carpal tunnel syndrome a analysis

8/3/2019 Carpal Tunnel Syndrome a Analysis.

1/12

Original article doi:10.1093/rheumatology/ker108

Carpal tunnel syndrome and its relationship to

occupation: a meta-analysis

Annica Barcenilla1, Lyn M. March1, Jian Sheng Chen1 and Philip N. Sambrook1

Abstract

Objective. To examine the association between work place exposure and CTS by meta-analysis, includ-ing analyses with respect to exposure to hand force, repetition, vibration and wrist posture.

Methods. All relevant peer-reviewed articles published between January 1980 and December 2009 wereidentified by a systematic search using the MEDLINE, CINAHL and PubMed databases. Papers werecritiqued independently by two researchers and the relevant exposure information was extracted. Usingthe raw data of exposed and unexposed cases, a cumulative effect of specific exposure risks werecalculated for hand force, repetition, a combination of force and repetition, vibration and wrist posture

using the statistical program, Stata version 11 (StataCorp, College Station, TX, USA). Heterogeneity,meta-regression, publication bias and subgroup sensitivity analyses were performed.

Results. Thirty-seven studies from English-language literature met the inclusion criteria. Using NationalInstitute for Occupational Health and Safety criteria for case definition, a significant positive associationbetween CTS and hand force, repetition, use of vibratory tools and wrist posture was observed withapproximate doubling of risk for all exposures. Significant heterogeneity among studies was observedfor most exposures and metaregression analyses identified CTS case definition, study design, country andrisk of bias score to be the significant determinants. When a more conservative definition of CTS wasemployed to include nerve conduction abnormality with symptoms and/or signs, risk factors significantly

associated with an increased risk of CTS among exposed workers were: vibration [odds ratio (OR) 5.40;95% CI 3.14, 9.31], hand force (OR 4.23; 95% CI 1.53, 11.68) and repetition (OR 2.26; 95% CI 1.73, 2.94).There was a non-significant trend for the association between CTS and combined exposure to both forceand repetition (OR 1.85; 95% CI 0.99, 3.45) and wrist posture (OR 4.73; 95% CI 0.42, 53.32).

Conclusion. Occupational exposure to excess vibration, increased hand force and repetition increase therisk of developing CTS. Workplace strategies to avoid overexposure to these risk factors should be

implemented.

Key words: Carpal tunnel syndrome, Meta-analysis, Occupation, Work-related risks.

Introduction

CTS is the most common peripheral neuropathy and

arises from compression of the median nerve as it

passes through the carpal tunnel in the wrist. It is knownto be associated with age, gender and obesity and has

also been associated with a number of medical conditions

including RA, acromegaly, hypothyroidism, pregnancy

and trauma. Certain occupational activities have also

been associated with an increased risk of CTS in some

but not all studies and the association between CTS andoccupation still remains controversial. There has been one

previous published meta-analysis which found an associ-

ation between force and repetition and occupational fac-

tors [1] in studies that used the US National Institute for

Occupational Health and Safety (NIOSH) criteria for def-

inition of CTS. More recently, a systematic literature

review commissioned by the UK Industrial Injuries

Advisory Council found reasonable evidence that pro-

longed and highly repetitious flexion and extension at

1Department of Rheumatology, Royal North Shore Hospital, Institute ofBone and Joint Research, University of Sydney, Sydney, Australia.

Correspondence to: Philip N. Sambrook, Department ofRheumatology, Institute of Bone and Joint Research, University ofSydney, Level 4, Building 35, Royal North Shore Hospital, St Leonards,NSW 2065, Australia. E-mail: [email protected]

Submitted 8 November 2010; revised version accepted9 February 2011.

! The Author 2011. Published by Oxford University Press on behalf of the British Society for Rheumatology. All rights reserved. For Permissions, please email: [email protected] 1

RHEUMATOLOGY

Rheumatology Advance Access published May 17, 2011


2/12

the wrist, when allied with a forceful grip increased the risk

$2-fold, as did the use of hand-held vibratory tools [2].

However, there have been no further meta-analyses of

this potential association despite a number of publications

in recent years.

Since this relationship has important ongoing implica-

tions for individual workers, work practice and workers

compensation systems around the world, the aim of this

study was to provide an updated meta-analysis of studiesof CTS and work factors including potentially important

specific work exposure risks such as hand force, repeti-

tion, vibration and wrist posture. The meta-analysis of ob-

servational studies in epidemiology (MOOSE) criteria for

reporting of meta-analyses in observational studies [3]

was also followed.

Methods

The meta-analysis was based on studies published be-

tween January 1980 and 2009. Articles from two earlier

reports were identified: a comprehensive meta-analysis

that described the pattern of risk of work-related CTS

from published studies between 1980 and 1985 [1] anda systematic literature review of published studies that

included the period until 2004 [2].

To update the analysis, another systematic search was

conducted using MEDLINE, CINAHL and PubMed data-

bases up to December 2009 and CTS-related papers

were retrieved and critiqued. A specific search was com-

pleted based on the following key words and subject

headings: CTS, carpal tunnel syndrome, median nerve en-

trapment or neuropathy. To ensure completeness, a wider

spectrum search in all three databases was completed

using the additional less-specific terms: cumulative

trauma disorder (CTD), repetitive strain injury (RSI) and

occupational overuse syndrome. This search strategy

was similar to that described in the recent systematicreview [2].

In order to represent the relevant occupational expos-

ures, the following key words and subject headings were

used: work related, occupation, repetitive, repetitious,

RSI, cumulative trauma disorder, CTI, CTD. Specific oc-

cupational titles (including wildcard terms $) were also

used: poultry process, meat cut, dental worker, supermar-

ket worker, meat industry worker, slaughterhouse, assem-

bly line, assembly worker, packer, garment worker, meat

process, butcher, textile worker, forestry work, fish pro-

cess and musician. This review did not address specific-

ally key words related to computer use or keyboard use.

The search was restricted to English language papers

that included any of the specific terms among their keywords, title and abstract: risk, rate, odds, incidence,

prevalence, ratio, epidemiolo, casecontrol or cohort.

Studies were included in the analysis if they were original

articles that reported the measures of effect [e.g. odds

ratios (ORs) or relative risks] and the cases met the

NIOSH criteria for definition, namely the presence of:

(i) one or more symptoms indicative of CTS, e.g. para-

esthesiae, pain or numbness; and (ii) clinical signs that

included a positive Tinels sign or Phalens sign or nerve

conduction findings indicative of nerve dysfunction across

the carpal tunnel as well as (iii) evidence of work-

relatedness or the development of symptoms proceeding

after employment in a job involving one or more activities

such as the use of hand force, repetitive motion, use of

vibrating tools and awkward positions [1]. Papers that did

not include a control group were excluded from the ana-

lysis. Studies reporting a more conservative definition

were also examined. This definition included: (i) at leastthe presence of abnormal nerve conduction findings indi-

cative of median nerve dysfunction across the carpal

tunnel; and (ii) either symptoms indicative of CTS, e.g.

paraesthesia, pain or numbness or clinical signs that

included a positive Tinels sign or Phalens sign.

The titles resulting from the search were studied and

duplicates and/or irrelevant papers were eliminated. In

order to decide which papers to retrieve, the remaining

abstracts were read by two independent researchers

(P.N.S. and A.B.) and a consensus with a third reviewer

(L.M.M.) was used to resolve any differences of opinion.

The identified papers were evaluated, extracting details of

the study population, exposure details, estimates of

effect, bias risks, numbers of exposed and unexposedcases, relative risks and OR. In one instance, where

data for the measures of effect were unavailable, the

author [4] was successfully contacted. Papers were

given a (low, moderate or high) risk of bias rating using

Cochrane methodology [5]. This involved assessing each

study for the presence of the following bias types: selec-

tion (e.g. were there systematic differences between the

groups being compared other than the exposure?), per-

formance (e.g. were there systematic differences in med-

ical care apart from the condition of interest), attrition

(e.g. was there a differential in study follow-up between

the comparison groups?), detection or measurement

(e.g. was there a difference in how outcomes or the con-

dition of interest was measured between the comparisongroups?) and reporting (e.g. was there a difference be-

tween reported and unreported findings?). Two independ-

ent researchers (P.N.S. and A.B.) individually assessed

each study and a cumulative risk of bias rating was as-

signed. A third reviewer (L.M.M.) participated to reach

consensus when needed.

Statistical analysis

Summary effects (OR with 95% CI) of specific exposure

risks were calculated based on random-effects models for

hand force, repetition, a combination of force and repeti-

tion, vibration and wrist posture. Heterogeneity was quan-

tified using the 2

and I statistics. The method ofmeta-regression was used to identify sources of hetero-

geneity and to estimate the extent to which covariates

(defined at study level) in the model explained heterogen-

eity in the exposure risks. Between-studies variance (2)

was estimated using the restricted maximum-likelihood

method. Smaller values of 2 indicate less variability be-

tween studies. Subgroup sensitivity analyses were then

conducted within certain study characteristics that could

explain the heterogeneity such as differing definitions of

2 www.rheumatology.oxfordjournals.org

Annica Barcenilla et al.


3/12

CTS, study quality scores and country of origin.

Publication bias was examined using funnel plots and

the Harbord test. The Duval and Tweedie non-parametric

trim and fill method was then used to account for the

publication bias. All P-values are two-tailed. The analyses

were conducted using Stata version 11 (StataCorp,

College Station, TX, USA).

Results

Thirty-seven relevant papers from the two earlier reports

were identified for possible inclusion [642]. These papers

as well as four others [4346] were independently identi-

fied in a literature search up until 2004. A search was also

conducted for the period from 2005 to 2009 and this pro-

duced a possible 22 eligible papers [4, 4767]. Eighteen

identified from the earlier reports [6, 8, 10, 11, 14, 1821,

24, 25, 30, 32, 35, 36, 38, 39, 41] and eight papers identi-

fied between 2005 and 2009 were later excluded

[47, 5153, 58, 60, 62, 66] due to ineligibility or incomplete

data for the purposes of the analysis. A total of 37 original

articles were thus included in the analysis. In most stu-

dies, the diagnosis of CTS was based on a combination ofabnormal nerve conduction findings and a combination of

symptoms or signs. The characteristics of the 37 studies

are provided in Table 1 [4, 7, 9, 12, 13, 1517, 22, 23,

2629, 31, 33, 34, 37, 40, 4246, 4850, 5457, 59, 61,

6365, 67]. There were 28 cross-sectional studies,

4 cohort studies and 5 casecontrol studies.

Effect of exposure

Using the NIOSH criteria, a significant positive association

between CTS and occupational hand force (13 studies;

OR 2.18; 95% CI 1.47, 3.25; P


4/12


5/12


6/12

FIG. 1 Forest plot showing individual study and pooled ORs for the association of force and repetition in combination.

Study or subgroup

Abbas et al. [37]

Bonfiglioli et al. [54]

Chiang et al. [22]

Frost et al. [34]

Fung et al. [59]

Gell et al. [49]

Osorio et al. [26]

Shiri et al. [67]

Silverstein et al. [9]

Total (95% CI)

Total events

Heterogeneity: t = 0.06; c = 10.45, df = 8 (P= 0.24); I = 23

Test for overall effect: Z= 3.95 (P< 0.0001)

Events

1

18

8

44

17

9

1

72

8

178

Total

9

226

28

743

26

102

12

2118

157

3421

Events

5

8

5

6

149

20

1

67

1

262

Total

102

98

61

398

251

330

20

4025

157

5442

Weight, %

2.3

12.5

7.1

12.6

13.0

13.6

1.5

34.8

2.7

100.0

OR

M-H, Random, 95% CI

0.01 0.1 1 10 100

M-H, Random, 95% CI

2.42 (0.25, 23.35)

0.97 (0.41, 2.32)

4.48 (1.31, 15.30)

4.11 (1.74, 9.74)

1.29 (0.55, 3.01)

1.50 (0.66, 3.41)

1.73 (0.10, 30.45)

2.08 (1.48, 2.91)

8.38 (1.04, 67.78)

2.03 (1.43, 2.89)

OR

Protective effect Deleterious effect

CTS inexposed

CTS inunexposed

FIG. 2 Forest plot showing individual study and pooled ORs for the association of repetition using a conservative case

definition.

Study or subgroup

Abbas et al. [37]

Armstrong et al. [4]

Barnhart et al. [16]

Bonfiglioli et al. [55]

Latko et al. [44]

Maghsoudipour et al. [64]

Osorio et al. [26]

Roquelaure et al. [31]

Rosecrance et al. [40]

Wieslander et al. [12]

Yagev et al. [61]

Total (95% CI)

Total events



Events

10

3

27

22

16

41

2

62

26

14

94

317

Total

35

85

83

51

234

278

12

116

226

44

159

1323

Events

29

15

10

5

3

6

0

3

65

20

33

189

Total

163

986

55

55

118

117

20

14

889

133

70

2620

Weight, %

9.3

4.3

9.6

5.8

4.3

8.4

0.7

3.9

24.7

10.3

18.8

100.0

OR

M-H, Random, 95% CI

0.001 0.1 1 10 1000

M-H, Random, 95% CI

1.85 (0.80, 4.26)

2.37 (0.67, 8.35)

2.17 (0.95, 4.95)

7.59 (2.59, 22.19)

2.81 (0.80, 9.86)

3.20 (1.32, 7.76)

9.76 (0.43, 222.43)

4.21 (1.12, 15.88)

1.65 (1.02, 2.66)

2.64 (1.19, 5.83)

1.62 (0.92, 2.86)

2.26 (1.73, 2.94)

OR

CTS inexposed

CTS inunexposed


FIG. 3 Forest plot showing individual study and pooled ORs for the association of vibration using a conservative case

definition.

Study or subgroup


Maghsoudipour et al. [64]Wieslander et al. [12]

Total (95% CI)

Total events



Events

15

3413

62

Total

410

13136

577

Events

3

1321

37

Total

661

264141

1066

Weight, %

17.3

47.435.3

100.0

OR

M-H, Random, 95% CI

0.01 0.1 1 10 100

M-H, Random, 95% CI

8.33 (2.40, 28.95)

6.77 (3.43, 13.37)3.23 (1.42, 7.36)

5.40 (3.14, 9.31)

OR


CTS inexposed

CTS inunexposed

6 www.rheumatology.oxfordjournals.org

Annica Barcenilla et al.


7/12

repetition. The Beggs test funnel plot indicated a potential

absence of negative studies with small sample sizes for

repetition.

Discussion

CTS is the most common peripheral neuropathy and has

been related to occupational activities in some but not all

studies. Clarifying this relationship has important implica-

tions for workers compensation systems around the

world. It should also alert employers to implement work-

place strategies to avoid overexposure to the specific risk

factors. In this updated meta-analysis of studies of CTS

and work-related factors, there was a strong relationship

between the more stringent case definition of CTS (symp-

toms combined with nerve conduction abnormality) and

the use of vibratory tools with a 5-fold increased risk aswell as with hand force with a 4-fold increased risk and

with repetition and a combined exposure to both force

and repetition with a doubling of risk. While wrist posture

also showed a 4-fold increased risk it did not reach sig-

nificance, heterogeneity persisted and the assessment of

wrist posture was indirect in two of the studies [13, 26].

However, all these findings are consistent with the bio-

mechanical hypothesis that underlies such a potential as-

sociation. For example, previous experimental studies in

human cadavers and animals suggest that carpal tunnel

pressure is strongly influenced by hand force, repetition,

hand or wrist vibration and wrist posture [68, 69]. Similarly,animal studies of repetitive flexion and extension of the

wrist for prolonged periods resulted in induction of swel-

ling in the carpal tunnel and effects on median nerve con-

duction [70, 71].

Our findings are consistent with and update the one

previous published meta-analysis that found an associ-

ation between force and repetition and occupational fac-

tors [1]. Our findings are also consistent with a more

recent systematic literature review that found reasonable

evidence that prolonged with highly repetitious flexion and

extension at the wrist, when combined with a forceful grip

increased the risk about 2-fold, as did the use of

hand-held vibratory tools [2]. Although we presented initial

results using the NIOSH criteria, given that the earliermeta-analysis [1] used those same criteria, we would

recommend the more conservative definition requiring

both abnormal nerve conduction findings and the pres-

ence of either typical symptoms or signs of CTS, as this

is more likely to reflect real clinical practice and

resolved heterogeneity between studies for a number of

measures.

This study has a number of strengths and limitations.

Many of the studies included in this meta-analysis

FIG. 4 Forest plot showing individual study and pooled ORs for the association of force using a conservative case

definition.

Study or subgroup

Abbas et al. [37]



Roquelaure et al. [31]Rosecrance et al. [40]

Total (95% CI)

Total events

Heterogeneity: t = 1.01; c =25.17, df = 4 (P< 0.0001); I = 84

Test for overall effect: Z= 2.79 (P= 0.005)

Events

25

18

43

2623

135

Total

70

897

198

34257

1456

Events

14

0

4

3968

125

Total

128

174

197

96858

1453

Weight, %

23.3

8.7

20.8

22.125.1

100.0

OR

M-H, Random, 95% CI

0.01 0.1 1 10 100%

M-H, Random, 95% CI

4.52 (2.16, 9.48)

7.34 (0.44, 122.39)

13.39 (4.70, 38.10)

4.75 (1.95, 11.58)1.14 (0.70, 1.87)

4.23 (1.53, 11.68)

OR


CTS inexposed

CTS inunexposed

FIG. 5 Forest plot showing individual study and pooled ORs for the association of wrist posture using a conservative

case definition.

Study or subgroupDe Krom et al. [13]


Osorio et al. [26]

Total (95% CI)

Total events


Test for overall effect: Z= 1.26 (P= 0.21)

Events35

46

2

83

Total138

293

12

443

Events121

1

0

122

Total491

102

20

613

Weight, %42.0

33.1

24.9

100.0

OR

M-H, Random, 95% CI

0.01 0.1 1 10 100

M-H, Random, 95% CI1.04 (0.67, 1.61)

18.81 (2.56, 138.24)

9.76 (0.43, 222.43)

4.73 (0.42, 53.32)

OR


CTS inexposed

CTS inunexposed

www.rheumatology.oxfordjournals.org 7

CTS and its relationship to occupation


8/12

collected information about exposures retrospectively and

the ability to control for confounding factors varied be-

tween studies. Few prospective studies were identifiedand there was a trend for the summary OR to decrease,

indicating less of an effect for risk exposure in these stu-

dies. There was a wide variety of workplace environments

represented and highly significant heterogeneity evident

when the studies were combined. No single occupational

activity or duration or dose response could be identified

from these studies. Thus, it is not currently possible to

determine what level of exposure is safe and this is par-

ticularly the case when several exposures are combined

or not well defined. We did not address specifically the

relationship between CTS and computer or keyboard use,

which remains a controversial topic [72]. Differences in

diagnostic classification or criteria for CTS between stu-

dies also varied considerably. We could not addressthe effect of gender because there were insufficient

sex-specific data to address this issue. However, even

accounting for these potential limitations, the evidence

as a whole remains consistent in suggesting the risks

associated with occupational exposure via repetition,

force, vibration and wrist posture can be reasonably

linked to CTS. Given the significant implications these

findings have for workers and employers, there is a high

priority to conduct well-documented prospective studies

of inception cohorts commencing work in the high-risk

occupations to document exposure objectively and to

derive more valid rates and consequently more accurateassessment of the attributable fractions of the exposures

to the development of CTS. We would be then better

placed to identify what the modifiable factors and oppor-

tunities for intervention are.

Our findings have significant implications for preventive

measures in the workforce. The most common legal test

applied in deciding whether a potential occupational dis-

ease should be covered by workers compensation is the

balance of probabilities. In the UK, this requires that the

risk of the disease should be at least doubled as a con-

sequence of the occupational exposure since a relative

risk of two corresponds to an attributable fraction of

50% in exposed persons [2]. Our meta-analysis found

that the risk for the five exposures that we studiedranged from between a 2- and 5-fold increased risk rep-

resenting attributable fractions of 5080%. Based on

these findings until further prospective studies of high

quality become available, highly repetitive wrist or hand

work should be avoided with regular rotation of tasks and

appropriate rest periods. Prolonged use of hand-held vi-

bratory tools should also be monitored. Further studies of

the types of occupational exposure in terms of specific

activities in a particular job are required. Because of the

TABLE 2 Effects of covariates (at study level) on explaining heterogeneity of the exposure risks

Type of exposure Covariates in meta-regression model Coefficient (95% CI) P-value s2 a

Force Definition 0.89 (0.06, 1.72) 0.04 0.31Study type 0.22 ( 0.47, 0.91) 0.54 0.46Country 0.79 ( 0.16, 1.74) 0.10 0.35Bias score 1.46 (0.70, 2.22)


9/12

TABLE

3SensitivityanalysesandsummaryORsofCTSbyexposu

retype(withandwithoutfactorsaffect

ingheterogeneity)

T

ypeofexposure

No.

ofstudies

Sum

maryORrandom

model

Heterogeneity

OR

(95%

CI)

P-value

2

P-value

ForceAllstudies

13

2.18(1.47,3.25)

carpal tunnel syndrome a analysis

Documents