physiotherapy research society may 2012 stephen may, sheffield hallam university
TRANSCRIPT
The evolution of research into management of low back
pain
Physiotherapy Research SocietyMay 2012
Stephen May, Sheffield Hallam University
Low back pain is extremely common
60-80% of adults at some point in their lifetime
The most costly single health problem for which an economic analysis was available (Maniadakis & Gray 2000)
Background
15% CLBP account for about 70% of costs (Linton et al 1998)
But what is the optimal management?
Despite billions of $ / £ invested in research over the last few decades do not seem to be any nearer an answer
Background
Where research has come from?
What concepts have influenced clinical practice and the research agenda
Highlight some of the limitations / problems with these approaches
Structure of session
Think about some new directions that LBP research is going in
That hopefully my produce more positive findings
And influence clinical practice
Structure of session
Patho-anatomical disorder within a biomedical model
Facet joint, disc pain, instability, SIJ etc
But actually ability to identify these so-called pathologies from clinical examination is very limited (Hancock et al 2007)
The past
Disc
Facet
SIJ
Diagnosis disc / facet / SIJ
Centralisation◦ Spec – 0.7-0.9◦ Sens – 0.2-0.92
Not informative
>3 PPPP◦ Spec – 0.7-0.8◦ Sens – 0.7-0.89
4 articles with 11 clinical tests
Majority of tests demonstrated limited ability to make diagnosis (sensitivity)
Diagnosis of lumbar segmental instability (Alqarni et al 2011)
‘Common patho-anatomical findings such as degenerative disc disease, annular tears, fissures, facet joint arthrosis and disc bulges have been found to be not predictive of future LBP.’
‘Diagnostic labels such as ‘instability’ should be reserved solely for ‘unstable fractures’ and ‘unstable spondylolisthesis.’ (O’Sullivan 2012)
It’s time for a change with management of NSLBP
Identification can be done with radiographically controlled injections
But has not actually led to improved treatment outcomes
These are not commonly available
The past
Furthermore only 8-15% can be given a specific diagnosis
So > 85% labelled non-specific LBP
Research has focussed on NSLBP with little attempt to identify specific groups
The past
Non-specific LBP has been randomly allocated to 2 or more treatment arms
Mean difference between treatment arms calculated
Does not provide information on individual responses, only averages
Who will respond best?
Research in the past
Pragmatic multicentre RCT LBP < 12 weeks duration
402 / 544 recruited and 329 followed up at 12 months
Randomised to: 1. pain management (advice, coping strategies,
usual activities, general exercises) 2. physiotherapy package (manual therapy,
stabilisation exercises, advice)
Example – Hay et al (2005)
Hay et al (2005) Primary outcome =
RMDQ at 3 & 12 months
Change scores = mean (+/- SD)
No significant difference between groups
3/12 (1)
3 /12 (2)
12 / 12 (1)
12 /12 (2)
02468
1012141618
Hay et al (2005) Pain management (1) v
physiotherapy (2)
Self report at 12 / 12: A. Completely better B. Much better C. Better D. Same E. Worse F. Much worse Tx 1 Tx 2
0
10
20
30
40
50
60
ABCDEF
So overall no significant difference and no difference in self-reported outcomes
But SD displayed huge range of responses, bigger than the mean differences over time
Some patients had major changes on RMDQ (14-15) and others had minimal (<3)
But cannot tell who responded best to each Tx
Hay et al (2005)
RCT SR Guidelines
Now numerous guidelines exist
Guidelines based on randomly selected patients, rather than selected on appropriateness of Tx
Past recommendations
Review of17 guidelines for LBP (van Tulder et al 2004; Arnau et al 2006)
Classification and therapeutic recommendations were largely similar
But ‘quality of reporting of guidelines was disappointing’ against the AGREE tool for appraisal of guidelines
Past recommendations
Assessment: To exclude serious spinal pathology For prognostic factors (yellow flags)
No treatment has been shown to be consistently superior
Exercise, manipulation, and CBT-type of approaches
(Airaksinen et al 2006)
Past recommendations - CLBP
TENS, hot/cold, traction, laser, US, SWD, IF, massage, corsets, and acupuncture CANNOT be recommended
Airaksinen et al 2006
Past recommendations
NICE guidelines suggest course of exercise, manipulation, acupuncture (8-10 sessions of each!)
If one fails try the next, and so on
If all fails then try 100 hours of CBT!
Past recommendations
Guidelines often arbitrary division into acute / chronic LBP
Acute LBP EG had 4 major recommendations◦ Reassure patients◦ Do not prescribe bed rest◦ Advise patients to stay active◦ Regular pain relief
? Manipulation No specific exercise
Van Tulder et al (2006)
Past recommendations
But guidelines present one-size-fits-all generic management advice
Just like the past RCTs does not tell you what you need to do with the patient in front of you
Past recommendations
‘Greatest danger of flawed clinical guidelines is to patients’:
‘Suboptimal, ineffective or harmful practices’
What is best for patients overall....... May be inappropriate for individuals.’
‘May come at the expense of reducing individualised care for patients.’
(Woolf et al 1999, BMJ)
Guidelines – potential harms
Clinicians do not belief that NSLBP is an homogeneous group (Kent & Keating 2004)
In other words sub-groups may require different interventions
Sub-groups based on clinical presentations
Are we missing something?
Better attempt to sub-classify patients prior to treatment
Treatment according to classification
Growing evidence of the relevance of this approach
The future for better back care?
Treatment-based classification (Delitto, Fritz et al)
Directional preference / centralization (McKenzie, Long, Werneke et al)
STarT back screening tool (Hill, Main, Hay et al)
Examples of different approaches to basing Tx on sub-classifications
Benefits of sub-grouping
• 76 patients acute LBP randomised to AHCPR guidelines OR treatment by classification
(Fritz et al 2003)
• Oswestry scores0
5
10
15
20
25
30
35
40
45
50
0
4w*
52w
AHCPR
sub-group
Benefits of sub-grouping
• 76 patients acute LBP randomised to AHCPR guidelines OR treatment by classification
(Fritz et al 2003)
• Oswestry scores0
5
10
15
20
25
30
35
40
45
50
0
4w
*
52w
AHCPR
sub-group
Benefits of sub-grouping
• Brennan et al (2006) recruited 123 sub-acute LBP and analysed outcomes in 3 ways:
• Randomised: manip / stab ex / dir pref
• Sub-group – identified after & blind to randomisation by pre-established rules
• If sub-group matched / unmatched by randomisation
Acute LBP
Baseline evaluation
R
Manipulation Stabilisation Dir Pref Ex
M UnM M UnM M UnM
CPR criteria
Benefits of sub-grouping
• Outcomes = Oswestry 4w and 52w
• Randomised groups P = 0.37
• Classified groups P = 0.19
• Matched / unmatched P = 0.013
Clinical features sub-classification systems CPR
Clinical features associated with improvement if treated with SMT (Flynn et al. 2002, Childs et al. 2004, Fritz et al 2004, 2005, Brennan et al. 2006, Cleland et al. 2006)
Clinical features associated with improvement if treated with SE (Hicks et al. 2005, Brennan et al. 2006, Teyhen et al. 2007)
Cohort all treated
Good responders defined as at least 50% improvement on Oswestry
Characteristics of those who so improved
Then RCT
Clinical prediction rules
Clinical prediction rules SMT
Duration of symptoms < 16 days
One hip with > 35ᵒ MR Lumbar hypomobility No symptoms distal to
knee FABQ work score <19
SE
+tive prone instability test
Aberrant motion (painful arc, Gower sign, catch, reversal of rhythm)
SLR >90ᵒ Age <40 years
Following on from derivation these need several stages of validation
SE CPR has not been validated
SMT CPR has been validated in one study but not in another population
Clinical prediction rules
Directional preference
312 patients – mechanical evaluation (Long et al 2004)
230 (74%) DP◦ Ext 191 (83%), flex 16 (7%), lateral 23 (10%)
Randomised to matched or opposite exercise or EBM guideline group (nearly 70 each group)
Benefits of sub-grouping
• 312 patients assessed – 230 (74%) with DP
• 230 randomised to matched / opposite / non-specific guideline approach
• RMDQ
(Long et al 2004) 0
2
4
6
8
10
12
14
16
18
20
0 2w*
match
opp
control
Self-rated improvement at 2 weeks (Long et al 2004)
0
10
20
30
40
50
60
match opposite control
resolved
better
no change
worse
Secondary analysis - Long et al 2008
At 2 weeks those who were 'worse' 'unchanged' or wanted a different exercise were offered cross-over T
96 initially unmatched patients then got matched DP
Long et al 2008
2
2.5
3
3.5
4
4.5
5
5.5
6
6.5
baseline 2 weeks* 4 weeks*
Opposite Non-directional
0.5
1
1.5
2
2.5
3
3.5
4
Baseline 2 weeks 4 weeks*
Opposite Non-directional
Back Pain Leg Pain
Directional preference In those with DP – DP exercises significantly
better outcomes than other exercises
In initial trial and when swopped over from non-specific to DP exercises
74% of assessed patients had DP◦ Acute to chronic◦ LBP with / without leg pain
STarT Back - classification by risk Hill et al (2008, 2009, 2010, 2011)
developed and validated a 9-question tool for determining those at prognostic risk◦ leg pain◦ comorbid pain◦ disability (items)◦ bothersomeness◦ catastrophising ◦ fear ◦ anxiety◦ depression (psychosocial subscale)
STarT Back 0-3 = low risk Advice session
4 or 5 on psychosocial subscale = high riskpsychologically informed
physiotherapy
rest = medium risk Standardised physiotherapy
STarT Back RCT 851 patients randomly assigned to risk-
informed Tx or control
Control = initial advice session, with option referral for further physiotherapy
469 assessed at 12 months (76%), but ITT analysis used
STarT Back RCT 26% / 46% / 28% were low / medium / high
risk
In control group >1/3rd medium and high risk not referred; half of low risk were referred
Mean number of Tx sessions same
Changes in RMDQ: All / Low risk
p=0.0001 / 0.0095 p=0.22 / 0.846
base-line
4m 8m 12m0
2
4
6
8
10
12
12
base-line
4m 8m 12m0
0.51
1.52
2.53
3.54
4.55
12
Changes in RMDQ: Medium / High
p=0.0012 / 0.0253 p=0.0024 / 0.1547
base-line
4m 8m 12m0
2
4
6
8
10
12
12
base-line
4m 8m 12m0
2
4
6
8
10
12
14
16
12
Baseline scores
function, LBP, catastrophing leg pain, below knee
RMDQ
pain pcs0
5
10
15
20
25
30
lowmediumhigh
0
10
20
30
40
50
60
70
80
low mediumhigh
Appears to be a useful tool to help identify patients with risk of prognosis
Relates to previously known prognostic factors
Can help to direct scarce resources to those most in need of more involved interventions
STarT Back
In the past non-specific LBP led to failure to select
Patients randomised to Tx regardless of whether appropriate
Mean differences hid responders / non-responders
Some Tx appeared superior
Conclusions
But overall differences between groups were minimal
Is there a way forward to improve outcomes?
Sub-classification based on clinical presentations
Conclusions
TBC - DP, SMT, SE - based on clinical factors
DP / centralisation - based on repeated movements
Prognostic risk factors - based on questionnaire
Conclusions
All have shown significant and improved outcomes in different study designs
But still using RCT-methodology, which is the only way to determine Tx superiority
Conclusions