessing exposure to the risk factor of
interest in a chosen time period preceding the outcome (i.e., case period). One or more
control periods are then selected, and exposure to the risk factor of interest in the case
period is compared with exposure during these control periods in the same individual.
Control for risk factors that that are constant within an individual is achieved through
using analysis methods that condition on matched time periods,86 analogous to a
matched case-control study.
12
Start ofstudyperiod
Exposurevisit
End ofstudyperiod
Non-sampledperiod
Controlperiod 7d
Outcome index date
Caseperiod 7d
0Days before event index date
Exposurevisit
Start ofstudyperiod
First exposure
visit
End ofstudyperiod
Effect period ofexposure 0-2d
Effect period ofexposure 3-7d
Effect period ofexposure 8-14d
Unexposedperiod
Calendar days after start of study period
Outcome index date
Representation of the case-crossover design. Times are relative to the event index date and the data
are viewed retrospectively from the standpoint of the event index date.
Representation of the self-controlled case series design. Times are expressed in terms of calendar
days from the start of the study period and the data are viewed as in a prospective cohort. Incidence rate
ratios compare the within-person rate of events during exposed periods with the rate of events during
unexposed periods.
Figure 1.1. The case-crossover and self-controlled case series designs
The case-crossover and self-controlled case series designs were originally introduced
for the study of the acute effects of short-term exposures, such as the occurrence of
febrile convulsions after vaccination87 or triggering of myocardial infarction by recent
coffee consumption86. Both designs are appropriate when a brief exposure (e.g., a
chiropractic visit) causes a transient change in risk (i.e., effect period of exposure) of a
rare-onset disease (e.g., acute LDH).
13
For unbiased estimation, the standard self-controlled case series design, which includes
person-time both before and after the outcome event (Figure 1.1), requires that the
exposure distribution be unrelated to event times.88 Bias may be introduced in a
standard self-controlled case series analysis, if the outcome event influences the
likelihood of post-event exposures, as when the prescription of a given drug is
contraindicated by an outcome or disease diagnosis. Recently however, a modification
of the method that ignores post-event exposures,88 has been shown to validly estimate
incidence rate ratios in situations where occurrence of the outcome event affects post-
event exposures. The modified method applies a recursive approach to the problem in
that it starts with the last observed pre-event exposure and works back through the pre-
event exposures, ignoring post-event exposures.88 Given the possibility of the
occurrence of acute LDH influencing the likelihood of post-event visits to a chiropractor
or primary care physician, it is this modified self-controlled case series design that we
use in Chapter 4 of this thesis.
The fundamental advantage of the chosen study design is that the analysis is
conditioned on the individual, yielding only within-person comparisons (i.e., self-
controlled). This is a major strength for assessing potential treatment-induced harm on
a population level because of the concern about potential confounding by factors not
recorded in health care databases. By structuring the analysis so that each person
serves as their own control, we eliminate confounding and selection bias by constant
(time-invariant) characteristics, such as, sex, location, genetics, underlying state of
health, tendency to seek professional health care, average physical activity, long-term
diet, habitual health behaviours, long-past health events such as illnesses and injuries,
occupation, social support, ethnicity, smoking history, and body mass history.
The rationale for using the self-controlled case series design over the case-crossover
design is based on three reasons. First, the case-crossover design may be susceptible
to biased estimates due to time trends in the exposure distribution (i.e., exposure-trend
bias). The case-crossover design is vulnerable to exposure-trend bias because the
control periods always precede the case period; that is, bias may arise because of time
trends in exposure and due to the fixed ordering in time of case and control
periods.83,89,90 Since the self-controlled case series method adopts a prospective cohort
perspective based on using the actual exposure histories of each study participant, it is
14
less susceptible to exposure-trend bias than the case-crossover design as the time
trend in exposure probability can be included in the model.83,85 Second, unlike the
case-crossover design, which typically requires the selection of one or more referent
time periods to serve as a control, the self-controlled case series design makes use of
all available temporal information without the need for selection. This need to choose
comparator time periods in the case-crossover design may make it susceptible to
overlap bias (i.e., a version of the bias that arises from choosing non-disjoint strata to
partition the population in a matched case-control study).91 Unlike the case-crossover
design, the self-controlled case series design is not vulnerable to overlap bias. Third,
the self-controlled case series method is particularly suited to situations when the
investigator can accurately classify each person’s exposure timing.83 This aligns well
with our data given that the timing of exposures (based on billing claims for chiropractic
and primary care physician visits) was likely well captured by the health professionals
involved when claiming for OHIP reimbursement. Appendix B compares and contrasts
the case-crossover and self-controlled case series designs and highlights their
strengths and limitations.
15
1.8 Thesis Aim and Objectives
Overall, three consistent findings emerged from the review of the literature. First, the
epidemiology of LDH with radiculopathy is not well understood. Second, with respect to
serious adverse events following spinal manipulation and perceived risk, beliefs seem to
matter and warrant a better understanding. Third, no valid estimate of the association
between chiropractic treatment and the risk for acute LDH is currently available in the
scientific literature.
The general aim of this thesis is to advance knowledge of the risk for acute LDH
following chiropractic care. To meet this aim, the thesis uses a mixed methods
approach focusing on the following three specific objectives, with the goal of
addressing the current knowledge gaps described previously:
1. To assemble and synthesize the best available evidence on the incidence and
determinants of LDH with radiculopathy in adults. This will help clarify the
epidemiology of this important condition and advance understanding of the risk
factors for LDH with radiculopathy.
2. To describe and quantify clinicians' beliefs (expressed as probability distributions)
about chiropractic spinal manipulation as a risk factor for: (1) acute LDH, and (2)
acute severe LDH that is surgically managed. This involves a Bayesian approach
that can determine the magnitude of a potential risk expected by clinicians, describe
beliefs, and examine the presence of uncertainty regarding this exposure-outcome
association.
3. To investigate the association between chiropractic care and acute LDH with early
surgical intervention (≤8 weeks), and compare this to the association between
primary care physician care and acute LDH with early surgery. This will provide a
valid epidemiologic assessment of the association between chiropractic care and
acute LDH that is managed with early surgical intervention.
16
References 1. Ferreira ML, Ferreira PH, Latimer J, Herbert R, Maher CG. Efficacy of spinal
manipulative therapy for low back pain of less than three months' duration. J Manipulative Physiol Ther. 2003;26:593-601.
2. Bronfort G, Haas M, Evans RL, Bouter LM. Efficacy of spinal manipulation and mobilization for low back pain and neck pain: a systematic review and best evidence synthesis. Spine J. 2004;4:335-56.
3. Chou R, Huffman LH, American Pain Society, American College of Physicians. Nonpharmacologic therapies for acute and chronic low back pain: a review of the evidence for an American Pain Society/American College of Physicians clinical practice guideline. Ann Intern Med. 2007;147:492-504.
4. Furlan AD, Yazdi F, Tsertsvadze A, et al. A systematic review and meta-analysis of efficacy, cost-effectiveness, and safety of selected complementary and alternative medicine for neck and low-back pain. Evid Based Complement Alternat Med. 2012;2012:953139.
5. Burton AK, Tillotson KM, Cleary J. Single-blind randomised controlled trial of chemonucleolysis and manipulation in the treatment of symptomatic lumbar disc herniation. Eur Spine J. 2000;9:202-7.
6. Liu J, Zhang S. Treatment of protrusion of lumbar intervertebral disc by pulling and turning manipulations. J Tradit Chin Med. 2000;20:195-7.
7. Santilli V, Beghi E, Finucci S. Chiropractic manipulation in the treatment of acute back pain and sciatica with disc protrusion: a randomized double-blind clinical trial of active and simulated spinal manipulations. Spine J. 2006;6:131-7.
8. McMorland G, Suter E, Casha S, du Plessis SJ, Hurlbert RJ. Manipulation or microdiskectomy for sciatica? A prospective randomized clinical study. J Manipulative Physiol Ther. 2010;33:576-84.
9. Bronfort G, Hondras MA, Schulz CA, Evans RL, Long CR, Grimm R. Spinal manipulation and home exercise with advice for subacute and chronic back-related leg pain: a trial with adaptive allocation. Ann Intern Med. 2014;161:381-91.
10. Assendelft WJ, Bouter LM, Knipschild PG. Complications of spinal manipulation: a comprehensive review of the literature. J Fam Pract. 1996;42:475-80.
11. Oliphant D. Safety of spinal manipulation in the treatment of lumbar disk herniations: a systematic review and risk assessment. J Manipulative Physiol Ther. 2004;27:197-210.
12. Hebert JJ, Stomski NJ, French SD, Rubinstein SM. Serious adverse events and spinal manipulative therapy of the low back region: a systematic review of cases. J Manipulative Physiol Ther. 2013.
13. Hoy D, Brooks P, Blyth F, Buchbinder R. The epidemiology of low back pain. Best Pract Res Clin Rheumatol. 2010;24:769-81.
14. Kent PM, Keating JL. The epidemiology of low back pain in primary care. Chiropr Osteopat. 2005;13:13.
15. Walker BF. The prevalence of low back pain: a systematic review of the literature from 1966 to 1998. J Spinal Disord. 2000;13:205-17.
16. Andersson GB. Epidemiology of low back pain. Acta Orthop Scand Suppl. 1998;281:28-31.
17
17. Vos T, Flaxman AD, Naghavi M, et al. Years lived with disability (YLDs) for 1160 sequelae of 289 diseases and injuries 1990–2010: a systematic analysis for the Global Burden of Disease Study 2010. The Lancet. 2012;380:2163-96.
18. Kreiner DS, Hwang SW, Easa JE, et al. An evidence-based clinical guideline for the diagnosis and treatment of lumbar disc herniation with radiculopathy. Spine J. 2014;14:180-91.
19. Genevay S, Atlas SJ, Katz JN. Variation in eligibility criteria from studies of radiculopathy due to a herniated disc and of neurogenic claudication due to lumbar spinal stenosis: a structured literature review. Spine. 2010;35:803-11.
20. van der Windt DAWM, Simons E, Riphagen II, et al. Physical examination for lumbar radiculopathy due to disc herniation in patients with low-back pain. Cochrane Database Syst Rev. 2010:CD007431.
21. Vroomen PC, de Krom MC, Wilmink JT, Kester AD, Knottnerus JA. Diagnostic value of history and physical examination in patients suspected of lumbosacral nerve root compression. J Neurol Neurosurg Psychiatry. 2002;72:630-4.
22. Deyo RA, Loeser JD, Bigos SJ. Herniated lumbar intervertebral disk. Ann Intern Med. 1990;112:598-603.
23. Weber H. The natural history of disc herniation and the influence of intervention. Spine. 1994;19:2234-8; discussion 3.
24. Koes BW, van Tulder MW, Peul WC. Diagnosis and treatment of sciatica. BMJ. 2007;334:1313-7.
25. Al Nezari NH, Schneiders AG, Hendrick PA. Neurological examination of the peripheral nervous system to diagnose lumbar spinal disc herniation with suspected radiculopathy: a systematic review and meta-analysis. Spine J. 2013;13:657-74.
26. Iversen T, Solberg TK, Romner B, et al. Accuracy of physical examination for chronic lumbar radiculopathy. BMC Musculoskelet Disord. 2013;14:206.
27. Heliovaara M, Impivaara O, Sievers K, et al. Lumbar disc syndrome in Finland. J Epidemiol Community Health. 1987;41:251-8.
28. Kaila-Kangas L, Leino-Arjas P, Karppinen J, Viikari-Juntura E, Nykyri E, Heliovaara M. History of physical work exposures and clinically diagnosed sciatica among working and nonworking Finns aged 30 to 64. Spine. 2009;34:964-9.
29. Heliovaara M. Epidemiology of sciatica and herniated lumbar intervertebral disc [thesis]. Helsinki, Finland, The Social Insurance Institution; 1988.
30. Carey TS, Garrett J, Jackman A, McLaughlin C, Fryer J, Smucker DR. The outcomes and costs of care for acute low back pain among patients seen by primary care practitioners, chiropractors, and orthopedic surgeons. The North Carolina Back Pain Project. N Engl J Med. 1995;333:913-7.
31. Haas M, Goldberg B, Aickin M, Ganger B, Attwood M. A practice-based study of patients with acute and chronic low back pain attending primary care and chiropractic physicians: two-week to 48-month follow-up. J Manipulative Physiol Ther. 2004;27:160-9.
32. Konstantinou K, Dunn KM. Sciatica: review of epidemiological studies and prevalence estimates. Spine. 2008;33:2464-72.
33. Heliovaara M, Knekt P, Aromaa A. Incidence and risk factors of herniated lumbar intervertebral disc or sciatica leading to hospitalization. J Chronic Dis. 1987;40:251-8.
18
34. Mattila VM, Sillanpää P, Visuri T, Pihlajamäki H. Incidence and trends of low back pain hospitalisation during military service--an analysis of 387,070 Finnish young males. BMC Musculoskelet Disord. 2009;10:10.
35. Heliovaara M. Occupation and risk of herniated lumbar intervertebral disc or sciatica leading to hospitalization. J Chronic Dis. 1987;40:259-64.
36. Kelsey JL, Githens PB, Walter SD, et al. An epidemiological study of acute prolapsed cervical intervertebral disc. J Bone Joint Surg Am. 1984;66:907-14.
37. Kelsey JL, Githens PB, White AA, III, et al. An epidemiologic study of lifting and twisting on the job and risk for acute prolapsed lumbar intervertebral disc. J Orthop Res. 1984;2:61-6.
38. Kelsey JL, Hardy RJ. Driving of motor vehicles as a risk factor for acute herniated lumbar intervertebral disc. Am J Epidemiol. 1975;102:63-73.
39. Leino-Arjas P, Kaila-Kangas L, Kauppinen T, Notkola V, Keskimaki I, Mutanen P. Occupational exposures and inpatient hospital care for lumbar intervertebral disc disorders among Finns. Am J Ind Med. 2004;46:513-20.
40. Saftic R, Grgic M, Ebling B, Splavski B. Case-control study of risk factors for lumbar intervertebral disc herniation in Croatian island populations. Croat Med J. 2006;47:593-600.
41. Seidler A, Bolm-Audorff U, Siol T, et al. Occupational risk factors for symptomatic lumbar disc herniation; a case-control study. Occup Environ Med. 2003;60:821-30.
42. Heliovaara M. Body height, obesity, and risk of herniated lumbar intervertebral disc. Spine. 1987;12:469-72.
43. Kelsey JL, Githens PB, O'Conner T, et al. Acute prolapsed lumbar intervertebral disc. An epidemiologic study with special reference to driving automobiles and cigarette smoking. Spine. 1984;9:608-13.
44. Kelsey JL, Greenberg RA, Hardy RJ, Johnson MF. Pregnancy and the syndrome of herniated lumbar intervertebral disc; an epidemiological study. Yale J Biol Med. 1975;48:361-8.
45. Jhawar BS, Fuchs CS, Colditz GA, Stampfer MJ. Cardiovascular risk factors for physician-diagnosed lumbar disc herniation. Spine J. 2006;6:684-91.
46. Côté P, Cassidy JD, Carroll L. The treatment of neck and low back pain: who seeks care? who goes where? Med Care. 2001;39:956-67.
47. Senstad O, Leboeuf-Yde C, Borchgrevink CF. Side-effects of chiropractic spinal manipulation: types frequency, discomfort and course. Scand J Prim Health Care. 1996;14:50-3.
48. Leboeuf-Yde C, Hennius B, Rudberg E, Leufvenmark P, Thunman M. Side effects of chiropractic treatment: a prospective study. J Manipulative Physiol Ther. 1997;20:511-5.
49. Hurwitz EL, Chiang LM. A comparative analysis of chiropractic and general practitioner patients in North America: findings from the joint Canada/United States Survey of Health, 2002-03. BMC Health Serv Res. 2006;6:49.
50. Hurwitz EL. Epidemiology: spinal manipulation utilization. Journal of electromyography and kinesiology. 2012;22:648-54.
51. Edwards IR, Aronson JK. Adverse drug reactions: definitions, diagnosis, and management. The Lancet. 2000;356:1255-9.
52. Roelofs PD, Deyo RA, Koes BW, Scholten RJ, van Tulder MW. Nonsteroidal anti-inflammatory drugs for low back pain: an updated Cochrane review. Spine. 2008;33:1766-74.
19
53. Ofman JJ, Maclean CH, Straus WL, et al. Meta-analysis of dyspepsia and nonsteroidal antiinflammatory drugs. Arthritis Care & Research. 2003;49:508-18.
54. Hernandez-Diaz S, Rodriguez LA. Association between nonsteroidal anti-inflammatory drugs and upper gastrointestinal tract bleeding/perforation: an overview of epidemiologic studies published in the 1990s. Arch Intern Med. 2000;160:2093-9.
55. Ofman JJ, MacLean CH, Straus WL, et al. A metaanalysis of severe upper gastrointestinal complications of nonsteroidal antiinflammatory drugs. J Rheumatol. 2002;29:804-12.
56. Fosbol EL, Gislason GH, Jacobsen S, et al. Risk of myocardial infarction and death associated with the use of nonsteroidal anti-inflammatory drugs (NSAIDs) among healthy individuals: a nationwide cohort study. Clin Pharmacol Ther. 2009;85:190-7.
57. Graham DJ, Campen D, Hui R, et al. Risk of acute myocardial infarction and sudden cardiac death in patients treated with cyclo-oxygenase 2 selective and non-selective non-steroidal anti-inflammatory drugs: nested case-control study. Lancet. 2005;365:475-81.
58. Kearney PM, Baigent C, Godwin J, Halls H, Emberson JR, Patrono C. Do selective cyclo-oxygenase-2 inhibitors and traditional non-steroidal anti-inflammatory drugs increase the risk of atherothrombosis? Meta-analysis of randomised trials. BMJ. 2006;332:1302-8.
59. Senstad O, Leboeuf-Yde C, Borchgrevink C. Frequency and characteristics of side effects of spinal manipulative therapy. Spine. 1997;22:435-40; discussion 40-1.
60. Rubinstein SM. Adverse events following chiropractic care for subjects with neck or low-back pain: do the benefits outweigh the risks? J Manipulative Physiol Ther. 2008;31:461-4.
61. Gouveia LO, Castanho P, Ferreira JJ. Safety of chiropractic interventions: a systematic review. Spine. 2009;34:E405-13.
62. Snelling NJ. Spinal manipulation in patients with disc herniation: A critical review of risk and benefit. Int J Osteopath Med. 2006;9:77-84.
63. Cherkin DC, Sherman KJ, Deyo RA, Shekelle PG. A review of the evidence for the effectiveness, safety, and cost of acupuncture, massage therapy, and spinal manipulation for back pain. Ann Intern Med. 2003;138:898-906.
64. Huang SL, Liu YX, Yuan GL, Zhang J, Yan HW. Characteristics of lumbar disc herniation with exacerbation of presentation due to spinal manipulative therapy. Medicine. 2015;94:e661.
65. Powell FC, Hanigan WC, Olivero WC. A risk/benefit analysis of spinal manipulation therapy for relief of lumbar or cervical pain. Neurosurgery. 1993;33:73-8; discussion 8-9.
66. Schmidley JW, Koch T. The noncerebrovascular complications of chiropractic manipulation. Neurology. 1984;34:684-5.
67. Morandi X, Riffaud L, Houedakor J, Amlashi SF, Brassier G, Gallien P. Caudal spinal cord ischemia after lumbar vertebral manipulation. Joint Bone Spine. 2004;71:334-7.
68. Tamburrelli FC, Genitiempo M, Logroscino CA. Cauda equina syndrome and spine manipulation: case report and review of the literature. Eur Spine J. 2011;20 Suppl 1:S128-31.
20
69. Oppenheim JS, Spitzer DE, Segal DH. Nonvascular complications following spinal manipulation. Spine J. 2005;5:660-6; discussion 6-7.
70. Rebain R, Baxter GD, McDonough S. The passive straight leg raising test in the diagnosis and treatment of lumbar disc herniation: a survey of United kingdom osteopathic opinion and clinical practice. Spine. 2003;28:1717-24.
71. Szklo M, Nieto J. Epidemiology: beyond the basics. 2nd ed. Sudbury, MA: Jones and Bartlett Publishers; 2006.
72. Boucher P, Robidoux S. Lumbar disc herniation and cauda equina syndrome following spinal manipulative therapy: a review of six court decisions in Canada. Journal of forensic and legal medicine. 2014;22:159-69.
73. Jagbandhansingh MP. Most common causes of chiropractic malpractice lawsuits. J Manipulative Physiol Ther. 1997;20:60-4.
74. Spiegelhalter DJ. Incorporating Bayesian ideas into health-care evaluation. Statistical Science. 2004;19:156-74.
75. Spiegelhalter DJ, Myles JP, Jones DR, Abrams KR. Methods in health service research. An introduction to bayesian methods in health technology assessment. BMJ. 1999;319:508-12.
76. Spiegelhalter D, Rice K. Bayesian statistics. Scholarpedia. 2009;4(8):5230. http://www.scholarpedia.org/article/Bayesian_statistics. Accessed 2015-May-15.
77. O'Hagan A, Buck CE, Daneshkhah A, et al. Uncertain judgements: eliciting experts' probabilities. West Sussex: John Wiley & Sons Ltd; 2006.
78. Lilford RJ, Thornton JG, Braunholtz D. Clinical trials and rare diseases: a way out of a conundrum. BMJ. 1995;311:1621-5.
79. Dunson DB. Commentary: practical advantages of Bayesian analysis of epidemiologic data. Am J Epidemiol. 2001;153:1222-6.
80. Hampson LV, Whitehead J, Eleftheriou D, Brogan P. Bayesian methods for the design and interpretation of clinical trials in very rare diseases. Stat Med. 2014;33:4186-201.
81. Hughes MD, Williams PL. Challenges in using observational studies to evaluate adverse effects of treatment. N Engl J Med. 2007;356:1705-7.
82. Smeeth L, Douglas I, Hubbard R. Commentary: we still need observational studies of drugs--they just need to be better. Int J Epidemiol. 2006;35:1310-1.
83. Maclure M, Fireman B, Nelson JC, et al. When should case-only designs be used for safety monitoring of medical products? Pharmacoepidemiol Drug Saf. 2012;21 Suppl 1:50-61.
84. Farrington CP, Nash J, Miller E. Case series analysis of adverse reactions to vaccines: a comparative evaluation. Am J Epidemiol. 1996;143:1165-73.
85. Whitaker HJ, Hocine MN, Farrington CP. The methodology of self-controlled case series studies. Stat Methods Med Res. 2009;18:7-26.
86. Maclure M. The case-crossover design: a method for studying transient effects on the risk of acute events. Am J Epidemiol. 1991;133:144-53.
87. Farrington P, Pugh S, Colville A, et al. A new method for active surveillance of adverse events from diphtheria/tetanus/pertussis and measles/mumps/rubella vaccines. Lancet. 1995;345:567-9.
88. Farrington CP, Whitaker HJ, Hocine MN. Case series analysis for censored, perturbed, or curtailed post-event exposures. Biostatistics. 2009;10:3-16.
89. Greenland S. Confounding and exposure trends in case-crossover and case-time-control designs. Epidemiology. 1996;7:231-9.
21
90. Vines SK, Farrington CP. Within-subject exposure dependency in case-crossover studies. Stat Med. 2001;20:3039-49.
91. Janes H, Sheppard L, Lumley T. Overlap bias in the case-crossover design, with application to air pollution exposures. Stat Med. 2005;24:285-300.
92. Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 [updated March 2011]. In: Higgins JPT, Green S, eds: The Cochrane Collaboration; 2011: http://www.cochrane-handbook.org. Accessed 2015-May-15.
22
Chapter 2
Incidence and Determinants Of Lumbar Disc Herniation With 2Radiculopathy In Adults: A Systematic Review
Manuscript 1
Hincapié CA, Cassidy JD, Côté P, Rampersaud YR, Boyle E, Tomlinson GA, Jadad AR.
Incidence and determinants of lumbar disc herniation with radiculopathy in adults: a
systematic review. In preparation for submission, 2015
Incidence and determinants of lumbar disc herniation with radiculopathy in
adults: a systematic review
Authors: Cesar A. Hincapié, DC, MHSc,1,2 J. David Cassidy, PhD, DrMedSc,1,2,3 Pierre
Côté, DC, PhD,4 Y. Raja Rampersaud, MD, MSc,5 Eleanor Boyle, PhD,1,3 George A.
Tomlinson, PhD,1,6 Alejandro R. Jadad, MD, DPhil1,6
Affiliations: 1 Dalla Lana School of Public Health, University of Toronto; 2 Toronto
Western Research Institute, University Health Network; 3 Department of Sports Science
and Clinical Biomechanics, University of Southern Denmark; 4 Faculty of Health
Sciences, University of Ontario Institute of Technology; 5 Division of Orthopaedic
Surgery, Toronto Western Hospital; 6 Toronto General Research Institute, University
Health Network
Corresponding Author: Cesar A. Hincapié, DC, MHSc, University Health Network,
LuCliff Place – 700 Bay Street, Suite 602, Toronto, Ontario M5G 1Z6, Canada. E-mail:
Keywords: intervertebral disc displacement; radiculopathy; sciatica; low back pain;
epidemiology; incidence; risk factors; systematic review
Word Count: 6,146
23
Abstract
Background: Lumbar disc herniation (LDH) is the most common cause of lumbosacral
radiculopathy. Compared to nonspecific low back pain, LDH with radiculopathy is
associated with greater disability, healthcare use and intervention. However, little is
known about its epidemiology.
Purpose: To assemble and synthesize the best available evidence on the incidence
and determinants of LDH with radiculopathy in adults.
Study design: Systematic review.
Methods: Medline, Embase and 3 Cochrane Library databases were systematically
searched from 1970 to December 2010. Identified abstracts were screened for
relevance by two independent reviewers using pre-specified inclusion and exclusion
criteria. Eligible articles were critically appraised for quality and risk of bias by teams of
two independent reviewers using the SIGN criteria. Data from eligible studies were
abstracted into evidence tables.
Results: After 7,430 abstracts were screened, 35 studies were critically reviewed, and
26 (74%) were included as scientifically admissible. The annual incidence of
hospitalized or surgically managed LDH with radiculopathy in the general population
ranged from 0.2 to 1.3 per 1,000 persons. However, much LDH with radiculopathy is
not treated in hospitals. The annual incidence of clinically defined LDH with
radiculopathy varied from 6.2 per 1,000 persons (0.6%) among US female nurses, to 93
per 1,000 persons (9.3%) among forest industry workers in Finland. Nonmodifiable risk
factors included age and sex. Modifiable risk factors included lower education, higher
BMI and cardiovascular risk factors in women (diabetes, high cholesterol, hypertension,
and a family history of coronary heart disease), smoking, high cumulative occupational
lumbar load by forward bending (work postures with ≥20-degrees of forward bending or
extreme forward bending of >90-degrees) and manual materials handling, higher levels
of perceived risk of work injury, lower job control (independence and ability to influence
work methods and pace), regular or irregular three-shift work or regular night work in
women, and increased time pressure at work. We found preliminary evidence that
24
manual occupation, genetics, and previous back pain may be associated with LDH with
radiculopathy.
Conclusions: Although the literature is of varying quality and heterogeneous, there is
evidence that LDH with radiculopathy is an important source of pain and disability in
society. The annual incidence ranges roughly between 0.1% and 10%, depending on
case definition, and risk factors are varied. Our findings support the need to develop
standardized case and surveillance definitions that validly classify the clinical spectrum
of this condition. Future research should focus on prospective designs examining
potentially modifiable risk factors and prevention strategies.
2.1 Introduction
Low back pain is a leading cause of disability and associated with a large economic
burden on individuals, industry and society.1-5 Radiculopathy due to lumbar disc
herniation (LDH) is one of the most recognizable presentations of low back pain. The
diagnosis is typically based on a combination of symptoms and signs suggesting lumbar
spinal nerve root compression or irritation, such as nerve root tension signs, neurologic
deficits, and advanced imaging findings that correlate with the clinical syndrome.6-8
LDH, defined as the localized displacement of disc material beyond the margins of the
intervertebral disc space,9 is considered the most common cause of lumbosacral
radiculopathy.10-12 Compared with nonspecific low back pain without radiating leg pain,
LDH with radiculopathy is associated with greater pain, disability, healthcare use and
intervention.13-16 However, little is known about the epidemiology of this important
condition.
Previous studies of symptomatic LDH have reported a point prevalence of about 5%
among the general adult population,17,18 varying by sex and age. In people aged 25–55
years, 95% of symptomatic herniated discs occur at the lower lumbar spine (L4-L5 and
L5-S1 levels).19 Nonetheless, very little is known about the incidence of LDH with
radiculopathy, and consequently, risk factors are not well understood.
The objective of our systematic review was to synthesize the best available evidence on
the incidence and determinants of LDH with radiculopathy in adults. Our aim was to
25
create a baseline of the best scientific evidence to inform clinicians, researchers and
policymakers about the epidemiology of LDH with radiculopathy.
2.2 Methods
Our review was conducted in accordance with the Preferred Reporting Items for
Systematic Reviews and Meta-Analyses (PRISMA)20,21 statement. Details of our
protocol were registered on the Prospective Register of Systematic Reviews
(PROSPERO) (CRD42011001197).22
Data Sources and Searches: We systematically searched Medline, Embase,
Cochrane Central Trials Registry, Cochrane Database of Systematic Reviews, and
Database of Abstracts Related to Effects, from 1970 to December 2010. Search
strategies (Appendix C) combined terms from three key concepts: intervertebral disc
herniation, lumbar spine, and etiology, and were developed in consultation with an
experienced Information Specialist. The reference lists of all articles eligible were
examined for additional relevant studies.
Study Selection: A three-phase screening process was used to identify eligible articles
for our review on the basis of pre-specified inclusion and exclusion criteria. Eligible
were published reports in English, French or Spanish that used cohort, case-control, or
randomized controlled trial designs that examined the incidence and/or determinants
(i.e., risk factors) of LDH with radiculopathy in adults. We excluded studies of
asymptomatic LDH, or LDH due to neoplasm, infection, fracture, dislocation or spinal
cord injury. Narrative reviews, case reports or series, cross-sectional, cadaveric or
animal studies were ineligible due to their inherent limitations for estimating incidence
and inferring etiology. After consulting clinical experts, LDH with radiculopathy was
defined as lumbosacral radiculopathy, sciatica and/or clinically relevant neurologic
deficit, with or without advanced imaging (i.e., MRI or CT) confirmation of disc
herniation.
In the first-level screen, one reviewer (CAH) rated all the citations from the search as
relevant, unsure, or irrelevant. In the second-level screen, three reviewers (CAH, JDC,
PC) independently identified articles meeting the review inclusion and exclusion criteria
among those citations rated as relevant or unsure from the first-level screen (n=282). In
26
the third-level screen, each article considered possibly relevant to the review was
retrieved and underwent full-text analysis for eligibility by two independent reviewers
(CAH/JDC or CAH/PC). When necessary, discrepancies were resolved by consensus
and arbitration by a third reviewer.
Risk of Bias Assessment and Data Extraction: Eligible articles were assessed for
risk of bias by teams of 2 independent reviewers (CAH reviewed all articles; other
reviewers were JDC, PC, ARJ, YRR, and EB) and using checklists based on criteria
recommended by the Scottish Intercollegiate Guidelines Network (SIGN).23 The SIGN
checklists include key questions/items that focus on aspects of study design that have
been shown to influence the validity of the results reported (e.g., selection of subjects,
assessment of exposures and outcomes, confounding assessment), and have been
evaluated and adapted to meet a balance between methodological rigour and
practicality of use.23 An overall assessment of risk of bias was added and rated as low,
moderate or high. When further information was required from original authors, three
attempts were made to contact authors by email. We used consensus and arbitration
by a third reviewer, when necessary, to resolve discrepancies on risk of bias
assessment.
One reviewer (CAH) extracted data from eligible studies into evidence tables relating to
incidence and risk factors of LDH with radiculopathy in adults. We extracted data on
study and population characteristics, case definitions and study outcomes, risk factors
considered, and estimates of incidence and risk.
Data Synthesis and Analysis: We synthesized the literature according to the
principles of best evidence synthesis.24 In best evidence synthesis, scientifically
admissible studies are qualitatively synthesized, with more weight given to evidence
from studies judged to be least vulnerable to bias.25,26 The evidence tables summarize
our findings and form the basis of our recommendations. We defined low and moderate
risk of bias studies as scientifically admissible, but also tabled high risk of bias studies in
order to examine their findings and how they differ from admissible studies.
To better delineate the strength of the evidence, risk factor evidence was categorized as
phase I, II or III.25,27,28 This categorization distinguishes between 3 hierarchic phases of
research aimed at determining causal relationships between a risk factor and a health
27
outcome. Phase I studies describe crude associations between potential determinants
and the outcome of interest. Phase II studies are exploratory analyses that focus on
particular sets of risk factors or attempt to discover which risk factors predict the
development of a health outcome without an explicit attempt to control for confounding.
Phase III investigations are confirmatory studies of explicitly pre-stated hypotheses that
allow for the quantification of the strength, direction, and independence of the proposed
relationship between a risk factor and the development of LDH with radiculopathy. Both
Phase I and II studies provide preliminary evidence of a potential risk factor’s
association with the outcome, whereas Phase III studies provide confirmatory evidence.
Our evidence tables report annual incidence and risk factor effect estimates with 95%
confidence intervals (CIs) as reported in original studies, or calculated from raw data
presented in original studies. Risk factor effect estimates are reported as odds ratios
(ORs), risk ratios (RRs), or hazard rate ratios (HRRs). When necessary, we computed
these statistics using R, version 2.15.1.29 Confidence intervals were calculated using
standard methods.30
Role of the Funding Source: The sponsors of the study had no role in study design,
data collection, data analysis, data interpretation, the writing of the report, or in the
decision to submit the report for publication.
2.3 Results
Search, Selection and Critical Appraisal: Our literature search yielded 8,617
citations, of which 95 articles were deemed potentially eligible after removing duplicates
and screening for relevance; these underwent full-text analysis for eligibility (Figure
2.1). Of these, 46 articles (representing 35 studies) met our eligibility criteria and were
reviewed. After risk of bias assessment and critical appraisal, we rated 9 studies (26%)
as inadmissible (high risk of bias) and 26 studies (74%) as admissible (low or moderate
risk of bias). These 26 admissible studies (represented by 37 articles) are the basis for
our best evidence synthesis.
28
Figure 2.1. Flow diagram of information through phases of the systematic review.
Abbreviations: LDH, lumbar disc herniation; ROB, risk of bias.
Records identified through database searching
(n = 8,559) S
cree
ning
In
clud
ed
Elig
ibili
ty
Iden
tific
atio
n Additional records identified through other sources
(n = 58)
Records after duplicates removed (n = 7,430)
Records screened (n = 7,430)
Records excluded (n = 7,335)
Full-text articles assessed for eligibility
(n = 95)
Full-text articles excluded (n = 49)
No symptomatic LDH (n=20)
Ineligible design (n=20) Not on incidence or risk (n=5) Wrong publication type (n=2)
Ineligible population (n=1) No estimates (n=1)
Articles eligible for review (n = 46) (representing 35
studies)
Low ROB (n=10) Moderate ROB (n=27)
High ROB (n=9)
Admissible studies (n=26) (represented by 37 articles)
Low ROB studies (n=8) Moderate ROB studies (n=18)
Incidence studies (n=15) Risk factor studies (n=22)
Inadmissible (high ROB) studies (n=9)
Incidence studies (n=2) Risk factor studies (n=9)
29
Study Characteristics: Table 2.1 presents a summary of characteristics of the 35
eligible studies, and Table 2.2 provides summary details of the eligible studies. Overall,
8 studies (23%) were rated as having a low risk of bias; 18 studies (51%), a moderate
risk of bias; and, 9 studies (26%), a high risk of bias. The most common sources of bias
included poor or inadequate description of sampling methods, exposure measurement,
and consideration of confounding. Appendix D, Table D.1 provides full descriptions of
the 17 eligible studies (15 admissible, 2 inadmissible) that reported on incidence.31-48
Appendix D, Table D.2 details the 31 eligible studies (22 admissible, 9 inadmissible)
that reported on risk factors.31,33,34,36,37,39,40,42-76
Table 2.1. Characteristics of eligible studies examining the incidence of and/or risk factors for lumbar
disc herniation with radiculopathy in adults
By admissibility*
Admissible studies by
focus†
Characteristic
All Studies
(n=35)
Admissible
(n=26)
Inadmissible
(n=9)
Incidence
(n=15) Risk (n=22)
Risk of bias – n (%)
Low 8 (23) 8 (31) na 8 (53) 5 (23)
Moderate 18 (51) 18 (69) na 7 (47) 17 (77)
High 9 (26) na 9 (100) na na
Focus – n (%)
Incidence only 4 (11) 4 (15) 0 4 (27) na
Risk only 18 (51) 11 (42) 7 (78) na 11 (50)
Both incidence and risk 13 (37) 11 (42) 2 (22) 11 (73) 11 (50)
Study design – n (%)
Cohort 18 (51) 15 (58) 3 (33) 15 (100) 11 (50)
Case-control 17 (49) 11 (42) 6 (67) 0 11 (50)
Case definition type – n (%)
Surgical or hospital 16 (46) 10 (38) 6 (67) 9 (60) 7 (32)
Clinical 15 (43) 13 (50) 2 (22) 6 (40) 12 (55)
Mixed surgical/hospital/clinical 4 (11) 3 (12) 1 (11) 0 3 (14)
30
By admissibility*
Admissible studies by
focus†
Characteristic
All Studies
(n=35)
Admissible
(n=26)
Inadmissible
(n=9)
Incidence
(n=15) Risk (n=22)
Population source – n (%)
General 6 (17) 6 (23) 0 6 (40) 4 (18)
Occupational 12 (34) 8 (31) 4 (44) 8 (53) 7 (32)
Health care 17 (49) 12 (46) 5 (56) 1 (7) 11 (50)
Sex – n (%)
Men only 9 (26) 6 (23) 3 (33) 4 (27) 5 (23)
Women only 2 (6) 2 (8) 0 2 (13) 2 (9)
Both men and women 24 (69) 18 (69) 6 (67) 9 (60) 15 (68)
Incidence estimate type‡ – n (%) (n=17) (n=15) (n=2) (n=15)
Cumulative incidence only 11 (65) 11 (73) 0 11 (73) na
Incidence rate only 5 (29) 4 (27) 1 (50) 4 (27) na
Both estimate types 1 (6) 0 1 (50) 0 na
Risk factors considered§ – n (%) (n=31) (n=22) (n=9) (n=22)
Sociodemographic 24 (77) 19 (86) 5 (56) na 19 (86)
General health, pain, comorb 17 (55) 11 (50) 6 (67) na 11 (50)
Health behaviours 15 (48) 12 (55) 3 (33) na 12 (55)
Personal – Physical 9 (29) 6 (27) 3 (33) na 6 (27)
Personal – Psychosocial 7 (23) 7 (32) 0 na 7 (32)
Workplace – Physical 15 (48) 11 (50) 4 (44) na 11 (50)
Workplace – Psychosocial 6 (19) 5 (23) 1 (11) na 5 (23)
Phase of evidence|| – n (%) (n=31) (n=22) (n=9) (n=22)
I 12 (39) 7 (32) 5 (56) na 7 (32)
II 15 (48) 11 (50) 4 (44) na 11 (50)
III 2 (6) 2 (9) 0 na 2 (9)
Mixed (Phase II and III) 2 (6) 2 (9) 0 na 2 (9)
31
Abbreviation: comorb, comorbidities; na, not applicable
* Admissible studies were those rated has having a low or moderate risk of bias; inadmissible studies
were those rated as having a high risk of bias
† Categories are not mutually exclusive
‡ Proportions among studies on incidence; denominators indicated in header row
§ Proportions among studies on risk factors and not mutually exclusive; denominators indicated in
header row
|| Hierarchical categories of risk factor evidence: phase I and II is preliminary evidence for an
association, phase III is confirmatory evidence from better quality studies
The 26 scientifically admissible studies consisted of 15 cohort studies,31-46,49,50,60 and 11
case-control studies.51-59,61-69 Most of the admissible studies were conducted in Finland
(46%) and the US (23%). Ten studies (38%) used case definitions based either on
surgery or hospital treatment, 13 (50%) used definitions based on clinical signs and
symptoms, and 3 (12%) used mixed case definitions combining clinical, hospital and
surgical definitions.
32
Table 2.2. Summary of eligible studies examining the incidence of and/or risk factors for lumbar disc herniation with radiculopathy in adults
First author,
Year
published
Focus Country
Study
design
Source
population
Participants and setting
Follow-up
Age
range
(mean)
N (% female);
Participation
% Study outcome
Case
definition
type ROB Phase
Heliövaara,
1987 31,49,50
Both*
Finland Cohort
with
case-
control
risk
analyses
General Nationwide cohort linked to
the National Hospital
Discharge Register
Follow-up: 11y
≥15y
(NR)
57,000 (48%);
83%
Hospitalized LDH
or sciatica
Hospital Low II
Burske-
Hohlfeld,
1990 32
Incidence
US Cohort General All LDH surgeries of
Olmsted County residents
Follow-up: 30y
15-78y
(42)
1,028 (NR);
88% incident
surgery
LDH surgery Surgical Low NA
Zitting, 1998 33
Both
Finland Cohort General Birth cohort from 2
provinces linked to the
National Hospital
Discharge Register
Follow-up: 28y
15-28y
among
cases
(NR)
12,058 (NR);
92%
Hospitalized LDH Hospital Low II
Miranda, 2002 34
Both
Finland Cohort Occupational Forest industry workers
responding to a
questionnaire
Follow-up: 1y
NR (45) 2,077 (26%);
77%
Sciatica Clinical Low II
33
First author,
Year
published
Focus Country
Study
design
Source
population
Participants and setting
Follow-up
Age
range
(mean)
N (% female);
Participation
% Study outcome
Case
definition
type ROB Phase
Jarvik, 2005 35
Incidence
US Cohort Health care Veterans Affairs outpatients
at the Puget Sound Health
Care System, Seattle
Division
Follow-up: 3y
35-70y
(median
, 53)
148 (13%);
89%
Clinical LDH Clinical Low NA
Jhawar, 2006 36
Both
US Cohort Occupational Female nurses from the
Nurses’ Health Study
responding to a
questionnaire
Follow-up: 16y
30-55y in
1976
(NR)
98,407
(100%);
≥85%
Clinical LDH Clinical Low III
Mattila, 2008 37
Both
Finland Cohort General Nationwide adolescent
cohort linked to the
National Hospital
Discharge Register
Follow-up: 651,027 person-
years
15-41y
among
cases
(27 at
surgery
)
57,408 (54%);
79%
LDH surgery Surgical Low II
34
First author,
Year
published
Focus Country
Study
design
Source
population
Participants and setting
Follow-up
Age
range
(mean)
N (% female);
Participation
% Study outcome
Case
definition
type ROB Phase
Mattila, 2009 38
Incidence
Finland Cohort Occupational All male military conscripts
linked to the National
Hospital Discharge
Register
Follow-up: 267,700 person-
years
18-29y
(20)
387,070 (0%);
100%
Hospitalized LDH Hospital Low NA
Kelsey, 1975 51-55
Risk
US Case-
control
Health care Cases: 223 radiology and
hospital patients with
LDH, 1971-1973
Controls: 217 controls
matched on age, sex, and
medical setting; 494
unmatched controls
Follow-up: NA
20-64y
(39
among
cases)
934 (41%);
78%
Combined
clinical,
hospitalized
and surgical
LDH
Clinical,
hospital
and
surgical
Mod II
Hurme, 1983 39
Both
Finland Cohort General AII LDH surgeries from
surgical registers of the
Turku University Central
Hospital area
Follow-up: 5y
15-80y
among
cases
(42 at
surgery
)
1,011
surgeries
(44%);
79% incident
surgery
LDH surgery Surgical Mod I
35
First author,
Year
published
Focus Country
Study
design
Source
population
Participants and setting
Follow-up
Age
range
(mean)
N (% female);
Participation
% Study outcome
Case
definition
type ROB Phase
Kelsey, 1984 56,57
Risk
US Case-
control
Health care Cases: 325 orthopaedic,
neurosurgical, and
hospital patients with
LDH, 1979-1981
Controls: 241 controls
matched on age, sex, and
medical setting.
Follow-up: NA
20-64y
(NR)
566 (45%);
72-79%
Combined
clinical,
hospitalized
and surgical
LDH
Clinical,
hospital
and
surgical
Mod II
Riihimäki,
1989 40
Both
Finland Cohort Occupational Male concrete
reinforcement workers
and house painters
responding to a
questionnaire
Follow-up: 5y
25-54y at
baselin
e (NR)
178 (0%);
77-80%
Sciatica Clinical Mod II
Heikkilä, 1989 41
Incidence
Finland Cohort General Finnish adult twin cohort of
the same sex linked to the
National Hospital
Discharge Register
Follow-up: 14y
24-60y+
(NR)
18,730 (55%);
100%
Hospitalized
sciatica
Hospital Mod NA
36
First author,
Year
published
Focus Country
Study
design
Source
population
Participants and setting
Follow-up
Age
range
(mean)
N (% female);
Participation
% Study outcome
Case
definition
type ROB Phase
Mundt, 1993 58,59
Risk
US Case-
control
Health care Cases: 297 orthopaedic,
neurosurgical, and
hospital patients with
LDH, 1986-1988
Controls: 287 clinical and
hospital controls
Follow-up: NA
20-64y
(NR)
585 (41%);
76-79%
Combined
clinical,
hospitalized
and surgical
LDH
Clinical,
hospital
and
surgical
Mod II
Jørgensen,
1994 42
Both
Denmark Cohort Mixed Assistant nurses and all
females linked to the
Danish National Registry
of Hospitalized Patients
Follow-up: 1y
20-69y
(NR)
1,681,152
(100%);
100%
LDH surgery Surgical Mod II
Riihimäki,
1994 43;
Pietri-Taleb,
1995 60
Both
Finland Cohort Occupational Male machine operators,
carpenters and office
workers responding to a
questionnaire
Follow-up: 3y
25-49y
(37)
1,149 (0%);
83%
Sciatica Clinical Mod II
37
First author,
Year
published
Focus Country
Study
design
Source
population
Participants and setting
Follow-up
Age
range
(mean)
N (% female);
Participation
% Study outcome
Case
definition
type ROB Phase
a Leino-Arjas,
2002 44 b Leino-Arjas,
2004 45
Both
Finland Cohort Occupational Nationwide workforce linked
to the National Hospital
Discharge Register
Follow-up: 1y
a 20-64y
(NR) b 25-64y
(NR)
a 2,409,319
(NR);
100% b 1,783,616
(51%);
100%
Hospitalized LDH
and LDH
surgery
Hospital
and
surgical
Mod a II b III
Leclerc, 2003 46
Both
France Cohort Occupational Male workers in the national
electricity and gas
company responding to a
questionnaire
Follow-up: 2y
40-50y at
baselin
e (NR)
841 (0%);
65%
Sciatica Clinical Mod II
Siedler, 2003 61
Risk
Germany Case-
control
Health care Cases: 225 male patients
with acute LDH from
Frankfurt/Main
Controls: 107 population
and 90 hospital controls
Follow-up: NA
25-65y
(42)
437 (0%);
66-93%
Clinical LDH Clinical Mod III
38
First author,
Year
published
Focus Country
Study
design
Source
population
Participants and setting
Follow-up
Age
range
(mean)
N (% female);
Participation
% Study outcome
Case
definition
type ROB Phase
Noponen-
Hietela,
2005 62
Risk
Finland Case-
control
Health care Cases: 155 unrelated
Finnish patients with
sciatica from the Oulu
University Hospital area,
1997-1998
Controls: 179 unrelated
University of Oulu
employees and students
(all Finnish)
Cases:
19-78y
(44)
Controls:
20-69y
(39)
334 (39%
among
cases, 69%
among
controls);
NR
Sciatica Clinical Mod I
Mio, 2007 63
Risk
Japan Case-
control
Health care Cases: 823 hospital patients
of Japanese origin with
LDH
Controls: 841 hospital
controls of Japanese
origin
Follow-up: NA
Cases:
11-83y
(36)
Controls:
13-87y
(61)
1,664 (41%
among
cases, 63%
among
controls;
NR
Clinical LDH Clinical Mod I
39
First author,
Year
published
Focus Country
Study
design
Source
population
Participants and setting
Follow-up
Age
range
(mean)
N (% female);
Participation
% Study outcome
Case
definition
type ROB Phase
Virtanen, 2007 64
Risk
Finland Case-
control
Health care Cases: 243 unrelated
Finnish patients with
sciatica from the Oulu
University Hospital area
Controls: 259 unrelated
Finnish persons from the
same catchment area
Follow-up: NA
NR (NR) 502 (45%);
NR
Sciatica Clinical Mod I
Hirose, 2008 65
Risk
Japan Case-
control
Health care Cases: 847 hospital patients
of Japanese origin with
LDH, 2001-2007
Controls: 896 Japanese
persons from the same
catchment area
Follow-up: NA
Cases:
NR (39)
Controls:
NR (62)
1,743 (38%);
NR
Clinical LDH Clinical Mod I
40
First author,
Year
published
Focus Country
Study
design
Source
population
Participants and setting
Follow-up
Age
range
(mean)
N (% female);
Participation
% Study outcome
Case
definition
type ROB Phase
Karasugi,
2009 66
Risk
Japan
and
Finland
Case-
control
Health care Japan
Cases: 862 hospital patients
of Japanese origin with
LDH, 2001-2007
Controls: 896 Japanese
persons from the same
catchment area
Finland
Cases: 257 unrelated
Finnish patients with
sciatica from the Oulu
University Hospital area
Controls: 249 unrelated
Finnish persons from the
same area
Follow-up: NA
Japan
Cases:
NR (39)
Controls:
NR (62)
Finland
NR (NR)
Japan
1,758 (38%);
NR
Finland
506 (NR);
NR
Clinical LDH and
sciatica
Clinical Mod I
41
First author,
Year
published
Focus Country
Study
design
Source
population
Participants and setting
Follow-up
Age
range
(mean)
N (% female);
Participation
% Study outcome
Case
definition
type ROB Phase
a Siedler, 2009 67
b Schumann,
2010 68
Risk
Germany Case-
control
Health care Cases: 564 patients with
hospital treatment for LDH
pain in 4 regions of
Germany
Controls: 901 population
controls
Follow-up: NA
25-70y
(48)
1,816 (50%);
53-66%
LDH hospital
treatment
Hospital Mod a III b II
Cong, 2010 69
Risk
China Case-
control
Health care Cases: 70 male hospital
patients of Chinese Han
origin with LDH
Controls: 14 male spinal
trauma controls and 113
male healthy blood donor
controls
Follow-up: NA
Cases:
14-41y
(33)
Controls:
20-49y
(38)
197 (0%);
NR
Clinical LDH Clinical Mod I
42
First author,
Year
published
Focus Country
Study
design
Source
population
Participants and setting
Follow-up
Age
range
(mean)
N (% female);
Participation
% Study outcome
Case
definition
type ROB Phase
Hrubec, 1975 70
Risk
US Case-
control
Occupational Cases: 1,132 first admission
records to Army hospitals
for LDH, 1944-1945
Controls: 1,095 records of
Army National Service Life
Insurance policyholders
matched on age and
period of military service
Follow-up: NA
18-56y
(NR)
1,095 case-
control pairs
(0%);
97%
Hospitalized LDH Hospital High II
Bongers, 1988 47
Both
Holland Cohort Occupational Disability pension due to
LDH in male crane
workers and floor workers
at a steel company
Follow-up: 10y
<25-60y+
(NR)
1,405 (0%);
71%
Disability
pension due to
LDH
Clinical High II
Netterstrøm,
1989 48
Both
Denmark Cohort Occupational All full-time male bus drivers
employed by 3 urban bus
companies linked to the
Danish National Patient
Register
Follow-up: 7y
20-69y
(NR)
2,465 (0%);
100%
Hospitalized LDH Hospital High I
43
First author,
Year
published
Focus Country
Study
design
Source
population
Participants and setting
Follow-up
Age
range
(mean)
N (% female);
Participation
% Study outcome
Case
definition
type ROB Phase
An, 1994 71
Risk
US Case-
control
Health care Cases: 163 consecutive
surgical patients with
LDH, 1987-1988
Controls: 205 inpatient
controls matched on sex
and age
Follow-up: NA
16-78y
(45)
368 (39%);
NR
LDH surgery Surgical High I
Chibnall, 2006 72
Risk
US Cohort Occupational African American and non-
Hispanic white workers’
compensation claimants
who filed low back injury
claims
Follow-up: NR
18-55y
(NR)
2,934 (38%);
50%
Clinical and
surgical LDH
Clinical
and
surgical
High II
44
First author,
Year
published
Focus Country
Study
design
Source
population
Participants and setting
Follow-up
Age
range
(mean)
N (% female);
Participation
% Study outcome
Case
definition
type ROB Phase
Saftic, 2006 73
Risk
Croatia Case-
control
Health care Cases: 67 adults from 9
villages on the Croatian
islands with a history of
LDH surgery
Controls: 268 adults
matched on age, gender,
and village of
residence/immigrant
status
Follow-up: NA
≥18y
(NR)
365 (NR); NR LDH surgery Surgical High I
Lee, 2006 74
Risk
Korea Case-
control
Health care Cases: 119 LDH levels in
111 adult patients who
underwent LDH surgery,
2000-2002
Controls: 82 normal disc
levels adjacent to the
herniated levels in the
same patients
Follow-up: NA
40-49y
(NR)
201 adult disc
levels (37%);
NR
LDH surgery Surgical High I
45
First author,
Year
published
Focus Country
Study
design
Source
population
Participants and setting
Follow-up
Age
range
(mean)
N (% female);
Participation
% Study outcome
Case
definition
type ROB Phase
Kunakornsawa
t, 2007 75
Risk
Thailand Case-
control
Health care Cases: 34 LDH levels in 34
adult patients who
underwent LDH surgery,
2001-2003
Controls: 34 normal disc
levels adjacent to the
herniated levels in the
same patients
Follow-up: NA
23-45y
(34)
68 disc levels
(35%);
NR
LDH surgery Surgical High I
Zhang, 2009 76
Risk
China Case-
control
Health care Cases: 2,010 orthopaedic
hospital patients with
LDH, 2005-2007
Controls: 2,170 randomly
selected hospital controls
matched on race, gender,
age, and living area
Follow-up: NA
<30-50y+
(46)
4,180 (40%);
NR
Clinical LDH Clinical High II
Abbreviations: LDH, lumbar disc herniation; Mod, Moderate; N, study size; NA, not applicable; NR, not reported; ROB, risk of bias; y, years
* Study provides data on both incidence and risk factors
46
Incidence: We accepted 15 cohort studies (represented by 16 articles) on the incidence
of LDH with radiculopathy in adults (summarized in Tables 2.1 and 2.2, and fully
detailed in Appendix D, Table D.1).
Ten of the 15 studies (67%) were conducted in Finland,31,33,34,37-41,43-45 3 (20%) in the
United States,32,35,36 and one each in Denmark42 and France.46
Source populations varied. Six studies estimated the incidence in general
populations,31-33,37,39,41 eight studies examined occupational populations,34,36,38,40,42-46
and one study investigated incidence in a health care population.35 The majority of
studies included both genders and a broad age range.
The incidence of LDH with radiculopathy varied by case definition. Among six general
population studies using a surgical or hospital-based case definition, the annual
incidence ranged from 0.2 per 1,000 persons to 1.3 per 1,000 persons (Appendix D,
Table D.1).31-33,37,39,41 In two studies investigating occupational populations and using a
case definition based on surgery or hospitalization, the annual incidence of LDH with
radiculopathy ranged between 1.0 per 1,000 persons and 2.5 per 1,000 persons.42,44,45
One study from Finland found a higher incidence rate of 7.8 per 1,000 person-years of
hospitalized LDH in male military conscripts performing compulsory service between
1990 and 2002.38
Studies using clinical signs and symptoms as the basis for case definition of LDH with
radiculopathy found higher estimates of incidence, suggesting that much symptomatic
LDH is not treated in hospitals. Among female nurses from the US-based Nurses’
Health Study, the annual incidence of clinical LDH with radiculopathy was estimated at
6.2 per 1,000 person-years.36 Five other cohort studies described the incidence of
clinical LDH with radiculopathy in electrical and gas workers,46 Veterans Affairs
outpatients,35 house painters and concrete workers,40 forest industry workers,34 and
office workers, machine operators, and carpenters.43 These found annual incidence
estimates ranging from a low of 28 per 1,000 persons (2.8%) among electrical and gas
workers in France,46 to a high of 93 per 1,000 persons (9.3%) among forest industry
workers in Finland34 (Appendix D, Table D.1).
47
Given the heterogeneity in inception periods, diverse source populations, varied case
definitions for LDH with radiculopathy, and the different methods used to report
incidence in the admissible studies, the variability in incidence estimates presented in
Appendix D, Table D.1 is not surprising, and an overall summary incidence estimate
would not be very informative.
Risk Factors: Our best evidence synthesis included 11 cohort studies (represented by
15 articles) and 11 case-control studies (represented by 18 publications) that
investigated the risk factors for LDH with radiculopathy in adults (summarized in Tables
2.1 and 2.2, and fully detailed in Appendix D, Table D.2).
Ten of the 22 admissible studies (45%) were conducted in Finland,31,33,34,37,39,40,43-
45,49,50,60,62,64 4 (18%) in the US,36,51-59 3 (14%) in Japan,63,65,66 2 (9%) in Germany,61,67,68
and 1 each in China,69 Denmark,42 and France.46
Source populations varied, with 4 studies investigating risk factors in general
populations;31,33,37,39,49,50 7 studies examining occupational populations,34,36,40,42-46,60 and
11 studies focusing on health care populations.51-59,61-69
We present our findings prioritized by the hierarchical phases of evidence as described
in the Methods section, with Phase III studies presenting the strongest evidence of
association, and preliminary evidence provided by Phase II and I studies, in descending
order of strength.
Sociodemographic Characteristics:
Age. Evidence from one Phase III,45 and four Phase II studies31,34,42,44 indicates that
the incidence of LDH with radiculopathy increases with age, peaking about the fourth
and fifth decades of life, and then decreases in later life. Two Phase II studies reported
no association between age and the incidence of clinical LDH (sciatica) among male
worker cohorts.40,46
Gender. We found evidence of LDH with radiculopathy occurring more commonly in
men than in women. One Phase III45 study and 6 Phase II studies31,33,37,39,44,51,52
reported an increased risk for men. However, one Phase II study34 found no
48
association between gender and the development of clinical LDH with radiculopathy
(sciatica) in Finnish forestry workers.
Education. Evidence from one Phase III45 and one Phase II44 study suggests that the
incidence of LDH with radiculopathy decreases with years of education. Two Phase II
studies43,57 found no association between education and the risk of LDH.
Income. The evidence linking income and the risk of hospitalized LDH varies in one
Phase III and one Phase II study. In their Phase III analysis, Leino-Arjas and
colleagues found an increased incidence of hospital care due to LDH across lower
quintiles of personal net income compared to the highest quintile.45 However, in their
Phase II analysis, they reported a greater risk of hospitalized LDH among those with the
highest quartile of personal net income compared to those with the lowest quartile.44
Socioeconomic status. The evidence varies in three Phase II studies investigating
socioeconomic status and the incidence of LDH with radiculopathy. One study found a
higher risk among middle class men.31 Another study reported some indication of an
association between higher social class and LDH in women, but no association in
men.51,52 A third study found no link between socioeconomic background and the
incidence of LDH surgery in men or women.37
Race. Preliminary evidence from two Phase II studies suggests that race is not
associated with the incidence of LDH with radiculopathy.52,53,57
Occupation. We found preliminary evidence that manual occupations are associated
with the risk of LDH with radiculopathy. The studies reported in the Incidence section
above and in Appendix D, Table D.1 show that the incidence varies across
occupations. In addition, Phase II studies reported a greater risk for LDH with
radiculopathy in male concrete workers compared with male house painters;40 in female
assistant nurses compared with all Danish females;42 in machine operators and
carpenters compared with office workers;43 and in manual occupations compared with
upper white-collar occupations.44 However, a French Phase II study by Leclerc et al.
found no significant difference in the risk of developing clinical LDH with radiculopathy
among male managers, supervising employees and operating employees in a national
electricity and gas company.46
49
Body Mass Index, Anthropometrics and Genetics:
Body mass index and obesity. The association between obesity and the risk of LDH
with radiculopathy seems to differ in women and men. In women, evidence from two
Phase III studies36,45 and two Phase II studies37,68 indicates that the incidence of LDH
with radiculopathy increases with obesity. Another Phase II study found no association
in women.49 In men, evidence from one Phase III study45 and two Phase II studies37,46
suggests no association between body mass index and the risk of LDH with
radiculopathy. However, three Phase II studies found a positive association in
men.40,49,68 In addition, three Phase II studies that did not report sex-specific results
found no difference in the incidence of LDH with radiculopathy on the basis of obesity or
body mass index.34,52,57
Height. The evidence linking height and the risk of LDH with radiculopathy varies in 6
Phase II studies. Two studies reported a positive association. Heliövaara and
colleagues found that height (≥180 cm in men and ≥170 cm in women) was associated
with an increased risk for LDH with radiculopathy in Finnish men and women,49 while
Leclerc et al. reported a greater risk for LDH in French male electricity and gas
company workers >180 cm in height.46 However, three Phase II studies reported no
association between height and the incidence of LDH.34,40,57 Finally, one Phase II US
study of radiology and hospitalized LDH patients reported some indication of a positive
association in women, but a negative association in men.52
Genetics. We found preliminary evidence of a genetic predisposition for LDH with
radiculopathy. Five Phase I studies reported an increased risk of LDH for
polymorphisms and sequence variations (mutations) involving inflammatory mediator
genes;62 cartilage collagen genes;63 intervertebral disc extracellular matrix protein
genes;65 the human sickle tail gene;66 and, the cartilage intermediate layer protein
gene.64 One Phase I study in Chinese Han men found an increased risk of LDH for two
allele polymorphisms (A21 and A25 alleles) of the aggrecan gene, and a decreased risk
for a third allele polymorphism (A29 allele) of the aggrecan gene.69
50
General Health, Prior Pain and Comorbidities:
Self-rated health status. The evidence varies in two Phase II studies examining self-
rated health status and the risk of LDH with radiculopathy. In their study of French male
electricity and gas workers, Leclerc et al. found that those who reported poor general
health were more likely to develop clinical LDH with radiculopathy (sciatica).46 In
Finland, Mattila et al. found no association between perceived health status and the
incidence of LDH surgery.37
Muscular strength. Preliminary evidence from one Phase II study suggests that back
and abdominal musculature strength is not associated with the incidence of LDH with
radiculopathy.40
History of musculoskeletal symptoms and injuries. We found preliminary evidence
that a history of back pain increases the risk of LDH with radiculopathy. Three Phase II
studies reported that those with a history of back symptoms or low back pain were at
greater risk of developing clinical LDH (sciatica) than those without a history of back
pain.40,43,46 Three Phase II studies reported no association between a history of back
accidents or low back injuries and clinical LDH with radiculopathy (sciatica).34,40,43 In
addition, in their study of French male electricity and gas workers, Leclerc et al. found
no association between the presence of neck pain and the incidence of sciatica.46
Cardiovascular risk factors. There is evidence from one Phase III study that
cardiovascular risk factors, including diabetes, high cholesterol, hypertension, and a
family history of coronary heart disease, are associated with an increased risk of LDH
with radiculopathy in women.36 We found no study that examined these associations in
men.
Other comorbidities. We found preliminary evidence from two Phase II studies that
lumbar spine degeneration40 or chronic disease or disability37 are not associated with
the incidence of LDH with radiculopathy. However, the evidence linking chronic cough
or respiratory symptoms and the risk of LDH varies in three Phase II studies. One study
found a positive association in women;53 two other studies reported no association in
men or women.31,57
51
Health Behaviours:
Smoking. Evidence from one Phase III study,36 and 6 Phase II studies34,37,43,57,58,68
suggests that the incidence of LDH with radiculopathy increases with smoking.
However, 4 Phase II studies,31,40,46,52 and one crude analysis in a Phase III study45
(using an ecological measure of smoking) found no association between smoking and
the risk of LDH with radiculopathy.
Physical or sports activity. The evidence linking physical or sports activity to the risk
of LDH with radiculopathy varies in 9 Phase II studies. One study reported an
increased incidence of clinical LDH (sciatica) associated with walking, but a decreased
incidence associated with jogging.34 Another study found a positive association
between the frequency of participation in sports clubs (4-5 times per week) and risk of
LDH surgery in women, but not in men.37 In their study of male machine operators,
carpenters and office workers, Riihimäki and colleagues reported an indication of a
positive association between the frequency of physical exercise and the incidence of
clinical LDH (sciatic pain).43 Two other studies reported indications of negative
associations between physical activity (measured as time spent doing housework and
gardening52 and cumulative body building and strength training68) and LDH with
radiculopathy. Finally, 4 other studies found no association between physical or sports
activity and the incidence of LDH with radiculopathy.31,46,57,59
Other health behaviour characteristics. We found preliminary evidence from one
Phase II study that some medication use (not specified) may decrease the risk of
hospitalized LDH in women, but not in men.31 The same study also reported that
frequent use of analgesics was associated with an increased risk of hospitalized LDH,
again in women, but not in men. Another Phase II study of a young general population
in Finland, found no association between drinking style (i.e., abstinence, occasional
drinking, recurrent drinking, recurring drunkenness) and the incidence of LDH surgery.37
Personal Physical Factors:
Non-occupational lifting, bending and other activities. Preliminary evidence from
one Phase II study suggests that off-the-job lifting with the knees straight, back bent,
and starting and ending lifts at the waist may be positively associated with an increased
52
incidence of LDH with radiculopathy.58 In this study, Mundt and colleagues also found
trends toward positive associations between LDH with radiculopathy and lifting children
≥25 pounds, as well as with 2 or more days per week of repeated bending while doing
off-the-job activities. They also reported a tendency toward a negative association
between snow shoveling and LDH, and no association for off-the-job stretching and
carrying.
Car driving and motor vehicle characteristics. The evidence linking personal car
driving and motor vehicle characteristics to the risk of LDH with radiculopathy varies in 5
Phase II studies. In two older studies by Kelsey and colleagues, an increased incidence
of LDH with radiculopathy was found in persons driving non-Japanese and non-Swedish
cars,57 and in those exposed to driving other than on the job.52,54 However, no
association was reported for driving pattern characteristics, such as use of local roads,
highways, bucket seats, regular seats, automatic or manual transmission, and being the
driver or passenger.57 Three, more recent, Phase II studies reported no association
between car driving and LDH with radiculopathy in workers,34,43 and in health care
workers.58
Other personal physical factors. There is preliminary evidence from one Phase II
study reporting some indication of a positive association between non-professional
home construction/do-it-yourself activities and the risk of LDH with radiculopathy.46 In
addition, no association was reported between the incidence of LDH with radiculopathy
and non-occupational vibration (two Phase II studies)57,58 and the frequency of wearing
shoes with high heels (one Phase II study).57
Personal Psychological Factors:
Psychological stress and personality. The evidence varies in 7 Phase II studies
examining mental stress and the risk for LDH with radiculopathy. Three studies from
Finland found an increased incidence of LDH with radiculopathy in those who reported:
mental stress “to some extent” or “rather much or much”;34 hysteria in blue-collar, but
not in white-collar workers;60 and a greater number of psychological distress symptoms
in women, but not in men.31 However, 4 other studies reported no associations
between the risk of LDH with radiculopathy and personal mental stress;37,40
psychological well-being;46 or, number of stressful life events in the previous year.52
53
Workplace Physical Factors:
Forward bending postures. We found evidence from two Phase III studies that
occupational lumbar load from forward bending is positively associated with the risk of
LDH with radiculopathy. In two German studies, Seidler et al. reported an increased
incidence of LDH with radiculopathy in: i) both men and women with high cumulative
lumbar load by intensive-load postures (postures with trunk inclination of ≥20-degrees)
at work;67 and, ii) men with high cumulative work hours of extreme forward bending
(>90-degrees trunk flexion).61 However, two Phase II studies of worker cohorts found
no association between LDH and working with the trunk forward flexed,34 or bending
forward and backward.46
Other occupational postures. We found evidence from one Phase III study
suggesting no association between inconvenient work postures and LDH with
radiculopathy.45 However, the evidence varies in three Phase II studies examining trunk
twisting movements/rotations and the risk of LDH.34,46,56 In addition, no association was
reported between the incidence of LDH with radiculopathy and sedentary/sitting work
(one Phase III study;45 2 Phase II studies34,46); video display terminal work (one Phase
III study45); working in a kneeling or squatting position (2 Phase II studies34,46); twisted
or bent occupational postures (one Phase II study43); working with hands above
shoulders (one Phase II study34); and prolonged standing at work (one Phase II
study46).
Manual materials handling. There is evidence from two Phase III studies linking
manual materials handling with an increased risk of LDH with radiculopathy. In their
case-control studies, Seidler and colleagues reported strong positive associations
between LDH with radiculopathy and cumulative lumbar load by manual materials
handling and forward bending postures in both women and men.61,67 Their analysis of
exposure by manual materials handling alone, revealed that men that reported higher
cumulative lumbar load by manual materials handling (i.e., lifting, carrying, pushing,
pulling, throwing, catching or shoveling of objects weighing ≥5 kilograms), compared
with those that reported lower cumulative lumbar load by manual materials handling,
were more likely to experience LDH; however, they found no such association in
women.67 One Phase II study also reported positive associations for the frequency of
54
occupational lifting, carrying, lifting while twisting, and lifting and twisting with knees not
bent.56,57 However, two Phase III and three Phase II studies reported no association
between manual materials handling and LDH in women34,45,52,53 and men.34,45,46,52,53,61
Physical workload or strenuousness of work. The evidence varies in 5 studies
examining work physical strenuousness and the risk of LDH with radiculopathy. One
Phase III study from Germany, found some indication of a positive association between
working ≥10 years in occupations with high physical workload and LDH.61 No
association between physical workload and LDH with radiculopathy was reported in one
Phase III study of the general Finnish population,45 one Phase II study of Finnish
forestry workers,34 and one Phase I study of the southwest region of Finland.39
However, one Phase II study of the general Finnish population found a positive
association in women, but no association in men.50
Other workplace physical factors. There is evidence from one Phase III study
indicating a tendency of a positive association between exposure to whole body
vibration and the risk of LDH with radiculopathy.61 Although, one Phase II study found
no such association for vibration.43 We also found preliminary evidence in one Phase II
study that workplace exposure to draft or cold is not associated with the incidence of
clinical LDH (sciatica).43 Finally, the evidence varies in three Phase II studies assessing
the link between occupational driving or truck driving and the risk of LDH with
radiculopathy (2 reported a positive association46,52-54 and 1 reported no association34).
Workplace Psychosocial Factors:
Time pressure, job control and other psychosocial factors at work. We found
evidence in one Phase III study indicating a link between time pressure at work and the
risk of LDH with radiculopathy. In Germany, Seidler and colleagues found that men
who reported a greater number of work years with a high degree of time pressure were
more likely to experience LDH with radiculopathy compared to those who reported no
work years with high time pressure.61 This same study found no association for other
psychosocial characteristics of work, such as monotonous, boring, opportunities to use
knowledge and skills, information about future plans, satisfaction with supervisor,
satisfaction with workmates, psychic strain through contact with clients, and too much
responsibility.61
55
Evidence from one Phase III study of a national Finnish workforce cohort suggests that
job control is negatively associated with the risk of hospitalized LDH with radiculopathy
in both women and men.45 Leino-Arjas and colleagues also reported an increased
incidence of hospitalized LDH with radiculopathy in women who reported having a “3-
shift” work schedule (defined by the authors as “regular or irregular three-shift work or
regular night work”) compared with women with regular daytime work; and in both
women and men who reported higher levels of work injury risk compared with those
who reported no work injury risk. They reported no association for challenging work
tasks and social demands of work.45
Three Phase II studies of Finnish worker cohorts found no associations between the
incidence of LDH with radiculopathy and job satisfaction;34,46 work psychosocial
overload;34 high work pace, monotonous work, or problems with workmates or
superiors;43 and perceived risk of work injury.34
2.4 Discussion
Summary of the main findings: The annual incidence of hospitalized or surgically
managed LDH with radiculopathy in the general population ranged from 0.2 to 1.3 per
1,000 persons. However, much LDH with radiculopathy is not treated in hospitals. The
annual incidence of LDH with radiculopathy defined on the basis of clinical signs and
symptoms varied from 6.2 per 1,000 persons (0.6%) among female nurses in the US, to
93 per 1,000 persons (9.3%) among forest industry workers in Finland.
The evidence suggests that the etiology of this condition is multifaceted. While there is
evidence that several occupational factors are important contributors to the
development of LDH with radiculopathy, it is also clear that other individual and
behavioural factors may contribute to the development of the condition. For instance,
although high cumulative lumbar load from occupational forward bending is a strong risk
factor for LDH with radiculopathy, not every person exposed to high cumulative lumbar
workload will develop LDH.34,61,67 Instead, combinations of risk factors are necessary to
cause LDH with radiculopathy, and the specific combination of risk factors leading to an
episode of the condition likely varies between persons.
56
Our systematic review suggests that LDH with radiculopathy results from complex
relationships between individual, behavioural, and work-related variables. We found
evidence (Phase III) that age, sex, education, BMI and cardiovascular risk factors in
women, smoking, occupational lumbar load by forward bending postures and manual
materials handling, perceived risk of work injury, decision latitude at work, regular or
irregular three-shift work or regular night work in women, and time pressure at work are
associated with the development of LDH with radiculopathy. We also found preliminary
evidence suggesting that manual occupation, genetics, and previous back pain may
contribute to the development of LDH with radiculopathy in adults.
Limitations of the published literature: We accepted 37 (80%) of the 46 eligible
articles that we reviewed. This literature has many limitations, and it is difficult to draw
consistent conclusions from it. Some of these problems are inherent in all studies of
LDH with radiculopathy. For example, there is no universally accepted definition of LDH
with radiculopathy, and the studies are so heterogeneous that it is difficult to compare
incidence rates and risk factors.
Two key challenges are that authors use different criteria to ascertain and define cases
of LDH with radiculopathy, and the lack of a standard definition for the lower range of
severity for LDH with radiculopathy. This leads to case definitions that are susceptible
to information bias, and results in the misclassification of cases. There is a need for
workable clinical and surveillance definitions of LDH with radiculopathy and subsequent
studies to validate various methods of ascertaining cases.6,77 Until there is some
consistency of definitions and their appropriate validation, studies of the incidence of
LDH with radiculopathy will remain so heterogeneous that comparing the incidence
estimates and risk factors will continue to prove challenging.
Another obvious problem is that many studies do not identify the population at risk that
should form the denominator in an incidence calculation. Many studies report the
number of cases admitted to hospitals over a specified time period, but do not provide
information on the population at risk for being admitted. In some cases, it might not be
possible to determine a denominator, since it is difficult to define the population that is
exclusively served by these institutions.
57
Our accepted studies also have variable inclusion and exclusion criteria that impact on
the interpretation and comparability of the findings. Many of the studies include hospital
admissions only. This is problematic since hospital admission policy for LDH with
radiculopathy can vary over time and jurisdictions.31 In addition, these studies typically
fail to capture cases of LDH with radiculopathy treated at the emergency department,
but not admitted to hospital. Studies that use hospital discharge data may be limited by
this selection bias. Even studies that capture emergency department cases miss those
that are treated at outpatient clinics, or receive no treatment at all.
Despite the limitations of this literature, there are some important conclusions that we
can make. The incidence of hospital-treated LDH with radiculopathy in the general
population ranged from 0.2 to 1.3 per 1,000 persons; however, little is known about the
general population incidence of LDH with radiculopathy not treated in hospital. We
found no admissible study estimating the incidence of clinically defined LDH with
radiculopathy in the general population. Our findings suggest that the incidence is
higher among worker populations, and is highly dependent on the method of case
definition and ascertainment. For example, in two studies that examined the incidence
among nurses in Denmark42 and the US36, the incidence of surgical LDH was much
lower than the incidence of clinical LDH (1.3 per 1,000 persons42 versus 6.2 per 1,000
persons36, Appendix D, Table D.1). There is evidence that the incidence of LDH with
radiculopathy is underestimated when rates are based only on those presenting to or
treated at hospital.
Of the 22 admissible studies that investigated risk factors, 7 reported crude associations
(Phase I studies) and 11 reported associations from exploratory multivariable analyses
(Phase II studies); only 4 studies reported associations from confirmatory Phase III
analyses. Thus, we have limited information on risk factors for LDH with radiculopathy
in adults. Nonetheless, the strongest evidence (Phase III studies) suggests that age
and sex are nonmodifiable risk factors for LDH and radiculopathy, and modifiable risk
factors include lower education, higher BMI and the presence of cardiovascular risk
factors in women, smoking, greater cumulative occupational lumbar load by forward
bending postures and manual materials handling, higher levels of perceived risk of work
injury, lower job control or decision latitude at work, regular or irregular three-shift work
or regular night work in women, and increased time pressure at work. We found
58
preliminary evidence that manual occupation, genetics, previous back pain, and lifting
technique may contribute in the development of LDH with radiculopathy in adults.
Limitations of our review: Our systematic review has two main limitations. First,
although all studies in our review were judged to be scientifically admissible, the
strength of our findings is limited by the considerable variation in the methods of the
primary literature. In particular, the source populations, case definitions, and
assessment and control of confounding were very heterogeneous. We tried to minimize
the effect of these potential sources of bias by rating the overall risk of bias as high,
moderate and low, and by classifying studies as Phase I, II and III; giving greater weight
to confirmatory Phase III studies, followed by preliminary Phase II and I studies.
Second, when assessing the evidence about certain risk factors, we judged the
evidence as varying between studies. However, it is possible that this observed
variation might have been due to us combining studies that did not necessarily
contradict each other, but rather provided valid evidence of differing effects in different
populations (i.e., population-specific effects).
Future recommendations: Our review highlights the many existing gaps in knowledge
of the epidemiology of LDH with radiculopathy in adults. Further descriptive and
analytic investigations are needed as the incidence and determinants of this disabling
and costly neuromusculoskeletal condition have not been adequately described or
examined. We found no admissible study that investigated the incidence of clinically
defined LDH with radiculopathy in the general population. To date, the effect of
cardiovascular risk factors in the development of LDH with radiculopathy has yet to be
examined in men. We also found no admissible evidence examining the link between
trauma and LDH with radiculopathy, be it from motor vehicle injury or clinical
iatrogenesis such as spinal manipulation treatment. A number of case reports have
suggested this as a possible risk factor for LDH with radiculopathy;78-85 however, our
best evidence synthesis suggests that further study using appropriate study designs
and methods is warranted in these areas.
Several methodologic issues are important for future research. We recommend that
authors explicitly describe their source and target populations and discuss
characteristics and issues related to their sampling frames. Producing valid incidence
59
estimates requires accurate and complete ascertainment of both the cases and the at-
risk population. There is a need for workable clinical and surveillance definitions of LDH
with radiculopathy and subsequent validation studies. In addition, incidence density
rates, a more accurate measure of the population-time at risk, are preferable to
cumulative incidence proportions. Risk factors should be studied in cohort or case-
control study designs using multivariable statistical analysis that allows for the
identification of independent risk factors after adjustment for other important covariates.
We recommend that better Phase III studies target modifiable risk factors to inform the
design and testing of evidence-based prevention and intervention programs. In sum,
better understanding of the incidence and determinants of LDH with radiculopathy
demands more careful attention to issues of bias and research methods.
2.5 Conclusion
Although the literature is of varying quality and heterogeneous, there is evidence that
LDH with radiculopathy is an important source of pain and disability in society. The
annual incidence ranges roughly between 0.1% and 10%, depending on case definition
and source population, and risk factors are varied. Our findings support the need to
develop standardized case and surveillance definitions that validly classify the clinical
spectrum of this important condition. Future research should focus on prospective
designs examining modifiable risk factors and prevention strategies.
Funding Sources
Dr. Cesar Hincapié was supported by a Fellowship Award in the Area of Knowledge
Translation from the Canadian Institutes of Health Research, with additional support
from the Canadian Chiropractic Research Foundation.
Acknowledgements
We thank Marina Englesakis, Information Specialist at the University Health Network,
for her support and contribution with the review search strategy.
60
References
1. Hoy D, Brooks P, Blyth F, Buchbinder R. The epidemiology of low back pain. Best Pract Res Clin Rheumatol. 2010;24:769-81.
2. Hoy D, March L, Brooks P, et al. The global burden of low back pain: estimates from the Global Burden of Disease 2010 study. Ann Rheum Dis. 2014;73:968-74.
3. Cassidy JD, Côté P, Carroll LJ, Kristman V. Incidence and course of low back pain episodes in the general population. Spine. 2005;30:2817-23.
4. Hincapié CA, Cassidy JD, Côté P. Is a history of work-related low back injury associated with prevalent low back pain and depression in the general population? BMC Musculoskelet Disord. 2008;9:22.
5. Vos T, Flaxman AD, Naghavi M, et al. Years lived with disability (YLDs) for 1160 sequelae of 289 diseases and injuries 1990–2010: a systematic analysis for the Global Burden of Disease Study 2010. The Lancet. 2012;380:2163-96.
6. Genevay S, Atlas SJ, Katz JN. Variation in eligibility criteria from studies of radiculopathy due to a herniated disc and of neurogenic claudication due to lumbar spinal stenosis: a structured literature review. Spine. 2010;35:803-11.
7. van der Windt DA, Simons E, Riphagen II, et al. Physical examination for lumbar radiculopathy due to disc herniation in patients with low-back pain. Cochrane Database Syst. Rev. 2010;2:CD007431.
8. Vroomen PC, de Krom MC, Wilmink JT, Kester AD, Knottnerus JA. Diagnostic value of history and physical examination in patients suspected of lumbosacral nerve root compression. J Neurol Neurosurg Psychiatry. 2002;72:630-4.
9. Fardon DF, Milette PC. Nomenclature and classification of lumbar disc pathology. Recommendations of the Combined task Forces of the North American Spine Society, American Society of Spine Radiology, and American Society of Neuroradiology. Spine. 2001;26:E93-E113.
10. Frymoyer JW. Back pain and sciatica. N Engl J Med. 1988;318:291-300. 11. Tarulli AW, Raynor EM. Lumbosacral radiculopathy. Neurol Clin. 2007;25:387-
405. 12. Valat JP, Genevay S, Marty M, Rozenberg S, Koes B. Sciatica. Best Pract Res
Clin Rheumatol. 2010;24:241-52. 13. Selim AJ, Ren XS, Fincke G, et al. The importance of radiating leg pain in
assessing health outcomes among patients with low back pain. Results from the Veterans Health Study. Spine. 1998;23:470-4.
14. BenDebba M, Torgerson WS, Long DM. A validated, practical classification procedure for many persistent low back pain patients. Pain. 2000;87:89-97.
15. Tubach F, Beaute J, Leclerc A. Natural history and prognostic indicators of sciatica. J Clin Epidemiol. 2004;57:174-9.
16. Hill JC, Konstantinou K, Egbewale BE, Dunn KM, Lewis M, van der Windt D. Clinical outcomes among low back pain consulters with referred leg pain in primary care. Spine. 2011;36:2168-75.
17. Heliövaara M, Impivaara O, Sievers K, et al. Lumbar disc syndrome in Finland. J Epidemiol Community Health. 1987;41:251-8.
18. Kaila-Kangas L, Leino-Arjas P, Karppinen J, Viikari-Juntura E, Nykyri E, Heliovaara M. History of physical work exposures and clinically diagnosed sciatica among working and nonworking Finns aged 30 to 64. Spine. 2009;34:964-9.
61
19. Jordan J, Konstantinou K, O'Dowd J. Herniated lumbar disc. Clin Evid (Online). 2011. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3275148/. Accessed 2015-May-15.
20. Moher D, Liberati A, Tetzlaff J, Altman DG, Group P. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med. 2009;6:e1000097.
21. Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. PLoS Med. 2009;6:e1000100.
22. Hincapié CA, Cassidy JD, Côté P, et al. The incidence and determinants of symptomatic lumbar disc herniation: a systematic review. PROSPERO. 2011. http://www.crd.york.ac.uk/PROSPERO/full_doc.asp?ID=CRD42011001197.
23. Scottish Intercollegiate Guidelines Network. Section 6: systematic literature review. SIGN 50: a guideline developer's handbook 2008; http://www.sign.ac.uk/guidelines/fulltext/50/section6.html. Accessed 2015-May-15.
24. Slavin RE. Best evidence synthesis: an intelligent alternative to meta-analysis. J Clin Epidemiol. 1995;48:9-18.
25. Cassidy JD, Carroll LJ, Peloso PM, et al. Incidence, risk factors and prevention of mild traumatic brain injury: results of the WHO Collaborating Centre Task Force on Mild Traumatic Brain Injury. J Rehabil Med. 2004:28-60.
26. Côté P, van der Velde G, Cassidy JD, et al. The burden and determinants of neck pain in workers: results of the Bone and Joint Decade 2000-2010 Task Force on Neck Pain and Its Associated Disorders. Spine. 2008;33:S60-S74.
27. Altman DG, Lyman GH. Methodological challenges in the evaluation of prognostic factors in breast cancer. Breast Cancer Res Treat. 1998;52:289-303.
28. Côté P, Cassidy JD, Carroll L, Frank JW, Bombardier C. A systematic review of the prognosis of acute whiplash and a new conceptual framework to synthesize the literature. Spine. 2001;26:E445-58.
29. R: A language and environment for statistical computing [computer program]. Vienna, Austria: R Foundation for Statistical Computing; 2012.
30. Szklo M, Nieto J. Epidemiology: beyond the basics. 2nd ed. Sudbury, MA: Jones and Bartlett Publishers; 2006.
31. Heliövaara M, Knekt P, Aromaa A. Incidence and risk factors of herniated lumbar intervertebral disc or sciatica leading to hospitalization. J Chronic Dis. 1987;40:251-8.
32. Bruske-Hohlfeld I, Merritt JL, Onofrio BM, et al. Incidence of lumbar disc surgery. A population-based study in Olmsted County, Minnesota, 1950-1979. Spine. 1990;15:31-5.
33. Zitting P, Rantakallio P, Vanharanta H. Cumulative incidence of lumbar disc diseases leading to hospitalization up to the age of 28 years. Spine. 1998;23:2337-43; discussion 43-4.
34. Miranda H, Viikari-Juntura E, Martikainen R, Takala EP, Riihimaki H. Individual factors, occupational loading, and physical exercise as predictors of sciatic pain. Spine. 2002;27:1102-9.
35. Jarvik JG, Hollingworth W, Heagerty PJ, Haynor DR, Boyko EJ, Deyo RA. Three-year incidence of low back pain in an initially asymptomatic cohort: clinical and imaging risk factors. Spine. 2005;30:1541-8; discussion 9.
62
36. Jhawar BS, Fuchs CS, Colditz GA, Stampfer MJ. Cardiovascular risk factors for physician-diagnosed lumbar disc herniation. Spine J. 2006;6:684-91.
37. Mattila VM, Saarni L, Parkkari J, et al. Early risk factors for lumbar discectomy: an 11-year follow-up of 57,408 adolescents. Eur Spine J. 2008;17:1317-23.
38. Mattila VM, Sillanpää P, Visuri T, Pihlajamäki H. Incidence and trends of low back pain hospitalisation during military service--an analysis of 387,070 Finnish young males. BMC Musculoskelet Disord. 2009;10:10.
39. Hurme M, Alaranta H, Törmä T, Einola S. Operated lumbar disc herniation: epidemiological aspects. Ann Chir Gynaecol. 1983;72:33-6.
40. Riihimäki H, Wickström G, Hänninen K, Luopajärvi T. Predictors of sciatic pain among concrete reinforcement workers and house painters--a five-year follow-up. Scand J Work Environ Health. 1989;15:415-23.
41. Heikkilä JK, Koskenvuo M, Heliövaara M, et al. Genetic and environmental factors in sciatica. Evidence from a nationwide panel of 9365 adult twin pairs. Ann Med. 1989;21:393-8.
42. Jørgensen S, Hein HO, Gyntelberg F. Heavy lifting at work and risk of genital prolapse and herniated lumbar disc in assistant nurses. Occup Med. 1994;44:47-9.
43. Riihimäki H, Viikari-Juntura E, Moneta G, Kuha J, Videman T, Tola S. Incidence of sciatic pain among men in machine operating, dynamic physical work, and sedentary work. A three-year follow-up. Spine. 1994;19:138-42.
44. Leino-Arjas P, Kaila-Kangas L, Keskimäki I, Notkola V, Mutanen P. Inpatient hospital care for lumbar intervertebral disc disorders in Finland in relation to education, occupational class, income, and employment. Public Health. 2002;116:272-8.
45. Leino-Arjas P, Kaila-Kangas L, Kauppinen T, Notkola V, Keskimäki I, Mutanen P. Occupational exposures and inpatient hospital care for lumbar intervertebral disc disorders among Finns. Am J Ind Med. 2004;46:513-20.
46. Leclerc A, Tubach F, Landre MF, Ozguler A. Personal and occupational predictors of sciatica in the GAZEL cohort. Occup Med. 2003;53:384-91.
47. Bongers PM, Boshuizen HC, Hulshof CT, Koemeester AP. Back disorders in crane operators exposed to whole-body vibration. Int Arch Occup Environ Health. 1988;60:129-37.
48. Netterstrøm B, Juel K. Low back trouble among urban bus drivers in Denmark. Scand J Soc Med. 1989;17:203-6.
49. Heliövaara M. Body height, obesity, and risk of herniated lumbar intervertebral disc. Spine. 1987;12:469-72.
50. Heliövaara M. Occupation and risk of herniated lumbar intervertebral disc or sciatica leading to hospitalization. J Chronic Dis. 1987;40:259-64.
51. Kelsey JL, Ostfeld AM. Demographic characteristics of persons with acute herniated lumbar intervertebral disc. J Chronic Dis. 1975;28:37-50.
52. Kelsey JL. An epidemiological study of acute herniated lumbar intervertebral discs. Rheumatol Rehabil. 1975;14:144-59.
53. Kelsey JL. An epidemiological study of the relationship between occupations and acute herniated lumbar intervertebral discs. Int J Epidemiol. 1975;4:197-205.
54. Kelsey JL, Hardy RJ. Driving of motor vehicles as a risk factor for acute herniated lumbar intervertebral disc. Am J Epidemiol. 1975;102:63-73.
63
55. Kelsey JL, Greenberg RA, Hardy RJ, Johnson MF. Pregnancy and the syndrome of herniated lumbar intervertebral disc; an epidemiological study. Yale J Biol Med. 1975;48:361-8.
56. Kelsey JL, Githens PB, White AA, III, et al. An epidemiologic study of lifting and twisting on the job and risk for acute prolapsed lumbar intervertebral disc. J Orthop Res. 1984;2:61-6.
57. Kelsey JL, Githens PB, O'Conner T, et al. Acute prolapsed lumbar intervertebral disc. An epidemiologic study with special reference to driving automobiles and cigarette smoking. Spine. 1984;9:608-13.
58. Mundt DJ, Kelsey JL, Golden AL, et al. An epidemiologic study of non-occupational lifting as a risk factor for herniated lumbar intervertebral disc. The Northeast Collaborative Group on Low Back Pain. Spine. 1993;18:595-602.
59. Mundt DJ, Kelsey JL, Golden AL, et al. An epidemiologic study of sports and weight lifting as possible risk factors for herniated lumbar and cervical discs. The Northeast Collaborative Group on Low Back Pain. Am J Sports Med. 1993;21:854-60.
60. Pietri-Taleb F, Riihimäki H, Viikari-Juntura E, Lindström K, Moneta GB. The role of psychological distress and personality in the incidence of sciatic pain among working men. Am J Public Health. 1995;85:541-5.
61. Seidler A, Bolm-Audorff U, Siol T, et al. Occupational risk factors for symptomatic lumbar disc herniation; a case-control study. Occup Environ Med. 2003;60:821-30.
62. Noponen-Hietala N, Virtanen I, Karttunen R, et al. Genetic variations in IL6 associate with intervertebral disc disease characterized by sciatica. Pain. 2005;114:186-94.
63. Mio F, Chiba K, Hirose Y, et al. A functional polymorphism in COL11A1, which encodes the alpha 1 chain of type XI collagen, is associated with susceptibility to lumbar disc herniation. Am J Hum Genet. 2007;81:1271-7.
64. Virtanen IM, Song YQ, Cheung KMC, et al. Phenotypic and population differences in the association between CILP and lumbar disc disease. J Med Genet. 2007;44:285-8.
65. Hirose Y, Chiba K, Karasugi T, et al. A functional polymorphism in THBS2 that affects alternative splicing and MMP binding is associated with lumbar-disc herniation. Am J Hum Genet. 2008;82:1122-9.
66. Karasugi T, Semba K, Hirose Y, et al. Association of the tag SNPs in the human SKT gene (KIAA1217) with lumbar disc herniation. J Bone Miner Res. 2009;24:1537-43.
67. Seidler A, Bergmann A, Jäger M, et al. Cumulative occupational lumbar load and lumbar disc disease--results of a German multi-center case-control study (EPILIFT). BMC Musculoskelet Disord. 2009;10:48.
68. Schumann B, Bolm-Audorff U, Bergmann A, et al. Lifestyle factors and lumbar disc disease: results of a German multi-center case-control study (EPILIFT). Arthritis Res Ther. 2010;12:R193.
69. Cong L, Pang H, Xuan D, Tu GJ. Association between the expression of aggrecan and the distribution of aggrecan gene variable number of tandem repeats with symptomatic lumbar disc herniation in Chinese Han of Northern China. Spine. 2010;35:1371-6.
70. Hrubec Z, Nashold BS, Jr. Epidemiology of lumbar disc lesions in the military in World War II. Am J Epidemiol. 1975;102:367-76.
64
71. An HS, Silveri CP, Simpson JM, et al. Comparison of smoking habits between patients with surgically confirmed herniated lumbar and cervical disc disease and controls. J Spinal Disord. 1994;7:369-73.
72. Chibnall JT, Tait RC, Andresen EM, Hadler NM. Race differences in diagnosis and surgery for occupational low back injuries. Spine. 2006;31:1272-5.
73. Saftic R, Grgic M, Ebling B, Splavski B. Case-control study of risk factors for lumbar intervertebral disc herniation in Croatian island populations. Croat Med J. 2006;47:593-600.
74. Lee DY, Ahn Y, Lee SH. The influence of facet tropism on herniation of the lumbar disc in adolescents and adults. J Bone Joint Surg Br. 2006;88:520-3.
75. Kunakornsawat S, Ngamlamaidt K, Tungsiripat R, Prasartritha T. The relationship of facet tropism to lumbar disc herniation. J Med Assoc Thai. 2007;90:1337-41.
76. Zhang Y-g, Sun Z, Zhang Z, Liu J, Guo X. Risk factors for lumbar intervertebral disc herniation in Chinese population: a case-control study. Spine. 2009;34:E918-22.
77. Lin CW, Verwoerd AJ, Maher CG, et al. How is radiating leg pain defined in randomized controlled trials of conservative treatments in primary care? A systematic review. Eur J Pain. 2014;18:455-64.
78. Huang SL, Liu YX, Yuan GL, Zhang J, Yan HW. Characteristics of lumbar disc herniation with exacerbation of presentation due to spinal manipulative therapy. Medicine. 2015;94:e661.
79. Assendelft WJ, Bouter LM, Knipschild PG. Complications of spinal manipulation: a comprehensive review of the literature. J Fam Pract. 1996;42:475-80.
80. Oliphant D. Safety of spinal manipulation in the treatment of lumbar disk herniations: a systematic review and risk assessment. J Manipulative Physiol Ther. 2004;27:197-210.
81. Powell FC, Hanigan WC, Olivero WC. A risk/benefit analysis of spinal manipulation therapy for relief of lumbar or cervical pain. Neurosurgery. 1993;33:73-8; discussion 8-9.
82. Schmidley JW, Koch T. The noncerebrovascular complications of chiropractic manipulation. Neurology. 1984;34:684-5.
83. Morandi X, Riffaud L, Houedakor J, Amlashi SF, Brassier G, Gallien P. Caudal spinal cord ischemia after lumbar vertebral manipulation. Joint Bone Spine. 2004;71:334-7.
84. Rivett DA, Milburn P. Complications arising from spinal manipulative therapy in New Zealand. Physiotherapy. 1997;83:626-32.
85. Tamburrelli FC, Genitiempo M, Logroscino CA. Cauda equina syndrome and spine manipulation: case report and review of the literature. Eur Spine J. 2011;20 Suppl 1:S128-31.
65
Chapter 3
Chiropractic Spinal Manipulation Treatment and the Risk For 3Acute Lumbar Disc Herniation: A Belief Elicitation Study
Manuscript 2
Hincapié CA, Cassidy JD, Rampersaud YR, Côté P, Jadad AR, Tomlinson GA.
Chiropractic spinal manipulation treatment and the risk for acute lumbar disc herniation:
a belief elicitation study. In preparation for submission, 2015
Chiropractic spinal manipulation treatment and
the risk for acute lumbar disc herniation: a belief elicitation study
Authors: Cesar A. Hincapié, DC, MHSc,1,2 J. David Cassidy, PhD, DrMedSc,1,2,3 Y.
Raja Rampersaud, MD, MSc,4 Pierre Côté, DC, PhD,5 Alejandro R. Jadad, MD, DPhil,1,6
George A. Tomlinson, PhD,1,7
Affiliations: 1 Dalla Lana School of Public Health, University of Toronto; 2 Toronto
Western Research Institute, University Health Network; 3 Department of Sports Science
and Clinical Biomechanics, University of Southern Denmark; 4 Division of Orthopaedic
Surgery, Toronto Western Hospital; 5 Faculty of Health Sciences, University of Ontario
Institute of Technology; 6 Centre for Global eHealth Innovation and Techna, University
Health Network; 7 Toronto General Research Institute and Department of Medicine,
University Health Network
Corresponding Author: Cesar A. Hincapié, DC, MHSc, University Health Network,
LuCliff Place – 700 Bay Street, Suite 602, Toronto, Ontario M5G 1Z6, Canada. E-mail:
Key Words: chiropractic; spinal manipulation; risk; lumbar disc herniation; beliefs;
Bayesian; priors; belief elicitation
Word Count: 3,996
66
Abstract
Background: Chiropractic spinal manipulation treatment is common for low back pain
and may increase the risk for acute lumbar disc herniation (LDH), but little is known
about this potential link. In the absence of definitive scientific evidence, clinicians can
have knowledge and beliefs about the risk of using spinal manipulation gained through
clinical experience. Our main aim was to elicit those beliefs among clinicians, while
generating new insights to enrich future Bayesian approaches to interpreting outcomes
after spinal manipulation treatment.
Objective: To describe and quantify clinicians' beliefs (expressed as probability
distributions) about chiropractic spinal manipulation as a risk factor for: 1) acute LDH,
and 2) acute severe LDH that is surgically managed.
Study design: Cross-sectional belief elicitation study of chiropractors, family physicians
and spine surgeons in Ontario, Canada.
Methods: Structured interviews were conducted using a validated approach to belief
elicitation with 47 clinicians that treat patients with back pain and disc herniation.
Participants were asked to estimate the 2-month incidence of acute LDH among an
average group of newly diagnosed patients with acute low back pain (without current
LDH) treated with chiropractic spinal manipulation, relative to the incidence in a
counterfactual group not treated with chiropractic spinal manipulation. Probability
distributions for the relative risk (RR) for acute LDH associated with chiropractic spinal
manipulation treatment were derived.
Results: Combining all clinicians, belief of the risk for acute LDH associated with
chiropractic spinal manipulation treatment was neutral (median RR, 0.97; IQR, 0.57-
1.20; 47% probability for a RR>1). Beliefs varied across clinician groups. Chiropractors
expressed the most optimistic belief (median RR, 0.56; IQR, 0.39-1.03; 27% probability
for a RR>1). Family physicians were more neutral in their belief (median RR, 0.97; IQR,
0.64-1.21; 47% probability for a RR>1). Spine surgeons expressed a slightly more
pessimistic belief (median RR, 1.07; IQR, 0.95-1.29; 65% probability for a RR>1).
67
Conclusions: Clinicians’ beliefs about the risk for acute LDH associated with
chiropractic spinal manipulation varied, with important differences among professions.
Our probability distributions quantify and describe these differing beliefs and could be
used as prior probabilities in Bayesian risk analyses of this exposure-outcome
association.
3.1 Introduction
Spinal manipulation treatment is commonly used for back pain and generally considered
to be safe, although some concern has been raised about its potential link with lumbar
disc herniation (LDH).1,2 Several systematic reviews have found that spinal
manipulation can benefit low back pain with little evidence of serious harm.3,4 In
addition, randomized clinical trial evidence supports the use of spinal manipulation for
the treatment of LDH with radiculopathy.5-9 On the other hand, some believe, on the
basis of case reports and small case series, that spinal manipulation should be
contraindicated for the treatment of disc herniation and have raised the hypothesis of
causal harm.10-16 In sum, there seems to be a disconnect between the scientific
evidence and some beliefs in this area.
Today, no valid estimate of the risk for acute disc herniation following chiropractic spinal
manipulation treatment is available in the scientific literature. Attempts to investigate
serious adverse events related to spinal manipulation face several challenges. First,
the valid study of acute LDH with radiculopathy is limited by the rarity of this outcome.
Well-designed prospective studies are impractical due to the inability to recruit and
follow the numbers of patients needed to ensure that sufficient events are observed. In
addition, this is a sensitive topic open to controversy and litigation,17,18 with a very
limited evidence base, and thus, susceptible to opinion and anecdotal information.
A Bayesian approach to health services research and evaluation presents an
opportunity to advance knowledge despite the above challenges.19,20 Bayesian
methods start with existing “prior” beliefs, formally quantified as probability distributions,
and update these using new data to arrive at “posterior” beliefs, which may be used as
the basis for inferential decisions.21 Quantifying currently held beliefs can determine the
magnitude of a potential risk expected by experts and describe the presence of
68
uncertainty or clinical equipoise about an exposure-outcome association. Experts in a
field can have knowledge of the risk of using a treatment through years of clinical
experience. When quantified, the knowledge gained from their clinical experience can
be included in models estimating risk, which may help to bridge the gap between beliefs
and evidence.
A particular strength of this approach is that probability distributions obtained through
the elicitation of beliefs can be used to augment scarce data,19,20,22,23 and be formally
incorporated into Bayesian risk analyses. Although infrequently used in epidemiologic
research, a Bayesian approach may have important utility in the study of treatment-
related rare serious adverse events, particularly in the absence of definitive scientific
evidence.19,20,24,25
No previous study has assessed interprofessional clinician beliefs regarding chiropractic
spinal manipulation and the risk for acute LDH. The objective of our study was to
describe and quantify clinicians' beliefs about the association between chiropractic
spinal manipulation and the development of: (1) acute LDH, and (2) acute severe LDH
that is surgically managed. We elicited beliefs from three groups of clinicians that treat
patients with back pain and disc herniation: chiropractors, family physicians and spine
surgeons. We hypothesized that beliefs about the risk following chiropractic care would
vary by clinician group, with chiropractors believing spinal manipulation to be less risky
than spine surgeons.
3.2 Methods
Study Participants: Since belief elicitation is best conducted as a face-to-face
interaction between participant and investigator,22,26 we considered eligible only those
participants available for an interview within a 2-hour driving radius of the City of
Toronto. The three clinician groups were defined as: (1) chiropractors – sampled from a
list of active registered members with the College of Chiropractors of Ontario within the
Greater Toronto Area (sampling strategy detailed below); 2) family physicians –
selected from a list of referring primary care physicians to a spine surgery service at a
tertiary care hospital in Toronto and active registered members with the College of
Physicians and Surgeons of Ontario; 3) spine surgeons – invited to participate were all
69
eligible active members of the Ontario Division of the Canadian Spine Society. We also
imposed a requirement (self-reported) that the clinician’s practice involve caring for
patients with low back pain and LDH.
There is no consensus on the preferred sampling method and sample size estimate for
a belief elicitation study.22,26,27 A systematic review of belief elicitation methods found a
median sample size in elicitation studies of 11 participants, and suggested that it may
be preferable to include participants chosen non-randomly (e.g., purposive sampling) in
order to capture a range of opinions of the target population.26 We implemented a
pragmatic purposive sampling strategy of the three clinician groups, with the goal of
interviewing 15 members of each group, for a total of 45 respondents. We
hypothesized that beliefs may vary on the basis of clinician group, years of clinical
experience or gender, and attempted to balance these characteristics in our samples.
Case Definitions: Acute LDH was defined as acute onset of lumbar radiculopathy due
to LDH. Acute severe LDH that is surgically managed was defined as acute onset of
severe lumbar radiculopathy (i.e., severe intractable pain and neurologic deficit, but not
cauda equina syndrome) due to LDH and necessitating surgical management.
Study Design: We conducted a belief elicitation study using a computer adaptation of a
validated approach to clinician belief elicitation.27,28 A face-to-face interactive interview
was conducted by a single interviewer (CAH) with each participant, using a
standardized interview questionnaire (Appendix E), script (Appendix F) and elicitation
software. The laptop-based elicitation software (modified SHELF29 application in R30)
allowed us to use real-time graphical representation of a participant’s probability
distributions to provide feedback and ensure that the probability distributions were a true
representation of the participant’s beliefs (Appendix G provides screenshots of the
computer interface and details of the modifications we made to the base SHELF
software). The interview, script and software were pilot tested on a group of 10
clinicians (chiropractors, family physicians and spine surgeons) for face and content
validity, time feasibility (15-minute duration) and clarity. We obtained ethics approval for
our study from the research ethics board of the University Health Network (REB #11-
0240-AE), and administrative approval from the Office of Research Ethics at the
University of Toronto (protocol reference #26588).
70
Belief Elicitation Interview: The interview was developed by applying best practices
for belief elicitation22,26,27 and in consultation with an expert in Bayesian methods (GT).
It aimed at deriving valid probability distributions to describe clinicians' current beliefs
about the effect of chiropractic spinal manipulation on the risk of developing: (1) acute
LDH, and (2) acute severe LDH that is surgically managed.
The belief elicitation interview began with a short introduction, followed by informed
written consent. The participant was then guided through a worked example (Appendix
E, Example Scenario, with Appendix F script) of a belief elicitation to become familiar
with the elicitation process and computer software. The belief elicitation involved two
clinical scenarios, one for each of the outcomes of interest, through which the study
data were collected. For each scenario, a participant was asked to consider the
following at-risk population: a hypothetical average group of 1,000 newly diagnosed
acute low back pain patients without current LDH. The rationale for using 1,000 as the
denominator for this elicitation was based on it being an observable and conceivable
sample of patients that clinicians would be able to imagine. This facilitates the
probabilistic mental manipulations that have to be made during the elicitation interview;
increases the reliability of the elicitation; and reduces some of the common elicitation
biases (e.g., tendency to think in terms of percentages).22,26,27 Our approach meant that
the smallest ‘unit’ of probability was 1/1000 or 0.001. Use of a larger denominator (e.g.,
10,000 or 100,000) would have allowed for a more precise belief to be quantified, but
literature suggests that it is unlikely that people could easily conceive of the difference
between a probability of 0.001 (1/1000) or 0.0001 (1/10,000) or 0.00001 (1/100,000).22
The outcomes of interest were clinicians’ incidence estimates of acute LDH and acute
severe LDH that is surgically managed, among an average group of 1,000 newly
diagnosed acute low back pain patients (without current LDH) that is treated with
chiropractic spinal manipulation, relative to incidence estimates among a counterfactual
group not treated with chiropractic spinal manipulation.
In the final section of the interview, we also collected demographic and additional
characteristics of the participants: gender, age group, clinical background, number of
years in practice treating low back pain patients, number of new low back pain patients
seen per year, history and level of formal statistical training, history of referral for
71
chiropractic care in practice for treatment of low back pain, sources of information used
in formulating beliefs about the effect of chiropractic spinal manipulation on the risk of
developing acute LDH, and history of having seen a patient in clinical practice whose
disc herniation was attributed to chiropractic treatment.
Belief Elicitation Software: We used a modified computer protocol (modified SHELF29
application in R30) that allowed participants to graphically specify a point estimate of
incidence and a plausible range of incidence values that they believed likely for each of
the two clinical scenarios (Appendix G). They indicated their weight of belief for the
incidence values along their specified range by allocating 20 virtual “chips”, each
representing 5% probability, thereby creating a probability distribution or “belief curve”.
This allowed us to provide participants with real-time graphical representation of
participants’ probability distributions. Participants were asked to review the shape and
distribution of their belief curve to ensure that it accurately reflected their true belief prior
to proceeding in the interview.
Statistical Analysis: Medians and proportions were used to summarize characteristics
of the respondents. On each respondent, the probability distribution (i.e., belief) for the
relative risk (RR) for acute LDH was derived by dividing values of the incidence of acute
LDH among patients treated with chiropractic spinal manipulation (the values on the x-
axis in Appendix G) by the point estimate of incidence of acute LDH among patients
not treated with chiropractic spinal manipulation. A similar calculation was carried out
for the second outcome to derive the probability distribution for the RR for surgically
managed acute LDH among patients treated with chiropractic spinal manipulation
compared with those not treated with chiropractic care.
We aggregated all individual belief elicitations to derive an overall distribution of belief
among participants for the RR for each of the two outcomes of interest. We also
aggregated elicitations within each of the clinician groups. Finally, optimistic and
pessimistic probability distributions were estimated by aggregating respondents within
the lowest and highest quartiles of mean RR estimates, respectively.
Prespecified secondary analyses were undertaken to describe and quantify beliefs by
gender, age groups (clinicians ≤50 years of age and clinicians ≥51 years of age), and
72
years of clinical experience (tertiles of years treating patients with back pain: <12, 12-24
and ≥25 years). We used R, version 2.15.130 to carry out our statistical analysis.
Role of the Funding Source: The sponsors of the study had no role in study design,
data collection, data analysis, data interpretation, the writing of the report, or in the
decision to submit the report for publication.
3.3 Results
Study Participants: Of 165 eligible potential participants invited, 87 did not respond
and 31 declined to participate (reason not given). Overall participation was 28%
(47/165) and varied by clinician group: 37% among chiropractors (16/43); 15% among
family physicians (15/99); and, 70% among spine surgeons (16/23).
Table 3.1 describes characteristics of the study participants. All spine surgeons were
men and on average older than the other clinician groups. Most participants (94%)
reported some formal post-secondary statistical training. The median number of years
treating low back pain patients among all participants was 18 (IQR, 10-26 years), with
chiropractors having a lower median of 13 years (IQR, 8-27 years), and spine surgeons
a higher median of 23 years (IQR, 10-26 years).
73
Table 3.1. Characteristics of belief elicitation study participants
Characteristic
All clinicians
(n=47)
Family
physicians
(n=15)
Chiropractors
(n=16)
Spine
surgeons
(n=16)
Men – N (%) 35 (74) 7 (47) 12 (75) 16 (100)
Age group – N (%)
21-40 years 13 (28) 4 (27) 8 (50) 1 (6)
41-60 years 28 (59) 10 (66) 7 (44) 11 (69)
≥61 years 6 (13) 1 (7) 1 (6) 4 (25)
N with post-secondary statistical training –
N (%)
44 (94) 13 (87) 15 (94) 16 (100)
N of years treating LBP patients – Median
(IQR)
18 (10-26) 18 (13-26) 13 (8-27) 23 (10-26)
N of new LBP patients seen per year* –
Median (IQR)
150 (50-500) 150 (55-500) 65 (40-150) 550 (240-
1060)
N that refer for chiropractic care in clinical
practice – N (%)
34 (72) 11 (73) 16 (100) 7 (44)
Elicited incidence of acute LDH among
acute LBP patients not treated with
chiropractic care – Median % (IQR)
5.0 (2.3-10.0) 5.0 (3.8-10.0) 5.0 (3.0-10.0) 3.3 (1.9-10.0)
Elicited incidence of surgically managed
acute LDH among acute LBP patients
not treated with chiropractic care –
Median % (IQR)
1.0 (0.5-2.0) 1.0 (0.3-1.0) 1.5 (0.5-2.6) 1.0 (0.4-1.4)
Abbreviations: IQR, interquartile range; LDH, lumbar disc herniation; LBP, low back pain; N, number
* Self-reported estimate
Clinicians’ Beliefs About the Risk For Acute LDH Associated With Chiropractic
Care: Combining all clinicians, the probability distribution for the risk for acute LDH
associated with chiropractic spinal manipulation was evenly divided between a
decrease in risk and an increase in risk (Figure 3.1.a). The perceived median RR for
74
acute LDH across all clinicians was 0.97 (IQR, 0.57-1.20), with a probability for a RR >1
of 46.7% (Table 3.2).
Beliefs varied by clinician group (Figure 3.1.b). Chiropractors reported the most
optimistic belief of the risk for acute LDH associated with chiropractic care, with a
perceived median RR of 0.56 (IQR, 0.39-1.03) and probability for a RR >1 of 26.7%
(Table 3.2). Family physicians held a more neutral belief, with a perceived median RR
of 0.97 (IQR, 0.64-1.21) and probability for a RR >1 of 47.1%. Spine surgeons reported
a slightly more pessimistic belief, with a perceived median RR of 1.07 (IQR, 0.95-1.29)
and probability for a RR >1 of 64.8%.
Bimodal probability distributions were observed for both chiropractors and family
physicians (Figure 3.1.b), suggesting that within each of these clinician groups there
was a subset of clinicians that believed that there is a protective effect of spinal
manipulation on the risk for acute LDH, and another subset that believed that there is
“no effect” with a chance of harm. The unimodal distribution for the spine surgeons
showed a more homogeneous belief of no benefit on average, with more chance of
harm than benefit.
Optimistic and Pessimistic Beliefs: Optimistic clinicians—those in the lowest quartile
of RR estimates—included 3 family physicians and 9 chiropractors. These clinicians
expressed a belief that chiropractic spinal manipulation reduces the incidence of acute
LDH by about 60% (median RR, 0.42; IQR, 0.29-0.53) (Figure 3.1.c and Table 3.2).
Pessimistic clinicians—those in the highest quartile of RR estimates—included 6 spine
surgeons, 3 family physicians and 3 chiropractors, who believed that chiropractic care
increases the incidence of acute LDH by about 30% (median RR, 1.29; IQR, 1.11-1.59).
Compared to optimists, pessimists were more commonly men, older, spine surgeons,
and reported a greater number of new low back patients per year (Table 3.3). More
pessimists than optimists also reported having seen a patient whose disc herniation was
attributed to chiropractic spinal manipulation.
75
Figure 3.1. Beliefs regarding the risk for acute lumbar disc herniation associated with chiropractic spinal manipulation treatment: a. among all clinicians, b.
among clinician groups, c. among optimistic versus pessimistic clinicians
0.0 0.5 1.0 1.5 2.0 2.5 3.0
0.0
0.5
1.0
1.5
2.0
2.5
a. All participants belief
Relative risk
Wei
ght o
f bel
ief
P(RR<1)=53.3% P(RR>1)=46.7%
0.0 0.5 1.0 1.5 2.0 2.5 3.0
0.0
0.5
1.0
1.5
2.0
2.5
b. Clinician group beliefs
Relative risk
Wei
ght o
f bel
ief
Chiropractors (n=15)Family physicians (n=15)Spine surgeons (n=16)
P(RR>1)=64.8%
P(RR>1)=47.1%
P(RR>1)=26.7%
0.0 0.5 1.0 1.5 2.0 2.5 3.0
0.0
0.5
1.0
1.5
2.0
2.5
c. Optimistic−pessimistic beliefs
Relative risk
Wei
ght o
f bel
ief
Optimists (n=12)Pessimists (n=12)
P(RR>1)=1%
P(RR>1)=85.1%
76
Table 3.2. Clinicians’ beliefs about the risk for acute lumbar disc herniation and surgically managed
acute lumbar disc herniation associated with chiropractic spinal manipulation treatment
Beliefs
Acute LDH Acute LDH-surgery
Median RR*
(IQR) Prob RR>1
Median RR
(IQR) Prob RR>1
All clinicians (n=46) 0.97 (0.57-1.20) 46.7% 1.00 (0.61-1.40) 49.9%
Clinician group
Chiropractors (n=15) 0.56 (0.39-1.03) 26.7% 0.63 (0.35-1.08) 28.3%
Family physicians (n=15) 0.97 (0.64-1.21) 47.1% 1.02 (0.73-1.35) 53.0%
Spine surgeons (n=16) 1.07 (0.95-1.29) 64.8% 1.18 (0.93-1.58) 67.4%
Optimistic-pessimistic beliefs
Optimistic belief (lower quartile of RR)
(n=12)
0.42 (0.29-0.53) 1.0% 0.42 (0.22-0.64) 6.5%
Pessimistic belief (upper quartile of
RR) (n=12)
1.29 (1.11-1.59) 85.1% 1.66 (1.36-2.21) 90.5%
Gender
Women (n=11) 0.77 (0.41-1.10) 34.2% 0.66 (0.30-1.08) 31.0%
Men (n=35) 1.00 (0.68-1.23) 50.6% 1.07 (0.77-1.50) 56.6%
Age group
≤50 years (n=24) 0.96 (0.51-1.24) 46.8% 0.97 (0.57-1.49) 47.4%
≥51 years (n=22) 0.98 (0.66-1.17) 46.7% 1.02 (0.70-1.35) 52.6%
Clinical experience (years treating LBP
patients)
Low tertile – <12 years (n=15) 0.98 (0.54-1.26) 48.8% 0.91 (0.50-1.72) 46.8%
Mid tertile – 12-24 years (n=14) 0.97 (0.52-1.24) 46.9% 0.98 (0.64-1.24) 47.2%
High tertile – ≥25 years (n=17) 0.97 (0.65-1.14) 44.7% 1.04 (0.77-1.36) 54.9%
Abbreviations: IQR, interquartile range; LBP, low back pain; LDH, lumbar disc herniation; n, sample size;
Prob; probability; RR, relative risk
* Perceived estimates of median RRs derived from the elicited incidence probabilities for the outcomes
with and without exposure to chiropractic spinal manipulation treatment
77
Table 3.3. Characteristics of clinicians with optimistic versus pessimistic beliefs about the risk for acute
lumbar disc herniation and surgically managed acute lumbar disc herniation associated with
chiropractic spinal manipulation treatment
Acute LDH Acute LDH-surgery
Characteristic
Optimists
(n=12)
Pessimists
(n=12)
Optimists
(n=12)
Pessimists
(n=12)
Men – N (%) 7 (58) 11 (92) 5 (42) 11 (92)
Age group – N (%)
21-40 years 5 (41) 2 (17) 7 (58) 3 (25)
41-60 years 6 (50) 7 (58) 5 (41) 7 (58)
≥61 years 1 (8) 3 (25) 0 2 (17)
Clinician group – N (%)
Family physicians 3 (25) 3 (25) 3 (25) 3 (25)
Chiropractors 9 (75) 3 (25) 9 (75) 3 (25)
Spine surgeons 0 6 (50) 0 6 (50)
Number with post-secondary statistical
training – N (%)
11 (92) 11 (92) 11 (92) 11 (92)
Number of years treating LBP patients –
Median (IQR)
14 (9-25) 20 (9-26) 12 (7-19) 12 (7-22)
Number of new LBP patients per year –
Median (IQR)
135 (72-187) 250 (50-600) 120 (83-150) 225 (50-763)
Number that refer for chiropractic care in
clinical practice – N (%)
11 (92) 8 (67) 12 (100) 7 (58)
Number that have seen a patient whose
disc herniation was attributed to
chiropractic spinal manipulation – N (%)
1 (8) 7 (58) 0 6 (50)
Abbreviations: IQR, interquartile range; LBP, low back pain; LDH, lumbar disc herniation; N, number
78
Beliefs By Gender, Age Groups and Years of Clinical Experience: Table 3.2 and
Figure 3.2.a show that some differences in belief about the risk for acute LDH
associated with chiropractic care were found between women (n=11) and men (n=35)
clinicians, with women reporting a more optimistic belief (median RR, 0.77; IQR, 0.41-
1.10; 34% probability for a RR >1) and men, a more neutral belief (median RR, 1.00;
IQR, 0.68-1.23; 51% probability for a RR >1).
There were few differences in beliefs according to age group (participants aged ≤50
years, n=24 vs. participants aged ≥51 years, n=22) or tertiles of years of clinical
experience treating patients with low back pain (<12 years, n=15 vs. 12-24 years, n=14
vs. ≥25, n=17) (Table 3.2 and Figures 3.2.b and 3.2.c).
Clinicians’ Beliefs About the Risk For Surgically Managed Acute LDH Associated
With Chiropractic Care: Tables 3.2 and 3.3 show that findings for surgically managed
acute LDH were similar to those detailed above. Compared with RR estimates for the
belief regarding risk for acute LDH associated with chiropractic spinal manipulation, a
small increase in the perceived risk estimates was found among all three clinician
groups. Of note, pessimistic clinicians reported a belief that chiropractic care increases
the incidence of acute LDH that is surgically managed by about 65% (median RR, 1.66;
IQR, 1.36-2.21), with a probability for a RR >1 of over 90%.
79
Figure 3.2. Beliefs regarding the risk for acute lumbar disc herniation associated with chiropractic spinal manipulation treatment: a. by gender, b. by age
group, c. by clinical experience
0.0 0.5 1.0 1.5 2.0 2.5 3.0
0.0
0.5
1.0
1.5
2.0
2.5
a. Beliefs by gender
Relative risk
Wei
ght o
f bel
ief
Women (n=11)Men (n=35)
P(RR>1)=50.6%
P(RR>1)=34.2%
0.0 0.5 1.0 1.5 2.0 2.5 3.0
0.0
0.5
1.0
1.5
2.0
2.5
b. Beliefs by age group
Relative risk
Wei
ght o
f bel
ief
<=50 years of age (n=24)>=51 years of age (n=22)
P(RR>1)=46.8%
P(RR>1)=46.7%
0.0 0.5 1.0 1.5 2.0 2.5 3.0
0.0
0.5
1.0
1.5
2.0
2.5
c. Beliefs by clinical experience
Relative risk
Wei
ght o
f bel
ief
<12 years experience (n=15)12−24 years experience (n=14)>=25 years experience (n=17)
P(RR>1)=48.8%
P(RR>1)=46.9%
P(RR>1)=44.7%
80
3.4 Discussion
We carried out a belief elicitation study to quantify, describe and gain important insights
into beliefs about the risk for acute LDH associated with chiropractic spinal manipulation
held by clinicians experienced with treating patients with back pain and disc herniation.
Clinician beliefs about the incidence of acute LDH and surgically managed acute LDH
with and without chiropractic spinal manipulation treatment varied, with interesting and
important differences observed across professions and specializations.
Our study advances knowledge of the potential link between chiropractic spinal
manipulation and acute LDH in two important respects. First, our study is the first to
examine and provide estimates of clinician beliefs regarding this potential exposure-
outcome association. There are no published estimates of the association between
chiropractic care and the incidence of acute LDH with which to compare our findings. In
addition, given the rarity of acute symptomatic LDH and modest utilization of
chiropractic care in the population, well-designed prospective observational studies may
be challenging to perform. Nonetheless, experts in a field can have knowledge of the
risk of using a treatment through years of clinical experience. By quantifying currently
held clinician beliefs, our results shed light on the magnitude of a potential risk expected
by experts and describe the presence of uncertainty about this exposure-outcome
association.
Among the back pain clinicians in this study, there was a range of beliefs about the risk
associated with chiropractic care. On average, clinicians believed that the incidence of
acute LDH is about the same with or without chiropractic spinal manipulation treatment;
in other words, “no effect” of exposure to chiropractic care on the risk for acute LDH.
However, it is striking to note that optimistic clinicians believed that treatment of acute
low back pain with chiropractic spinal manipulation reduces the incidence of acute LDH
by 58% (IQR, 47-71%), whereas pessimistic clinicians believed that chiropractic care
increases the incidence of acute LDH that necessitates surgical management by 66%
(IQR, 36-121%) (Table 3.2). In addition, the bimodal distributions observed in our
graphical analyses (Figures 3.1 and 3.2) indicate distinct “pools” of belief, with some
clinicians believing in a beneficial effect of chiropractic care and others mainly
81
expressing a belief of no effect with a possibility of harm. Our findings suggest a
divergence of opinion within these back pain clinicians, indicating the presence of
community uncertainty.
Another advance comes as a result of the usefulness of the elicited beliefs under a
Bayesian inferential paradigm. The probability distributions derived from this study may
be relevant to researchers investigating the risk for these rare serious adverse events
following chiropractic spinal manipulation. Using a Bayesian approach, elicited beliefs
in the form of prior probability distributions can be used to augment limited data.22-24,31,32
This could facilitate a Bayesian approach to risk analyses of these exposure-outcome
associations in which informative clinical prior beliefs about the potential risk can be
explicitly and transparently specified as Bayesian priors. This view appears particularly
appropriate in the context of investigating contentious potential treatment-related rare
serious adverse events in the absence of compelling scientific data and evidence.19,20
Our study demonstrates that this belief elicitation method can be used to describe and
quantify clinicians’ beliefs about the safety of an intervention or treatment exposure.
This methodology is not limited to the study of beliefs about the safety of manual
therapy such as spinal manipulation. For example, it has been used to elicit beliefs
regarding: treatment efficacy in pulmonary arterial hypertension28 and pediatric intestinal
failure-associated liver disease;33 and risk factors for falls in community-dwelling older
people.34 This belief elicitation methodology could be generalized to the study of many
uncommon diseases and conditions,22 and could have potential web-based research
applications.35
Our study has limitations. This is a preliminary study involving a relatively small number
of clinicians that may not be representative of the clinician group source populations.
However, our main objective was to describe the beliefs of our study sample and
generate hypotheses, rather than to make strong generalizable inferences about beliefs
in these clinician source populations. The main limitations of our study may relate to
characteristics of the study respondents. Since there is no formal definition for back
pain clinician expert, we defined an expert as a chiropractor, family physician or spine
surgeon with a self-reported active practice treating patients with low back pain. The
large number of years treating patients with low back pain and large number of new
82
patients with low back pain seen per year (Table 3.1) support the notion that the
participants were experienced clinician experts. Our findings are limited to the three
clinician groups interviewed and do not include information on beliefs among physical
therapists. Nonresponse bias may have occurred as a result of clinicians not
responding to our invitation to participate. This may have led to bias if the reasons for
study participation were associated with beliefs regarding chiropractic care safety. We
hypothesize that respondents may have had more extreme beliefs than nonrespondents
(i.e., those without strong opinions might be less likely to take part). If this had been the
case, our estimates may overestimate clinician group beliefs, but on average, would be
similar to our current finding of a neutral belief among all participants. In addition, we
would have preferred a more balanced study sample with respect to gender and age
distribution. However, this proved challenging, as the spine surgery field in Ontario is
predominantly male and older, compared with the more balanced family physician
population and the younger population of chiropractors in the Greater Toronto Area
(Table 3.1). We acknowledge that patient mix and clinical experience likely varied
across clinician groups. We attempted to examine the impact of years of clinical
experience, age and gender by estimating probability distributions on the basis of these
characteristics. Elicited beliefs were similar across age groups and levels of clinical
experience, although differences in beliefs were found between women and men, with
women expressing a more optimistic belief.
Finally, it is also important to note that participants were asked to estimate the incidence
of acute LDH with and without chiropractic spinal manipulation treatment, rather than
provide their belief of the relative risk directly. The probability distribution for the relative
risk was derived from the individual distributions of incidence with and with chiropractic
care. Our rationale for this approach was based on evidence that it is much more
difficult for respondents to directly specify relative risk distributions as these are not
directly observable.22 A clinician can reflect on their knowledge and experience to
estimate a frequency representation of the probability for acute LDH (e.g., “Of 1000
patients with acute low back pain treated with chiropractic spinal manipulation, how
many will go on to develop acute LDH?”), however, it is a much more challenging task
to abstract that information and elicit a valid probability distribution for the ratio of
incidence probabilities (i.e., the relative risk).
83
Our hypothesis-generating study raises some interesting questions warranting further
investigation. What are the reasons for the differences in clinician beliefs that we
found? Do these clinicians interpret the same evidence, as limited as it is, differently?
Do they see different types of patients, so that their beliefs actually reflect their personal
clinical experience? Does spine surgeons’ experience with the more severe end of the
clinical spectrum raise their index of suspicion for viewing spinal manipulation as
potentially harmful? Do they have in-built biases based on their professions and
specializations? Would similar beliefs be observed in health care settings where the
education of these clinician groups is more integrated (e.g., Denmark,36 Switzerland37),
or where family physicians and spine surgeons deliver spinal manipulation? In
Germany, for example, general practitioners and ambulatory orthopaedic surgeons are
often specially trained in spinal manipulation and offer it to their patients.38 This
highlights the need for future well-designed studies to examine determinants of beliefs,
confirm clinicians’ beliefs about the safety of chiropractic spinal manipulation in other
settings, and investigate the association between chiropractic care and acute LDH in
the population.
With regard to public health significance, this study represents an important first step in
the process of understanding current perceptions about the safety of chiropractic spinal
manipulation. It may provide opportunities for interdisciplinary educational initiatives
among the various clinician groups, as well as knowledge translation with patients,
policymakers and other relevant stakeholders with an interest in the safety of
chiropractic care.
3.5 Conclusion
Our study is the first to describe and evaluate clinician beliefs about the association
between chiropractic spinal manipulation and the development of acute LDH and acute
LDH that necessitates surgical intervention. We found that beliefs were diverse, with
important interprofessional differences, indicating the presence of community
uncertainty. Our probability distributions quantify and describe these differing beliefs
and could be used as prior probabilities in Bayesian risk analyses of this exposure-
outcome.
84
Funding Sources
Dr. Cesar Hincapié was supported by a Fellowship Award in the Area of Knowledge
Translation from the Canadian Institutes of Health Research, with additional support
from the Canadian Chiropractic Research Foundation.
Acknowledgements
We thank Drs. Sindhu Johnson and Ivan Diamond for their assistance with the
development of the belief elicitation interview and questionnaire. We acknowledge Drs.
Jeremy Oakley and Anthony O’Hagan for their development and open source sharing of
the SHELF elicitation framework. We thank all the clinician experts who participated in
the belief elicitation interviews.
References
1. Gouveia LO, Castanho P, Ferreira JJ. Safety of chiropractic interventions: a systematic review. Spine. 2009;34:E405-13.
2. Snelling NJ. Spinal manipulation in patients with disc herniation: A critical review of risk and benefit. Int J Osteopath Med. 2006;9:77-84.
3. Cherkin DC, Sherman KJ, Deyo RA, Shekelle PG. A review of the evidence for the effectiveness, safety, and cost of acupuncture, massage therapy, and spinal manipulation for back pain. Ann Intern Med. 2003;138:898-906.
4. Furlan AD, Yazdi F, Tsertsvadze A, et al. A systematic review and meta-analysis of efficacy, cost-effectiveness, and safety of selected complementary and alternative medicine for neck and low-back pain. Evid Based Complement Alternat Med. 2012:953139.
5. Burton AK, Tillotson KM, Cleary J. Single-blind randomised controlled trial of chemonucleolysis and manipulation in the treatment of symptomatic lumbar disc herniation. Eur Spine J. 2000;9:202-7.
6. Liu J, Zhang S. Treatment of protrusion of lumbar intervertebral disc by pulling and turning manipulations. J Tradit Chin Med. 2000;20:195-7.
7. Santilli V, Beghi E, Finucci S. Chiropractic manipulation in the treatment of acute back pain and sciatica with disc protrusion: a randomized double-blind clinical trial of active and simulated spinal manipulations. Spine J. 2006;6:131-7.
8. McMorland G, Suter E, Casha S, du Plessis SJ, Hurlbert RJ. Manipulation or microdiskectomy for sciatica? A prospective randomized clinical study. J Manipulative Physiol Ther. 2010;33:576-84.
9. Bronfort G, Hondras MA, Schulz CA, Evans RL, Long CR, Grimm R. Spinal manipulation and home exercise with advice for subacute and chronic back-related leg pain: a trial with adaptive allocation. Ann Intern Med. 2014;161:381-91.
85
10. Huang SL, Liu YX, Yuan GL, Zhang J, Yan HW. Characteristics of lumbar disc herniation with exacerbation of presentation due to spinal manipulative therapy. Medicine. 2015;94:e661.
11. Assendelft WJ, Bouter LM, Knipschild PG. Complications of spinal manipulation: a comprehensive review of the literature. J Fam Pract. 1996;42:475-80.
12. Powell FC, Hanigan WC, Olivero WC. A risk/benefit analysis of spinal manipulation therapy for relief of lumbar or cervical pain. Neurosurgery. 1993;33:73-8; discussion 8-9.
13. Schmidley JW, Koch T. The noncerebrovascular complications of chiropractic manipulation. Neurology. 1984;34:684-5.
14. Morandi X, Riffaud L, Houedakor J, Amlashi SF, Brassier G, Gallien P. Caudal spinal cord ischemia after lumbar vertebral manipulation. Joint Bone Spine. 2004;71:334-7.
15. Tamburrelli FC, Genitiempo M, Logroscino CA. Cauda equina syndrome and spine manipulation: case report and review of the literature. Eur Spine J. 2011;20 Suppl 1:S128-31.
16. Oppenheim JS, Spitzer DE, Segal DH. Nonvascular complications following spinal manipulation. Spine J. 2005;5:660-6; discussion 6-7.
17. Boucher P, Robidoux S. Lumbar disc herniation and cauda equina syndrome following spinal manipulative therapy: a review of six court decisions in Canada. Journal of forensic and legal medicine. 2014;22:159-69.
18. Jagbandhansingh MP. Most common causes of chiropractic malpractice lawsuits. J Manipulative Physiol Ther. 1997;20:60-4.
19. Spiegelhalter DJ. Incorporating Bayesian ideas into health-care evaluation. Statistical Science. 2004;19:156-74.
20. Spiegelhalter DJ, Myles JP, Jones DR, Abrams KR. Methods in health service research. An introduction to bayesian methods in health technology assessment. BMJ. 1999;319:508-12.
21. Spiegelhalter D, Rice K. Bayesian statistics. Scholarpedia. 2009;4(8):5230. http://www.scholarpedia.org/article/Bayesian_statistics. Accessed 2015-May-15.
22. O'Hagan A, Buck CE, Daneshkhah A, et al. Uncertain judgements: eliciting experts' probabilities. West Sussex: John Wiley & Sons Ltd; 2006.
23. Lilford RJ, Thornton JG, Braunholtz D. Clinical trials and rare diseases: a way out of a conundrum. BMJ. 1995;311:1621-5.
24. Dunson DB. Commentary: practical advantages of Bayesian analysis of epidemiologic data. Am J Epidemiol. 2001;153:1222-6.
25. Hampson LV, Whitehead J, Eleftheriou D, Brogan P. Bayesian methods for the design and interpretation of clinical trials in very rare diseases. Stat Med. 2014;33:4186-201.
26. Johnson SR, Tomlinson GA, Hawker GA, Granton JT, Feldman BM. Methods to elicit beliefs for Bayesian priors: a systematic review. J Clin Epidemiol. 2010;63:355-69.
27. Johnson SR, Tomlinson GA, Hawker GA, Granton JT, Grosbein HA, Feldman BM. A valid and reliable belief elicitation method for Bayesian priors. J Clin Epidemiol. 2010;63:370-83.
28. Johnson SR, Granton JT, Tomlinson GA, Grosbein HA, Hawker GA, Feldman BM. Effect of warfarin on survival in scleroderma-associated pulmonary arterial hypertension (SSc-PAH) and idiopathic PAH. Belief elicitation for Bayesian priors. J Rheumatol. 2011;38:462-9.
86
29. SHELF: the Sheffield Elicitation Framework (version 2.0) [computer program]. Version 2.0. Sheffield, UK: School of Mathematics and Statistics, University of Sheffield; 2010.
30. R: A language and environment for statistical computing [computer program]. Vienna, Austria: R Foundation for Statistical Computing; 2012.
31. Billingham L, Malottki K, Steven N. Small sample sizes in clinical trials: a statistician’s perspective. Clinical Investigation. 2012;2:655-7.
32. Chaloner K, Rhame FS. Quantifying and documenting prior beliefs in clinical trials. Stat Med. 2001;20:581-600.
33. Diamond IR, Grant RC, Feldman BM, et al. Expert beliefs regarding novel lipid-based approaches to pediatric intestinal failure-associated liver disease. JPEN. Journal of parenteral and enteral nutrition. 2014;38:702-10.
34. Deandrea S, Negri E, Ruggeri F. Integrating clinicians’ opinion in the Bayesian meta-analysis of observational studies: the case of risk factors for falls in community-dwelling older people. Epidemiology, Biostatistics and Public Health. 2014;11:e8909.
35. Morris DE, Oakley JE, Crowe JA. A web-based tool for eliciting probability distributions from experts. Environmental Modelling & Software. 2014;52:1-4.
36. Myburgh C, Mouton J. The development of contemporary chiropractic education in Denmark: an exploratory study. J Manipulative Physiol Ther. 2008;31:583-92.
37. Humphreys BK, Peterson CK, Muehlemann D, Haueter P. Are Swiss chiropractors different than other chiropractors? Results of the job analysis survey 2009. J Manipulative Physiol Ther. 2010;33:519-35.
38. Chenot JF, Becker A, Leonhardt C, et al. Use of complementary alternative medicine for low back pain consulting in general practice: a cohort study. BMC Complement Altern Med. 2007;7:42.
87
Chapter 4
Chiropractic Care and Risk For Acute Lumbar Disc 4Herniation: A Population-Based Self-Controlled Case Series
Study
Manuscript 3
Hincapié CA, Tomlinson GA, Côté P, Rampersaud YR, Jadad AR, Cassidy JD.
Chiropractic care and risk for acute lumbar disc herniation: a population-based self-
controlled case series study. In preparation for submission, 2015
Chiropractic care and risk for acute lumbar disc herniation:
a population-based self-controlled case series study
Authors: Cesar A. Hincapié, DC, MHSc,1,2 George A. Tomlinson, PhD,1,3 Pierre Côté,
DC, PhD,4 Y. Raja Rampersaud, MD, MSc,5 Alejandro R. Jadad, MD, DPhil,1,3 J. David
Cassidy, PhD, DrMedSc1,2,6
Affiliations: 1 Dalla Lana School of Public Health, University of Toronto; 2 Toronto
Western Research Institute, University Health Network; 3 Toronto General Research
Institute, University Health Network; 4 Faculty of Health Sciences, University of Ontario
Institute of Technology; 5 Division of Orthopaedic Surgery, Toronto Western Hospital; 6
Department of Sports Science and Clinical Biomechanics, University of Southern
Denmark
Corresponding Author: Cesar A. Hincapié, DC, MHSc, University Health Network,
LuCliff Place – 700 Bay Street, Suite 602, Toronto, Ontario M5G 1Z6, Canada. E-mail:
Key Words: chiropractic; spinal manipulation; intervertebral disc displacement;
epidemiologic methods; risk; adverse events; self-controlled case series
Word Count: 4,966
88
Abstract
Background: Chiropractic care is popular for low back pain, but may increase the risk
for acute lumbar disc herniation (LDH). Low back pain is a common early symptom of
LDH, which commonly precedes LDH diagnosis.
Objective: To investigate associations between chiropractic visits and acute LDH with
early surgical intervention (early surgery, ≤8 weeks), and compare this to associations
between primary care physician (PCP) visits and acute LDH with early surgery.
Study design and setting: Self-controlled case series method using 4 population-
based health care databases in Ontario, Canada.
Methods: All patients with a recorded acute LDH with incident early surgery from April
1994 to December 2004 were selected for the study. Health care exposure visits to
chiropractors and PCPs in the year before the acute LDH event index date were
determined from health billing records. Incidence ratios for acute LDH with early
surgery in exposed periods after chiropractic visits relative to unexposed periods were
estimated within individuals, and compared with incidence ratios for acute LDH with
early surgery following PCP visits.
Results: We identified 195 cases of acute LDH with early surgery in a population of
more than 100 million person-years. We found evidence of strong positive associations
between both chiropractic and PCP visits, and acute LDH. Compared with the risk for
acute LDH with early surgery associated with PCP visits, there was no increase in the
risk associated with chiropractic visits. Lumbar spine-related health care visits were
highly associated with subsequent acute LDH with early surgery.
Conclusion: Acute LDH with early surgical intervention is an exceedingly rare event in
Ontario. Both chiropractic and primary medical care may be associated with an
increase in the risk for acute LDH with early surgery. However, these positive
associations are likely due to patients with low back pain from a developing disc
herniation seeking health care before full clinical expression of acute LDH. We found
no evidence of excess risk for acute LDH with early surgery associated with chiropractic
care compared with PCP care.
89
4.1 Introduction
Low back pain is recognized as an important public health concern around the world
because it is associated with considerable disability and burden to individuals, industry
and society.1-5 It affects about 70% of all people in their lifetime, and between 15% to
30% on any given day with varying types of clinical presentations.6-8
Symptomatic lumbar disc herniation (LDH) can be one of the most recognizable
presentations of low back pain. The diagnosis is typically based on a combination of
symptoms and signs suggesting lumbar spinal nerve root compression or irritation, such
as: lumbosacral radiculopathy (i.e., radicular leg pain or sciatica), nerve root tension
signs, neurologic deficits (i.e., muscle weakness and reflex changes), and advanced
imaging (i.e., MRI or CT) findings that correlate with the clinical syndrome.9-11 However,
many patients present with a less clear clinical picture, involving low back pain in the
early (prodromal) phase that then progresses to radicular leg pain with or without
neurologic signs.12,13 In addition, diagnostic imaging may only be indicated in the
prodromal clinical course if there is suspicion of underlying pathology (e.g., infection,
malignancy) other than disc herniation.14 All these factors contribute in making the
diagnosis during the early phase of the syndrome especially difficult.
In North America, primarily medical physicians, chiropractors and physical therapists
clinically manage back pain.15 Approximately 12% of American and Canadian adults
seek chiropractic care annually, and about 95% of chiropractic visits involve spinal
manipulation treatment.16-19 Chiropractic patients, compared to those seeking medical
care for back pain, tend to be younger and have higher socioeconomic status and fewer
health problems.15,16 Several systematic reviews suggest that chiropractic spinal
manipulation can benefit low back pain, but the summarized studies are of varying
quality and too small to evaluate the risk for rare serious adverse events.20-23 In
addition, there is evidence from randomized clinical trials showing some benefit of
spinal manipulation for the management of LDH,24-28 yet little is known about the safety
of this treatment.
Today, no valid estimate of the risk for acute disc herniation following chiropractic
treatment is available in the scientific literature. The current literature presents case
90
reports and small case series linking lumbar spine manipulation to disc herniation and
cauda equina syndrome.29 However, case reports offer the lowest level of scientific
evidence with regard to the determination of risk and cannot be used to make valid
inferences about the lack of safety of a treatment. They have, however, raised the
hypothesis of potential harm.
Individuals in the early phases of a symptomatic LDH often complain of back pain.12,13
As the condition progresses, most develop leg pain or sciatica. At different points in
time along this course they may seek health care for assessment and intervention. If
chiropractic treatment occurs before the disc herniation progresses to the presence of
radiculopathy or neurologic deficits and is thus diagnosed, then the treatment itself can
be erroneously blamed for causing the disc herniation, since the patient may have
progressed regardless of treatment. This systematic error (protopathic bias) is a type of
reverse-causality bias due to processes that occur before a diagnosed or measured
outcome event.30,31 Given that LDH can initially present as low back pain, it is possible
that these patients seek chiropractic care in the prodrome or early-stage of their disc
herniation,32 implying that an observed association between chiropractic care and acute
LDH may not be causal. Since patients also commonly see primary care physicians
(PCPs) for back pain and this health care encounter is unlikely to cause disc herniation,
an observed association between PCP visits and acute LDH could be attributed to care-
seeking for initial symptoms of LDH (protopathic bias).
Rigorous epidemiological studies are needed to assess the possibility of increased risk
for acute LDH following chiropractic care. The purpose of our study was to investigate
the association between chiropractic care and acute LDH with incident early surgery,
and compare it to the association between PCP care and acute LDH with incident early
surgery by using population-based health care data from Ontario, Canada and the self-
controlled case series method. We hypothesized that evidence that chiropractic care
increases the risk for acute LDH would be present if the measured association between
chiropractic visits and acute LDH exceeds the association between PCP visits and
acute LDH.
91
4.2 Methods
Data Sources: We combined data from 4 large population-based health care
databases, totaling over 100 million person-years of observation: 1) the Ontario
Registered Persons Database; 2) the Discharge Abstract Database (DAD); 3) the
National Ambulatory Care Reporting System (NACRS); and 4) the Ontario Health
Insurance Plan Billings Database (OHIP). These health administrative databases
include information on patients’ hospitalizations with diagnosis and surgical intervention
coding, emergency department (ED) visits with diagnosis and intervention coding, and
practitioner (chiropractor and physician) and other health services utilization as
documented by fee-for-service billings with diagnosis coding. Linkage between
datasets was done using encrypted health card numbers as unique identifiers, and all
health information obtained was anonymous. We requested and received these data
from the Health Data Branch of the Ontario Ministry of Health and Long-Term Care.
Research ethics board approval was received at the University Health Network (REB
#09-0668-AE), and administrative approval from the Office of Research Ethics at the
University of Toronto (protocol reference #24781).
Source Population: The source population was all Ontario residents, aged 18 years or
older, who were covered by the provincial universal health care system (Ontario Health
Insurance Plan) between April 1, 1994 and November 30, 2004. We could not study
chiropractic care after November 30, 2004, as this was the last date that chiropractic
care was covered under OHIP and visits after this date could not be identified for the
purpose of our study. Members of the armed forces, federal inmates, the Royal
Canadian Mounted Police, and First Nation individuals living on reserves were ineligible
for the study, as they are not covered by the provincial health care system. Participants
with a history of long term care service in the 2 years prior to their surgery were also
excluded.
Outcome and Case Definition: Eligible cases were all cases of acute LDH with
incident early surgical intervention (≤8 weeks) between April 1, 1994 and November 30,
2004, with at least 2 years of health care coverage prior to their surgery to ensure that
visits to chiropractors and physicians (exposures) could be identified. We used the
92
following 4-step approach to define cases of acute LDH with incident early surgery
based on the coding detailed in Appendix H, Table H.1:
1. Initial cases of incident LDH surgery were identified using at least one disc surgery
intervention code and one LDH diagnosis code from DAD or NACRS data. Hospital
discharge ICD-9 codes for LDH have been shown to have a high positive predictive
value when compared to chart review.33
2. Preliminary cases of acute LDH with incident early surgery were then identified by
including persons (from Step 1 above) who presented to a hospital ED for LDH within 8
weeks prior to their LDH surgery using OHIP and NACRS data.
The 8-week ED visit window (ED window) prior to LDH surgery was chosen in
consultation with a spine surgery expert (YRR) and data suggesting early surgery for
severe or intractable radiculopathy due to disc herniation can occur up to 6 to 12 weeks
following the onset of symptoms.34,35 Given the typically long surgical wait times in the
Ontario health system during our study period and the inherent limitations of
administrative health data,36-40 it was felt that this case scenario (early surgery following
an ED visit for LDH) would most reliably represent a clinically relevant and significant
acute LDH event, and possible rare serious adverse event following chiropractic
treatment. The event index date was defined as this date of service for LDH at the ED
within 8 weeks prior to the LDH surgery date.
3. Using DAD and NACRS data, we excluded persons who had a diagnosis of LDH or
other conditions potentially associated with LDH, 21 months prior to their event index
date. Excluded also were those with other spine surgery interventions on or prior to their
date of incident LDH surgery.
4. Using OHIP data, we excluded persons who had, within 21 months prior to their
event index date, a diagnosis of LDH or other conditions potentially associated with
LDH, specialist visits to neurosurgeons, orthopedic surgeons, neurologists, physiatrists
and rheumatologists, or advanced spine imaging or diagnostic testing related to LDH.
Our outcome of interest was acute LDH with incident early surgical intervention. We did
not include cauda equina syndrome as a target for our case definition because this
93
more severe serious adverse event was investigated in another study41 that formed part
of an overall epidemiologic research program on the safety of chiropractic care.
Exposures: All ambulatory care billing records for visits to chiropractors and PCPs
were identified in the OHIP data (codes in Appendix H, Table H.2) starting from the
one-year period before the event index date (start of the observation period). Multiple
billings on the same date were counted as one visit. Lumbar spine-related chiropractic
visits were defined using diagnosis codes: C07-C09, lumbar, lumbosacral, sacroiliac
and coccyx subluxation; C13-15, multiple site subluxation; C20-24, acquired, postural or
congenital spine curvature; C31-C32, lumbosacral sprain/strain. For PCP visits, we
included community medicine physicians if they submitted ambulatory fee codes to
OHIP. PCP visits for natal care, ophthalmology care, genetic screening and group
counselling were excluded. Lumbar spine related PCP visits were identified using
diagnosis codes: 724, lumbar strain, lumbago, coccydynia, sciatica; and 847, low back,
coccyx sprain/strain. When addressing repeat exposures (repeat chiropractic or PCP
visits), we assumed the risk to remain constant after each exposure, thus not allowing
for a dose effect.
Study Design: The self-controlled case series method is a type of a cohort study in
which the relative risk is based on within-person comparisons rather than between-
person comparisons, with each person contributing to both the exposed and unexposed
observation time.42,43 This means the potential confounding effects of differences
between patients are removed if these factors are fixed (time-invariant) over the
observation period.
We examined the risk for acute LDH after exposure to chiropractic care and PCP care.
Persons who have had health care encounters (chiropractic or medical) may differ from
those who have not in ways that can be difficult to measure and control for. Some of
these differences may also be associated with the future risk for disc herniation, which
makes a conventional observational (case-control or cohort) design a less valid source
of information on this association. Therefore, we used the self-controlled case series
method, which relies on within-person comparisons in a sample of persons, all of which
have had the outcome of interest (cases-only). This design is most appropriate when a
brief, well-defined exposure (chiropractic visit) causes a transient change in risk of a
94
rare-onset disease (acute LDH with early surgery). The main advantage of this design
is that inference is within a person, so that fixed confounders (those that do not vary
with time during the observation period) are controlled for by design. We calculated
incidence rate ratios of events occurring during prespecified risk periods after an
exposure (0-2 days after a visit, 0-7 days after a visit, 0-14 days after a visit, and 0-28
days after a visit), relative to all other observed time (baseline unexposed periods) for
each person. Figure 4.1 illustrates the self-controlled cases series design.
Figure 4.1. Representation of the self-controlled case series design. Times are expressed in terms
of calendar days from the start of the study period and the data are viewed as in a prospective cohort.
Incidence rate ratios compare the rate of events during exposed periods with the rate of events during
unexposed periods. Two possible scenarios (each representing a single patient) for the timing of
chiropractic or primary care physician visits and acute LDH events are shown. A. Patient has three risk
periods (each after a chiropractic or primary care physician visit) and has an acute LDH event at baseline
between the second and third visits. B. Patient has two visits and has an acute LDH during the second
risk period. Each risk period begins the day of a visit and lasts up to 28 days after a visit (risk periods are
not drawn to scale relative to length of baseline periods). Four risk periods were prespecified: 0-2, 0-7, 0-
14, and 0-28 days after a visit
Statistical Analysis: An important assumption underpinning the standard self-
controlled case series method is that the occurrence of an event must not alter the
Start of study period End of study period
Baseline unexposed period
Risk period after a chiropractic or primary care physician visit
Visit Visit Acute LDH event
Acute LDH event
Visit
A
B
95
probability of subsequent exposure. This assumption does not hold for health care
utilization exposures (chiropractic or PCP visits) because the occurrence of a LDH likely
influences the probability of post-event visits to a chiropractor or PCP. When this
assumption fails, a valid and reasonable strategy is to use a recently developed
modification of the method (the pseudo-likelihood approach) that uses counterfactual
modeling and only exposures experienced prior to the event to estimate the relative
incidence.42,44,45 We therefore used this modified self-controlled case series method for
our study.
The observation time in the year prior to the index ED visit was divided into risk periods
and unexposed time. We ran analyses with risk periods of four different lengths: (i) 0-2
days after a visit, (ii) 0-7 days after a visit, (iii) 0-14 days after a visit, and (iv) 0-28 days
after a visit; the baseline period comprised remaining unexposed time. To estimate the
associations between acute LDH and chiropractic care, and acute LDH and PCP care,
we used conditional Poisson regression to calculate incidence rate ratios (IRRs) and
95% confidence intervals for events occurring within each risk period compared to
baseline.42,44-46 Separate analyses were conducted for exposure to lumbar spine related
chiropractic and PCP visits.
We examined the annual incidence of events by fiscal year of health data (1-Apr to 31-
Mar) and observed an unequal distribution of events by fiscal year corresponding to the
implementation of the NACRS database for ED services in Ontario in 2002. To assess
and control for event incidence temporality that may have been related to the initiation
of NACRS on April 1, 2002, we also calculated NACRS-adjusted risk estimates for all
analyses.
Several other sensitivity analyses were undertaken: (1) To assess the impact of our
case conceptualization of early (≤8 weeks) surgical intervention after an ED LDH visit,
all analyses were performed varying the time interval between date of presentation to
the ED for acute LDH and LDH surgery date (in addition to the primary analysis time
interval of up to 8-weeks between an ED LDH visit and LDH surgery date, we examined
time intervals of up to 4- and 12-weeks). (2) To investigate the effect of potential
misclassification of the event diagnosis, we repeated analyses using a broader coding
approach for LDH diagnosis on the event index date (detailed in Appendix H, Table
96
H.1). (3) To examine the effect of event-day visits to chiropractors and PCPs (possible
reverse-causality), we ran analyses with a 1-day lag on exposures in order to estimate
incidence ratios excluding event-day exposure visits. (4) Where possible, we also
performed analyses in subgroups of patients whom only had chiropractic care and those
whom only had PCP care during their observation period. All analyses were performed
using R, version 2.15.147
Finally, we used the nonparametric bootstrap for three additional analyses: (1) To check
the model-based confidence intervals from the conditional Poisson regression for our
primary model, we took 2000 bootstrap resamples from the cases of acute LDH with
early surgery and refitted the models for both DC visits and PCP visits to each sample.
The 2.5th and 97.5th percentiles of the 2000 estimates of the IRRs were used to estimate
the 95% confidence interval for the IRR; (2) For each of the bootstrap samples, we
divided the IRR for a DC visit by the IRR for a PCP visit, thereby estimating the specific
effect on the risk for acute LDH with early surgery of seeing a chiropractor relative to
seeing a primary care physician; (3) We made a scatterplot of the 2000 bootstrapped
pairs of IRRs and a kernel density estimate of the ratio in (2) to better visualize the
relationship between the risks for acute LDH with early surgery associated with each
type of healthcare visit.
Role of the Funding Source: The sponsors of the study had no role in study design,
data collection, data analysis, data interpretation, the writing of the report, or in the
decision to submit the report for publication.
4.3 Results
Identification of Acute LDH Cases: A total of 36,745 persons were identified from the
Ministry of Health data with incident disc surgery during the study period. Of these, we
excluded 423 that were under 18 years of age at the time of their surgery, 180 with a
history of long term care service in the 2 years prior to their surgery, 9,576 with no LDH
diagnosis linked to their disc surgery record, 25,323 with no ED LDH visit within 8
weeks prior to their surgery date, and 1,048 with prevalent LDH or associated
diagnoses and interventions within 21 months prior to the ED event index date. A flow
diagram showing how cases were selected is shown in Figure 4.2. For our primary
97
analysis, 195 incident cases of acute LDH with early surgery met our inclusion/exclusion
criteria over the 11-year study period, and are described in Table 4.1. The mean age of
cases was 43 years at the LDH surgery date, and 60% were male. Of the 195 cases,
72 (37%) had visited a chiropractor during the observation period starting 12 months
prior to their event index date, while 186 (95%) had visited a PCP within that time.
Comorbidities were approximately as common in patients with chiropractic visits as in
patients with PCP visits.
98
Figure 4.2. Case definition flow diagram.
36,745 persons who had incident disc surgery from 1 April 1994 to 30 November 2004
General eligibility exclusions (n=603)
423 under 18 years of age on disc surgery date
180 with long-term care service within 2 years prior to disc surgery date
26,566 initial cases of incident LDH surgery
Excluded persons with no LDH diagnosis linked to disc surgery (n=9,576)
Top 5 diagnoses among 23,830 excluded hospitalization records 722.0 – cervical disc herniation (n=2,175; 9.1%)
724.0 – spinal stenosis other than cervical (n=1,720; 7.2%)
722.5 – degeneration of thoracic or lumbar IVD (n=1,342; 5.6%)
401.9 – unspecified essential hypertension (n=807; 3.4%)
721.1 – cervical spondylosis with myelopathy (n=601; 2.5%)
344.6 & G83.4 – cauda equine syndrome (n=223; 0.9%)
36,142 eligible cases of incident disc surgery
1,243 preliminary cases of acute LDH with incident early surgery
Excluded persons with no ED LDH visit within 8 weeks prior to LDH surgery
date (n=25,323)
195 cases of acute LDH with incident early surgery; 72 with DC visits, 186 with PCP visits
Excluded persons who had other spine surgery interventions on or prior to
LDH surgery date; and, persons with a diagnosis of LDH or other
conditions potentially associated with LDH within 21 months prior to the
event index date, using DAD and NACRS data (n=617)
Excluded persons who had (within 21 months prior to the event index date)
a diagnosis of LDH or other conditions potentially associated with LDH,
specialist visits, or advanced spine imaging or diagnostic testing indicating
LDH, using OHIP data (n=431)
99
Table 4.1. Characteristics of study participants at the time of the recording of an acute lumbar disc
herniation with incident early surgery
Characteristic
All cases
(n=195)*
Cases with
DC visits
(n=72)†
Cases with
PCP visits
(n=186)
Men – n (%) 116 (59.5) 47 (65.3) 110 (59.1)
Age at the time of LDH surgery – mean years (SD) 42.6 (12.9) 42.1 (11.9) 42.5 (12.7)
Comorbidities‡ – n (%)
Diabetes 11 (5.6) 6 (8.3) 11 (5.9)
Hypertension 33 (16.9) 15 (20.8) 32 (17.2)
Coronary heart disease 26 (13.3) 9 (12.5) 25 (13.4)
High cholesterol 17 (8.7) 8 (11.1) 17 (9.1)
Osteoarthritis 28 (14.4) 12 (16.7) 28 (15.1)
Rheumatoid arthritis 0 0 0
Abbreviations: DC, chiropractor; LDH, lumbar disc herniation; PCP, primary care physician
* Persons included in the primary analysis of acute LDH with early surgery
† The majority of cases (n not reportable) had both DC and PCP visits during their observation period
‡ Comorbid conditions identified using OHIP billing diagnosis codes during 2 years prior to the LDH
surgery: diabetes (code 250), hypertension (codes 401 to 403), coronary heart disease (codes 410 to
415 and 426 to 429), high cholesterol (code 272), osteoarthritis (code 715), rheumatoid arthritis (code
714)
Description of Visits: Overall, 22% of cases had a chiropractic visit within 14 days
prior to the event index date, while 59% of cases had a PCP visit within that same time
interval (Table 4.2). A total of 57 cases were identified during the first 8 years of the
study period (April 1, 1994 to March 31, 2002), while 138 cases were identified during
the balance of the study period (April 1, 2002 to November 30, 2004). We found an
uneven distribution of event incidence by fiscal year of Ministry of Health data, which
corresponded with the initiation of the NACRS database for ED services in Ontario, on
April 1, 2002. For cases whose event occurred prior to the 2002 fiscal year (pre-
NACRS), 9 cases (15.8%) had consulted a chiropractor within 7 days of their index
date, and 27 cases (47.4%) had consulted a PCP within the same interval. Among
cases whose event occurred after the start of the NACRS database (event fiscal year
100
≥2002), 14.5% had 4 or more chiropractic visits within the month before the index date,
while 9.4% had 4 or more PCP visits within the same time period.
Table 4.2. Distribution of chiropractor and primary care physician visits before the event index date for
all cases and stratified by event fiscal year*
Exposure
All cases
(n=195)
Event fiscal
year <2002
(n=57)
Event fiscal
year ≥2002
(n=138)
Most recent DC visit before event – n cases (%)
0-2 days 23 (11.8) † 19 (13.8)
0-7 days 38 (19.5) 9 (15.8) 29 (21.0)
0-14 days 42 (21.5) 11 (19.3) 31 (22.5)
Most recent PCP visit before event – n cases (%)
0-2 days 59 (30.3) 19 (33.3) 40 (29.0)
0-7 days 87 (44.6) 27 (47.4) 60 (43.5)
0-14 days 114 (58.5) 34 (59.6) 80 (58.0)
Number of DC visits in month before event – n cases (%)
None 147 (75.4) 44 (77.2) 103 (74.6)
1 to 3 21 (10.8) 6 (10.5) 15 (10.9)
4 or more 27 (13.8) 7 (12.3) 20 (14.5)
Number of PCP visits in month before event – n cases (%)
None 59 (30.3) 17 (29.8) 42 (30.4)
1 to 3 121 (62.1) 40 (70.2)†† 83 (60.1)
4 or more 15 (7.7) † 13 (9.4)
Abbreviations: DC, chiropractor; LDH, lumbar disc herniation; PCP, primary care physician
* We found an unequal distribution of events by fiscal year of incidence corresponding to the
implementation of the NACRS database for ED services in Ontario, Canada in 2002.
† Cell sizes with fewer than six (6) cases were suppressed to ensure confidentiality.
†† Cell was merged with the “4 or more” group to ensure confidentiality.
101
Self-Controlled Cases Series Results: Using the modified self-controlled cases series
method, there was evidence of an increased association between chiropractic visits and
acute LDH with early surgery regardless of the risk period examined (Table 4.3). For
exposed periods 0 to 7 days after a chiropractor visit, there was a cumulative 9 years of
exposed time and 38 patients experienced an incident acute LDH during this exposed
period (adjusted IRR, 12.91; 95% CI, 7.16-23.28).
There was similar evidence of an increased risk of acute LDH for all risk periods
examined up to 28 days after a visit to a PCP. A total of 28 patient-years of exposed
time accumulated for risk periods 0 to 7 days after a PCP visit and 87 persons
experienced an acute LDH within this risk period (adjusted IRR, 14.48; 95% CI, 9.90-
21.18). For patients with PCP visits, 39 had their acute LDH event occur on the same
day as a visit.
When restricting the analyses to visits related to lumbar spine complaints, we observed
sharp increases in associations for PCP visits, but not for associations for chiropractic
visits, which remained about the same (Table 4.3). Adjustment for the initiation of the
NACRS database (April 1, 2002) attenuated the observed associations, however, our
estimates remained consistently high for all risk periods examined for both chiropractic
and PCP visits, with substantial overlap of confidence intervals.
102
Table 4.3. Incidence rate ratios for an acute lumbar disc herniation with first early surgery event in exposed periods following chiropractic and primary
care physician visits
DC Care (n=72) PCP Care (n=186)
Exposure
Patient
years
n
Acute
LDHs*
Crude IRR for Acute
LDH (95% CI)
NACRS-adjusted IRR
for Acute LDH
(95% CI)
Patient
years
n
Acute
LDHs*
Crude IRR for Acute
LDH (95% CI)
NACRS-adjusted IRR
for Acute LDH
(95 % CI)
Any visit
Unexposed Baseline Baseline Baseline Baseline
0-2 days after visit 5.03 23 9.46 (4.87 - 18.41) 8.26 (4.35 - 15.67) 11.42 59 18.00 (11.80 - 27.45) 15.20 (10.17 - 22.73)
0-7 days after visit 9.11 38 15.70 (8.85 - 27.86) 12.91 (7.16 - 23.28) 28.35 87 17.50 (11.72 - 26.15) 14.48 (9.90 - 21.18)
0-14 days after visit 12.74 42 14.06 (7.80 - 25.35) 11.80 (6.42 - 21.67) 47.66 114 19.77 (12.18 - 32.10) 15.72 (9.96 - 24.80)
0-28 days after visit 18.60 48 14.41 (7.66 - 27.11) 11.72 (6.06 - 22.68) 76.82 136 33.57 (17.39 - 64.78) 25.67 (13.56 - 48.60)
Lumbar spine visit
Unexposed Baseline Baseline Baseline Baseline
0-2 days after visit 4.18 20 11.27 (5.86 - 21.68) 9.72 (5.11 - 18.48) 4.48 43 45.94 (26.52 - 79.56) 40.61 (24.15 - 68.31)
0-7 days after visit 7.45 33 16.80 (9.05 - 31.21) 13.98 (7.36 - 26.57) 11.08 63 44.89 (25.07 - 80.36) 37.92 (21.78 - 66.01)
0-14 days after visit 10.47 36 14.91 (7.96 - 27.90) 12.22 (6.32 - 23.63) 18.51 79 54.68 (28.62 - 104.50) 44.12 (23.42 - 83.13)
0-28 days after visit 15.47 41 15.48 (7.91 - 30.29) 12.25 (5.96 - 25.18) 29.38 93 85.29 (41.31 - 176.10) 67.02 (31.37 - 143.18)
Abbreviations: CI, confidence interval; DC, chiropractor; IRR, incidence rate ratio; LDH, lumbar disc herniation; NACRS, national ambulatory care
reporting system; PCP, primary care physician.
* Number of acute LDH with incident early surgery events that occurred during the exposed period.
103
Results of Sensitivity and Secondary Analyses: Qualitatively, the sensitivity
analyses gave similar results to those obtained in the primary analysis (Table 4.4). The
results of the sensitivity analyses showed a consistent trend of strong positive
associations between both chiropractic and PCP visits and acute LDH with incident
early surgical intervention; with increases in associations between acute LDH and PCP
lumbar spine visits.
The bootstrap estimates of the IRRs in the primary analysis of any visit exposure with a
risk period of 0-7 days was in line with our model estimates: adjusted IRR for acute LDH
with early surgery, 9.50 (95%CI, 3.01-35.51) for chiropractic care, compared with 14.45
(95%CI, 7.60-29.28) for PCP care (Figure 4.3).
Finally, the bootstrap analysis of the ratio of IRRs suggested a positive safety profile for
chiropractic care relative to the baseline risk represented by PCP care (Figure 4.4), with
a median ratio of the two IRRs of 0.65 (95%CI, 0.19-2.43). The majority (72%) of the
2000 bootstrapped values had a smaller incidence ratio for a DC visit than for a PCP
visit (Figure 4.3).
104
Table 4.4. Results of sensitivity analyses compared to primary analysis of an acute lumbar disc herniation with first early surgery event in risk periods 0
to 7 days after chiropractor and primary care physician visits*
DC Care PCP Care
Patient
years
n Acute
LDHs†
NACRS-adjusted
IRR for Acute LDH
(95% CI)
Patient
years
n
Acute
LDHs†
NACRS-adjusted
IRR for Acute LDH
(95 % CI)
Primary analysis‡ (n=195)
Any visit 9.11 38 12.91 (7.16 – 23.28) 28.35 87 14.48 (9.90 – 21.18)
Lumbar spine visit 7.45 33 13.98 (7.36 – 26.57) 11.08 63 37.92 (21.78 – 66.01)
Sensitivity analyses by specified criteria
Broad inclusion coding for index ED LDH visit§ (n=724)
8-week ED window, broad coding, any visit 38.39 123 10.79 (7.80 – 14.93) 107.58 333 18.94 (14.67 – 24.44)
8-week ED window, broad coding, lumbar spine visit 29.48 102 12.10 (8.42 – 17.39) 41.54 252 63.49 (43.95 – 91.72)
ED LDH visit within 4-weeks (n=258) and 12-weeks (n=961) prior to LDH surgery date
4-week ED window, narrow coding, any visit 6.11 27 13.35 (6.53 – 27.28) 17.45 59 14.00 (8.77 – 22.36)
4-week ED window, narrow coding, lumbar spine visit 4.82 24 15.72 (6.96 – 35.52) 6.10 46 43.68 (22.28 – 85.63)
12-week ED window, narrow coding, any visit 13.41 50 9.93 (5.82 – 16.95) 40.63 117 14.01 (9.96 – 19.71)
12-week ED window, narrow coding, lumbar spine visit 10.88 41 9.86 (5.51 – 17.62) 16.52 84 37.10 (22.96 – 59.95)
Exposure visits lagged by 1 day (exclude potential event-day reverse-causality) (n=195)
Lagged exposures, 8-week ED window, narrow coding, any visit 9.11 39 14.15 (7.73 – 25.90) 28.38 74 10.85 (7.43 – 15.85)
Lagged exposures, 8-week ED window, narrow coding, lumbar spine
visit 7.46 34 15.60 (8.04 – 30.27) 11.08 51 31.77 (17.96 – 56.19)
105
DC Care PCP Care
Patient
years
n Acute
LDHs†
NACRS-adjusted
IRR for Acute LDH
(95% CI)
Patient
years
n
Acute
LDHs†
NACRS-adjusted
IRR for Acute LDH
(95 % CI)
Patients with only DC care versus patients with only PCP care
8-week ED window, narrow coding, any visit ND ND ND 6.28 33 19.45 (11.89 – 31.82)
8-week ED window, narrow coding, lumbar spine visit ND ND ND 5.65 30 40.58 (18.99 – 86.74)
12-week ED window, broad coding, any visit 2.10 6 8.94 (2.16 – 37.02) 86.61 234 18.49 (13.97 – 24.45)
12-week ED window, broad coding, lumbar spine visit 1.65 || 10.56 (1.77 – 63.04) 33.26 163 62.09 (43.17 – 89.32)
Abbreviations: CI, confidence interval; DC, chiropractor; ED, emergency department; IRR, incidence rate ratio; LDH, lumbar disc herniation; NACRS,
national ambulatory care reporting system; ND, no data; PCP, primary care physician.
* Exposed periods 0-7 days after a visit compared with baseline. Baseline period is all observation time except for the 8-day risk period after a visit.
† Number of acute LDH with incident early surgery events that occurred during the exposed period.
‡ 8-week ED window, narrow inclusion coding (see Step 2 of Appendix H, Table H.1), IRR for acute LDH within exposed periods 0-7 days after visit.
§ In addition to the ED LDH diagnosis codes used for the primary analysis (narrow coding) (Step 2 of Appendix H, Table H.1), broad coding for ED LDH
diagnosis included codes 724, 847 and M545.
|| Cell sizes with fewer than six (6) cases were suppressed to ensure confidentiality.
106
Figure 4.3. Bootstrap incidence rate ratios (IRRs) for primary analysis of any visit exposure with a risk
period of 0-7 days following chiropractic versus primary care physician care.
Figure 4.4. Kernel density estimate of the ratio of bootstrap incidence rate ratios for chiropractic versus
primary care physician care.
107
4.4 Discussion
We investigated the relationship between chiropractic care and the risk for acute LDH
with early surgical intervention in adults by performing a self-controlled case series
study. Our study showed that among adults, chiropractic care may be positively
associated with acute LDH that is managed with early surgery.
We also found strong positive associations between PCP visits and subsequent acute
LDH that is surgically managed. We acknowledge that protopathic bias cannot be
excluded as a plausible explanation for this finding, by which some patients with low
back pain, a common early symptom of LDH,12,13 may have sought health care due to
this prodromal symptom prior to LDH being diagnosed. Since it is unlikely that PCPs
cause LDH through their care for these patients, we posit that that the observed
association between recent PCP visits and early surgically managed acute LDH
represents the background risk associated with patients seeking care for early
prodromal symptoms of LDH. Because the associations between chiropractic care and
acute LDH were not greater than the associations between PCP visits and acute LDH,
there appears to be no excess risk for acute LDH following chiropractic care greater
than the baseline risk associated with primary medical care seeking.
In studies examining the risk for acute LDH following a health care visit exposure, the
potential for confounding is great because persons who seek health care may differ
from those who do not in ways that may be difficult to measure and control for. A major
strength of our study is that we used the self-controlled case series method, in which
within-person comparisons are done, thereby controlling for both known and unknown
confounding factors that do not change over time and which could affect the risk for
acute LDH. This is important because smoking, obesity, average occupational lumbar
spine load and other important risk factors for symptomatic LDH are not commonly
recorded in health administrative data.
Our findings are similar to a previous study examining the association between
chiropractic care and vertebrobasilar stroke, in which similar associations were found
between both chiropractic care and PCP care, and the outcome of vertebrobasilar
stroke.48 In their study, Cassidy and colleagues reported no evidence of excess risk for
108
vertebrobasilar stroke associated with chiropractic care compared to primary medical
care. They attributed their observed associations to the effect of patients seeking health
care, from both chiropractors and PCPs, for prodromal neck pain and headache before
receiving a diagnosis of vertebrobasilar stroke.48
Strengths and Limitations: Our study population included the entire population
registered in Ontario’s provincial health care system over an 11-year period,
representing over 100 million person-years of observation. This allowed us to identify
all early surgically managed cases of acute LDH, as well as OHIP-insured visits to
chiropractors and PCPs. Yet, we found only 195 early-surgically managed acute LDH
cases for our primary analysis, limiting the precision of our estimates. In particular, our
NACRS-adjusted analyses were based on smaller numbers of exposed cases, and
further stratification by lumbar spine related visit diagnosis codes yielded wider
confidence intervals. Nonetheless, there are few other jurisdictions in the world where it
would be possible to carry out a population study linking incident surgically managed
acute LDH with chiropractic and PCP visits.
A notable strength of our study was the use of the self-controlled case series method
and within-person comparisons. This ensured that confounding by time-invariant
factors was controlled for by design. Confounding could have occurred only if patients
had risk factors for surgically managed acute LDH that changed over time, and if these
factors were also associated with the timing of visits to chiropractors and PCPs, and if
these time-dependent effects existed for a high proportion of study participants.
Another strength was that exposures were measured independently of the outcome.
This eliminated the possibility of exposure recall bias. With respect to chiropractic care,
the exposure measure does not necessarily indicate a treatment of lumbar spine
manipulation, although there is evidence that about 95% of chiropractic visits involve
spinal manipulation treatment.16-19
There was some overlap between chiropractic care and PCP care. In our principal
analysis, fewer than six cases (exact number suppressed to ensure confidentiality) had
seen only a chiropractor; and 115 (59.0%) had seen only a PCP. In a sensitivity
analysis we performed using up to a 12-week time interval between ED LDH visit and
LDH surgery date and a broader coding approach for ED LDH diagnosis, we identified a
109
total of 961 cases of acute LDH with incident early surgery, of which 18 cases (1.9%)
had seen only a chiropractor during their observation period, and 514 cases (53.5%)
had seen only a PCP. Although the small number of cases with only chiropractic care
limited the precision of the estimates, this sensitivity analysis made no material
difference to our findings (Table 4.4).
Key limitations of our study should be considered. Misclassification bias is a major
limitation of using administrative health data. Our case conceptualization is an
algorithm of surgical intervention codes, hospital diagnosis codes, and health services
diagnosis and fee codes, from multiple databases and over an 11-year study period.
The reliability and validity of our case definition is not known; however, because of the
high costs of chart abstraction, an attempt to validate the case definition was not
undertaken. It is possible that our case definition algorithm resulted in an overinclusive
or underinclusive cohort of cases, and our definition does not capture events that were
not managed with early surgery. To investigate the impact of our definition of early
surgery, we carried out sensitivity analyses varying the time interval between the LDH
surgery date and the preceding ED visit for disc herniation; from an ED visit within 4
weeks prior to LDH surgery date, to an ED visit within 12 weeks prior to the surgery
date. We also examined the effect of potential misclassification due to the event index
diagnostic coding by repeating analyses using a broader coding approach for ED LDH
diagnosis on the event index date. These sensitivity analyses yielded results similar to
our primary analysis and did not change our conclusions (Table 4.4).
In our study, the exposed period starts on the same day as a visit to a chiropractor or
PCP; therefore, the day of a health encounter visit contributes to the risk period. This
leads to the challenge of events that occurred on the same day as an exposure visit and
the possibility that the exposure visit may have taken place after the ED LDH visit,
which could have plausibly been the case for both chiropractic and PCP event-day
visits. This potential reverse-causality bias would lead to our estimates overestimating
the associations between acute LDH and both chiropractic and PCP care. To assess
the effect of event-day visits to chiropractors and PCPs, we ran a sensitivity analysis
with a 1-day lag on exposure visit dates in order to estimate incidence ratios excluding
event-day exposure visits. This, in effect, re-specified the risk period intervals as: 1-3,
1-8, 1-15, and 1-29 days after a visit, thereby removing the effect of event-day exposure
110
visits on our estimates. Our analysis showed some attenuation of the PCP estimates,
but made no material difference to our findings (Table 4.4).
Although our study period could only extend up until December 2004 due to the
limitation of OHIP coverage for chiropractic care in Ontario, we believe that our findings
are likely still relevant today as the practice of chiropractic care for acute low back pain
and the indications for early LDH surgery (i.e., severe intractable pain and/or neurologic
deficit) have not changed materially over ensuing time period.
Our study is the first epidemiologic investigation of the association between chiropractic
care and acute LDH with early surgical intervention. Our findings should be interpreted
cautiously and considered within clinical context. We have not excluded spinal
manipulation as a possible cause of some cases of acute LDH. Our results suggest
that there is a positive association between a patient seeking chiropractic care and
subsequent acute LDH that is surgically managed. However, the association was not
greater than the association we found between PCP care and surgically managed acute
LDH. The observed associations are likely explained by patients seeking health care
for prodromal symptoms of LDH (i.e., protopathic bias). Chiropractic care is not likely to
be a major cause of these rare events. Nevertheless, we cannot exclude the possibility
that chiropractic spinal manipulation, or even simple physical examination by any health
care practitioner, could cause an exacerbation of a developing or underlying disc
herniation leading to full clinical expression of the pre-existing condition.
Regrettably, there are no clinical screening tests to identify patients with back pain that
may be at increased risk of developing acute disc herniation,49 and current evidence
indicates poor diagnostic performance of most physical tests used to identify LDH.10,50
Future studies would need to be multi-centered and undertake prospective, unbiased
and detailed ascertainment of patients’ reasons for seeking health care. With respect to
clinical practice in light of our current state of knowledge, decisions on how to treat
patients with low back pain should continue to be guided primarily by clinical
effectiveness and current best practice.51,52
111
4.5 Conclusion
Our population-based self-controlled case series study found positive associations
between health care visits to both chiropractors and PCPs, and acute LDH with early
surgical intervention. This suggests that patients with prodromal back pain related to a
developing disc herniation seek health care from both chiropractors and PCPs before
full clinical expression of acute LDH that is subsequently managed with early surgery.
Funding Sources
Dr. Cesar Hincapié was supported by a Fellowship Award in the Area of Knowledge
Translation from the Canadian Institutes of Health Research, with additional support
from the Canadian Chiropractic Research Foundation. The study received additional
support from the Canadian Chiropractic Protective Association.
Acknowledgements
We thank Drs. Heather Whitaker and Ronny Kuhnert for their advice about the self-
controlled case series design and analysis of this study. We acknowledge the Ontario
Ministry of Health and Long-Term Care for support with data acquisition and linkage.
References
1. Hoy D, Brooks P, Blyth F, Buchbinder R. The epidemiology of low back pain. Best Pract Res Clin Rheumatol. 2010;24:769-81.
2. Hoy D, March L, Brooks P, et al. The global burden of low back pain: estimates from the Global Burden of Disease 2010 study. Ann Rheum Dis. 2014;73:968-74.
3. Cassidy JD, Côté P, Carroll LJ, Kristman V. Incidence and course of low back pain episodes in the general population. Spine. 2005;30:2817-23.
4. Hincapié CA, Cassidy JD, Côté P. Is a history of work-related low back injury associated with prevalent low back pain and depression in the general population? BMC Musculoskelet Disord. 2008;9:22.
5. Vos T, Flaxman AD, Naghavi M, et al. Years lived with disability (YLDs) for 1160 sequelae of 289 diseases and injuries 1990–2010: a systematic analysis for the Global Burden of Disease Study 2010. The Lancet. 2012;380:2163-96.
6. Kent PM, Keating JL. The epidemiology of low back pain in primary care. Chiropr Osteopat. 2005;13:13.
7. Walker BF. The prevalence of low back pain: a systematic review of the literature from 1966 to 1998. J Spinal Disord. 2000;13:205-17.
8. Andersson GB. Epidemiology of low back pain. Acta Orthop Scand Suppl. 1998;281:28-31.
112
9. Genevay S, Atlas SJ, Katz JN. Variation in eligibility criteria from studies of radiculopathy due to a herniated disc and of neurogenic claudication due to lumbar spinal stenosis: a structured literature review. Spine. 2010;35:803-11.
10. van der Windt DAWM, Simons E, Riphagen II, et al. Physical examination for lumbar radiculopathy due to disc herniation in patients with low-back pain. Cochrane Database Syst Rev. 2010:CD007431.
11. Vroomen PC, de Krom MC, Wilmink JT, Kester AD, Knottnerus JA. Diagnostic value of history and physical examination in patients suspected of lumbosacral nerve root compression. J Neurol Neurosurg Psychiatry. 2002;72:630-4.
12. Deyo RA, Loeser JD, Bigos SJ. Herniated lumbar intervertebral disk. Ann Intern Med. 1990;112:598-603.
13. Weber H. The natural history of disc herniation and the influence of intervention. Spine. 1994;19:2234-8; discussion 3.
14. Koes BW, van Tulder MW, Peul WC. Diagnosis and treatment of sciatica. BMJ. 2007;334:1313-7.
15. Côté P, Cassidy JD, Carroll L. The treatment of neck and low back pain: who seeks care? who goes where? Med Care. 2001;39:956-67.
16. Hurwitz EL, Chiang LM. A comparative analysis of chiropractic and general practitioner patients in North America: findings from the joint Canada/United States Survey of Health, 2002-03. BMC Health Serv Res. 2006;6:49.
17. Hurwitz EL, Coulter ID, Adams AH, Genovese BJ, Shekelle PG. Use of chiropractic services from 1985 through 1991 in the United States and Canada. Am J Public Health. 1998;88:771-6.
18. Senstad O, Leboeuf-Yde C, Borchgrevink CF. Side-effects of chiropractic spinal manipulation: types frequency, discomfort and course. Scand J Prim Health Care. 1996;14:50-3.
19. Leboeuf-Yde C, Hennius B, Rudberg E, Leufvenmark P, Thunman M. Side effects of chiropractic treatment: a prospective study. J Manipulative Physiol Ther. 1997;20:511-5.
20. Ferreira ML, Ferreira PH, Latimer J, Herbert R, Maher CG. Efficacy of spinal manipulative therapy for low back pain of less than three months' duration. J Manipulative Physiol Ther. 2003;26:593-601.
21. Bronfort G, Haas M, Evans RL, Bouter LM. Efficacy of spinal manipulation and mobilization for low back pain and neck pain: a systematic review and best evidence synthesis. Spine J. 2004;4:335-56.
22. Chou R, Huffman LH, American Pain Society, American College of Physicians. Nonpharmacologic therapies for acute and chronic low back pain: a review of the evidence for an American Pain Society/American College of Physicians clinical practice guideline. Ann Intern Med. 2007;147:492-504.
23. Furlan AD, Yazdi F, Tsertsvadze A, et al. A systematic review and meta-analysis of efficacy, cost-effectiveness, and safety of selected complementary and alternative medicine for neck and low-back pain. Evid Based Complement Alternat Med. 2012:953139.
24. Burton AK, Tillotson KM, Cleary J. Single-blind randomised controlled trial of chemonucleolysis and manipulation in the treatment of symptomatic lumbar disc herniation. Eur Spine J. 2000;9:202-7.
25. Liu J, Zhang S. Treatment of protrusion of lumbar intervertebral disc by pulling and turning manipulations. J Tradit Chin Med. 2000;20:195-7.
113
26. Santilli V, Beghi E, Finucci S. Chiropractic manipulation in the treatment of acute back pain and sciatica with disc protrusion: a randomized double-blind clinical trial of active and simulated spinal manipulations. Spine J. 2006;6:131-7.
27. McMorland G, Suter E, Casha S, du Plessis SJ, Hurlbert RJ. Manipulation or microdiskectomy for sciatica? A prospective randomized clinical study. J Manipulative Physiol Ther. 2010;33:576-84.
28. Bronfort G, Hondras MA, Schulz CA, Evans RL, Long CR, Grimm R. Spinal manipulation and home exercise with advice for subacute and chronic back-related leg pain: a trial with adaptive allocation. Ann Intern Med. 2014;161:381-91.
29. Hebert JJ, Stomski NJ, French SD, Rubinstein SM. Serious adverse events and spinal manipulative therapy of the low back region: a systematic review of cases. J Manipulative Physiol Ther. 2013.
30. Maclure M, Fireman B, Nelson JC, et al. When should case-only designs be used for safety monitoring of medical products? Pharmacoepidemiol Drug Saf. 2012;21 Suppl 1:50-61.
31. Salas M, Hofman A, Stricker BH. Confounding by indication: an example of variation in the use of epidemiologic terminology. Am J Epidemiol. 1999;149:981-3.
32. Crawford CM, Hannan RF. Management of acute lumbar disk herniation initially presenting as mechanical low back pain. J Manipulative Physiol Ther. 1999;22:235-44.
33. Faciszewski T, Broste SK, Fardon D. Quality of data regarding diagnoses of spinal disorders in administrative databases. A multicenter study. J Bone Joint Surg Am. 1997;79:1481-8.
34. Peul WC, van den Hout WB, Brand R, Thomeer RT, Koes BW, Leiden-The Hague Spine Intervention Prognostic Study G. Prolonged conservative care versus early surgery in patients with sciatica caused by lumbar disc herniation: two year results of a randomised controlled trial. BMJ. 2008;336:1355-8.
35. Peul WC, van Houwelingen HC, van der Hout WB, et al. Prolonged conservative treatment or 'early' surgery in sciatica caused by a lumbar disc herniation: rationale and design of a randomized trial [ISRCT 26872154]. BMC Musculoskelet Disord. 2005;6:8.
36. Chan BTB, Schull MJ, Schultz SE. Emergency Department Services in Ontario. Toronto: Institute for Clinical Evaluative Services; 2001.
37. Iron K, Jaakkimainen L, Rothwell DM, Ping L, Laupacis A. Investigation of acute lower back pain in Ontario: Are guidelines being followed? Toronto, Ontario: Institute for Clinical Evaluative Sciences; 2004.
38. Jaakkimainen L, Klein-Geltink JE, Guttmann A, et al. Indicators of primary care based on administrative data. In: Jaakkimainen L, Upshur RE, Klein-Geltink JE, et al., eds. Primary care in Ontario: ICES Atlas. Toronto: Institute for Clinical Evaluative Sciences; 2006.
39. Jaakkimainen L, Schultz SE, Klein-Geltink JE, Thiruchelvam D, Kopp A. Ambulatory Physician Care for Adults. In: Jaakkimainen L, Upshur RE, Klein-Geltink JE, et al., eds. Primary care in Ontario: ICES Atlas. Toronto: Institute for Clinical Evaluative Sciences; 2006.
40. Juurlink D, Preyra C, Croxford R, et al. Canadian Institute for Health Information Discharge Abstract Database: A Validation Study. Toronto: Institute for Clinical Evaluative Sciences; 2006.
114
41. Côté P, Boyle E, Choi S, Rampersaud YR, Hincapié CA, Cassidy JD. Chiropractic care and the risk for cauda equina syndrome: A population-based case-crossover study. In Progress. 2015.
42. Whitaker HJ, Hocine MN, Farrington CP. The methodology of self-controlled case series studies. Stat Methods Med Res. 2009;18:7-26.
43. Whitaker HJ, Farrington CP, Spiessens B, Musonda P. Tutorial in biostatistics: the self-controlled case series method. Stat Med. 2006;25:1768-97.
44. Farrington CP, Whitaker HJ, Hocine MN. Case series analysis for censored, perturbed, or curtailed post-event exposures. Biostatistics. 2009;10:3-16.
45. Hua W, Sun G, Dodd CN, et al. A simulation study to compare three self-controlled case series approaches: correction for violation of assumption and evaluation of bias. Pharmacoepidemiol Drug Saf. 2013;22:819-25.
46. Kunhert R, Whitaker HJ. adSCCS: Case series analysis for censored, perturbed or curtailed post-event exposures [computer program]. R package version 1.5. 2012.
47. R: A language and environment for statistical computing [computer program]. Vienna, Austria: R Foundation for Statistical Computing; 2012.
48. Cassidy JD, Boyle E, Côté P, et al. Risk of vertebrobasilar stroke and chiropractic care: results of a population-based case-control and case-crossover study. Spine. 2008;33:S176-83.
49. Hartvigsen L, Kongsted A, Hestbaek L. Clinical examination findings as prognostic factors in low back pain: a systematic review of the literature. Chiropr Man Therap. 2015;23:13.
50. Al Nezari NH, Schneiders AG, Hendrick PA. Neurological examination of the peripheral nervous system to diagnose lumbar spinal disc herniation with suspected radiculopathy: a systematic review and meta-analysis. Spine J. 2013;13:657-74.
51. Koes BW, van Tulder M, Lin CW, Macedo LG, McAuley J, Maher C. An updated overview of clinical guidelines for the management of non-specific low back pain in primary care. Eur Spine J. 2010;19:2075-94.
52. Dagenais S, Tricco AC, Haldeman S. Synthesis of recommendations for the assessment and management of low back pain from recent clinical practice guidelines. Spine J. 2010;10:514-29.
115
Chapter 5
Summary and Synthesis 5
The research reported in this thesis aimed to advance knowledge of the risk for acute
lumbar disc herniation (LDH) following chiropractic care through a mixed methods
scientific approach. First, we synthesized the current evidence on the incidence and
determinants of LDH with radiculopathy in adults. Second, we described and quantified
clinicians’ beliefs about the risk for acute LDH associated with chiropractic spinal
manipulation treatment. Finally, we investigated associations between chiropractic care
and incident acute LDH with early surgical intervention. In this chapter, the main
findings of our studies are summarized, synthesized and placed in broader clinical,
research and public health context. We elaborate on the implications and limitations of
our work and provide guidance for future research.
5.1 Main Findings
In Chapter 2, the incidence and risk factors of LDH with radiculopathy in adults were
examined through a best evidence synthesis of the literature. This chapter provides
evidence of the varying quality and heterogeneity of the literature on the epidemiology
of LDH. The annual incidence ranges roughly between 0.1% and 10%, and is highly
dependent on case definition and source population. Risk factors varied and our
systematic review points to a multifaceted etiology involving relationships between
individual, behavioural and work-related characteristics. We call for the development of
standardized case and surveillance definitions that validly classify the clinical spectrum,
and the need for further well-designed descriptive and analytic studies investigating the
epidemiology of LDH with radiculopathy.
In Chapter 3, we determined clinicians’ beliefs regarding the effect of chiropractic spinal
manipulation on the risk of developing acute LDH and acute LDH that necessitates early
surgical management using a belief elicitation study. Beliefs were diverse, with
interesting and important interprofessional differences, indicating the presence of
community uncertainty. Optimistic clinicians believed that treatment of acute low back
pain with chiropractic spinal manipulation is associated with a 58% decreased risk for
116
acute LDH, while pessimistic clinicians believed that chiropractic care is associated with
a 66% increased risk for surgically managed acute LDH. Our probability distributions
quantify and describe varying clinician beliefs and could be used as prior probabilities in
Bayesian models estimating this exposure-outcome association.
In Chapter 4, the association between chiropractic care and incident acute LDH was
investigated and compared to the association between primary care physician care and
incident early surgically managed acute LDH. Using Ontario administrative health data
and a self-controlled case series design, we found strong positive associations between
both chiropractic and primary care physician visits, and acute LDH with early surgical
intervention. Compared with the risk associated with primary care physician visits, there
was no increase in the risk associated with chiropractic visits. We believe that this
suggests that patients with prodromal back pain related to a developing disc herniation
seek health care from both chiropractors and primary care physicians before full clinical
expression of acute LDH that is subsequently surgically managed.
5.2 Synthesis and Relevance of Findings
Epidemiology of Lumbar Disc Herniation With Radiculopathy: Our systematic
review highlights the many existing gaps in knowledge of the epidemiology of LDH with
radiculopathy in adults. The incidence of hospital-treated LDH with radiculopathy in the
general population ranged from 0.2 to 1.3 per 1,000 person-years, but little is known
about the incidence of clinically defined LDH with radiculopathy in the general
population. As a point of reference, general population studies involving clinical
assessment have estimated the point prevalence of lumbar disc syndrome at about
5%,1,2 varying by sex and age. Our study shows that the incidence of LDH with
radiculopathy is higher among worker populations, ranging from 6.2 per 1,000 person-
years (0.6%) among female nurses in the US to 93 per 1,000 person-years (9.3%)
among forest industry workers in Finland. One other systematic review has reported
even more marked variation in prevalence estimates of sciatica, ranging from 1.6% in
the general population to 43% in a selected working population.3
This study is the first to systematically gather, appraise and synthesize information on
the incidence and determinants of LDH with radiculopathy. It is difficult to draw
117
conclusions about risk factors. These varied and included individual, behavioural and
work-related variables. The best available evidence points to the following modifiable
targets for possible risk reduction: higher BMI, cardiovascular risk factors (in women),
smoking, greater cumulative occupational lumbar load by forward bending postures and
manual materials handling, higher levels of perceived risk of work injury, lower job
control or decision latitude at work, regular or irregular three-shift work or regular night
work in women, and increased time pressure at work.
There is an important need for consensus on clinical and surveillance definitions and
subsequent validation of methods of ascertaining cases. It is encouraging to note the
developing literature in this area,4-10 although difficulties with clinical diagnosis,
prognosis and classification remain. Until there is some consistency of definitions and
their appropriate validation, studies of LDH with radiculopathy will remain so
heterogeneous that comparing incidence estimates and risk factors will continue to
prove challenging.
Our study helps clarify the epidemiology of this important condition and advance
understanding of the risk factors for LDH with radiculopathy. It is a first step in clarifying
what is currently known about the incidence and determinants, so as to inform future
work and understanding. The review advances knowledge for the entire back pain
community including patients, health care practitioners, and back pain researchers.
Clinician Beliefs Regarding Chiropractic Spinal Manipulation and the Risk For
Acute Lumbar Disc Herniation: We have shown that clinician beliefs about
chiropractic treatment as a risk factor for acute symptomatic LDH vary, indicating
uncertainty within a community of back pain clinician experts. Our study supports the
notion of distinct “pools” of belief among chiropractors, family physicians and spine
surgeons, with some clinicians believing in a beneficial effect of chiropractic care and
others mainly expressing a belief of no effect with a chance of harm.
We hypothesize that differences in beliefs might be related to: (1) interprofessional
differences in the interpretation of a very limited body of evidence; (2) the influence of
varying patient-mix and clinical experience across the health professionals; (3) learned
biases as a result of the professionalization and education of health care professionals.
There is evidence from the clinical guideline development literature that clinician
118
judgments (i.e., beliefs) can diverge from research evidence for reasons related to
clinical experience and a weak or limited body of literature.11 In addition, others have
reported on the not surprising tendency for specialists to express overconfidence in the
effectiveness and appropriateness of their intervention.12,13
Using a Bayesian inferential paradigm, elicited beliefs in the form of prior probability
distributions can be used to augment limited epidemiologic data.14-18 The probability
distributions from our study create an opportunity for a Bayesian approach to models
estimating risk, in which informative clinical prior beliefs about the association can be
explicitly and transparently specified as Bayesian priors. This view has been proposed
as a way forward in the field of health services research.19,20 We believe it is particularly
reasonable in the context of investigating contentious potential treatment-related rare
serious adverse events in the absence of compelling scientific data and evidence.
Our study is the first to describe and evaluate clinician beliefs about the association
between chiropractic spinal manipulation and the development of acute LDH. With
respect to clinical and public health significance, this study represents an important first
step in the process of understanding current perceptions about the safety of chiropractic
care. Our hope is that it facilitates interdisciplinary educational initiatives among the
various health professions, as well as knowledge translation with patients, policymakers
and other relevant stakeholders with an interest in the safety of chiropractic care.
Association Between Chiropractic Care and Risk For Acute Lumbar Disc
Herniation: Our study suggests that the risk for acute LDH with early surgery following
chiropractic care is not greater than the baseline risk for acute LDH with early surgery
associated with primary care physician care. We identified 195 cases of early surgically
managed acute LDH in a population of more than 100 million person-years. Using a
self-controlled case series study design, strong positive associations between both
chiropractic and primary care physician visits, and acute LDH were found.
Our study is the first epidemiologic investigation of the association between chiropractic
care and acute LDH that is managed with early surgery. Our findings should be
interpreted cautiously and considered within clinical context. Persons in the early
phases of a symptomatic LDH often complain of back pain, followed by leg pain or
radiculopathy.21,22 With poorer prognosis and quality of life, they have greater pain
119
severity and incur more work loss, medication use, primary health care use and surgery
than those with uncomplicated low back pain.23-27 At different points in time along the
course of their developing disc herniation they may seek health care. If chiropractic
treatment occurs before the disc herniation progresses to the presence of radiculopathy
and is thus diagnosed, then the treatment itself could be erroneously blamed for
causing the disc herniation, since the patient may have progressed regardless of care.
Given our findings of strong positive associations with no excess risk associated with
chiropractic care, we believe that this suggests that patients with prodromal back pain
related to a developing disc herniation seek health care from both chiropractors and
primary care physicians before full clinical expression of acute LDH that subsequently
undergoes early surgical intervention. This is the current best available epidemiologic
assessment of the association between chiropractic care and acute LDH with early
surgical intervention.
5.3 Implications
Clinical Implications: An accurate evaluation of the value of any health care treatment
requires knowledge of both clinical and cost effectiveness, as well as information on the
treatment’s safety profile including valid estimates of the risk for serious adverse events
and the magnitude of risk compared to other treatment alternatives. Our finding of no
excess risk for acute LDH associated with chiropractic care compared to primary
medical care suggests that spinal manipulation should not be considered
contraindicated for the treatment of low back pain. This conclusion, however, should be
qualified by our inability to exclude spinal manipulation as a possible cause of some,
albeit likely very few cases of acute LDH. Nonetheless, in light of the RCT evidence
showing benefit of spinal manipulation for low back pain and LDH with radiculopathy
and our finding of no excess risk associated with chiropractic care, we believe spinal
manipulation should be considered a viable treatment option for these common
disorders. During the clinical management of patients with low back pain with or without
radiculopathy, attention should be paid to presenting signs and symptoms that may
indicate a developing disc herniation and the treatment plan adjusted with the aim of
maximizing potential benefit while reducing potential harm or exacerbation.
120
Methodological Implications: The belief elicitation methodology that we applied to
describe and quantify clinician beliefs is pragmatic, reproducible and not unique to this
clinical area. This method could be translated to other clinical research questions
relevant to the safety profile of chiropractic care (e.g., beliefs and uncertainty
surrounding chiropractic neck manipulation) and generalized to other disciplines as well.
Our belief elicitation method could be used by other researchers who wish to elicit and
quantify belief about risk or treatment effects to develop informative clinical priors for
inclusion in Bayesian models.
The self-controlled case series method that we used is useful for assessing risk and
safety not only for chiropractic care, but also for all types of different care. Given the
dearth of evidence of chiropractic’s safety profile, this method could be used to assess
associations between chiropractic care and cauda equine syndrome,28 and undertake
replication studies to assess the consistency of associations between chiropractic care
and vertebrobasilar stroke examined previously.29 With respect to risk assessments of
other health care, there is already a rich literature of the usefulness of this method.30-32
Finally, our development work for the application of the self-controlled case series
method on the R platform33 may prove useful to other researchers wanting to carry out a
self-controlled case series analysis using the R platform.
Policy Implications: In our self-controlled case series study, we used administrative
health data from the Ontario Ministry of Health and Long-Term Care to carry out our
epidemiologic assessment of chiropractic care. Regrettably, the chiropractic billing data
are no longer available for current or future health services research programs. With
the delisting of chiropractic services from the Ontario Health Insurance Plan (OHIP) as
of December 1, 2004, data collection capturing chiropractic visits and services across
the province ended. Given the current fiscal challenges faced by the Ontario health
system, it is unlikely that chiropractic care will be relisted for coverage through OHIP.
Nonetheless, it would behoove the provincial, national and scientific leadership of the
chiropractic profession with the support and assistance of the provincial and federal
health ministries to attempt to develop a high-quality database that can begin to gather
important clinical practice data once again to facilitate patient-centered research and
improved patient outcomes. It is encouraging to note recent developments on this front,
with the creation of a Canadian Chiropractic Practice-Based Research Network.34
121
Our belief elicitation study represents an important step in the process of understanding
current perceptions about the safety of chiropractic care. This study uses a
characteristic of the Bayesian paradigm—acknowledgement of subjectivity and
context19—as part of the rationale for describing and quantifying clinicians’ beliefs as
probability distributions. The use of subjective priors has a tradition of controversy.
Some object to their use;35 others see value in their explicit and transparent
specification.36 Bayesian methods allow for the possibility that the conclusions of an
analysis may depend on who is conducting it and their available evidence and opinion;
therefore, understanding of subjectivity and context is essential.19
We believe this approach may be particularly useful in the context of investigating
contentious potential treatment-related rare serious adverse events in the absence of
compelling scientific data and evidence. We hope that our findings facilitate
interdisciplinary educational initiatives among the various health professions, and
knowledge translation with policymakers and other relevant stakeholders with an
interest in the safety of chiropractic care.
Legal Implications: There are notable medicolegal implications related to this thesis.
Disc herniation is the leading cause of malpractice claims against chiropractors in
Canada37 and the US.38 Expert opinion is often used on behalf of both plaintiffs and
defendants to provide testimony on scientific and clinical evidence and on causation. In
their recent narrative review of six LDH and cauda equine syndrome court decisions in
Canada, Boucher and Robidoux37 highlighted that all expert witnesses agreed that “no
large scale epidemiological study has yet been conducted to help understand the
relationship between disc herniation and [spinal manipulation].” Our findings fill in this
gap and will likely inform legal proceedings in this area.
5.4 Limitations
Selection Bias: Selection bias is a systematic error due to differences in characteristics
between those who take part in a study and those who do not. The clinicians that
participated in our belief elicitation study (Chapter 3) are a purposive sample. They
may not be representative of the source population of family physicians, chiropractors
and spine surgeons within 2 hours driving distance of the Greater Toronto Area. This
122
limits the external validity of our results, but does not threaten the internal validity of our
study. We hypothesize that respondents may have had more extreme beliefs than
nonrespondents (i.e., those with strong opinions might be more likely to participate), but
had no recourse to assess this possibility. We can, however, consider the implications
on our findings, if this had been the case. If nonparticipants had less extreme beliefs
than participants, there would be little change in the average belief among all
participants, since we found, on average, a neutral belief of “no effect” of chiropractic
manipulation on the risk for disc herniation. Clinician group findings would revert to less
extreme probability distributions, and this would attenuate the chiropractors’ belief of
benefit, as they held the most extreme belief of the three groups. Our study is coherent
with the Bayesian idea that there is value in explicit and transparent formulation of a
“community of priors,”19, and advances knowledge of different viewpoints on the safety
of chiropractic care.
In Chapter 2 and 4, we believe the possibility of selection bias was reduced through our
chosen study designs and research methods. In our systematic review (Chapter 2), we
systematically identified, appraised and synthesized the literature on the incidence and
determinants of LDH with radiculopathy in adults. We applied a best evidence
synthesis methodology to increase the internal validity of our review, which was a
priority given the heterogeneity and limited quality of this body of literature. In Chapter
4, we used a self-controlled case series design. The main strength of this method is
that it uses within-person comparisons, thereby eliminating selection bias and
confounding by constant (time-invariant) characteristics (e.g., smoking history, body
mass history, healthcare-seeking behaviour, average occupational lumbar load) that are
typically not well documented in administrative health data.
Information Bias: Information bias is a systematic error that results because the
information collected about or from study participants is erroneous. These errors may
result in misclassification of exposure status, outcome status, or both. Misclassification
of participants for either exposure or outcome can be either nondifferential or
differential. When the misclassification of outcome is independent of the exposure
status, and vice versa, it is called nondifferential misclassification, and tends to result in
effect estimates that are diluted or attenuated (i.e., biased toward the null value). When
the misclassification of outcome is related to exposure status, or the misclassification of
123
exposure is different for those with and without the outcome, it is differential
misclassification, and the estimate can either exaggerate or underestimate an effect.
Misclassification bias is a major limitation of using administrative health data. In
Chapter 4, we undertook a population-based data linkage study using four Ontario
health care databases. A key limitation is nondifferential misclassification of case
status. The reliability and validity of our case definition is not known. Due to the high
costs of chart abstraction, an attempt to validate the case definition was not undertaken.
Nonetheless, the quality of our case selection was bolstered by a number of strategies.
First, a prespecified 4-step algorithm was developed with careful consideration of the
target outcome of acute LDH with incident early surgical intervention. We reviewed the
methodologies used by previous Ontario health services reports of acute low back
pain,39 primary care services,40,41 emergency department services,42 and the Discharge
Abstract Database,43 and multiple experts in epidemiology and spine surgery were
consulted. Second, we prespecified sensitivity analyses of the case definition to
examine the effect of clinically reasoned variations in our approach to case selection.
Despite our best efforts, it is possible that our case definition was either overinclusive or
underinclusive. However, as the misclassification is nondifferential, this may have lead
to attenuation of our estimates, but would not change our conclusions. Nondifferential
misclassification of the exposure measures was also likely. The reliability and validity of
the codes to classify lumbar spine visits to chiropractors and primary care physicians is
not known. Again, the net result is that risk estimates for acute LDH with early surgery
may be biased toward the null, but this does not affect the conclusions of the study.
Some chiropractic visits (for persons that exhausted the $150 of annual OHIP coverage)
may not have been captured in the OHIP data. However, the proportion of missing
visits is likely small, since the majority of patients have fewer than 15 chiropractic visits
per year.44
In Chapter 3, our findings on clinician beliefs regarding the association between
chiropractic spinal manipulation and acute LDH are preliminary. The reliability and
validity of our approach is not known. We grounded our approach in current best
practice in elicitation, a belief elicitation method45 that has been validated in
rheumatology clinicians, and consulted experts in belief elicitation and the SHELF46
124
elicitation framework in the development of our elicitation interview questionnaire and
methods.
Reverse-Causality Bias: Protopathic bias is a term used if the first (prodromal) signs
and symptoms of a condition are the reason for initial use of a treatment.47-49 This is a
type of reverse-causality bias due to processes that occur before a diagnosed or
measured outcome event.31,49 In Chapter 4, we compared estimates of our focal
association of interest (chiropractic care and acute LDH with early surgery) with
estimates of association between primary care physician care and acute LDH with early
surgery to assess the potential role of protopathic bias. Since patients commonly see
primary care physicians for back pain and these visits are unlikely to cause disc
herniation, we hypothesized that an observed association between primary care
physician visits and acute LDH could be attributed to care-seeking for initial symptoms
of LDH. The consistency of our findings in multiple sensitivity analyses of these
associations gave support to our overall conclusion that the positive associations are
likely due to patients with prodromal back pain from a developing disc herniation
seeking health care before full clinical expression of acute LDH (i.e., protopathic bias).
Chapter 4 also presented us with the additional challenge of assessing reverse-
causality due to event-day exposure visits for both chiropractors and primary care
physicians, and the inability to firmly establish the temporal sequence of a health care
visit (to either chiropractors or primary care physicians) occurring before the ED LDH
visit on the event index date. We applied a statistical solution of specifying a one-day
lag on the exposure visits; thereby removing event-day exposure visits from our
estimates. This sensitivity analysis resulted in attenuation of our estimates, but did not
change our conclusions.
Confounding: Confounding refers to the mixing of effects when an exposure effect is
mixed together with the effect of another variable (confounder) leading to systematic
error in the measurement of the exposure-outcome association. Our studies in
Chapters 2 and 3 are descriptive in nature, so that confounding is not an issue. In
Chapter 4, the main strength of the chosen study design was that comparisons were
within-person, thereby eliminating between-person confounding as well as within-person
confounding due to characteristics (within an individual) that were likely not to change
during the observation period of the analysis. This is an excellent design for assessing
125
potential treatment-induced harm on a population level because of the concern about
potential confounding by factors not recorded in administrative health databases. By
structuring the analysis so that each person serves as their own control, we eliminated
confounding by time-invariant characteristics, such as, sex, location, genetics,
underlying state of health, tendency to seek professional health care, average physical
activity, long-term diet, habitual health behaviours, long-past health events such as
illnesses and injuries, occupation, social support, ethnicity, smoking history, and body
mass history. Nonetheless, the possibility of residual confounding by within-person
factors that do vary with time is still a possibility. For example, a high instantaneous
occupational lumbar load (confounding factor) leading to prodromal back pain and
health-care seeking, which is eventually diagnosed as LDH.
Random Error: Random error is the variability in the data that cannot be explained by
systematic error (bias). This is related to sample size and is reflected in the precision of
the estimates generated by the data. When the sample size of the study is large, the
width of the confidence intervals for estimates from those data is smaller than those
generated from a smaller study. In Chapter 4, the incidence rate ratios (IRR) are based
on small numbers, and the confidence intervals around the estimates are large. This
makes the interpretation of the IRR a little more difficult. Even though many of the
chiropractic care point estimates were smaller than those for primary care physician
care, interpretation and comparison of the associations was informed more by the
relationship of confidence intervals (overlapping vs. nonoverlapping interval estimates)
and thus lead to our conclusion of similar positive associations for both types of health
care visits and acute LDH.
5.5 Future Research
Our work raises some research questions for future studies. Using a Bayesian
approach to health services research, one logical avenue of inquiry would be to
undertake a Bayesian analysis of the association between chiropractic care and acute
LDH with early surgical intervention. Our objective would be to integrate the probability
distributions from Chapter 3 (Bayesian priors) with the findings of Chapter 4 (the
likelihood), to estimate the posterior probability of the association using Bayes’s
theorem. This would allow us to communicate direct probabilities for acute LDH with
126
early surgery given different prior opinions of the clinicians we elicited beliefs from in
Chapter 3 (e.g., optimists vs. pessimists). As an example, Figure 5.1 provides a
preliminary plot of the results of this Bayesian analysis using the mean and standard
deviation of bootstrap log incidence rate ratios for chiropractic versus primary care
physician care from Chapter 4 to form a normal likelihood, and normal approximations
of the optimistic and pessimistic probability distributions from Chapter 3 as prior
distributions on the log ratio of these incidence rate ratios.
Figure 5.1. Preliminary Bayesian triplots of risk for acute lumbar disc herniation with early surgery
associated with chiropractic care.
The knowledge translation value of this approach to interprofessional education and
communication, and open and transparent treatment of prior beliefs would be interesting
to explore and presents a possible area of novel and innovative scientific study.
127
Given the paucity of evidence and available data that can be used to further examine
the association between chiropractic care and the risk for serious adverse events, it
would seem that this line of inquiry may be limited, at least on the epidemiologic front.
Many research questions with respect to perceptions and beliefs regarding the safety of
chiropractic spinal manipulation can be formulated: What are patients’, clinicians’ and
other stakeholders’ beliefs regarding spinal manipulation as a risk factor for disc
herniation or stroke? What are the determinants of beliefs? Can beliefs change and be
“updated” (as Bayes’s theorem proposes) based on limited evidence? Developing web-
based approaches to belief elicitation may allow future elicitation studies to determine
the beliefs of larger samples of participants in valid and reliable ways, but this is a fairly
nascent area of research.50
With respect to lumbar radiculopathy and symptomatic LDH, many epidemiologic and
clinical questions remain to be addressed as our Chapter 2 highlights. What is the
incidence of and risk factors for lumbar radiculopathy in the general population? Are
cardiovascular risk factors (diabetes, high cholesterol, hypertension, and a family history
of coronary heart disease) determinants of LDH in men? What are the prognostic
factors that determine the course of radiculopathy and symptomatic disc herniation?
Furthermore, there is a great need to develop and assess therapeutic approaches that
could benefit those patients with lumbar radiculopathy and symptomatic disc herniation.
These measures might include conventional clinical therapies, but consideration should
also be given to innovative psychosocial and work-place interventions, given our best
evidence synthesis in Chapter 2.
5.6 Conclusion
In this dissertation, we have met our goal of advancing knowledge of the risk for acute
LDH following chiropractic care through an integrated mixed methods approach. Using
the principles and methods of best evidence synthesis, belief elicitation and
epidemiologic investigation, we have successfully integrated methods that use multiple
types of information of an exposure-outcome association with very limited evidence and
great uncertainty, to make more valid estimates of association and perceptions, and
reduce uncertainty. Our work provides evidence of the varying quality and
heterogeneity of the literature on the epidemiology of LDH with radiculopathy; shows
128
that clinician beliefs about the association between chiropractic care and acute LDH are
diverse, with important interprofessional differences, indicating the presence of
community uncertainty; and reports no evidence of excess risk for acute LDH with early
surgical intervention associated with chiropractic care compared with primary medical
care.
References 1. Heliövaara M, Impivaara O, Sievers K, et al. Lumbar disc syndrome in Finland. J
Epidemiol Community Health. 1987;41:251-8. 2. Kaila-Kangas L, Leino-Arjas P, Karppinen J, Viikari-Juntura E, Nykyri E,
Heliovaara M. History of physical work exposures and clinically diagnosed sciatica among working and nonworking Finns aged 30 to 64. Spine. 2009;34:964-9.
3. Konstantinou K, Dunn KM. Sciatica: review of epidemiological studies and prevalence estimates. Spine. 2008;33:2464-72.
4. Lin CW, Verwoerd AJ, Maher CG, et al. How is radiating leg pain defined in randomized controlled trials of conservative treatments in primary care? A systematic review. Eur J Pain. 2014;18:455-64.
5. Al Nezari NH, Schneiders AG, Hendrick PA. Neurological examination of the peripheral nervous system to diagnose lumbar spinal disc herniation with suspected radiculopathy: a systematic review and meta-analysis. Spine J. 2013;13:657-74.
6. Genevay S, Atlas SJ, Katz JN. Variation in eligibility criteria from studies of radiculopathy due to a herniated disc and of neurogenic claudication due to lumbar spinal stenosis: a structured literature review. Spine. 2010;35:803-11.
7. Konstantinou K, Hider SL, Vogel S, Beardmore R, Somerville S. Development of an assessment schedule for patients with low back-associated leg pain in primary care: a Delphi consensus study. Eur Spine J. 2012;21:1241-9.
8. Kreiner DS, Hwang SW, Easa JE, et al. An evidence-based clinical guideline for the diagnosis and treatment of lumbar disc herniation with radiculopathy. Spine J. 2014;14:180-91.
9. van der Windt DAWM, Simons E, Riphagen II, et al. Physical examination for lumbar radiculopathy due to disc herniation in patients with low-back pain. Cochrane Database Syst Rev. 2010:CD007431.
10. Haswell K, Gilmour J, Moore B. Clinical decision rules for identification of low back pain patients with neurologic involvement in primary care. Spine (Phila Pa 1976). 2008;33:68-73.
11. Raine R, Sanderson C, Hutchings A, Carter S, Larkin K, Black N. An experimental study of determinants of group judgments in clinical guideline development. The Lancet. 2004;364:429-37.
12. Fitch K, Lázaro P, Aguilar MD, Martín Y, Bernstein SJ. Physician recommendations for coronary revascularization. The European Journal of Public Health. 1999;9:181-7.
13. Herrin J, Etchason JA, Kahan JP, Brook RH, Ballard DJ. Effect of panel composition on physician ratings of appropriateness of abdominal aortic
129
aneurysm surgery: elucidating differences between multispecialty panel results and specialty society recommendations. Health Policy. 1997;42:67-81.
14. Billingham L, Malottki K, Steven N. Small sample sizes in clinical trials: a statistician’s perspective. Clinical Investigation. 2012;2:655-7.
15. Chaloner K, Rhame FS. Quantifying and documenting prior beliefs in clinical trials. Stat Med. 2001;20:581-600.
16. Dunson DB. Commentary: practical advantages of Bayesian analysis of epidemiologic data. Am J Epidemiol. 2001;153:1222-6.
17. O'Hagan A, Buck CE, Daneshkhah A, et al. Uncertain judgements: eliciting experts' probabilities. West Sussex: John Wiley & Sons Ltd; 2006.
18. Lilford RJ, Thornton JG, Braunholtz D. Clinical trials and rare diseases: a way out of a conundrum. BMJ. 1995;311:1621-5.
19. Spiegelhalter DJ. Incorporating Bayesian ideas into health-care evaluation. Statistical Science. 2004;19:156-74.
20. Spiegelhalter DJ, Myles JP, Jones DR, Abrams KR. Methods in health service research. An introduction to bayesian methods in health technology assessment. BMJ. 1999;319:508-12.
21. Deyo RA, Loeser JD, Bigos SJ. Herniated lumbar intervertebral disk. Ann Intern Med. 1990;112:598-603.
22. Weber H. The natural history of disc herniation and the influence of intervention. Spine. 1994;19:2234-8; discussion 3.
23. Hill JC, Konstantinou K, Egbewale BE, Dunn KM, Lewis M, van der Windt D. Clinical outcomes among low back pain consulters with referred leg pain in primary care. Spine (Phila Pa 1976). 2011;36:2168-75.
24. Kent PM, Keating JL. Can we predict poor recovery from recent-onset nonspecific low back pain? A systematic review. Man Ther. 2008;13:12-28.
25. Kongsted A, Kent P, Albert H, Jensen TS, Manniche C. Patients with low back pain differ from those who also have leg pain or signs of nerve root involvement - a cross-sectional study. BMC Musculoskelet Disord. 2012;13:236.
26. Konstantinou K, Hider SL, Jordan JL, Lewis M, Dunn KM, Hay EM. The impact of low back-related leg pain on outcomes as compared with low back pain alone: a systematic review of the literature. The Clinical journal of pain. 2013;29:644-54.
27. Selim AJ, Ren XS, Fincke G, et al. The importance of radiating leg pain in assessing health outcomes among patients with low back pain. Results from the Veterans Health Study. Spine (Phila Pa 1976). 1998;23:470-4.
28. Côté P, Boyle E, Choi S, Rampersaud YR, Hincapié CA, Cassidy JD. Chiropractic care and the risk for cauda equina syndrome: A population-based case-crossover study. In Progress. 2015.
29. Cassidy JD, Boyle E, Côté P, et al. Risk of vertebrobasilar stroke and chiropractic care: results of a population-based case-control and case-crossover study. Spine. 2008;33:S176-83.
30. Weldeselassie YG, Whitaker HJ, Farrington CP. Use of the self-controlled case-series method in vaccine safety studies: review and recommendations for best practice. Epidemiol Infect. 2011;139:1805-17.
31. Maclure M, Fireman B, Nelson JC, et al. When should case-only designs be used for safety monitoring of medical products? Pharmacoepidemiol Drug Saf. 2012;21 Suppl 1:50-61.
32. Whitaker HJ. The self-controlled case series method website. 2011; http://statistics.open.ac.uk/sccs. Accessed 2015-May-15.
130
33. R: A language and environment for statistical computing [computer program]. Vienna, Austria: R Foundation for Statistical Computing; 2012.
34. Bussières A, Côté P, French S, et al. Creating a Chiropractic Practice-Based Research Network (PBRN): Enhancing the management of musculoskeletal care. The Journal of the Canadian Chiropractic Association. 2014;58:8-15.
35. Gelman A. Objections to Bayesian statistics. 2008:445-9. 36. Diamond GA, Kaul S. Prior convictions: bayesian approaches to the analysis and
interpretation of clinical megatrials. Journal of the American College of Cardiology. 2004;43:1929-39.
37. Boucher P, Robidoux S. Lumbar disc herniation and cauda equina syndrome following spinal manipulative therapy: a review of six court decisions in Canada. Journal of forensic and legal medicine. 2014;22:159-69.
38. Jagbandhansingh MP. Most common causes of chiropractic malpractice lawsuits. J Manipulative Physiol Ther. 1997;20:60-4.
39. Iron K, Jaakkimainen L, Rothwell DM, Ping L, Laupacis A. Investigation of acute lower back pain in Ontario: Are guidelines being followed? Toronto, Ontario: Institute for Clinical Evaluative Sciences; 2004.
40. Jaakkimainen L, Klein-Geltink JE, Guttmann A, et al. Indicators of primary care based on administrative data. In: Jaakkimainen L, Upshur RE, Klein-Geltink JE, et al., eds. Primary care in Ontario: ICES Atlas. Toronto: Institute for Clinical Evaluative Sciences; 2006.
41. Jaakkimainen L, Schultz SE, Klein-Geltink JE, Thiruchelvam D, Kopp A. Ambulatory Physician Care for Adults. In: Jaakkimainen L, Upshur RE, Klein-Geltink JE, et al., eds. Primary care in Ontario: ICES Atlas. Toronto: Institute for Clinical Evaluative Sciences; 2006.
42. Chan BTB, Schull MJ, Schultz SE. Emergency Department Services in Ontario. Toronto: Institute for Clinical Evaluative Services; 2001.
43. Juurlink D, Preyra C, Croxford R, et al. Canadian Institute for Health Information Discharge Abstract Database: A Validation Study. Toronto: Institute for Clinical Evaluative Sciences; 2006.
44. Rothwell DM, Bondy SJ, Williams JI. Chiropractic manipulation and stroke: a population-based case-control study. Stroke. 2001;32:1054-60.
45. Johnson SR, Tomlinson GA, Hawker GA, Granton JT, Grosbein HA, Feldman BM. A valid and reliable belief elicitation method for Bayesian priors. J Clin Epidemiol. 2010;63:370-83.
46. SHELF: the Sheffield Elicitation Framework (version 2.0) [computer program]. Version 2.0. Sheffield, UK: School of Mathematics and Statistics, University of Sheffield; 2010.
47. Jick H, Vessey MP. Case-control studies in the evaluation of drug-induced illness. Am J Epidemiol. 1978;107:1-7.
48. Lanes SF, Delzell E, Dreyer NA, Rothman KJ. Analgesics and kidney disease. Int J Epidemiol. 1986;15:454-5.
49. Salas M, Hofman A, Stricker BH. Confounding by indication: an example of variation in the use of epidemiologic terminology. Am J Epidemiol. 1999;149:981-3.
50. Morris DE, Oakley JE, Crowe JA. A web-based tool for eliciting probability distributions from experts. Environmental Modelling & Software. 2014;52:1-4.
131
Chapter 6
Appendices
Appendix A. Risk of bias assessment summary (Chapter 1) 132
Appendix B. Comparison of the case-crossover (CCO) and self-controlled
case series (SCCS) designs (Chapter 1)
133
Appendix C. Systematic review search strategies (Chapter 2) 134
Appendix D. Detailed evidence tables for systematic review (Chapter 2) 143
Appendix E. Belief elicitation questionnaire (Chapter 3) 168
Appendix F. Belief elicitation script (Chapter 3) 172
Appendix G. Belief elicitation computer protocol (Chapter 3) 178
Appendix H. Case and exposure definition tables for self-controlled case
series study (Chapter 4)
179
132
Appendix A. Risk of bias assessment summary (Chapter 1)
Ade
quat
e se
quen
ce g
ener
atio
n?
Allo
catio
n co
ncea
lmen
t?
Blin
ding
?
Inco
mpl
ete
outc
ome
data
add
ress
ed?
Free
of s
elec
tive
repo
rting
?
Free
of o
ther
bia
s?
Santilli et al7 Y Y Y Y Y U
McMorland et al8 Y Y U Y Y U
Bronfort et al9 Y Y U Y Y U
Abbreviations: Y = yes; N = no; U = unclear
* Low risk of bias assessed using the risk-of-bias assessment recommended
by the Cochrane Collaboration.92 For a study to be considered low risk of
bias none of the 6 domains recommended by Cochrane (randomization,
allocation concealment, blinding, incomplete outcome data, selective
outcome reporting, other) may be rated “no” and no more than 2 may be
rated “unclear.” Because of the inherent complexities involved in blinding
manual treatments, trials that did not blind providers or participants were
rated unclear, provided that study personnel were unlikely to influence
outcome assessments.
Figure. Risk of bias assessment summary
133
Appendix B. Comparison of the case-crossover (CCO) and self-controlled case series (SCCS) designs (Chapter 1)
Characteristic CCO SCCS
“Zero” time Outcome index date Start of observation period prior to the outcome index date
Analytical approach Matched case-control logic Matched cohort logic
Temporal perspective Data are viewed retrospectively from the outcome index date
Data are viewed prospectively from the “zero” time of the study
Statistical model Conditional logistic regression used to estimate the exposure odds ratio
Poisson regression used to estimate the incidence rate ratio
Comparison of interest Exposure status between a time period just before occurrence of the outcome (i.e., hazard period) and one or more referent time periods (i.e., control periods) prior to the hazard period when the outcome did not occur
The likelihood of the outcome occurring in a time period following exposure (i.e., “exposed” period) is compared with the likelihood of the outcome occurring in “unexposed” periods
Handling of temporal data
Time windows (i.e., hazard and control periods) are selected and exposure status is assessed only within the selected time windows
Makes use of all available temporal information without the need for selection of a comparator time window
Strengths Improved control over time-invariant confounders
Suitable for measuring transient effects of accurately recorded brief exposures (e.g., a chiropractic visit) on the immediate risk of a rare-onset disease (e.g., acute lumbar disc herniation)
Improved control over time-invariant confounders
Suitable for measuring transient effects of accurately recorded brief exposures (e.g., a chiropractic visit) on the immediate risk of a rare-onset disease (e.g., acute lumbar disc herniation)
Modified SCCS validly estimates incidence rate ratios in situations where occurrence of the outcome event affects post-event exposures
Reduced susceptibility to exposure-trend bias
Not vulnerable to overlap bias
Weaknesses Susceptible to exposure-trend bias due to the fixed ordering in time of control and case periods (i.e., control periods always precedes the hazard period
Vulnerable to overlap bias (i.e., a version of the bias that arises from choosing non-disjoint strata to partition the population in a matched case-control study)
Bias may be introduced in a standard SCCS analysis if the outcome event influences the likelihood of post-event exposures
134
Appendix C. Systematic review search strategies (Chapter 2)
Ovid MEDLINE(R) 1950 to November Week 3 2010
# Searches Results Search Type
1 Intervertebral Disk Displacement/ 14310 Advanced
2 Intervertebral Disk Displacement?.mp. 14312 Advanced
3 Intervertebral Disc Displacement?.mp. 12 Advanced
4 Inter-vertebral Disk Displacement?.mp. 0 Advanced
5 Inter-vertebral Disc Displacement?.mp. 1 Advanced
6 Sciatica/ 3930 Advanced
7 sciatica.mp. 4846 Advanced
8 ((hernia* or ruptur* or prolaps* or protru* or bulg* or slip* or displac*) adj3 (disc? or disk? or nuclei or nucleus)).mp. 17799 Advanced
9 ((extru* or sequestra* or migrat*) adj2 (disc? or disk? or nuclei or nucleus)).mp. 751 Advanced
10 (discal adj1 hernia*).mp. 162 Advanced
11 (discus adj1 hernia*).mp. 11 Advanced
12 ischialgi*.mp. 151 Advanced
13 lumboischialgi*.mp. 69 Advanced
14 lumbo-ischialgi*.mp. 19 Advanced
15 cauda equina syndrome?.mp. 783 Advanced
16 or/1-15 22004 Advanced
17 Lumbar Vertebrae/ 33587 Advanced
18 Intervertebral Disk/ 9220 Advanced
19 Lumbosacral Region/ 8354 Advanced
20 ((lumbar or lumbosacral or lumbo-sacral) adj1 (spine or region or area or vertebr* or disk? or disc?)).mp. 53048 Advanced
21 discus intervertebra*.mp. 8 Advanced
22 (spinal adj1 (disk? or disc?)).mp. 138 Advanced
23 (low back adj1 (spine or region or area or vertebr*)).mp. 45 Advanced
24 Intervertebral Disk?.mp. 21747 Advanced
25 Intervertebral Disc?.mp. 5565 Advanced
26 Inter-vertebral Disk?.mp. 4 Advanced
27 Inter-vertebral Disc?.mp. 31 Advanced
28 or/17-27 64882 Advanced
29 16 and 28 16624 Advanced
30 incidence/ 142494 Advanced
31 exp risk/ 651849 Advanced
32 exp Cohort Studies/ 803357 Advanced
33 exp Case-Control Studies/ 495517 Advanced
135
34 epidemiologic studies/ 5002 Advanced
35 incidence.mp. 470248 Advanced
36 risk factor?.mp. 549307 Advanced
37 cohort?.mp. 219800 Advanced
38 case control?.mp. 147466 Advanced
39 risk*.mp. 1189985 Advanced
40 between group*.tw. 48950 Advanced
41 relative risk*.tw. 44395 Advanced
42 etiolog*.mp. 149098 Advanced
43 exp Epidemiology/ 18044 Advanced
44 Epidemiologic Methods/ 26634 Advanced
45 epidemiolog*.mp. 250922 Advanced
46 ep.fs. 966847 Advanced
47 or/30-46 2999198 Advanced
48 29 and 47 4835 Advanced
49 limit 48 to yr="1970 -Current" 4770 Advanced
50 remove duplicates from 49 4442 Advanced
51 limit 50 to humans 4299 Advanced Ovid MEDLINE(R) In-Process & Other Non-Indexed Citations December 21, 2010
# Searches Results Search Type
1 Intervertebral Disk Displacement/ 7 Advanced
2 Intervertebral Disk Displacement?.mp. 7 Advanced
3 Intervertebral Disc Displacement?.mp. 0 Advanced
4 Inter-vertebral Disk Displacement?.mp. 0 Advanced
5 Inter-vertebral Disc Displacement?.mp. 0 Advanced
6 Sciatica/ 2 Advanced
7 sciatica.mp. 111 Advanced
8 ((hernia* or ruptur* or prolaps* or protru* or bulg* or slip* or displac*) adj3 (disc? or disk? or nuclei or nucleus)).mp. 447 Advanced
9 ((extru* or sequestra* or migrat*) adj2 (disc? or disk? or nuclei or nucleus)).mp. 32 Advanced
10 (discal adj1 hernia*).mp. 0 Advanced
11 (discus adj1 hernia*).mp. 0 Advanced
12 ischialgi*.mp. 2 Advanced
13 lumboischialgi*.mp. 4 Advanced
14 lumbo-ischialgi*.mp. 0 Advanced
15 cauda equina syndrome?.mp. 47 Advanced
16 or/1-15 581 Advanced
17 Lumbar Vertebrae/ 41 Advanced
18 Intervertebral Disk/ 13 Advanced
136
19 Lumbosacral Region/ 6 Advanced
20 ((lumbar or lumbosacral or lumbo-sacral) adj1 (spine or region or area or vertebr* or disk? or disc?)).mp. 1200 Advanced
21 discus intervertebra*.mp. 1 Advanced
22 (spinal adj1 (disk? or disc?)).mp. 13 Advanced
23 (low back adj1 (spine or region or area or vertebr*)).mp. 7 Advanced
24 Intervertebral Disk?.mp. 63 Advanced
25 Intervertebral Disc?.mp. 294 Advanced
26 Inter-vertebral Disk?.mp. 0 Advanced
27 Inter-vertebral Disc?.mp. 1 Advanced
28 or/17-27 1479 Advanced
29 16 and 28 254 Advanced
30 incidence/ 262 Advanced
31 exp risk/ 1339 Advanced
32 exp Cohort Studies/ 1342 Advanced
33 exp Case-Control Studies/ 1110 Advanced
34 epidemiologic studies/ 10 Advanced
35 incidence.mp. 17608 Advanced
36 risk factor?.mp. 13224 Advanced
37 cohort?.mp. 10007 Advanced
38 case control?.mp. 2774 Advanced
39 risk*.mp. 47523 Advanced
40 between group*.tw. 2707 Advanced
41 relative risk*.tw. 1309 Advanced
42 etiolog*.mp. 5891 Advanced
43 exp Epidemiology/ 26 Advanced
44 Epidemiologic Methods/ 20 Advanced
45 epidemiolog*.mp. 8393 Advanced
46 ep.fs. 1744 Advanced
47 or/30-46 79129 Advanced
48 29 and 47 45 Advanced EMBASE Classic+EMBASE 1947 to 2010 December 21
# Searches Results Search Type
1 exp intervertebral disk hernia/ or lumbar disk hernia/ 17656 Advanced
2 Intervertebral Disk Displacement?.mp. 622 Advanced
3 Intervertebral Disc Displacement?.mp. 11 Advanced
4 Inter-vertebral Disk Displacement?.mp. 1 Advanced
5 Inter-vertebral Disc Displacement?.mp. 1 Advanced
6 ((hernia* or ruptur* or prolaps* or protru* or bulg* or slip* or displac*) adj3 22852 Advanced
137
(disc? or disk? or nuclei or nucleus)).mp.
7 ((extru* or sequestra* or migrat*) adj2 (disc? or disk? or nuclei or nucleus)).mp. 996 Advanced
8 (discal adj1 hernia*).mp. 352 Advanced
9 (discus adj1 hernia*).mp. 24 Advanced
10 ischialgia/ 6839 Advanced
11 ischialgi*.mp. 6956 Advanced
12 lumboischialgi*.mp. 126 Advanced
13 lumbo-ischialgi*.mp. 32 Advanced
14 sciatica.mp. 4581 Advanced
15 cauda equina syndrome/ 1267 Advanced
16 cauda equina syndrome?.mp. 1702 Advanced
17 or/1-16 30028 Advanced
18 lumbar vertebra/ 14625 Advanced
19 intervertebral disk/ 9886 Advanced
20 back/ 7733 Advanced
21 ((lumbar or lumbosacral or lumbo-sacral) adj1 (spine or region or area or vertebr* or disk? or disc?)).mp. 58172 Advanced
22 discus intervertebra*.mp. 15 Advanced
23 (spinal adj1 (disk? or disc?)).mp. 192 Advanced
24 (low back adj1 (spine or region or area or vertebr*)).mp. 71 Advanced
25 Intervertebral Disk?.mp. 27270 Advanced
26 Intervertebral Disc?.mp. 8682 Advanced
27 Inter-vertebral Disk?.mp. 10 Advanced
28 Inter-vertebral Disc?.mp. 65 Advanced
29 lumbar spine/ 23300 Advanced
30 or/18-29 83215 Advanced
31 17 and 30 21422 Advanced
32 risk:.mp. or exp methodology/ or exp epidemiology/ 5090165 Advanced
33 (cohort or relative risk:).tw. 221597 Advanced
34 incidence/ or standardized incidence ratio/ 165939 Advanced
35 exp risk/ or risk factor/ 977563 Advanced
36 cohort analysis/ 91803 Advanced
37 exp case control study/ 53633 Advanced
38 epidemiology/ 164156 Advanced
39 (incidence or risk factor? or cohort? or case control).mp. 1316967 Advanced
40 etiology/ 333051 Advanced
41 epidemiolog*.mp. 410418 Advanced
42 ep.fs. 786591 Advanced
43 or/32-42 5960537 Advanced
44 31 and 43 6311 Advanced
138
45 limit 44 to yr="1970 -Current" 5974 Advanced
46 remove duplicates from 45 5879 Advanced
47 limit 46 to medline 1784 Advanced
48 46 not 47 4095 Advanced EBM Reviews - Cochrane Database of Systematic Reviews 2005 to December 2010
# Searches Results Search Type
1 [Intervertebral Disk Displacement/] 0 Advanced
2 Intervertebral Disk Displacement?.mp. 6 Advanced
3 Intervertebral Disc Displacement?.mp. 0 Advanced
4 Inter-vertebral Disk Displacement?.mp. 0 Advanced
5 Inter-vertebral Disc Displacement?.mp. 0 Advanced
6 [Sciatica/] 0 Advanced
7 sciatica.mp. 50 Advanced
8 ((hernia* or ruptur* or prolaps* or protru* or bulg* or slip* or displac*) adj3 (disc? or disk? or nuclei or nucleus)).mp. 33 Advanced
9 ((extru* or sequestra* or migrat*) adj2 (disc? or disk? or nuclei or nucleus)).mp. 2 Advanced
10 (discal adj1 hernia*).mp. 0 Advanced
11 (discus adj1 hernia*).mp. 0 Advanced
12 ischialgi*.mp. 16 Advanced
13 lumboischialgi*.mp. 1 Advanced
14 lumbo-ischialgi*.mp. 0 Advanced
15 cauda equina syndrome?.mp. 8 Advanced
16 or/1-15 71 Advanced
17 [Lumbar Vertebrae/] 0 Advanced
18 [Intervertebral Disk/] 0 Advanced
19 [Lumbosacral Region/] 0 Advanced
20 ((lumbar or lumbosacral or lumbo-sacral) adj1 (spine or region or area or vertebr* or disk? or disc?)).mp. 119 Advanced
21 discus intervertebra*.mp. 0 Advanced
22 (spinal adj1 (disk? or disc?)).mp. 2 Advanced
23 (low back adj1 (spine or region or area or vertebr*)).mp. 3 Advanced
24 Intervertebral Disk?.mp. 12 Advanced
25 Intervertebral Disc?.mp. 10 Advanced
26 Inter-vertebral Disk?.mp. 0 Advanced
27 Inter-vertebral Disc?.mp. 2 Advanced
28 or/17-27 126 Advanced
29 16 and 28 47 Advanced
30 [incidence/] 0 Advanced
31 [exp risk/] 0 Advanced
139
32 [exp Cohort Studies/] 0 Advanced
33 [exp Case-Control Studies/] 0 Advanced
34 [epidemiologic studies/] 0 Advanced
35 incidence.mp. 3422 Advanced
36 risk factor?.mp. 1230 Advanced
37 cohort?.mp. 1242 Advanced
38 case control?.mp. 521 Advanced
39 risk*.mp. 6338 Advanced
40 between group*.tw. 6559 Advanced
41 relative risk*.tw. 3145 Advanced
42 etiolog*.mp. 751 Advanced
43 [exp Epidemiology/] 0 Advanced
44 [Epidemiologic Methods/] 0 Advanced
45 epidemiolog*.mp. 1480 Advanced
46 [ep.fs.] 0 Advanced
47 or/30-46 6595 Advanced
48 29 and 47 47 Advanced EBM Reviews - Cochrane Central Register of Controlled Trials 4th Quarter 2010
# Searches Results Search Type
1 Intervertebral Disk Displacement/ 466 Advanced
2 Intervertebral Disk Displacement?.mp. 481 Advanced
3 Intervertebral Disc Displacement?.mp. 0 Advanced
4 Inter-vertebral Disk Displacement?.mp. 0 Advanced
5 Inter-vertebral Disc Displacement?.mp. 0 Advanced
6 Sciatica/ 157 Advanced
7 sciatica.mp. 303 Advanced
8 ((hernia* or ruptur* or prolaps* or protru* or bulg* or slip* or displac*) adj3 (disc? or disk? or nuclei or nucleus)).mp. 880 Advanced
9 ((extru* or sequestra* or migrat*) adj2 (disc? or disk? or nuclei or nucleus)).mp. 8 Advanced
10 (discal adj1 hernia*).mp. 3 Advanced
11 (discus adj1 hernia*).mp. 3 Advanced
12 ischialgi*.mp. 55 Advanced
13 lumboischialgi*.mp. 7 Advanced
14 lumbo-ischialgi*.mp. 2 Advanced
15 cauda equina syndrome?.mp. 6 Advanced
16 or/1-15 1110 Advanced
17 Lumbar Vertebrae/ 1394 Advanced
18 Intervertebral Disk/ 171 Advanced
19 Lumbosacral Region/ 226 Advanced
140
20 ((lumbar or lumbosacral or lumbo-sacral) adj1 (spine or region or area or vertebr* or disk? or disc?)).mp. 3215 Advanced
21 discus intervertebra*.mp. 0 Advanced
22 (spinal adj1 (disk? or disc?)).mp. 8 Advanced
23 (low back adj1 (spine or region or area or vertebr*)).mp. 5 Advanced
24 Intervertebral Disk?.mp. 651 Advanced
25 Intervertebral Disc?.mp. 209 Advanced
26 Inter-vertebral Disk?.mp. 0 Advanced
27 Inter-vertebral Disc?.mp. 0 Advanced
28 or/17-27 3536 Advanced
29 16 and 28 775 Advanced
30 incidence/ 5250 Advanced
31 exp risk/ 19132 Advanced
32 exp Cohort Studies/ 83768 Advanced
33 exp Case-Control Studies/ 6223 Advanced
34 epidemiologic studies/ 28 Advanced
35 incidence.mp. 36403 Advanced
36 risk factor?.mp. 19045 Advanced
37 cohort?.mp. 9285 Advanced
38 case control?.mp. 3182 Advanced
39 risk*.mp. 54509 Advanced
40 between group*.tw. 241268 Advanced
41 relative risk*.tw. 4158 Advanced
42 etiolog*.mp. 5723 Advanced
43 exp Epidemiology/ 15 Advanced
44 Epidemiologic Methods/ 594 Advanced
45 epidemiolog*.mp. 5160 Advanced
46 ep.fs. 19657 Advanced
47 or/30-46 300521 Advanced
48 29 and 47 597 Advanced
49 limit 48 to medline records 462 Advanced
50 limit 48 to embase records 89 Advanced
51 49 or 50 551 Advanced
52 48 not 51 46 Advanced EBM Reviews - Database of Abstracts of Reviews of Effects 4th Quarter 2010
# Searches Results Search Type
1 [Intervertebral Disk Displacement/] 0 Advanced
2 Intervertebral Disk Displacement?.mp. 21 Advanced
3 Intervertebral Disc Displacement?.mp. 0 Advanced
141
4 Inter-vertebral Disk Displacement?.mp. 0 Advanced
5 Inter-vertebral Disc Displacement?.mp. 0 Advanced
6 [Sciatica/] 0 Advanced
7 sciatica.mp. 29 Advanced
8 ((hernia* or ruptur* or prolaps* or protru* or bulg* or slip* or displac*) adj3 (disc? or disk? or nuclei or nucleus)).mp. 43 Advanced
9 ((extru* or sequestra* or migrat*) adj2 (disc? or disk? or nuclei or nucleus)).mp. 1 Advanced
10 (discal adj1 hernia*).mp. 0 Advanced
11 (discus adj1 hernia*).mp. 0 Advanced
12 ischialgi*.mp. 0 Advanced
13 lumboischialgi*.mp. 0 Advanced
14 lumbo-ischialgi*.mp. 0 Advanced
15 cauda equina syndrome?.mp. 4 Advanced
16 or/1-15 61 Advanced
17 [Lumbar Vertebrae/] 0 Advanced
18 [Intervertebral Disk/] 0 Advanced
19 [Lumbosacral Region/] 0 Advanced
20 ((lumbar or lumbosacral or lumbo-sacral) adj1 (spine or region or area or vertebr* or disk? or disc?)).mp. 118 Advanced
21 discus intervertebra*.mp. 0 Advanced
22 (spinal adj1 (disk? or disc?)).mp. 1 Advanced
23 (low back adj1 (spine or region or area or vertebr*)).mp. 0 Advanced
24 Intervertebral Disk?.mp. 29 Advanced
25 Intervertebral Disc?.mp. 4 Advanced
26 Inter-vertebral Disk?.mp. 0 Advanced
27 Inter-vertebral Disc?.mp. 0 Advanced
28 or/17-27 129 Advanced
29 16 and 28 31 Advanced
30 [incidence/] 0 Advanced
31 [exp risk/] 0 Advanced
32 [exp Cohort Studies/] 0 Advanced
33 [exp Case-Control Studies/] 0 Advanced
34 [epidemiologic studies/] 0 Advanced
35 incidence.mp. 1669 Advanced
36 risk factor?.mp. 1193 Advanced
37 cohort?.mp. 1426 Advanced
38 case control?.mp. 713 Advanced
39 risk*.mp. 5161 Advanced
40 between group*.tw. 6841 Advanced
41 relative risk*.tw. 1445 Advanced
142
42 etiolog*.mp. 1577 Advanced
43 [exp Epidemiology/] 0 Advanced
44 [Epidemiologic Methods/] 0 Advanced
45 epidemiolog*.mp. 1313 Advanced
46 [ep.fs.] 0 Advanced
47 or/30-46 9322 Advanced
48 29 and 47 27 Advanced
143
Appendix D. Detailed evidence tables for systematic review (Chapter 2)
Table D.1. Characteristics of eligible studies examining the incidence of lumbar disc herniation with radiculopathy in adults
First author, Year published Country Study design
Study population and setting Follow-up N (% female) Age range (mean) Participation %
Study outcome Case definition type Case definition ROB Incidence estimates (95% CI)
Burske-Hohlfeld, 1990 32 USA Cohort
All Olmsted County residents followed from 1950-1979 for any form of LDH surgery
Follow-up: 30y N=1,028 (NR) 15-78y (42) 88% incident surgery
LDH surgery Surgical Any form of back surgery for suspected LDH,
HNP, or fragments of disc material identified by review of complete (inpatient and outpatient) medical records
“Proven” LDH was a protruded, extruded or sequestered disc seen during surgery
“Suspected” LDH was a disc described as bulging or degenerated
Low Incidence rate of first surgery for suspected LDH: age-adjusted in men, 0.57 (0.52-0.62) per 1,000 person-years; age-adjusted in women, 0.36 (0.32-0.40) per 1,000 person-years; age- and sex-adjusted, 0.46 (0.43-0.49) per 1,000 person-years
Incidence rate of first surgery for proven LDH: age-adjusted in men, 0.51 (0.47-0.56) per 1,000 person-years; age-adjusted in women, 0.31 (0.27-0.34) per 1,000 person-years; age- and sex-adjusted, 0.41 (0.38-0.44) per 1,000 person-years
Raw data: numerators: 538, 371, 909, 485, 316, 801;
denominators: NR, but census data used to derive denominator person-years
Zitting, 1998 33 Finland Cohort
Birth cohort from 2 provinces of Finland followed from 1966-1994 for hospitalized LDH in the National Hospital Discharge Register
Follow-up: 28y N=12,058 (NR) 15-28y among cases (NR) 92%
Hospitalized LDH Hospital ICD-9 diagnosis codes and corresponding ICD-
8 codes used to identify all possible lumbar disc disease cases: 7221, 7227, 7244, 7225, 7242, 7245, 7561, 7384, 7385, 7213, 7214
“Confirmed” LDH was a reliable description of herniation in a surgical report or reliable evidence of a herniation on MRI, CT or myelogram with appropriate symptoms (i.e., on the same side) described in the medical records
Low Cumulative incidence of hospitalized LDH: in men, 9.1 (6.9-12.0) per 1,000 persons; in women, 4.2 (2.8-6.4) per 1,000 persons
Average annual incidence: in men, 0.33 (0.25-0.43) per 1,000 persons in women, 0.15 (0.10-0.23) per 1,000 persons
Raw data: in men, 50/5,474; in women, 22/5,218
144
First author, Year published Country Study design
Study population and setting Follow-up N (% female) Age range (mean) Participation %
Study outcome Case definition type Case definition ROB Incidence estimates (95% CI)
Mattila, 2008 37 Finland Cohort
Nationwide cohort of adolescents aged 14-18y, without LDH and prior hospitalized nonspecific back-related diagnosis followed up to December 31, 2001, in the National Hospital Discharge Register
Follow-up: 651,027 person-years (11y average follow-up)
N=57,408 (54%) 15-41y among cases (27 at surgery) 79%
LDH surgery Surgical Lumbar discectomy defined as ICD code 9211
between 1979 and 1996, and then with new codes ABC07, ABC16 and ABC26 from 1997 onwards
Low Incidence rate of LDH surgery: 0.41 (0.36-0.46) per 1,000 person-years; in men, 0.55 (0.46-0.63) per 1,000 persons-years; in women, 0.25 (0.19-0.30) per 1,000 persons-years
Raw data: numerators: 251, 166, 85; denominators: NR
Heliövaara, 1987 31 Finland Cohort
Nationwide cohort of Finnish adults followed from 1970-1980 for hospitalized LDH or sciatica in the National Hospital Discharge Register
Follow-up: 11y N=57,000 (48%) ≥15y (NR) 83%
Hospitalized LDH or sciatica Hospital ICD-8 codes indicating principal diagnosis for
hospitalization: 725.10 or 725.19 for LDH; 353.99 for sciatica
Low 11-year cumulative incidence of hospitalized LDH or sciatica: 14.5 (13.5-15.5) per 1,000 persons; of hospitalized LDH: 8.0 (7.3-8.7) per 1,000 persons; of hospitalized sciatica: 6.5 (5.9-7.2) per 1,000 persons
Average annual incidence of hospitalized LDH or sciatica: 1.32 (1.23-1.41) per 1,000 persons; of hospitalized LDH: 0.72 (0.66-0.79) per 1,000 persons; of hospitalized sciatica: 0.59 (0.53-0.65) per 1,000 persons
Raw data: 825/57,000, 454/57,000, 371/57,000
Mattila, 2009 38 Finland Cohort
All male military conscripts without severe back diseases and performing compulsory service between 1990-2002, were followed during their service period by record linkage to the National Hospital Discharge Register
Follow-up: 267,700 person-years; 6- to 12-month military service period
N=387,070 (0%) 18-29y (20) 100%
Hospitalized LDH Hospital Hospitalized LDH (lumbar and other
intervertebral disc disorders with radiculopathy) was ICD-10 diagnosis code M51.1 (lumbar and other intervertebral disc disorders with radiculopathy), and ICD-9 diagnosis codes 7227C and 3539X
Low Incidence rate of hospitalized LDH: 7.8 (6.7-8.3) per 1,000 person-years
Incidence rates ranged from a high of 12.3 (9.6-13.3) per 1,000 person-years in 1993, to a low of 4.2 (3.0-5.0) per 1,000 person-years 2001
Raw data: NR
145
First author, Year published Country Study design
Study population and setting Follow-up N (% female) Age range (mean) Participation %
Study outcome Case definition type Case definition ROB Incidence estimates (95% CI)
Leino-Arjas, 2002 44; Leino-Arjas, 2004 45
Finland Cohort
Nationwide workforce cohort followed in 1996 for lumbar intervertebral disc disorders by record linkage to the National Hospital Discharge Register
Follow-up: 1y N=2,409,319 (NR) 20-64y (NR) 100% a n=2,409,319 (NR); 20-64y b n=1,783,616 (51% women); 25-64y
Hospitalized LDH and LDH surgery Hospital and Surgical Hospitalized LDH was primary diagnosis ICD-
10 codes M51.1-M51.9 (intervertebral disc disorders other than those of the cervical spine)
LDH surgery was those with information on surgical decompression or spondylodesis of the lumbar spine during any of the admissions in 1996
Mod 1-year cumulative incidence of hospitalized LDH: among entire 20-64y workforce, 1.9 (1.9-2.0) per 1,000 persons; among those gainfully employed throughout prior year, 2.2 (2.1-2.2) per 1,000 persons; among men gainfully employed throughout prior year, 2.5 (2.4-2.7) per 1,000 persons; among women gainfully employed throughout prior year, 1.8 (1.7-1.9) per 1,000 persons
1-year cumulative incidence of LDH surgery: among entire 20-64y workforce, 1.0 (0.9-1.0) per 1,000 persons; among those gainfully employed throughout prior year, 1.1 (1.1-1.2) per 1,000 persons
Raw data: 4,643/2,409,319 entire workforce; 3,863/1,783,616
gainfully employed workforce; 2,211/868,876 in men gainfully employed; 1,652/914,740 in women gainfully employed; 2,368/2,409,319 surgical cases in entire workforce; 2,015/1,783,616 surgical cases in gainfully employed workforce
Jørgensen, 1994 42 Denmark Cohort
Occupationally active assistant nurses followed for LDH surgery in 1988 by record linkage to the Danish National Registry of Hospitalized Patients, and compared to all Danish females
Follow-up: 1y N=1,681,152 (100%) 20-69y (NR) 100%
LDH surgery Surgical LDH surgery was ICD-8 surgical codes: 82073,
ablatio prolapsus disci intervertebralis lumbalis; 82173, evacuatio disci intervertebralis lumbalis
Mod 1-year cumulative incidence of LDH surgery among nurses aged 30-69y: 1.3 (1.0-1.8) per 1,000 persons
1-year cumulative incidence of LDH surgery among Danish women aged 30-69y: 0.78 (0.74-0.84) per 1,000 persons
Raw data: 37/28,008; 969/1,235,038
Hurme, 1983 39 Finland Cohort
AII lumbar disc herniation surgeries performed in South-West Finland from 1975-1979, based on surgical department registers of the Turku University Central Hospital area (mean area population during study period, 455,000)
Follow-up: 5y N=1,011 surgeries (44%) 15-80y among cases (42 at surgery) 79% incident surgery
LDH surgery Surgical No explicit case definition provided Included all operations performed for LDH,
classified as first operations and reoperations
Mod 5-year cumulative incidence of first LDH surgery: 1.7 (1.6-1.8) per 1,000 persons
Average annual incidence: 0.34 (0.32-0.37) per 1,000 persons Raw data: 778/455,000
146
First author, Year published Country Study design
Study population and setting Follow-up N (% female) Age range (mean) Participation %
Study outcome Case definition type Case definition ROB Incidence estimates (95% CI)
Heikkilä, 1989 41 Finland Cohort
Finnish twin cohort consisting of 9,365 adult pairs of the same sex followed from 1972-1985 for hospitalized sciatica by record linkage to the National Hospital Discharge Register
Follow-up: 14y N=18,730 (55%) 24-60y+ (NR) 100%
Hospitalized sciatica Hospital Sciatica was ICD-8 discharge diagnosis code:
353
Mod 14-year cumulative incidence of hospitalized sciatica among twins: 16.5 (14.8-18.5) per 1,000 persons
Average annual incidence: 1.2 (1.1-1.3) per 1,000 persons Raw data: 304/18,370
Jhawar, 2006 36 USA Cohort
Female nurses from the Nurses’ Health Study without prior LDH in 1976, were followed up in 1992
Follow-up: 16y N=98,407 (100%) 30-55y in 1976 (NR) ≥85%
Clinical LDH Clinical Self-reported physician-diagnosis of LDH that
was confirmed by MRI or CT, in follow-up mailed questionnaire
Low Incidence rate: 6.2 (6.0-6.5) per 1,000 person-years Raw data: 2,727/438,662
Miranda, 2002 34 Finland Cohort
Forest industry workers who had no sciatica during the past 12 months in 1994
Follow-up: 1y N=2,077 (26%) NR (45) 77%
Sciatica Clinical Sciatica was more than 7 days of low back pain
radiating below the knee during the preceding 12 months, with a manikin used to denote the anatomic area in mailed questionnaire
Low 1-year cumulative incidence of sciatica: 93.4 (81.6-106.7) per 1,000 persons
Raw data: 194/2,077
Jarvik, 2005 35 USA Cohort
Outpatients without LBP or sciatica in the past 4 months from four clinics at the Veterans Affairs Puget Sound Health Care System, Seattle Division
Follow-up: 3y N=148 (13%) 35-70y (median, 53) 89%
Clinical LDH Clinical LDH was MRI-confirmed disc protrusion or
extrusion, with pain frequency for low back or buttock pain rated as more than “some of the time” [and] sciatic leg pain; or numbness or tingling in the leg, foot, or groin; or weakness in leg or foot, rated as more than “none”
Low 3-year cumulative incidence among persons without MRI disc-findings at baseline: 89.4 (50.7-153.1) per 1,000 persons
Average annual incidence: 29.8 (16.9-51.0) per 1,000 persons Raw data: 11/123
147
First author, Year published Country Study design
Study population and setting Follow-up N (% female) Age range (mean) Participation %
Study outcome Case definition type Case definition ROB Incidence estimates (95% CI)
Riihimäki, 1989 40 Finland Cohort
Male concrete reinforcement workers and house painters without prior history of sciatic pain in 1977 followed up in 1982
Follow-up: 5y N=178 (0%) 25-54y at baseline 77-80%
Sciatica Clinical Sciatica was defined as back pain radiating to a
leg in follow-up mailed questionnaire
Mod 5-year cumulative incidence of sciatica: among concrete reinforcement workers, 343 (244-457) per 1,000 persons; among house painters, 229 (159-318) per 1,000 persons
Average annual incidence of sciatica: among concrete reinforcement workers, 68.5 (48.8-91.3) per 1,000 persons; among house painters, 45.7 (31.7-63.5) per 1,000 persons
Raw data: 25/73, 24/105
Leclerc, 2003 46 France Cohort
Male workers in the French national electricity and gas company without LBP during the past 12 months in 1992 were followed up in 1994
Follow-up: 2y N=841 (0%) 40-50y at baseline (NR) 65%
Sciatica Clinical Sciatica was pain, discomfort or stiffness in the
low back region at least 1 day in the previous 12 months, with radiating symptoms in the leg, in follow-up mailed questionnaire
Mod 2-year cumulative incidence of sciatica: 55.9 (42.3-73.5) per 1,000 persons; among those with a history of LBP before 1992, 101.9 (63.7-159.1) per 1,000 persons; among those with no history of LBP before 1992, 45.3 (32.1-63.6) per 1,000 persons
Average annual incidence of sciatica: 27.9 (21.1-36.8) per 1,000 persons; among those with a history of LBP before 1992, 51.0 (31.9-79.5) per 1,000 persons; among those with no history of LBP before 1992, 22.6 (16.1-31.8) per 1,000 persons
Raw data: 47/841; 16/157; 31/684
Riihimäki, 1994 43 Finland Cohort
Male machine operators, carpenters and office workers without prior history of sciatica in 1984, followed up in 1987
Follow-up: 3y N=1,149 (0%); 25-49y (37) 83%
Sciatica Clinical Sciatica was LBP radiating to a leg in follow-up
mailed questionnaire
Mod 3-year cumulative incidence of sciatica: among office workers, 141 (111-177) per 1,000 persons; among machine operators, 220 (181-264) per 1,000 persons; among carpenters, 241 (198-290) per 1,000 persons
Average annual incidence of sciatic pain: among carpenters, 80.4 (66.1-96.5) per 1,000 persons; among machine operators, 73.2 (60.4-87.8) per 1,000 persons; among office workers, 46.9 (37.0-59.0) per 1,000 persons
Raw data: 81/336; 85/387; 60/426
148
First author, Year published Country Study design
Study population and setting Follow-up N (% female) Age range (mean) Participation %
Study outcome Case definition type Case definition ROB Incidence estimates (95% CI)
Netterstrøm, 1989 48 Denmark Cohort
All full-time male bus drivers employed on April 1, 1978 by 3 urban bus companies followed up from 1978-1984 for hospitalized LDH in the Danish National Patient Register
Follow-up: 7y N=2,465 (0%) 20-69y (NR) 100%
Hospitalized LDH Hospital ICD discharge diagnosis codes for LDH
(725.10-725.11)
High Incidence rate of hospitalized LDH: 3.6 (2.8-4.6) per 1,000 person-years
7-year cumulative incidence of hospitalized LDH: 14.6 (10.6-20.2) per 1,000 persons
Average annual incidence: 2.1 (1.5-2.9) per 1,000 persons Raw data: 62/17,122; 36/2,465
Bongers, 1988 47 Netherlands Cohort
All male crane workers and floor workers employed in the same departments at a steel company on January 1, 1975 and those hired up to the end of 1979, were followed up for incident disability pensions due to LDH up until December 31, 1984
Follow-up: 10y N=1,405 (0%) <25-60y+ (NR) 71%
Disability pension due to LDH Clinical Disability pension due to displacement of
intervertebral disc defined as ICD-9 diagnosis code 722.2, assigned by social insurance physician after evaluation of the social insurance medical records
High Incidence rate of LDH disability pension among crane operators, 2.9 (1.6-4.9) per 1,000 person-years; among floor workers, 1.3 (0.5-2.9) per 1,000 person-years
Raw data: 13/4,454.2; 5/3,805.8
Abbreviations: CT, computerized tomography; HNP, herniated nucleus pulposus; ICD, international classification of diseases; LBP, low back pain; LDH, lumbar disc herniation; Mod, moderate; MRI, magnetic resonance imaging; N, study size; NR, not reported; ROB, risk of bias; y, years
149
Appendix Table D.2. Characteristics of eligible studies examining risk factors of lumbar disc herniation with radiculopathy in adults
First author, Year of publication
Country Study design
Participants and setting Follow-up N (% female) Age range (mean) Participation %
Study outcome Case definition type Case definition ROB Risk factors considered
Phase of evidence Risk estimates (95% CI)
Zitting, 1998 33 Finland Cohort
Birth cohort from 2 provinces of Finland followed from 1966-1994 for hospitalized LDH in the National Hospital Discharge Register
Follow-up: 28 years N=12,058 (NR) 15-28y among cases (NR) 92%
Hospitalized LDH Hospital ICD-9 diagnosis codes and
corresponding ICD-8 codes used to identify all possible lumbar disc disease cases: 7221, 7227, 7244, 7225, 7242, 7245, 7561, 7384, 7385, 7213, 7214
“Confirmed” LDH was a reliable description of herniation in a surgical report or reliable evidence of a herniation on MRI, CT or myelogram with appropriate symptoms (i.e., on the same side) described in the medical records
Low Sex, age Phase II Sex: women RR 1.0; men RR 2.2 (1.3-3.6) Age: NR; in men, hospitalized LDH first occurred around 15y of
age, and incidence rose more sharply from 20y of age; in women, first cases of hospitalized LDH also appeared around 15y of age
Mattila, 2008 37 Finland Cohort
Nationwide cohort of adolescents aged 14-18y, without LDH and prior hospitalized nonspecific back-related diagnosis followed up to December 31, 2001, in the National Hospital Discharge Register
Follow-up: 651,027 person-years (11y average follow-up)
N=57,408 (54%) 15-41y among cases (27 at
surgery) 79%
LDH surgery Surgical Lumbar discectomy defined as ICD
code 9211 between 1979 and 1996, and then with new codes ABC07, ABC16 and ABC26 from 1997 on
Low Sex, socioeconomic background, perceived health status, chronic disease or disability, number of stress symptoms per week, timing of puberty, overweight, smoking, drinking style, frequency of participation in sports clubs, frequency of other leisure-time physical exercise, school success
Phase II Sex: women RR 1.0; men RR 2.2 (1.7-2.9) In men Smoking: not daily HR 1.0; daily HR 1.5 (1.1-2.2) Timing of puberty: early HR 1.0; normal HR 0.7 (0.5-1.1); late HR
0.6 (0.4-1.0) In women Overweight: no HR 1.0; yes HR 2.1 (1.1-4.1) Frequency of participation in sports clubs: never HR 1.0; 2-3 times
per week or less HR 1.5 (0.9-2.5); 4-5 times per week or more HR 2.7 (1.1-6.3)
Heliövaara, 1987 31,49,50
Finland Cohort with case-
control risk analyses
Nationwide cohort of Finnish adults followed from 1970-1980 for hospitalized LDH or sciatica in the National Hospital Discharge Register
Hospitalized LDH or sciatica Hospital ICD-8 codes indicating principal
diagnosis for hospitalization: 725.10 or 725.19 for LDH; 353.99 for sciatica
Low a Sex, age, geographic region, type of population, social class (men only), number of births (women only), marital status, leisure
Phase II Hospitalized LDH Sex: women RR 1.0; men RR 1.6 (p<0.001) Age at baseline (y): 15-19 RR 1.0; 20-29 RR 3.0 (p<0.001); 30-39
RR 5.5 (p<0.001); 40-49 RR 6.5 (p<0.001); 50-59 RR 3.0 (p<0.001); 60+ RR 1.3
150
First author, Year of publication
Country Study design
Participants and setting Follow-up N (% female) Age range (mean) Participation %
Study outcome Case definition type Case definition ROB Risk factors considered
Phase of evidence Risk estimates (95% CI)
Follow-up: 11y N=57,000 (48%) ≥15y (NR) 83% a n=2,732; 20-59y b n=1,537; 20-59y c n=2,732; 20-59y
time physical activity, smoking, chronic cough, number of psychological distress symptoms, medication use, frequent use of analgesics
b Height, BMI (as an overall measure of obesity), and triceps skinfold thickness
c Occupation, strenuousness of work
Type of population: urban RR 1.0; rural RR 0.8 (p<0.05); industrial RR 1.1
Hospitalized sciatica Sex: women RR 1.0; men RR 1.3 (p<0.05) Age at baseline (y): 15-19 RR 1.0; 20-29 RR 2.7 (p<0.05); 30-39
RR 8.6 (p<0.001); 40-49 RR 12.1 (p<0.001); 50-59 RR 7.3 (p<0.001); 60+ RR 4.3 (p<0.01)
Type of population: urban RR 1.0; rural RR 0.9; industrial RR 1.5 (p<0.001)
Case-control analyses in men Hospitalized LDH Social class: I OR 1.0; II OR 2.4 (p<0.05); III OR 2.6 (p<0.05); IV
OR 2.0; V OR 1.2 Height (cm): ≤169 OR 1.0; 170-174 OR 1.1 (0.7-1.7); 175-179 OR
1.1 (0.7-1.7); ≥180 OR 2.3 (1.4-3.9) BMI (kg/m2): <21.9 OR 1.0; 22.0-23.9 OR 2.4 (1.3-4.5); 24.0-25.9
OR 3.2 (1.7-6.1); 26.0-27.9 OR 3.1 (1.5-6.3); 28.0-29.9 OR 3.7 (1.7-8.0); ≥30.0 OR 2.3 (0.8-6.2)
Occupation: professional/white collar OR 1.0; intermediate non-manual OR 2.3 (p<0.05); forestry OR 2.9; farmer/agricultural OR 1.4; motor vehicle driver OR 2.9 (p<0.05); metal/machine worker OR 3.0 (p<0.01); construction worker OR 2.4 (p<0.05); chemical processor/paper worker OR 2.4; other industrial OR 2.2 (p<0.05); service and other groups OR 2.3
Hospitalized LDH or sciatica Social class: I OR 1.0; II OR 3.2 (p<0.001); III OR 3.0 (p<0.001); IV
OR 2.6 (p<0.01); V OR 1.5 Occupation: professional/white collar OR 1.0; intermediate non-
manual OR 2.8; forestry OR 3.1 (p<0.05); farmer/agricultural OR 2.5 (p<0.01); motor vehicle driver OR 4.6 (p<0.001); metal/machine worker OR 4.2 (p<0.001); construction worker OR 3.1 (p<0.001); chemical processor/paper worker OR 3.2 (p<0.001); other industrial OR 2.6 (p<0.01); service and other groups OR 3.1 (p<0.01)
Case-control analyses in women Hospitalized LDH Height (cm): ≤159 OR 1.0; 160-164 OR 1.0 (0.6-1.6); 165-169 OR
151
First author, Year of publication
Country Study design
Participants and setting Follow-up N (% female) Age range (mean) Participation %
Study outcome Case definition type Case definition ROB Risk factors considered
Phase of evidence Risk estimates (95% CI)
1.2 (0.6-2.3); ≥170 OR 3.7 (1.6-6.6) Number of psychological distress symptoms: 0 OR 1.0; 1 OR 2.0
(p<0.05); 2 OR 2.9 (p<0.001); 3-5 OR 1.2 Strenuousness of work: light or very light OR 1.0; normal OR 3.8
(p<0.05); heavy or very heavy OR 2.4 Hospitalized LDH or sciatica Number of psychological distress symptoms: 0 OR 1.0; 1 OR 1.7
(p<0.05); 2 OR 2.0 (p<0.01); 3-5 OR 1.8 (p<0.05) Medication use: None OR 1.0; Some OR 0.7 (p<0.05) Frequent use of analgesics: No OR 1.0; Yes OR 1.9 (p<0.01) Strenuousness of work: light or very light OR 1.0; normal OR 2.0
(p<0.05); heavy or very heavy OR 2.5 (p<0.05) Seidler, 2009 67;
Schumann, 2010 68
Germany Case-control
Cases: 564 patients with inpatient or outpatient hospital treatment for pain associated with LDH in 4 regions of Germany (Frankfurt/Main, Freiburg, Halle/Saale, Regensburg)
Controls: 901 persons randomly selected from a 1% random sample of residents from local population registration offices of the same 4 regions
Follow-up: NA 1,816 (50%) 25-70y (48) 53-66%
LDH hospital treatment Hospital Inpatient or outpatient hospital
treatment because of LDH with radiculopathy, sensory and/or motor deficits (neurological findings), and confirmation of LDH by MRI or CT scan
Mod a Occupational: cumulative lumbar load by manual materials handling (lifting, carrying, pushing, pulling, throwing, catching or shoveling of objects weighing ≥5 kg), cumulative lumbar load by intensive-load postures (postures with trunk inclination of ≥20 degrees)
b Lifestyle factors: BMI,
smoking, and sports activities (endurance sports, ball sports, athletic sports, body building sports)
a Phase III In men Cumulative lumbar load by manual materials handling and/or
intensive-load postures (Nh): 0-<5.0*106 OR 1.0; 5.0-<21.51*106 OR 1.7 (1.1-2.7); ≥21.51*106 OR 3.4 (2.2-5.0)
Cumulative lumbar load by manual materials handling (Nh): 0-<2.34*106 OR 1.0; 2.34-<8.98*106 OR 1.2 (0.7-2.0); ≥8.98*106 OR 2.0 (1.2-3.5)
Cumulative lumbar load by intensive-load postures (Nh): 0 OR 1.0; >0-<4.85*106 OR 1.1 (0.6-2.0); 4.85-14.62*106 OR 1.7 (0.9-3.2); ≥14.62*106 OR 1.9 (1.0-3.5)
In women Cumulative lumbar load by manual materials handling and/or
intensive-load postures (Nh): 0 OR 1.0; 0-<4.04*106 OR 1.6 (1.1-2.7); 4.04-<14.47*106 OR 2.4 (1.6-3.8); ≥14.47*106 OR 2.3 (1.5-3.6)
Cumulative lumbar load by intensive-load postures (Nh): 0 OR 1.0; >0-<2.77*106 OR 1.9 (1.0-3.7); 2.77-8.83*106 OR 2.4 (1.2-4.6); ≥8.83*106 OR 3.2 (1.6-6.3)
b Phase II In men BMI (kg/m2): <21.88 OR 1.0; ≥21.88-<24.30 OR 1.4 (0.8-2.4); ≥24.30-<29.21 OR 2.1 (1.3-3.6); ≥29.21 OR 1.6 (0.7-3.8)
Smoking (pack-years): 0 OR 1.0; >0-<8 OR 1.0 (0.6-1.7); ≥8-<20 OR 1.2 (0.8-1.9); ≥20-<40 OR 1.6 (1.0-2.5); ≥40 OR 0.8 (0.4-
152
First author, Year of publication
Country Study design
Participants and setting Follow-up N (% female) Age range (mean) Participation %
Study outcome Case definition type Case definition ROB Risk factors considered
Phase of evidence Risk estimates (95% CI)
1.5) Cumulative hours of body building sports (h): 0 OR 1.0; >0-<400
OR 1.0 (0.5-2.4); 400-<1,350 OR 1.0 (0.5-2.2); ≥1,350 OR 0.5 (0.2-1.1)
In women BMI (kg/m2): <21.88 OR 1.0; ≥21.88-<24.30 OR 1.3 (0.9-1.8); ≥24.30-<29.21 OR 1.3 (0.8-1.9); ≥29.21 OR 2.0 (1.1-3.7)
Smoking (pack-years): 0 OR 1.0; >0-<8 OR 1.0 (0.7-1.6); ≥8-<20 OR 1.5 (1.0-2.4); ≥20-<40 OR 1.0 (0.6-1.7); ≥40 OR 1.4 (0.4-4.5)
Leino-Arjas, 2002 44; Leino-Arjas, 2004 45
Finland Cohort
a Nationwide workforce cohort followed in 1996 for lumbar intervertebral disc disorders by record linkage to the National Hospital Discharge Register
Follow-up: 1y N=2,409,319 (NR) 20-64y (NR) 100% b Nationwide occupationally
active workforce cohort followed in 1996 for lumbar intervertebral disc disorders by record linkage to the National Hospital Discharge Register
Follow-up: 1y N=1,783,616 (51%) 25-64y (NR) 100%
a,b Hospitalized LDH Hospital Hospitalized LDH was primary
diagnosis ICD-10 codes M51.1-M51.9 (intervertebral disc disorders other than those of the cervical spine)
Mod a Sex, age, employment status, education, occupational class, income
b Sex, age, education,
income, smoking and BMI
Occupational: physical workload, manual materials handling, accident risk, inconvenient work postures, sedentary work, video display terminal work, challenging tasks, social demands, job control, work time schedule
a Phase II Hospitalized LDH (crude analysis) Age (y): 20-34 RR 1.0; 35-44 RR 1.9 (1.7-2.0); 45-54 RR 1.8 (1.6-
1.9); 55-64 RR 0.9 (0.8-1.1) Employment status at end of year 1995: occupationally active RR
1.0; unemployed RR 0.8 (0.7-0.8); on unemployment pension RR 0.8 (0.5-1.1); on other pension RR 1.7 (1.6-1.9); student, at military service or other RR 0.5 (0.4-0.5)
Hospitalized LDH in those occupationally active throughout 1995 (multivariable analysis)
Sex: men RR 1.0; women RR 0.8 (0.8-0.9) Education: higher RR 1.0; secondary RR 1.3 (1.2-1.5); basic RR 1.5
(1.3-1.8) Occupational class: upper white-collar RR 1.0; lower white-collar
(in supervising position or with independent work tasks) RR 1.0 (0.8-1.1); lower white-collar (in dependent position or with routine work tasks) RR 1.2 (1.0-1.4); specialized manual workers RR 1.4 (1.2-1.6); non-specialized manual workers RR 1.5 (1.3-1.8); farmers RR 1.3 (1.1-1.5); other entrepreneurs RR 1.2 (1.0-1.4)
Income: highest quartile RR 1.0; 2nd RR 1.0 (1.0-1.1); 3rd RR 1.0 (0.9-1.1); lowest RR 0.7 (0.6-0.8)
b Phase III Sex: women RR 1.0; men RR 1.4 (1.3-1.5) In men Age (y): 25-34 RR 1.0; 35-44 RR 1.4 (1.2-1.5); 45-54 RR 1.3 (1.2-
1.5); 55-64 RR 1.1 (0.9-1.3) Education: higher RR 1.0; intermediate RR 1.4 (1.2-1.6); basic RR
153
First author, Year of publication
Country Study design
Participants and setting Follow-up N (% female) Age range (mean) Participation %
Study outcome Case definition type Case definition ROB Risk factors considered
Phase of evidence Risk estimates (95% CI)
1.7 (1.4-1.9) Personal income: highest quintile RR 1.0; 2nd RR 1.2 (1.0-1.5); 3rd
RR 1.5 (1.3-1.8); two lowest quintiles combined RR 1.4 (1.2-1.6) Accident risk: no RR 1.0; low RR 1.2 (1.0-1.5); high RR 1.5 (1.2-1.8) Job control: low RR 1.0; high RR 0.9 (0.8-1.0) In women Age (y): 25-34 RR 1.0; 35-44 RR 1.8 (1.5-2.0); 45-54 RR 1.7 (1.5-
2.0); 55-64 RR 1.2 (1.0-1.5) Education: higher RR 1.0; intermediate RR 1.6 (1.4-1.8); basic RR
1.8 (1.5-2.1) Personal income: highest quintile RR 1.0; 2nd RR 1.5 (1.3-1.8); 3rd
RR 1.5 (1.2-1.8); two lowest quintiles combined RR 1.5 (1.2-1.8) BMI: lowest tertile RR 1.0; middle tertile RR 1.1 (1.0-1.2); highest
tertile RR 1.3 (1.1-1.5) Accident risk: no RR 1.0; low RR 1.3 (1.1-1.5); high RR 1.4 (1.2-1.8) Job control: low RR 1.0; high RR 0.8 (0.8-0.9) Work time schedule: regular daytime work RR 1.0; two-shift work,
regular evening work, weekend work or other irregular work hours not including night work RR 1.1 (0.9-1.3); regular or irregular three-shift work or regular night work RR 1.3 (1.2-1.6)
Jørgensen, 1994 42
Denmark Cohort
Occupationally active assistant nurses followed for LDH surgery in 1988 by record linkage to the Danish National Registry of Hospitalized Patients, and compared to all Danish females
Follow-up: 1y N=1,681,152 (100%) 20-69y (NR) 100%
LDH surgery Surgical LDH surgery was ICD-8 surgical
codes: 82073, ablatio prolapsus disci intervertebralis lumbalis; 82173, evacuatio disci intervertebralis lumbalis
Mod Age, occupation Phase II Age: NR; incidence of LDH surgery increased with age up to the
45-49y age group in nurses, and the 50-54y age group in all Danish females
Occupation: all Danish females aged 30-69y RR 1.0; assistant nurses aged 30-69y RR 1.6 (1.2-2.2)
Hurme, 1983 39 Finland Cohort
AII lumbar disc herniation surgeries performed in South-West Finland from 1975-1979, based on surgical department registers of the Turku
LDH surgery Surgical No explicit case definition provided Included all operations performed for
LDH, classified as first operations and reoperations
Mod Sex, age, workload, home locality, seasonality
Phase I Sex: NR; 56% of surgeries done on men Age: NR; 90% of patients between 25 and 54 years of age Workload: NR; neither heavy nor light work were more common
among the incident surgery group than among the general population of Finland or South-West Finland
154
First author, Year of publication
Country Study design
Participants and setting Follow-up N (% female) Age range (mean) Participation %
Study outcome Case definition type Case definition ROB Risk factors considered
Phase of evidence Risk estimates (95% CI)
University Central Hospital area (mean area population during study period, 455,000)
Follow-up: 5y N=1,011 surgeries (44%) 15-80y among cases (42 at
surgery) 79% incident surgery
Jhawar, 2006 36 USA Cohort
Female nurses from the Nurses’ Health Study without prior LDH in 1976, were followed up in 1992
Follow-up: 16 years N=98,407 (100%) 30-55y in 1976 (NR) ≥85%
Clinical LDH Clinical Self-reported physician-diagnosis of
LDH that was confirmed by MRI or CT, in follow-up mailed questionnaire
Low Cardiovascular risk factors: BMI, smoking, diabetes, high cholesterol, hypertension, family history of coronary heart disease
Phase III High cholesterol: no RR 1.0; yes RR 1.3 (1.1-1.4) Diabetes: no RR 1.0; yes RR 1.5 (1.2-2.0) Hypertension: no RR 1.0; yes RR 1.3 (1.1-1.4) Family history: no RR 1.0; yes RR 1.1 (1.0-1.3) BMI (kg/m2): <21.9 RR 1.0; 22.0-24.9 RR 1.0 (0.9-1.1); 25.0-26.9
RR 1.1 (1.0-1.3); 27.0-28.9 RR 1.2 (1.0-1.4); 29+ RR 1.1 (1.0-1.3) Smoking: non-smoker RR 1.0; ex-smoker RR 1.1 (1.0-1.2); current
RR 1.4 (1.3-1.5)
Miranda, 2002 34 Finland Cohort
Forest industry workers who had no sciatica during the past 12 months in 1994
Follow-up: 1y N=2,077 (26%) NR (45) 77%
Sciatica Clinical Sciatica was more than 7 days of low
back pain radiating below the knee during the preceding 12 months, with a manikin used to denote the anatomic area
Low Individual: sex, age, height, BMI, mental stress, smoking, car driving, previous low back injuries
Physical exercise: frequency of physical exercise, sports activity, different types of sports
Work-related: amount of twisting movements of trunk, working with trunk forward flexed, working with a hand above shoulder level, working in sitting position, working in kneeling or squatting position, daily lifting of loads, operating a
Phase II Age (y): <35 OR 1.0; 35-44 OR 2.1 (1.1-4.1); 45-54 OR 2.0 (1.0-
4.0); ≥55 OR 3.3 (1.4-7.6) Smoking: non-smoker OR 1.0; ex-smoker OR 1.3 (0.9-2.0); current
smoker (1-15cig/d, 1-15y) OR 1.3 (0.6-3.1); current smoker (>15cig/d, 1-15y) OR 1.2 (0.3-4.6); current smoker (1-15cig/d, >15y) OR 2.2 (1.2-4.0); current smoker (>15cig/d, >15y) OR 2.3 (1.3-3.9)
Mental stress: not at all OR 1.0; only little OR 1.6 (0.9-2.9); to some extent OR 2.1 (1.2-3.7); rather much or much OR 3.0 (1.5-5.9)
Walking: not at all or only little OR 1.0; moderately OR 1.8 (1.2-2.8); actively OR 1.9 (1.2-3.0)
Jogging: not at all or only little OR 1.0; moderately or actively OR 0.5 (0.3-1.0)
Twisting movements of trunk during the work day: not at all or only little OR 1.0; moderately OR 1.6 (1.1-2.5); much OR 1.9 (1.1-3.2)
155
First author, Year of publication
Country Study design
Participants and setting Follow-up N (% female) Age range (mean) Participation %
Study outcome Case definition type Case definition ROB Risk factors considered
Phase of evidence Risk estimates (95% CI)
motor vehicle, physical strenuousness of work, job satisfaction, overload at work, risk of accident at work
Seidler, 2003 61 Germany Case-control
Cases: 225 male patients with acute LDH (94 LDH only; 131 LDH and osteochondrosis or spondylosis) from 2 orthopaedic practices and 4 neurosurgical or orthopaedic clinics in the Frankfurt/Main area
Controls: 107 population controls without chronic LBP and 90 patients hospitalized for urolithiasis treatment by lithotripsy without chronic LBP and osteochondrosis/ spondylosis (197 controls in total)
Excluded persons with Bechterew’s disease, spine fractures, malignancies of the spine, and poliomyelitis
Follow-up: NA 437 (0%) 25-65y (42) 66-93%
Clinical LDH Clinical LDH only cases defined as currently
symptomatic, radiographically confirmed (MRI and/or CT scan) LDH or protrusion
Mod Occupational: physical workload, lifting/carrying, extreme forward bending postures, whole body vibration, psychosocial work environment (monotonous, boring, opportunities to use knowledge and skills, information about future plans, satisfaction with supervisor, satisfaction with workmates, psychic strain through contact with clients, time pressure, too much responsibility)
Phase III Physical workload: always working in occupations with low physical
workload OR 1.0; ≥10y in occupations with medium physical workload OR 0.8 (0.4-1.7); ≥10y in occupations with high physical workload OR 2.1 (0.9-4.6)
Cumulative extreme (>90° trunk flexion) forward bending (h): 0 OR 1.0; >0-1,500 OR 1.4 (0.7-2.8); >1,500 OR 2.7 (1.2-6.4)
Cumulative whole body vibration (h): 0 OR 1.0; >0-1,800 OR 2.1 (0.9-4.8); >1,800 OR 1.9 (0.7-4.9)
Psychosocial time pressure (number of working years with high degree of time pressure, classified as 5 or 6): 0 OR 1.0; >0-<10 OR 1.2 (0.6-2.6); ≥10 OR 2.9 (1.3-6.3)
Riihimäki, 1989 40 Finland Cohort
Male concrete reinforcement workers and house painters without prior history of sciatic pain in 1977 followed up in 1982
Follow-up: 5y N=178 (0%)
Sciatica Clinical Sciatica was defined as back pain
radiating to a leg in follow-up mailed questionnaire
Mod Occupation, previous back symptoms, earlier back accidents, lumbar spine degeneration, back muscle strength, abdominal muscle strength, height, BMI,
Phase II Occupation: house painters OR 1.0; concrete reinforcement
workers OR 1.8 (1.2-2.9) Earlier back accidents: no OR 1.0; yes OR 1.6 (1.0-2.7) Previous back symptoms: no symptoms OR 1.0; lumbago or
nonspecific low back pain OR 1.8 (0.9-3.4) BMI (kg/m2): ≤23.9 OR 1.0; 24.0-27.9 OR 1.1 (0.6-2.1); ≥28.0 OR
156
First author, Year of publication
Country Study design
Participants and setting Follow-up N (% female) Age range (mean) Participation %
Study outcome Case definition type Case definition ROB Risk factors considered
Phase of evidence Risk estimates (95% CI)
25-54y at baseline 77-80%
stress, smoking 1.6 (0.8-3.1)
Leclerc, 2003 46 France Cohort
Male workers in the French national electricity and gas company without LBP during the past 12 months in 1992 were followed up in 1994
Follow-up: 2y N=841 (0%) 40-50y at baseline (NR) 65%
Sciatica Clinical Sciatica was pain, discomfort or
stiffness in the low back region at least 1 day in the previous 12 months, with radiating symptoms in the leg, in follow-up mailed questionnaire
Mod Individual: age, living alone, height, BMI, exercise and sports, gardening, non-professional home construction activities, do-it-yourself, smoking
Health: history of low back pain, presence of neck pain, self-rated general health, score of psychological and psychosomatic well-being
Occupational: occupational category, score of job satisfaction and psychosocial aspects, prolonged sitting, prolonged standing, carrying loads, pulling or pushing heavy loads, bending forward and backward, trunk rotations, kneeling and squatting, driving
Phase II Past history of LBP: no OR 1.0; yes OR 3.7 (1.9-7.4) Height (cm): ≤180 OR 1.0; >180 OR 2.7 (1.2-6.4) Driving for >2h: less than once per week OR 1.0; more than once
per week OR 2.7 (1.2-6.4); daily OR 2.0 (0.9-4.4) Self-rated health: good (score 1 or 2) OR 1.0; medium (score 3 or
4) OR 1.0 (0.5-2.0); bad (score 5-8) OR 2.9 (1.2-7.1) Do-it-yourself activities: no OR 1.0; yes OR 2.0 (0.9-4.8)
Riihimäki, 1994 43; Pietri-Taleb, 1995 60
Finland Cohort
Male machine operators, carpenters and office workers without prior history of sciatica in 1984, followed up in 1987
Follow-up: 3y N=1,149 (0%); 25-49y (37) 83%
Sciatica Clinical Sciatica was LBP radiating to a leg in
follow-up mailed questionnaire
Mod a Age, occupation, seniority in occupation, education level, car driving, physical exercise, smoking, twisted or bent occupational postures, high pace of work, monotonous work, problems with workmates or
Phase II a Occupation: office workers OR 1.0; machine operators OR 1.4
(1.0-1.9); carpenters OR 1.5 (1.1-2.1) Physical exercise: maximum once per week OR 1.0; more than
once per week OR 1.3 (1.0-1.6) Smoking: non-smokers OR 1.0; smokers and ex-smokers OR 1.3
(1.0-1.7) History of other low back pain: no OR 1.0; mild OR 2.7 (1.7-4.2);
severe OR 4.5 (2.7-7.6) b In blue-collar workers
157
First author, Year of publication
Country Study design
Participants and setting Follow-up N (% female) Age range (mean) Participation %
Study outcome Case definition type Case definition ROB Risk factors considered
Phase of evidence Risk estimates (95% CI)
superiors, draft, cold, vibration, history of back accidents, history of lumbago or other LBP
b Psychological distress
and personality (subscales of the Middlesex Hospital Questionnaire and Maudsley Personality Inventory)
Hysteria (quadratic term): OR per one unit increase, 1.3 (1.1-1.7)
Mio, 2007 63 Japan Case-control
Cases: 823 patients of Japanese origin with LDH from hospitals in the Toyama, Tokyo, and Kyoto areas
Controls: 841 hospital patients of Japanese origin who received medical examinations
Excluded patients with spinal canal stenosis, spondylolisthesis, spondylosis, synovial cysts, spinal tumor, and trauma; and those with occupational and/or habitual risk factors, such as heavy manual laborers, occupational drivers, and heavy smokers
Follow-up: NA 1,664 (41% among cases,
63% among controls Cases:11-83y (36); controls:
13-87y (61) NR
Clinical LDH Clinical LDH was unilateral pain radiating
from the back along the femoral or sciatic nerve to the corresponding dermatome of the nerve root of >3 months duration, with positive MRI findings
Mod Cartilage collagen genes SNPs in 3 type XI
collagen genes: COL11A1, COL11A2, COL2A1
Phase I COL11A1 (rs1676486) SNP, c.4603C→T in exon 62, OR 1.4
(1.2-1.7)
158
First author, Year of publication
Country Study design
Participants and setting Follow-up N (% female) Age range (mean) Participation %
Study outcome Case definition type Case definition ROB Risk factors considered
Phase of evidence Risk estimates (95% CI)
Hirose, 2008 65 Japan Case-control
Cases: 847 patients of Japanese origin with LDH from 19 hospitals between 2001-2007
Controls: 896 Japanese persons from the same catchment area
Excluded patients with synovial cyst, spinal tumor, spondylosis, spondylolisthesis, trauma, and inflammatory disease
Follow-up: NA 1,743 (38%) Cases: NR (39); controls: NR
(62) NR
Clinical LDH Clinical LDH defined by meeting 3 criteria: i)
a history of unilateral pain radiating from the back along the femoral or sciatic nerve to the corresponding dermatome of the nerve root for >3 months; ii) diagnosis of LDH by MRI; and iii) treatment and monitoring for >1 year by an orthopaedic surgeon
Mod Intervertebral disc extracellular matrix protein genes
SNPs in the 2 thrombospondin genes: THBS1, THBS2
SNPs in 2 matrix metalloproteinases: MMP2, MMP9
Phase I THBS2 SNP (rs9406328), IVS10-8C→T, OR 1.4 (1.2-1.6) MMP9 SNP (rs17576), Q279R, OR 1.3 (1.1-1.5) Genotypes for combined effect of THBS2 (TT/TC/CC) and MMP9
(GG/GA/AA): CC and AA, OR 1.0; TC and GA, OR 1.8 (0.9-3.3); TT and GG, OR 3.0 (1.6-5.8)
Karasugi, 2009 66 Japan & Finland Case-control
Japan Cases: 862 patients of
Japanese origin with LDH from 20 hospitals between 2001-2007
Controls: 896 Japanese persons from the same catchment area
Excluded patients with synovial cyst, spinal tumor, spondylosis, spondylolisthesis, trauma, and inflammatory disease
1,758 (NR) Cases: NR (39); controls: NR
(62) 38% Finland Cases: 257 unrelated Finnish
patients with sciatica from the Oulu University Hospital catchment area
Japan Clinical LDH Clinical LDH was defined by meeting 3
criteria: i) a history of unilateral pain radiating from the back along the femoral or sciatic nerve to the corresponding dermatome of the nerve root for >3 months; ii) diagnosis of LDH by MRI; and iii) treatment and monitoring for >1 year by an orthopaedic surgeon
Finland Sciatica Clinical Disabling unilateral shooting band-like
sciatic pain referring from the back to below the knee (dermatomes L4, L5, and S1) from 3 weeks to 6 months, nonresponsive to nonsteroidal anti-inflammatory agents, with clinical presentation
Mod Human sickle tail (SKT) gene
Japan: 68 tag SNPs of the SKT gene
Finland: SKT SNP (rs16924573) only
Phase I LDH in Japan SKT SNP (rs16924573), OR 1.3 (1.1-1.6) SKT SNP (rs2285592), OR 1.3 (1.1-1.5) SKT SNP (rs17469499), OR 1.2 (1.0-1.5) Sciatica in Finland SKT SNP (rs16924573), OR 2.8 (1.1-7.2)
159
First author, Year of publication
Country Study design
Participants and setting Follow-up N (% female) Age range (mean) Participation %
Study outcome Case definition type Case definition ROB Risk factors considered
Phase of evidence Risk estimates (95% CI)
Controls: 249 unrelated Finnish persons from the same area
Follow-up: NA 506 (NR) NR (NR) NR
concordant with MRI findings
Cong, 2010 69 China Case-control
Cases: 70 male patients of Chinese Han origin with LDH, from the First Affiliated Hospital of China Medical University
Controls: 14 male spinal trauma patients who underwent surgery at the same hospital and 113 male healthy blood donors without LDH symptoms
Excluded persons with occupational and lifestyle risk factors for LDH, such as heavy manual labour, occupational driving and heavy smoking
Follow-up: NA 197 (0%) Cases: 14-41y (33); controls:
20-49y (38) NR
Clinical LDH Clinical Radicular pain with signs of positive
nerve root tension or neurologic deficit, a confirmatory imaging study (MRI and/or CT scan) indicating LDH corresponding to the symptoms and presence of symptoms for ≥6 weeks
Mod Aggrecan gene Aggrecan gene VNTR
polymorphism Expression of aggrecan
Phase I Aggrecan VNTR polymorphism, A25 allele OR 2.1 (1.1-4.0); A21
allele OR 11.8 (2.1-65.5); A29 allele OR 0.2 (0.1-0.8)
Noponen-Hietela, 2005 62
Finland Case-control
Cases: 155 unrelated Finnish patients with sciatica from Oulu University Hospital area, 1997-1998
Controls: 179 unrelated University of Oulu employees and students (all Finnish), no information on possible MSK disorders
Sciatica Clinical Disabling unilateral shooting band-like
sciatic pain referring from the back to below the knee (dermatomes L4, L5, and S1) from 3 weeks to 6 months, nonresponsive to nonsteroidal anti-inflammatory agents, with clinical presentation
Mod Inflammatory mediator genes
Sequence variations (mutations) in 3 interleukin (IL) genes and 1 tumor necrosis factor (TNF) gene: IL1A, IL1B, IL6, TNFA
16 SNPs in 10 candidate
Phase I Genotypes of the IL6 SNP, T15A in exon 5, (AA and AT v TT)
OR 4.4 (1.2-15.7) Genotypes (haplotype pairs) of IL6, (GGGA/GGGA or
GGGA/other v other/other) OR 5.4 (1.5-19.2) Other findings No mutations identified in the IL1A, IL1B, IL6 or TNFA genes
were associated with sciatica
160
First author, Year of publication
Country Study design
Participants and setting Follow-up N (% female) Age range (mean) Participation %
Study outcome Case definition type Case definition ROB Risk factors considered
Phase of evidence Risk estimates (95% CI)
Excluded patients with chronic pain syndromes and neurological disorders
Follow-up: NA 334 (39% among cases, 69%
among controls) Cases: 19-78y (44); controls:
20-69y (39) NR
concordant with MRI findings cytokine genes: IL1A, IL1B, IL1 receptor antagonist (IL1RN), IL2, IL4, IL4R, IL6, IL10, TNFA, and interferon γ (IFNG)
Virtanen, 2007 64 Finland & China Case-control
Finland Cases: 243 unrelated Finnish
patients with sciatica from Oulu University Hospital area
Controls: 259 unrelated Finnish persons from the same area
502 (45%) NR (NR) NR China Irrelevant outcome
Sciatica Clinical Disabling unilateral shooting band-like
sciatic pain referring from the back to below the knee (dermatomes L4, L5, and S1) from 3 weeks to 6 months, nonresponsive to nonsteroidal anti-inflammatory agents, with clinical presentation concordant with MRI findings
Mod Cartilage intermediate layer protein (CILP) gene
Functional SNP (rs2073711), +1184T→C, in exon 8 of the CILP gene
Other SNPs in the CILP gene
Phase I CILP SNP (rs2073711), +1184T→C in exon 8, OR 1.4 (1.0-1.9)
Mundt, 1993 58,59 USA Case-control
Cases: 297 patients with confirmed and unconfirmed LDH between 1986-1988, from 38 orthopaedic and neurosurgical practices and 5 hospitals in Massachusetts, New Jersey, and New York
Controls: 287 patients admitted to the same hospital service or practice for a condition not related to the back or neck, matched on age, sex, geographic location, and
Combined clinical, hospitalized and surgical LDH
Clinical, hospital and surgical Confirmed case was herniation,
prolapse, rupture, protrusion, extrusion, extradural defect, or free fragment noted on the surgical report, myelogram, CT scan, or MRI as reported in the medical record
Unconfirmed case was probable or possible herniation based on signs and symptoms consistent with herniation
Probable case was pain, numbness or tingling radiating to the hip,
Mod a In 2y before index date Occupational activities,
use of motor vehicles, whole body and arm vibration, riding in planes, trains, buses, subways and motorcycles, non-occupational lifting (inanimate objects and children), carrying, stretching, bending, shovelling, pregnancy history, smoking
Lifetime history Occupational lifting,
Phase II All cases Smoking (average 10cigs/d): OR per one unit increase, 1.4 (1.2-
1.6) Shovelling: no OR 1.0; yes, 5-15 times OR 0.8 (0.6-1.2); yes, >15
times OR 0.7 (0.4-1.2) Repeated bending while doing off the job activities: no OR 1.0; yes, ≥2 days per week OR 1.4 (0.9-2.2)
Non-occupational lifting ≥11.3kg inanimate objects: no OR 1.0; yes, ≥1/wk for 6mo OR 1.1 (0.8-1.6); yes, knees bent, back straight OR 0.6 (0.4-0.9); yes, knees bent, back bent OR 1.2 (0.6-2.3); yes, knees straight, back bent OR 2.3 (1.1-4.7); yes, starting and ending the lift at the waist OR 2.0 (1.0-4.1)
Used free weights ≥10 times: no OR 1.0; yes OR 0.9 (0.6-1.4); yes, warmed up before workout <½ the time OR 0.5 (0.3-1.1); yes, warmed up before workout ≥½ the time OR 1.1 (0.7-1.7)
161
First author, Year of publication
Country Study design
Participants and setting Follow-up N (% female) Age range (mean) Participation %
Study outcome Case definition type Case definition ROB Risk factors considered
Phase of evidence Risk estimates (95% CI)
source of medical care Excluded those who had
experienced symptoms for >1 year before index date; those who had previous disc surgery >1 year before index date; those who had other conditions of the back or neck; and those who had experienced activity limitation >1y before index date from back, leg, neck or arm pain of ≥4 weeks duration
Follow-up: NA 585 (41%) 20-64y (NR) 76-79%
buttock, thigh or below the knee in a pattern consistent with nerve root impingement by the disc, with worsening of the symptoms with coughing, stretching the leg, or straining while moving the bowels
Possible case differed from probable LDH in that symptoms did not worsen with cough, stretch or strain; or symptoms referred to the thigh, and worsened with cough, stretch or strain, or positive straight leg raising indicated in the medical record; or symptoms were in the lower leg only, and worsened with cough, stretch or strain, or positive straight leg raising
bending, and twisting b In 2y before index date Participation in specific
sports (baseball or softball, golf, bowling, swimming, diving from a board at a height of at least 3 feet, jogging, aerobics, racquet sports), use of free weights, use of weight lifting equipment, warming up before workout
Used weight lifting equipment ≥10 times: no OR 1.0; yes OR 1.0 (0.7-1.5); yes, warmed up before workout <½ the time OR 0.5 (0.2-1.2); yes, warmed up before workout ≥½ the time OR 1.1 (0.7-1.7)
Confirmed cases Non-occupational lifting ≥11.3kg inanimate objects: no OR 1.0; yes, ≥1/wk for 6mo OR 1.6 (1.0-2.5); yes, knees bent, back straight OR 0.7 (0.4-1.3); yes, knees bent, back bent OR 2.0 (0.8-4.9); yes, knees straight, back bent OR 4.0 (1.6-10.0); yes, starting and ending the lift at the waist OR 2.5 (1.1-6.0); yes, arms extended when lift started ≥½ the time OR 1.9 (1.0-3.5); yes, twisted while lifting ≥½ the time OR 1.9 (0.9-3.9)
Kelsey, 1984 56,57 USA Case-control
Cases: 325 patients with surgical, probable or possible LDH after having lumbar spine x-rays or myelograms taken in 3 area hospitals, 1 neurosurgical practice, or 2 orthopaedic practices between 1979-1981, from the New Haven and Hartford, Connecticut areas
Controls: 241 patients admitted to the same hospital service or practice for a condition not related to the spine, matched on age, sex, and medical setting
Excluded those who had
Combined clinical, hospitalized and surgical LDH
Clinical, hospital and surgical Surgical cases were those in which: i)
the hospital chart indicated that the surgeon saw a herniated disc during surgery (descriptions included ruptured, free fragments, herniated, prolapsed, bulging and extruded, but not disc degeneration without evidence of nerve root involvement); and ii) the patient reported pain distributed along the sciatic nerve; and iii) the patient had a positive straight leg raising test, and/or increased pain in the low back or along the sciatic nerve when stretching or extending the leg from a sitting position, and/or
Mod a Smoking, motor vehicle type, size or model, driving pattern characteristics (local roads, highways, bucket seats, regular seats, driver, passenger, automatic transmission, manual transmission), occupational lifting while twisting body, race, education, marital status, residence, height, weight, number of pregnancies, number of children, respiratory symptoms (chronic phlegm, chronic bronchitis), vibration, frequency of wearing
Phase II All cases combined Smoking: never smoked OR 1.0; former smoker (not in the past
year) OR: 1.0 (0.6-1.7); current smoker (in past year) OR 1.7 (1.0-2.5)
Smoking (average 10cigs/d): OR per one unit increase, 1.2 (1.0-1.4)
Motor vehicle type (driven during the past 5y period): only Japanese or Swedish OR 1.0; other OR 3.0 (1.1-8.3)
Driving non-Japanese or non-Swedish car (average 5h/wk): OR per one unit increase, 1.3 (1.1-1.6)
Married v not married (in women) OR 2.2 (1.0-5.1) Rural residence v urban or suburban residence OR 1.7 (0.9-3.3) Lifting objects >11.3kg, >25 times per day while twisting body:
knees bent (yes v no) OR 1.9 (0.8-4.8); knees not bent (yes v no) OR 7.2 (2.0-25.8);
Lifting objects >11.3kg while twisting body (with and without knees bent): no lifting while twisting body OR 1.0; lifting while twisting body, knees bent OR 2.7 (0.9-7.9); lifting while twisting body, knees almost straight OR 6.1 (1.3-27.9)
162
First author, Year of publication
Country Study design
Participants and setting Follow-up N (% female) Age range (mean) Participation %
Study outcome Case definition type Case definition ROB Risk factors considered
Phase of evidence Risk estimates (95% CI)
previous prolapsed lumbar or cervical disc, or previous neck, back, leg or arm problems that caused activity restriction for more than 4 consecutive weeks, and all persons who had experienced symptoms for ≥1year prior to study entry
Follow-up: NA 566 (45%) 20-64y (NR) 72-79%
increased pain along the sciatic nerve when coughing, sneezing, or straining at stools
Probable cases were similar to surgical cases except that a herniation need not have been observed at surgery; and included cases in which the sciatic pain was felt in both the thigh and lower leg, and cases in which there was sciatic pain in part of the leg and numbness in another part
Possible cases differed from probable cases in that the sciatic pain was only in the thigh or the lower leg but not in both; or if the leg was numb but straight leg raising increased LBP
shoes with high heels, participation in baseball, golf, bowling, swimming, diving from a board, tennis, bicycling, or jogging
b Occupational: frequency
of lifting, frequency of carrying, frequency of twisting at waist, lifting while twisting body, holding while twisting body, bending knees while lifting and twisting
Frequency of lifting objects >11.3kg: not at all OR 1.0; <5 times/d OR 1.2 (0.7-2.0); 5-25 times/d OR 1.3 (0.7-2.5); >25 times/d OR 3.5 (1.5-8.5)
Frequency of carrying objects >11.3kg: not at all OR 1.0; <5 times/d OR 1.0 (0.6-1.9); 5-25 times/d OR 2.1 (1.0-4.3); >25 times/d OR 2.7 (1.2-5.8)
Lifting objects >11.3kg while twisting body: never or small amount OR 1.0; moderate amount OR 2.5 (0.9-6.8); large amount OR 3.1 (1.3-7.5)
Kelsey, 1975 51-55 USA Case-control
Cases: 223 patients with surgical, probable or possible LDH after having lumbar spine x-rays taken in 3 area hospitals or in the office of 2 private radiologists between 1971-1973, from New Haven, Connecticut
Controls (matched): 217 patients admitted to the same hospital service or radiologist office for a condition not related to the spine, matched on age, sex, and medical setting
Controls (unmatched): 494 patients who had lumbar spine x-rays and were not classified as cases
Excluded those who had
Combined clinical, hospitalized and surgical LDH
Clinical, hospital and surgical Surgical cases were those in which: i)
the hospital chart indicated that the surgeon saw a herniated disc during surgery (descriptions included ruptured, free fragments, herniated, prolapsed, bulging and extruded, but not disc degeneration without evidence of nerve root involvement); and ii) the patient reported pain distributed along the sciatic nerve; and iii) the patient had a positive straight leg raising test, and/or increased pain in the low back or along the sciatic nerve when stretching or extending the leg from a sitting position, and/or increased pain along the sciatic
Mod a,b Sex, age, race, social class, residence, marital status, number of children
b Month of onset of
symptoms, height, weight, body bulk (BMI for men, weight/height for women), weight gain/loss in previous year, respiratory symptoms (chronic phlegm, chronic bronchitis), smoking, physical activity and sports, stressful life events in the previous year
b,c Occupational:
Phase II All cases combined Sex: NR; among surgical cases, male to female ratio 2.1 (p<0.001);
among probable and possible cases, male to female ratio 1.1 (p>0.10)
Age: NR; LDH most common in the 30-49y age groups Residence (matched analysis): urban OR 1.0; suburban OR 1.5
(1.0-2.4) Sedentary jobs, age ≥35y (matched analysis) OR 2.4 (1.3-4.7) Jobs requiring driving (matched analysis) OR 2.8 (1.2-7.1) Truck driving job (matched analysis) OR 4.7 (1.3-25.3) Driving other than on job, drivers v nondrivers: (in men; matched
analysis) OR 2.7 (1.0-8.3); (in women; matched analysis) OR 1.9 (0.9-4.1); (in both men and women; matched analysis) OR 2.2 (1.2-3.9)
Chronic phlegm (in women; matched analysis) OR 2.8 (1.1-8.8) Chronic bronchitis (in women; matched analysis) OR 3.7 (1.0-20.5) Physical activity (matched analysis) OR 0.4 (0.2-1.0) Pregnancy: NR; cases had more pregnancies resulting in live births
than other women of their age Other findings
163
First author, Year of publication
Country Study design
Participants and setting Follow-up N (% female) Age range (mean) Participation %
Study outcome Case definition type Case definition ROB Risk factors considered
Phase of evidence Risk estimates (95% CI)
previous LDH or other serious back problems, and all persons who had experienced symptoms for ≥1year prior to study entry
Follow-up: NA 934 (41%) 20-64y (39 among cases) 78% a, b, c, d n=934; 41% female e n=407; 100% female
nerve when coughing, sneezing, or straining at stools
Probable cases were similar to surgical cases except that a herniation need not have been observed at surgery; and included cases in which the sciatic pain was felt in both the thigh and lower leg, and cases in which there was sciatic pain in part of the leg and numbness in another part
Possible cases differed from probable cases in that the sciatic pain was only in the thigh or the lower leg but not in both; or if the leg was numb but straight leg raising increased LBP
sedentary jobs (sitting ≥half the time), jobs requiring driving (sitting ≥half the time in a motor vehicle; men only), truck driving (men only), jobs involving any lifting, jobs involving any pushing, jobs involving any pulling, jobs involving any carrying, heavy lifting
b,d Driving: jobs requiring
driving (sitting ≥half the time in a motor vehicle; men only), truck driving (men only), driving other than on job
b,e Pregnancy: average
number of pregnancies, pregnancies resulting in live births, pregnancies not resulting in live births
Social class: NR; some indication of an association between higher social class and LDH in women, but no association in men
Sports: NR; some tendency for cases to play sports rarely and controls more frequently
Lifting on the job: no association found with heavy lifting or with frequent lifting
Hrubec, 1975 70 Risk
Cases: 1,132 first admission records to Army hospitals for LDH in 1944-1945
Controls: 1,095 records of Army National Service Life Insurance policyholders individually matched on age and period of military service
Follow-up: NA 1,095 case-control pairs (0%) 18-56y (NR)
Hospitalized LDH Hospital No explicit case definition provided Records of first admission for LDH
identified using punch cards from the Office of the Surgeon General
High Height, weight, posture, body frame, place of birth, place of residence, education, occupation, marital status, physical defects, military occupation specialty assignments, overseas service, blood type, religion, rank, combat credit (medals or decorations)
Phase II All statistically significant associations: Height (cm): 152-164 OR 0.4; 165-181 OR 0.9; 182-203 OR 2.6 Weight (kg): 45-58 OR 0.5; 59-81 OR 1.0; 82-136 OR 1.6 Craftsman, foreman or kindred occupation: OR 1.5 Clerical or kindred occupation: OR 0.6 Married or re-married at induction: OR 1.4 Place of birth <2500 population or rural: OR 1.6 Rural free delivery address: OR 1.6 Good posture at induction: OR 1.3 Defects at induction relevant to HNP affecting back or legs: OR
3.5
164
First author, Year of publication
Country Study design
Participants and setting Follow-up N (% female) Age range (mean) Participation %
Study outcome Case definition type Case definition ROB Risk factors considered
Phase of evidence Risk estimates (95% CI)
97% Heavy frame at induction: OR 1.5 Military occupation specialty, ground combat: OR 1.6 Combat credit, 2+ battle stars: OR 0.7 Rank, staff sergeant or sergeant: OR 1.3
Netterstrøm, 1989 48
Denmark Cohort
All full-time male bus drivers employed on April 1, 1978 by 3 urban bus companies followed up from 1978-1984 for hospitalized LDH in the Danish National Patient Register
Follow-up: 7y N=2,465 (0%) 20-69y (NR) 100%
Hospitalized LDH Hospital ICD discharge diagnosis codes for
LDH (725.10-725.11)
High Occupation Phase I Occupation: hospitalized LDH in male urban bus drivers compared
to all Danish men: SMR 137 (105-176)
An, 1994 71 Risk
Cases: 163 consecutive surgical patients with LDH at Pennsylvania Hospital between 1987-1988
Controls: 205 inpatients without lumbar disc disease from medical and surgical services at the same hospital, matched on sex and age
Follow-up: NA 368 (39%) 16-78y (45) NR
LDH surgery Surgical Confirmed LDH surgery case defined
by meeting 3 criteria: i) prolapsed lumbar intervertebral disc was the primary diagnosis; ii) all were symptomatic and admitted for surgical management (100% surgically confirmed); iii) all had radiographic evidence (myelogram, CT scan or MRI) of intervertebral herniation
High Smoking Phase I Smoking (current and ex-smokers v non-smokers) OR 2.2 Smoking (current smokers v non-smokers) OR 3.0 Other findings Incorrect analysis for a matched CC study; compares proportion of
current and ex-smokers among cases v controls instead of discordant pairs, results not sex- and age-adjusted
Saftic, 2006 73 Risk
Cases: 67 adults from 9 villages on the Croatian islands of Rab, Vis, Lastovo, and Mljet, who had a positive history of LDH surgery
Controls: 268 adults matched on age, gender, and village of residence/immigrant
LDH surgery Surgical LDH surgery was a positive history of
surgery due to lumbar intervertebral disc herniation at the level of L4/L5 or L5/S1 based on examination of medical histories and medical records
High BMI, occupation type, intensity of physical labor at work, intensity of physical labour at home, smoking index, claudication index, self-assessed limitation in physical activity, level of education,
Phase I BMI (kg/m2): ≥25.7 v <25.7 OR 2.8 (1.1-4.5) Occupation: involving hard physical labour (agriculture workers,
soldiers, construction workers, mechanics or fishermen) v involving sitting or standing (clerks, lawyers, economists, tailors, waiters, cooks, salespersons, teachers, policemen, electricians, and housewives) OR 1.9 (1.1-3.8)
Intensity of physical labour at work: hard v sitting, easy or moderate) OR 2.9 (1.1-4.8)
165
First author, Year of publication
Country Study design
Participants and setting Follow-up N (% female) Age range (mean) Participation %
Study outcome Case definition type Case definition ROB Risk factors considered
Phase of evidence Risk estimates (95% CI)
status on a 4:1 basis Excluded persons who
underwent surgery of the lower spine due to other causes, such as degenerative changes or stenosis
Follow-up: NA 365 (NR) ≥18y (NR) NR
socioeconomic status, and family history of lower spine surgery
Positive family history of spine surgery (having a parent who underwent spine surgery): yes v no OR 4.0 (1.9-6.1)
Lee, 2006 74 Risk
Cases: 119 herniated lumbar disc levels in 111 adult patients who underwent LDH surgery between 2000-2002
Controls: 82 normal discs levels adjacent to the herniated levels in the same adult patients
Excluded patients who had not had pre-operative CT scans
Follow-up: NA 201 adult disc levels (37%) 40-49y (NR) NR
LDH surgery Surgical Herniated disc level noted during
open discectomy or percutaneous endoscopic discectomy for radiculopathy
High Facet joint tropism Phase I NR; degree of facet tropism at the L4-L5 level was significantly
greater in herniated discs than in normal discs (6.9±5.5° v 3.6±3.0°, p<0.001)
No significant difference found between herniated and normal discs at the L3-L4 and L5-S1 levels
Kunakornsawat, 2007 75
Risk
Cases: 34 herniated lumbar disc levels in 34 adult patients who underwent discectomy at Lerdsin Hospital between 2001-2003
Controls: 34 normal discs levels adjacent to the herniated levels in the same patients
Excluded MRIs of persons
LDH surgery Surgical Patient aged <45y who underwent
conventional discectomy for LDH (single protrusion or extrusion disc at L3-L4, L4-L5 or L5-S1), without previous surgery, and with symmetrical MRI scan of each disc level
High Facet joint tropism Phase I Facet tropism at L4-L5 OR 1.8 (0.3-10) Facet tropism at L5-S1 OR 1.7 (0.3-10) Other findings No association found between facet tropism and LDH in all levels
166
First author, Year of publication
Country Study design
Participants and setting Follow-up N (% female) Age range (mean) Participation %
Study outcome Case definition type Case definition ROB Risk factors considered
Phase of evidence Risk estimates (95% CI)
who had associated spinal abnormality such as spina bifida, unilateral sacralization, enlarged transverse processes, or spondylolisthesis
Follow-up: NA 68 disc levels (35%) 23-45y (34) NR
Bongers, 1988 47 Netherlands Cohort
All male crane workers and floor workers employed in the same departments at a steel company on January 1, 1975 and those hired up to 1979, followed up for incident disability pensions due to LDH until December 1984
Follow-up: 10y N=1,405 (0%) <25-60y+ (NR) 71%
Disability pension due to LDH Clinical Disability pension due to
displacement of intervertebral disc defined as ICD-9 diagnosis code 722.2, assigned by social insurance physician after evaluation of the social insurance medical records
High Whole-body vibration Phase II Disability pension due to LDH in crane operators exposed to
whole-body vibration v floor workers, IDR 2.5 (90% CI, 1.2-12.5)
Zhang, 2009 76 Risk
Cases: 2,010 patients with LDH in orthopaedic departments at 3 hospitals between 2005-2007
Controls: 2,170 randomly selected patients (in-patients or medical exam) without current back/sciatic pain, or back/sciatic pain for >1 month ever, matched on race, gender, age, and living area
Excluded patients with lumbar spinal stenosis,
Clinical LDH Clinical No explicit case definition provided LDH diagnosis evaluated by ≥2
orthopaedic experts in terms of patient’s symptoms, signs and imaging examination (MRI and/or CT scan) among patients presenting with back-leg pain and typical sciatica
High Sex, age, height, weight, smoking, drinking, family history of lumbar disc herniation, bed characteristics (hard, soft), educational background (primary school, secondary school, high school, university), physical exercise, occupational lumbar load (quite light, light, medium, heavy), occupational character (non-manual, half non-
Phase II Among all persons Family history OR 3.6 (1.9-6.5) Lumbar load OR 2.1 (1.7-2.6) Hard-working OR 1.8 (1.1-2.5) Educational background OR 0.8 (0.6-1.0) Physical exercise OR 0.4 (0.2-0.8) Bed characteristics (hard v soft) OR 0.4 (0.2-0.6) Other findings Model omitted sex, age, BMI, occupational character, drinking,
smoking, and vocational activities
167
First author, Year of publication
Country Study design
Participants and setting Follow-up N (% female) Age range (mean) Participation %
Study outcome Case definition type Case definition ROB Risk factors considered
Phase of evidence Risk estimates (95% CI)
spinal congenital dysplasia, intraspinal tumor, spinal instability from trauma, scoliosis, and spondylolisthesis
Follow-up: NA 4,180 (40%) <30-50y+ (46) NR
manual/half manual, manual), psychosocial factors at work (monotonous, boring, time urgency, too much responsibility, life pressure, hard-working)
Chibnall, 2006 72 Risk
African American and non-Hispanic white workers’ compensation claimants who filed low back injury claims in St. Louis City or Kansas City, Missouri, and whose claims were settled between January 1, 2001, and June 1, 2002
Follow-up: NA 2,934 (38%) 18-55y (NR) 50%
Clinical and surgical LDH Clinical and surgical No explicit case definition provided From a series of medical diagnosis
questions, participants were categorized into 2 diagnosis groups: LDH v low back sprain/strain/pain (regional backache)
For claimants with a LDH diagnosis, surgery was classified as “no” v “yes”
High Sex, age, race, socioeconomic status, presence of lower extremity pain, lumbar degeneration, legal representation, legal representation due to dissatisfaction with medical treatment, current work status
Phase II LDH diagnosis (v regional backache diagnosis) Sex: women OR 1.0; men OR 1.7 (1.4-2.2) Age (y): OR per one unit increase, 1.2 (1.1-1.3) Race: white OR 1.0; African American OR 0.6 (0.4-0.7) Lower extremity pain: no OR 1.0; yes OR 3.3 (2.3-4.9) Lumbar degeneration: no OR 1.0; yes OR 2.1 (1.6-2.6) Legal representation: no OR 1.0; yes OR 1.0 (0.7-1.4) Legal representation due to dissatisfaction with medical treatment:
no OR 1.0; yes OR 0.6 (0.5-0.8) Surgery among those with LDH diagnosis Race: white OR 1.0; African American OR 0.3 (0.2-0.5) Socioeconomic status: OR per one unit increase, 1.1 (1.0-1.4) Lower extremity pain: no OR 1.0; yes OR 2.9 (1.2-6.4) Legal representation: no OR 1.0; yes OR 1.4 (0.9-2.3) Legal representation due to dissatisfaction with medical treatment:
no OR 1.0; yes OR 0.6 (0.4-0.8) Abbreviations: BMI, body mass index; cig, cigarettes; cm, centimeter; CT, computerized tomography; d, day; h, hour; ICD, international classification of diseases; kg, kilogram; LBP, low back pain;
LDH, lumbar disc herniation; m, meter; mo, month; MRI, magnetic resonance imaging; MSK, musculoskeletal; N, study size; NA, not applicable; Nh, Newton-hours; NR, not reported; OR, odds ratio; ROB, risk of bias; RR, relative risk; SMR, standardized morbidity ratio; SNP, single nucleotide polymorphism; v, versus; VNTR, variable number of tandem repeats; wk, week; y, year
168
Appendix E. Belief elicitation questionnaire (Chapter 3)
!"#$%&'(&& &
Version date: 2011-03-25 Belief Elicitation Questionnaire – Final 1&
Belief Elicitation – Example Scenario
Treatment: No health care Outcome: Tension-type headache E.1. Among an average group of 1,000 newly diagnosed acute low back pain patients
receiving no health care treatment, how many do you believe will go on to
develop tension-type headache within 2 months of diagnosis?
Estimate: 20 patients
E.2. There may be some uncertainty around your estimate. What do you believe is a
plausible range for the number of patients that will develop tension-type headache
within 2 months of diagnosis?
Lower bound (L): 10 patients Upper bound (U): 25 patients
E.3. You have been given 20 virtual chips. Each chip represents 5% probability, for a
total of 100% probability. Placing the chips in the bins, indicate the weight of belief
for your estimates of the number of patients developing tension-type headache
within 2 months of diagnosis.
Do you have any questions? Please review the shape and distribution of your answer. Does this reflect
what you truly believe? If not, please feel free to revise your placement of chips.
Do you have any questions?
169
!"#$%&'(&& &
Version date: 2011-03-25 Belief Elicitation Questionnaire – Final 2&
Belief Elicitation Questionnaire
Scenario 1 Treatment: Chiropractic care Outcome: Acute lumbar disc herniation 1. Among an average group of 1,000 newly diagnosed acute low back pain patients
not treated with chiropractic care, how many do you believe will go on to develop acute lumbar disc herniation within 2 months of diagnosis? Estimate:
2. Among an average group of 1,000 newly diagnosed acute low back pain patients
treated with chiropractic care, how many do you believe will go on to develop acute lumbar disc herniation within 2 months of diagnosis? Estimate:
3. There may be some uncertainty around your estimate. What do you believe is a
plausible range for the number of patients treated with chiropractic care that will develop acute lumbar spine disc herniation within 2 months of diagnosis? Lower bound (L): Upper bound (U):
4. You have been given 20 virtual chips. Each chip represents 5% probability, for a
total of 100% probability. Placing the chips in the bins, indicate the weight of belief for your estimates of the number of patients treated with chiropractic care that will develop acute lumbar disc herniation within 2 months of diagnosis.
Please review the shape and distribution of your answer. Does this reflect what you truly believe? If not, please feel free to revise your placement of chips. 5. Compared to primary medical care, what overall effect do you believe
chiropractic care has on the risk of developing acute lumbar disc herniation?
! Decreases the risk ! No effect on the risk ! Increases the risk
170
!"#$%&'(&& &
Version date: 2011-03-25 Belief Elicitation Questionnaire – Final 3&
Scenario 2 Treatment: Chiropractic care Outcome: Acute severe lumbar disc herniation that is surgically managed 6. Among an average group of 1,000 newly diagnosed acute low back pain patients
not treated with chiropractic care, how many do you believe will go on, within 2 months, to develop acute severe lumbar disc herniation that is surgically managed? Estimate:
7. Among an average group of 1,000 newly diagnosed acute low back pain patients
treated with chiropractic care, how many do you believe will go on, within 2 months, to develop acute severe lumbar disc herniation that is surgically managed? Estimate:
8. There may be some uncertainty around your estimate. What do you believe is a
plausible range for the number of patients treated with chiropractic care that will go on, within 2 months, to develop acute severe lumbar disc herniation that is surgically managed? Lower bound (L): Upper bound (U):
9. You have been given 20 virtual chips. Each chip represents 5% probability, for a
total of 100% probability. Placing the chips in the bins, indicate the weight of belief for your estimates of the number of patients treated with chiropractic care that will go on, within 2 months, to develop acute severe lumbar disc herniation that is surgically managed.
Please review the shape and distribution of your answer. Does this reflect what you truly believe? If not, please feel free to revise your placement of chips. 10. Compared to primary medical care, what overall effect do you believe
chiropractic care has on the risk of developing acute severe lumbar disc herniation that is surgically managed?
! Decreases the risk ! No effect on the risk ! Increases the risk
171
!"#$%&'(&& &
Version date: 2011-03-25 Belief Elicitation Questionnaire – Final 4&
Demographic information 11. Sex: ! Male ! Female 12. Age group (years): ! 21-30 ! 31-40 ! 41-50 ! 51-60 ! !61 13. Clinical background: ! Primary care medicine
! Chiropractic ! Spine surgery or Neurosurgery
14. How many years have you been treating low back pain patients? 15. How many new low back pain patients do you see per year? 16. Have you ever had formal statistical training? ! Yes ! No
Highest level of statistical training: ! None ! Graduate ! Undergraduate ! Postgraduate 17. Do you refer for chiropractic care in your practice to treat low back pain patients? ! Yes ! No Rationale: 18. Which source(s) of information do you use in formulating your beliefs about the
effect of chiropractic care on the risk of developing acute lumbar spine disc herniation? Check all that apply.
! Research evidence ! Clinical experience ! News media ! Other:
Comment:
19. In your clinical practice, have you ever seen a patient whose disc herniation was
due to chiropractic spinal manipulation? ! Yes ! No 20. Please feel free to provide any additional comments:
172
Appendix F. Belief elicitation script (Chapter 3)
Version date: 2011-03-25 Elicitation Script – Final 1!
Standardized Belief Elicitation Interview Script
Thank you for agreeing to meet with me. The purpose of this study is to document clinicians’ beliefs about the effect of chiropractic care on the risk of developing two outcomes: 1) acute lumbar disc herniation (aLDH); and 2) acute severe lumbar disc herniation that is surgically managed (aLDH-surgery). I am a PhD candidate in epidemiology at the University of Toronto. This study will form part of my PhD thesis. This interactive questionnaire will take approximately 15 minutes to complete. If you have any questions, I am here to assist you. Participation in this study is voluntary. Your responses will be kept anonymous, and will only be used for research purposes. You have been assigned a unique identifier code that will be written at the top of your questionnaire. The code will be kept confidential. You will not be named or identified in any reports, publications, or presentations that may come from this study. If you do not treat a low back pain patient population, or do not wish to participate in this study, please feel free to decline. Are you agreeable to participation in this study? If no—thank the participant for their time. If yes—proceed to the example. Before we begin the questionnaire, I have a worked example of what will be asked of you. For each scenario, you will be asked to consider an average group of 1,000 newly diagnosed acute low back pain patients without current lumbar disc herniation. Acute lumbar disc herniation is defined as acute onset of lumbar radiculopathy due to disc herniation. The health care treatment options under consideration include:
No health care No additional health care after LBP diagnosis Primary medical care Course of NSAIDs/painkillers prescribed by family physician Chiropractic care Course of spinal manipulation prescribed by chiropractor This sheet lists these patient and treatment characteristics so that they will be easy to recall. Provide sheet. In the case of the example, a hypothetical participant is asked about his beliefs about the development of tension-type headache (TTH) in acute low back pain patients that receive no health care treatment. The questions and process in this worked example are similar to the questionnaire you will be asked to complete. Do you have any questions? If yes—answer any questions. Once all questions have been addressed, give the participant the example questionnaire. If no—give the participant the example questionnaire and read each question aloud.
173
Version date: 2011-03-25 Elicitation Script – Final 2!
Let’s start with the example: E.1. Among an average group of 1,000 newly diagnosed acute low back pain patients receiving no
health care treatment, how many do you believe will go on to develop TTH within 2 months of diagnosis? Estimate: 20 patients In this example, the participant thought that the number of patients that would go on to develop TTH within 2 months was 20; so he gave an estimate of 20 patients.
E.2. There may be some uncertainty around your estimate. What do you believe is a plausible range for the number of patients that will develop TTH within 2 months of diagnosis? Lower bound (L): 10 patients Upper bound (U): 25 patients Here, the participant thought that a plausible range for the number of patients that would go on to develop TTH within 2 months could be as low as 10 and as high as 25; so he gave a lower bound of 10 and an upper bound of 25.
E.3. In Question 2, the participant indicated his range of plausible values for the number of patients developing TTH within 2 months. However, he may believe that some estimates along that range would be more likely than others. Using virtual chips, each representing 5% probability, he placed the chips in the bins to indicate his weight of belief for the estimates along his range. In this example, he had more belief in an estimate of 20 and less belief in estimates of 10 and 25. Do you have any questions? If yes—answer any questions. Once all questions have been addressed, continue example. If no—continue example. Once all the chips have been placed, you can see that it creates a shape and distribution; we’ll call this a “belief curve”. At this point, the participant is asked to take a moment and check if the shape and distribution of the curve is a fair representation of what he truly believes. If he feels that the curve does not reflect what he believes, he is able to rearrange his chips until he is satisfied that his true belief is represented.
Do you have any questions? If yes—answer any questions. Once all questions have been addressed, proceed to the elicitation questionnaire. If no—proceed to the elicitation questionnaire.
174
Version date: 2011-03-25 Elicitation Script – Final 3!
Standardized Interview Script – Belief Elicitation Scenarios You have been shown an example of this belief elicitation exercise using a worked example of no health care treatment and TTH. The structure of the questionnaire I am giving you now is very similar. You will be asked to consider 2 scenarios: 1) chiropractic care and the risk of developing aLDH, and 2) chiropractic care and the risk of developing aLDH-surgery. When answering these questions, remember to keep in mind the patient and treatment profiles listed on the sheet that you have in front of you: Patient Profile: An average group of 1,000 newly diagnosed acute low back pain patients without current lumbar disc herniation. Outcome Profile: Acute onset of lumbar radiculopathy due to lumbar disc herniation Treatment Profiles:
No health care No additional health care after LBP diagnosis Primary medical care Course of NSAIDs/painkillers prescribed by family physician Chiropractic care Course of spinal manipulation prescribed by chiropractor I am here to assist you throughout the elicitation. Do you have any questions? If yes—answer any questions. Once all questions have been addressed, give the participant the elicitation questionnaire and read each question aloud. If no— give the participant the elicitation questionnaire and read each question aloud. Treatment: Chiropractic care Outcome: Acute lumbar disc herniation 1. Among an average group of 1,000 newly diagnosed acute low back pain patients not treated
with chiropractic care, how many do you believe will go on to develop aLDH within 2 months of diagnosis? Estimate: Please provide your estimate of the number of patients developing aLDH within 2 months of diagnosis.
2. Among an average group of 1,000 newly diagnosed acute low back pain patients treated with
chiropractic care, how many do you believe will go on to develop aLDH within 2 months of diagnosis? Estimate: Please provide your estimate of the number of patients developing aLDH within 2 months of diagnosis.
3. There may be some uncertainty around your estimate. What do you believe is a plausible
range for the number of patients treated with chiropractic care that will develop aLDH within 2 months of diagnosis? Lower bound (L): Upper bound (U): Elicit the range of plausible values for the number of patients developing aLDH within 2 months. Identify a
175
Version date: 2011-03-25 Elicitation Script – Final 4!
range such that it is extremely unlikely (but not necessarily impossible) that the point estimate lies outside.
4. You have indicated that the lower bound of the range is L and the upper bound is U. I will divide the range from L to U into equal-width bins. I have placed 1 virtual chip, representing 5% probability, in each of the L and U bins. I will now give you 18 more chips, each representing 5% probability, for a total of 100%. Using these chips, please indicate the weight of belief for your estimates of the number of patients treated with chiropractic care that will develop aLDH within 2 months of diagnosis. Do you have any questions? If yes—answer any questions. Once all questions have been addressed, continue scenario. If no—continue scenario. Once all the chips have been placed and distribution fitted: Please review the shape and distribution of your answer. Does this reflect what you truly believe? If not, please feel free to revise your placement of chips until you are satisfied that the resulting distribution reflects your true belief. Do you have any questions? If yes—answer any questions. Once all questions have been addressed, proceed to question 5. If no— proceed to question 4.
5. Compared to primary medical care, what overall effect do you believe chiropractic care has on the risk of developing aLDH? Decreases the risk No effect on the risk Increases the risk
Do you have any questions? If yes—answer any questions. Once all questions have been addressed, proceed to next scenario. If no— proceed to next scenario.
Let’s move on. The outcome of interest now is aLDH-surgery: Treatment: Chiropractic care Outcome: Acute severe lumbar disc herniation that is surgically managed 6. Among an average group of 1,000 newly diagnosed acute low back pain patients not treated
with chiropractic care, how many do you believe will go on, within 2 months, to develop aLDH-surgery? Estimate: Please provide your estimate of the number of patients developing aLDH-surgery within 2 months.
176
Version date: 2011-03-25 Elicitation Script – Final 5!
7. Among an average group of 1,000 newly diagnosed acute low back pain patients treated with chiropractic care, how many do you believe will go on, within 2 months, to develop aLDH-surgery? Estimate: Please provide your estimate of the number of patients developing aLDH-surgery within 2 months.
8. There may be some uncertainty around your estimate. What do you believe is a plausible
range for the number of patients treated with chiropractic care that will go on, within 2 months, to develop aLDH-surgery? Lower bound (L): Upper bound (U): Elicit range of plausible values for the number of patients developing aLDH-surgery within 2 months. Identify a range such that it is extremely unlikely (but not necessarily impossible) that the point estimate lies outside.
9. You have indicated that the lower bound of the range is L and the upper bound is U. I will divide the range from L to U into equal-width bins. I have placed 1 virtual chip, representing 5% probability, in each of the L and U bins. I will now give you 18 more chips, each representing 5% probability, for a total of 100%. Using these chips, please indicate the weight of belief for your estimates of the number of patients treated with chiropractic care that will go on, within 2 months, to develop aLDH-surgery. Do you have any questions? If yes—answer any questions. Once all questions have been addressed, continue scenario. If no—continue scenario. Once all the chips have been placed and distribution fitted: Please review the shape and distribution of your answer. Does this reflect what you truly believe? If not, please feel free to revise your placement of chips until you are satisfied that the resulting distribution reflects your true belief. Do you have any questions? If yes—answer any questions. Once all questions have been addressed, proceed to question 10. If no— proceed to question 10.
10. Compared to primary medical care, what overall effect do you believe chiropractic care has on the risk of developing aLDH-surgery? Decreases the risk No effect on the risk Increases the risk
Do you have any questions? If yes—answer any questions. Once all questions have been addressed, proceed to next section. If no— proceed to next section.
The demographic section is the final part of this questionnaire. Please complete it now and let me know when you are finished.
177
Version date: 2011-03-25 Elicitation Script – Final 6!
After completion of the questionnaire: Thank you very much for sharing your beliefs with us and participating in this study. Your beliefs will be synthesized with that of your colleagues. If you would like, when the study is completed, we can email you the results of the study for your interest. Would this be of interest to you? If yes—record email address. If no—continue. Thank you for your time.
178
Appendix G. Belief elicitation computer protocol (Chapter 3)
Figure. Eliciting a probability distribution using the modified SHELF computer protocol. In the top screen, a respondent first expresses a point estimate, lower and upper bounds of their uncertainty, and indicates the distribution of the weight of their belief by interactively allocating “chips in bins”. In the bottom screen, a best fitting curve, from among 6 choices, is fitted to the chips in bins and presented back to the respondent, with the median and quartiles to ensure that the “belief curve” reflects their true belief. The software was modified so that: (1) the bins for the choices adapted to the initial range of possible values the respondent chose; (2) there was feedback on the number of chips placed and the number remaining; (3) the smoothed plot at the bottom gave a little more feedback on estimated mean, median IQR for the fitted function; (4) the entire set of data (number of chips in each bin, bin width, fitted functions, etc.) and plots of results could be saved with the respondent ID in the filenames.
179
Appendix H. Case and exposure definition tables for self-controlled case series study (Chapter 4)
Table H.1. Defining incident cases of acute LDH with early surgical intervention – Inclusion and exclusion
criteria and corresponding diagnostic/fee service codes for administrative health databases
Code Description Type Source
Step 1 – Initial cases of LDH surgery Initial cases of LDH surgery identified using at least one disc surgery intervention code and one LDH diagnosis
code from DAD and NACRS data
Disc surgery intervention codes Discectomy 92.31 Excision or destruction of intervertebral disc ICD-9 DAD Chemonucleolysis 1.SE.59 Destruction of intervertebral disc ICD-10 DAD, NACRS Discotomy/Excision 1.SE.87 Intervertebral disc discotomy/partial excision ICD-10 DAD, NACRS Discectomy/Excision/Laminectomy/La
minotomy 1.SE.89 Intervertebral disc discectomy/total excision/
laminectomy/laminotomy ICD-10 DAD, NACRS
LDH diagnosis codes Disc herniation 722.1 Displacement of thoracic or lumbar
intervertebral disc without myelopathy ICD-9 DAD
722.2 Displacement of intervertebral disc site unspecified without myelopathy
ICD-9 DAD
722.7 Intervertebral disc disorder with myelopathy ICD-9 DAD M51.0 Lumbar and other intervertebral disc
disorders with myelopathy ICD-10 DAD, NACRS
M51.1 Lumbar and other intervertebral disc disorders with radiculopathy
ICD-10 DAD, NACRS
M51.2 Other specified intervertebral disc displacement
ICD-10 DAD, NACRS
Sciatica 724.3 Syndrome characterized by pain radiating from the back into the buttock and into the lower extremity along its posterior or lateral aspect, and most commonly caused by protrusion of a lower lumbar intervertebral disc
ICD-9 DAD
Lumbosacral neuritis or radiculitis unspecified
724.4 Radicular syndrome of lower limbs ICD-9 DAD
Other relevant syndromes M54.1 Radiculopathy ICD-10 DAD, NACRS M54.3 Sciatica ICD-10 DAD, NACRS M54.4 Lumbago with sciatica ICD-10 DAD, NACRS
Step 2 – Preliminary cases of surgically managed acute LDH Preliminary cases of acute LDH with early surgery identified by including persons (from Step 1 above) who
presented to a hospital ED for LDH within 8 weeks prior to their LDH surgery using OHIP and NACRS data; this ED LDH visit date is defined as the event index date for the study
Emergency department LDH diagnosis codes (codes used for sensitivity analysis 2 as detailed in the Statistical Analysis section of the Methods; primary analysis does not include codes that may be less disc-specific as indicated below by *)
Diseases of musculoskeletal system 722 Intervertebral disc disorder MOH OHIP 724* Lumbar strain, lumbago, coccydynia, sciatica MOH OHIP Sprains, strains and other trauma 847* Low back, coccyx, neck MOH OHIP Disc herniation M51.0 Lumbar and other intervertebral disc
disorders with myelopathy ICD-10 NACRS
180
Code Description Type Source
M51.1 Lumbar and other intervertebral disc disorders with radiculopathy
ICD-10 NACRS
M51.2 Other specified intervertebral disc displacement
ICD-10 NACRS
Other relevant syndromes M54.1 Radiculopathy ICD-10 NACRS M54.3 Sciatica ICD-10 NACRS M54.4 Lumbago with sciatica ICD-10 NACRS M54.5* Low back pain ICD-10 NACRS
Emergency department fee codes H codes H101 Minor Assess.-G.P.-Emerg.-Dr.On Duty MOH OHIP H102 Comprehensive Assess. & Care MOH OHIP H103 Multiple Systems Assessment-G.P.-
Emergency-Dr. On Duty MOH OHIP
H104 GP-Reassess-Emerg Dept-Physician On Duty M-F Days
MOH OHIP
H121 Emerg.Dept.Phys.On Duty 12mn-8am Minor Assess.
MOH OHIP
H122 12 Midnight To 8:00 A.M. Comprehensive Assess. & Care
MOH OHIP
H123 Emerg.Dept.Phys.On Duty 12mn-8am Mult.Syst.Assess.
MOH OHIP
H124 Emerg.Dept.Phys.On Duty 12mn-8am Re-Assess.
MOH OHIP
H151 Emerg.Dept.Phys.On Duty Sat./Sun./Holiday Minor Assess.
MOH OHIP
H152 Sat/Sun & Holidays Comprehensive Assess. & Care
MOH OHIP
H153 Emerg.Dept.Phys.On Duty Sat./Sun/Holiday Mult.Syst.Assess.
MOH OHIP
H154 Emerg.Dept.Phys.On Duty Sat./Sun./Holiday Reassess.
MOH OHIP
H055 Emergency Department - Physician On Duty MOH OHIP H065 Emerg.Phys.Consult.(Non Spec. .In
Emerg.Med.) MOH OHIP
H100 Description Unknown MOH OHIP H105 Interim Inpatient Admission Orders MOH OHIP H106 Description Unknown MOH OHIP H107 Description Unknown MOH OHIP H112 Emerg.Dept.-Dr.On Duty-Extra To Proc.- 12
Mn To 8 Am. MOH OHIP
H113 Emerg.Dept.-Dr.On Duty-Extra To Proc.- Sat./Sun./Holidays.
MOH OHIP
H131 Physician On Duty ED - Evenings (18:00-24:00)
MOH OHIP
H132 Physician On Duty ED - Evenings (18:00-24:00)
MOH OHIP
H133 Physician On Duty ED - Evenings (18:00-24:00)
MOH OHIP
H134 GP-Reassess-Emerg Dept-Physician On Duty M-F Evenings
MOH OHIP
H980 Spec Vis Emerg Dept Phys., Wk/Daytime MOH OHIP H981 Spec Vis Emerg Dept Phys., Addit'l Pt., MOH OHIP
181
Code Description Type Source Wk/Daytime
H984 Spec Vis Emerg Dept Phys., Mon-Fri., Eve MOH OHIP H985 Spec Vis Emerg Dept Phys., - Addit'l Pt., Eve MOH OHIP H986 Spec Vis Emerg Dept Phys., Nights MOH OHIP H987 Spec Vis Emerg Dept Phys., Addit'l Pt.,
Nights MOH OHIP
H988 Spec Vis Emerg Dept Phys., Sat-Sun-Hols MOH OHIP H989 Spec Vis Emerg Dept Phys., Sat-Sun-Hols,
Addit'l Pt. MOH OHIP
K codes K990 Spec Vis Emerg Dept., Wk/Daytime MOH OHIP K991 Spec Vis Emerg Dept., Wk/Daytime, Addit'l
Pt. MOH OHIP
K992 Spec Vis Emerg Dept.-Sac.Off.Hrs. Wk/Daytime
MOH OHIP
K993 Spec Vis Emerg Dept.-Sac.Off.Hrs. Wk/Daytime Addit'l Pt.
MOH OHIP
K994 Spec Vis - Emerg Dept., Mon-Fri., Eve. MOH OHIP K995 Spec Vis - Emerg Dept., Mon-Fri., Eve.
Addit'l Pt. MOH OHIP
K996 Spec Vis - Emerg Dept., Nights MOH OHIP K997 Spec Vis - Emerg Dept., Nights Addit'l Pt. MOH OHIP K998 Spec Vis - Hosp Emerg Dept., Sat-Sun-Hols MOH OHIP K999 Spec Vis - Hosp Emerg Dept., Additional
Patients MOH OHIP
Step 3 – DAD and NACRS Exclusions Excluded persons who had other spine surgery interventions on or prior to their date of incident LDH surgery;
or, diagnosis of LDH or other conditions potentially associated with LDH within 21 months prior to the event index date using DAD and NACRS data
Other spine surgery codes Exploration and decompression of
spinal canal 16.01 Removal of foreign body from spinal canal ICD-9 DAD
16.02 Reopening of laminectomy site ICD-9 DAD 16.09 Other exploration and decompression ICD-9 DAD Division of intraspinal nerve root 16.1 ICD-9 DAD Chordotomy 16.2 ICD-9 DAD Excision or destruction of lesion 16.3 Excision or destruction of lesion of spinal
cord and spinal meninges ICD-9 DAD
Plastic operations 16.4x Plastic operations on spinal cord and spinal meninges
ICD-9 DAD
Freeing of adhesions 16.5 Freeing of adhesions of spinal cord and nerve roots
ICD-9 DAD
Shunt and drainage 16.6x Shunt and drainage by shunt of spinal theca ICD-9 DAD Injection of destructive agent 16.7 Injection of destructive agent into spinal
canal ICD-9 DAD
Invasive diagnostic procedures 16.8x Invasive diagnostic procedures on spinal cord and spinal canal structures
ICD-9 DAD
Other operations 16.9x Other operations on spinal cord and canal structures
ICD-9 DAD
Decompression/ Release
1.AW.72 Spinal cord decompression/release ICD-10 DAD, NACRS
Destruction/ 1.AW.59 Spinal cord destruction/rhizotomy ICD-10 DAD, NACRS
182
Code Description Type Source Rhizotomy
Discectomy 1.SC.75 Multilevel spine discectomy with interbody fusion
ICD-10 DAD, NACRS
1.SE.53 Discectomy with replacement of intervertebral spacer device
ICD-10 DAD, NACRS
Implantation 1.SE.53 Intervertebral disc device implantation ICD-10 DAD, NACRS Inspection 2.AX.70 Spinal canal inspection ICD-10 DAD, NACRS Laminectomy/
Laminotomy 1.SF.80 Sacral (for decompression of sacral spinal
canal) ICD-10 DAD, NACRS
1.SC.80 Spinal vertebrae (for decompression of spinal cord)
ICD-10 DAD, NACRS
1.SC.74 Spinal vertebrae with instrumentation or with discectomy and instrumentation
ICD-10 DAD, NACRS
1.SC.75 Spinal vertebrae with fusion or with discectomy and multilevel fusion
ICD-10 DAD, NACRS
Neuroectomy 1.AW.87 Spinal nerves [T2-S5] neuroectomy ICD-10 DAD, NACRS
Prior LDH diagnosis codes (exclude prior LDH as defined by Step 1 LDH diagnosis codes)
Disc herniation 722.1 Displacement of thoracic or lumbar intervertebral disc without myelopathy
ICD-9 DAD
722.2 Displacement of intervertebral disc site unspecified without myelopathy
ICD-9 DAD
722.7 Intervertebral disc disorder with myelopathy ICD-9 DAD M51.0 Lumbar and other intervertebral disc
disorders with myelopathy ICD-10 DAD, NACRS
M51.1 Lumbar and other intervertebral disc disorders with radiculopathy
ICD-10 DAD, NACRS
M51.2 Other specified intervertebral disc displacement
ICD-10 DAD, NACRS
Sciatica 724.3 Syndrome characterized by pain radiating from the back into the buttock and into the lower extremity along its posterior or lateral aspect, and most commonly caused by protrusion of a lower lumbar intervertebral disc
ICD-9 DAD
Lumbosacral neuritis or radiculitis unspecified
724.4 Radicular syndrome of lower limbs ICD-9 DAD
Other relevant syndromes M54.1 Radiculopathy ICD-10 DAD, NACRS M54.3 Sciatica ICD-10 DAD, NACRS M54.4 Lumbago with sciatica ICD-10 DAD, NACRS
Prior other diagnosis codes (spine-related disorders of the nervous and MSK systems, congenital anomalies, fractures)
Disorders of the nervous system 324.1 Intraspinal abscess ICD-9 DAD 334.8 Other spinocerebellar diseases ICD-9 DAD 334.9 Spinocerebellar disease unspecified ICD-9 DAD 335.x Anterior horn cell diseases ICD-9 DAD 336.x Other diseases of spinal cord ICD-9 DAD 340 Multiple sclerosis ICD-9 DAD 342.x Hemiplegia and hemiparesis ICD-9 DAD 344.x Other paralytic syndromes ICD-9 DAD 349.0 Reaction to spinal or lumbar puncture ICD-9 DAD
183
Code Description Type Source
349.81 Cerebrospinal fluid rhinorrhea ICD-9 DAD 353.x Nerve root and plexus disorders ICD-9 DAD 355.x Mononeuritis of lower limb and unspecified
site ICD-9 DAD
356.x Hereditary and idiopathic peripheral neuropathy
ICD-9 DAD
357.x Inflammatory and toxic neuropathy ICD-9 DAD 358.x Myoneural disorders ICD-9 DAD 359.x Muscular dystrophies and other myopathies ICD-9 DAD G06.1 Intraspinal abscess and granuloma ICD-10 DAD, NACRS G11.3 Cerebellar ataxia with defective DNA repair ICD-10 DAD, NACRS G11.8 Other hereditary ataxias ICD-10 DAD, NACRS G11.9 Hereditary ataxia, unspecified ICD-10 DAD, NACRS G12.x Spinal muscular atrophy and related
syndromes ICD-10 DAD, NACRS
G95.x Other and unspecified diseases of spinal cord
ICD-10 DAD, NACRS
G32.0 Subacute combined degeneration of spinal cord in diseases classified elsewhere
ICD-10 DAD, NACRS
G99.2 Myelopathy in diseases classified elsewhere ICD-10 DAD, NACRS G35 Multiple sclerosis ICD-10 DAD, NACRS G81.x Hemiplegia and hemiparesis ICD-10 DAD, NACRS G82.x Paraplegia (paraparesis) and quadriplegia
(quadriparesis) ICD-10 DAD, NACRS
G83.x Other paralytic syndromes ICD-10 DAD, NACRS G97.1 Other reaction to spinal and lumbar
puncture ICD-10 DAD, NACRS
G96.0 Cerebrospinal fluid leak ICD-10 DAD, NACRS G54.x Nerve root and plexus disorders ICD-10 DAD, NACRS G57.x Mononeuropathies of lower limb ICD-10 DAD, NACRS G60.x Hereditary and idiopathic neuropathy ICD-10 DAD, NACRS G61.x Inflammatory polyneuropathy ICD-10 DAD, NACRS G62.x Other and unspecified polyneuropathies ICD-10 DAD, NACRS G63.x Polyneuropathy in diseases classified
elsewhere ICD-10 DAD, NACRS
E08.42 Diabetes mellitus due to underlying condition with diabetic polyneuropathy
ICD-10 DAD, NACRS
E09.42 Drug or chemical induced diabetes mellitus with neurological complications with diabetic polyneuropathy
ICD-10 DAD, NACRS
E10.42 Type 1 diabetes mellitus with diabetic polyneuropathy
ICD-10 DAD, NACRS
E11.42 Type 2 diabetes mellitus with diabetic polyneuropathy
ICD-10 DAD, NACRS
E13.42 Other specified diabetes mellitus with diabetic polyneuropathy
ICD-10 DAD, NACRS
G70.x Myasthenia gravis and other myoneural disorders
ICD-10 DAD, NACRS
G71.x Primary disorders of muscles ICD-10 DAD, NACRS G72.x Other and unspecified myopathies ICD-10 DAD, NACRS G73.x Disorders of myoneural junction and muscle ICD-10 DAD, NACRS
184
Code Description Type Source in diseases classified elsewhere
Disorders of the MSK system
722.x Intervertebral disc disorders ICD-9 DAD
738.5 Other acquired deformity of back or spine ICD-9 DAD M51.x Thoracic, thoracolumbar, and lumbosacral
intervertebral disc disorders ICD-10 DAD, NACRS
M96.1 Postlaminectomy syndrome, not elsewhere classified
ICD-10 DAD, NACRS
M99.83 Other biomechanical lesions of lumbar region
ICD-10 DAD, NACRS
M99.84 Other biomechanical lesions of sacral region ICD-10 DAD, NACRS Congenital anomalies 740.x Anencephalus and similar anomalies ICD-9 DAD 741.x Spina bifida ICD-9 DAD 742.x Other congenital anomalies of nervous
system ICD-9 DAD
754.2 Congenital musculoskeletal deformities of spine
ICD-9 DAD
Q00.x Anencephaly and similar malformations ICD-10 DAD, NACRS Q05.x Spina bifida ICD-10 DAD, NACRS Q07.x Other congenital malformations of nervous
system ICD-10 DAD, NACRS
Q01.x Encephalocele ICD-10 DAD, NACRS Q02.x Microcephaly ICD-10 DAD, NACRS Q03.x Congenital hydrocephalus ICD-10 DAD, NACRS Q04.x Other congenital malformations of brain ICD-10 DAD, NACRS Q06.x Other congenital malformations of spinal
cord ICD-10 DAD, NACRS
Q67.5 Congenital deformity of spine ICD-10 DAD, NACRS Q76.x Congenital malformations of spine and bony
thorax ICD-10 DAD, NACRS
Fractures, dislocations, sprains and strains of the spine
805.x Fracture of vertebral column without mention of spinal cord injury
ICD-9 DAD
806.x Fracture of vertebral column with spinal cord injury
ICD-9 DAD
S32.x Fracture of lumbar spine and pelvis ICD-10 DAD, NACRS Injury To Nerves And Spinal Cord
952.x Spinal cord injury without evidence of spinal bone injury
ICD-9 DAD
953.x Injury to nerve roots and spinal plexus ICD-9 DAD S34.x Injury of lumbar and sacral spinal cord and
nerves at abdomen, lower back and pelvis level
ICD-10 DAD, NACRS
Step 4 – OHIP Exclusions Excluded persons who had, within 21 months prior to their event index date, a diagnosis of disc herniation or
other conditions potentially associated with LDH, specialist visits to neurosurgeons, orthopedic surgeons, neurologists, physiatrists or rheumatologists, or advanced spine imaging or diagnostic testing, using OHIP data
Prior disc disorder diagnosis codes Diseases of musculoskeletal system 722 Intervertebral disc disorder MOH OHIP Lumbar spine chiropractic diagnosis C42 Neuritis and neuralgia – Lumbar MOH OHIP C43 Neuritis and neuralgia – Pelvic, ileoiguinal MOH OHIP C45 Sciatic – acute MOH OHIP
185
Code Description Type Source
C46 Sciatic – chronic MOH OHIP C47 Sciatic – leg, other than sciatic MOH OHIP C48 Sciatic – discopathic or discogenic MOH OHIP C52 Radiculitis – lumbar, lumbosacral MOH OHIP C54 Radiculitis – vertebrogenic or pressure MOH OHIP
Prior other diagnosis codes Neoplasms 140 – 239 Malignant neoplasms, benign neoplasms,
carcinoma in situ, neoplasms of uncertain behaviour
MOH OHIP
Nervous system disorders 320 – 330 333 – 344 348 – 349
Central nervous system diseases MOH OHIP
353 – 359 Peripheral nervous system diseases MOH OHIP Congenital anomalies 741 – 759 Congenital anomalies MOH OHIP Fractures 805 Vertebral column - without spinal cord
damage MOH OHIP
806 Vertebral column - with spinal cord damage MOH OHIP 829 Other fractures MOH OHIP
Prior specialty visit codes Specialty of billing physician 04 Neurosurgery MOH OHIP 06 Orthopaedic surgery MOH OHIP 18 Neurology MOH OHIP 31 Physical medicine MOH OHIP 48 Rheumatology MOH OHIP
Prior spine imaging and diagnostic testing fee codes MRI X490 Limited spine (one segment) – multislice MOH OHIP X492 Limited spine (one segment) – repeat MOH OHIP X493 Intermediate spine (2 segments) – multislice MOH OHIP X495 Intermediate spine (2 segments) – repeat MOH OHIP X496 Complex spine – multislice sequence MOH OHIP X498 Complex spine – repeat MOH OHIP CT X415 Spine CT without IV contrast MOH OHIP X416 Spine CT with IV contrast MOH OHIP X417 3-dimensional spine CT acquisition
sequencing MOH OHIP
X128 Spine CT with and without IV contrast MOH OHIP EMG G455 Complete-Nerve Conduct.-Tech.Comp. MOH OHIP G456 EMG And Nerve Conduction Studies-P1 MOH OHIP G457 EMG And/Or Perf/Super/Interp. MOH OHIP G458 Single Fibre Electromyography MOH OHIP
G459 EMG And Nerve Conduction Studies-
Interpret-P2 MOH OHIP
G465 Thera. Proc.-Manipulation Joint-Major MOH OHIP
G466 Nerve Conduct Studies Tech.Comp.To
G469 Sched.B MOH OHIP
G467 Phys Med - Misc Therap Proc (Phys's Own
Patients) MOH OHIP
186
Code Description Type Source
G469 Phys.Med Ltd Nerve Conduct Studies/EMG
Interpr.Only Sched.B MOH OHIP
Other diagnostic testing G368 Chemonucl.If Lumbosacral Disc Incl. MOH OHIP G386 Inj./Infusion Lat. Discography Subs. MOH OHIP
J006 Diag. Radiol. Clinic. Proc.-Discogram-One
Disc MOH OHIP
J011 Diag. Radiol. Clinic. Proc.-Myelogram. MOH OHIP
J020 Diag. Radiol. Clinic. Proc.-Mylogram-
Post.Fossa Views MOH OHIP
J030 Diag. Radiol. Clinic. Proc.-Discogram-Ea.
Add. Disc. MOH OHIP
J038 Diag. Radiol. Clinic. Proc.-Myelogram/Supine
Vw. MOH OHIP
X164 Diag. Radiology-Discogram(S)-One/More
Levels MOH OHIP
X173 Diag. Radiology-Myelogram-Spine +/Or
Posterior Fossa MOH OHIP
Z454 Chemonucleolysis-Lateral Discography-1st
Disk MOH OHIP
187
Table H.2. Chiropractor and primary care physician exposure codes.
Specialist Specialty code Fee code Fee code description
Chiropractor* 59 - Chiropractor V101 Chiropractic subsequent visit office/institution V102 Chiropractic home service V103 Chiropractic initial service office/institution Primary Care Physician† 00 - Family practice and general practice 05 - Community medicine A001 Minor assessment A005 Consultation A003 General assessment A004 General re-assessment A006 Repeat consultation A007 Intermediate assessment A008 Mini assessment A050 Special community medicine consultation A051 Complex medical specific re-assessment A053 Medical specific assessment A054 Medical specific re-assessment A055 Community medicine consultation – office A056 Repeat consultation A058 Partial assessment A400 Comprehensive community medicine consultation A405 Community medicine limited consultation – office A901 General/family practice – house call assessment A905 General/family practice – limited consultation A911 General/family practice – special consultation A912 General/family practice – comprehensive consultation K005 Individual care K006 Hypnotherapy K007 Psychotherapy K013 Individual Care – first three units K017 Annual health examination K022 HIV primary care K028 STD management K030 Diabetic management assessment K033 Individual care – additional units K037 Fibromyalgia/chronic fatigue syndrome care
* Chiropractor visit was defined as a billing record with a combination of relevant specialty code and fee code as listed above
† Primary care physician visit was defined as a billing record with a combination of relevant specialty code and fee code as listed above