Question What is the prevalence of work-related musculoskeletal disorders among at-risk physicians (surgeons and interventionalists)?
Findings Among 21 articles (5828 physicians) included in this systematic review and meta-analysis, the 12-month prevalence estimates for neck, shoulder, back, and upper extremity pain were estimated. The career prevalence estimates for degenerative cervical spine disease, rotator cuff pathology, degenerative lumbar spine disease, and carpal tunnel syndrome were also calculated.
Meaning The prevalence of work-related musculoskeletal disorders among at-risk physicians is comparable to that reported among high-risk workers (eg, laborers); given the impending physician shortage, this problem warrants prompt attention and action.
Importance Physicians in procedural specialties are at high risk for work-related musculoskeletal disorders (MSDs). This has been called “an impending epidemic” in the context of the looming workforce shortage; however, prevalence estimates vary by study.
Objectives To estimate the prevalence of work-related MSDs among at-risk physicians and to evaluate the scope of preventive efforts.
Data Sources and Study Selection Systematic search in MEDLINE (Ovid), Embase (Elsevier), Web of Science, PubMed (National Center for Biotechnology Information), and 2 clinical trial registries, without language restriction, for studies reporting on the prevalence and prevention of work-related MSDs among at-risk physicians published until December 2016. The Meta-analysis of Observational Studies in Epidemiology (MOOSE) guidelines for meta-analyses and systematic reviews of observational studies were used. At-risk physicians were defined as surgeons and medical interventionalists. Studies reporting on specific disorders or pain assessed with validated instruments were included.
Data Extraction and Synthesis Study characteristics; disease prevalence for the neck, shoulder, back, and upper extremity; and measures of resulting disability were recorded. Study estimates were pooled using random-effects meta-analytic models.
Main Outcomes and Measures Career prevalence of injuries and 12-month prevalence of pain.
Results Among 21 articles (5828 physicians [mean age, 46.0 years; 78.5% male; 12.8 years in practice; 14.4 hours performing procedures per week]) included in this systematic review and meta-analysis, pooled crude prevalence estimates of the most common work-related MSDs were degenerative cervical spine disease in 17% (457 of 2406 physicians) (95% CI, 12%-25%), rotator cuff pathology in 18% (300 of 1513 physicians) (95% CI, 13%-25%), degenerative lumbar spine disease in 19% (544 of 2449 physicians) (95% CI, 5%-16%), and carpal tunnel syndrome in 9% (256 of 2449 physicians) (95% CI, 5%-16%). From 1997 to 2015, the prevalence of degenerative cervical spine disease and degenerative lumbar spine disease increased by 18.3% and 27%, respectively. Pooled prevalence estimates for pain ranged from 35% to 60% and differed by assessment instrument. Of those with a work-related MSD, 12% (277 of 2319 physicians) (95% CI, 7%-18%) required a leave of absence, practice restriction or modification, or early retirement. Heterogeneity was considerable for all crude analyses (mean I2 = 93.5%) but was lower for sensitivity analyses (mean I2 = 72.3%). Interventions focused on products and behaviors. Twelve at-risk specialties described a gross lack of awareness and an unmet need for ergonomics education.
Conclusions and Relevance Prevalence estimates of work-related MSDs among at-risk physicians appear to be high. Further research is needed to develop and validate an evidence-based applied ergonomics program aimed at preventing these disorders in this population.
Workers of many occupations bear a health burden associated with disabling musculoskeletal pain and injuries of a work-related etiology, collectively called work-related musculoskeletal disorders (MSDs). These are a group of preventable disorders affecting muscles, tendons, and nerves. Examples include carpal tunnel syndrome, tendinitis, degenerative spine disease, thoracic outlet syndrome, and tension neck syndrome.1 These diagnoses share common risk factors, such as sustained nonneutral postures and forceful repetitive tasks, often resulting from poor instrumentation design; they manifest with insidious pain that can result in temporary or permanent work disability, if not addressed.
Procedural physicians, such as surgeons and interventional medical specialists, have a high risk for work-related MSDs. This is due to long work hours involving repetitive movements, static and awkward postures, and challenges with instrument design, especially given the rapid rate of innovation in the setting of a diversifying workforce.1 Ergonomists have described the surgeon’s work environment and working conditions as equal to, if not at times harsher than, those of certain industrial workers.2 This observation is consistent with studies demonstrating higher prevalence estimates of work-related MSDs among at-risk physicians compared with the general population3 and even labor-intensive occupations, such as coal miners, manufacturing laborers, and physical therapists.4 Although great strides have been made in industrial ergonomics to reduce the burden of disease, medicine has proven to be a unique challenge and the lack of intervention in this group is now becoming apparent.2
The growing prevalence of work-related MSDs among at-risk physicians has been called “an impending epidemic”5 and “the tip of an iceberg.”6 Numerous cross-sectional studies5,7– 11 report that more than 80% of at-risk physicians experience significant pain when performing procedures; the prevalences of tendinitis11 and carpal tunnel syndrome12 appear to be high, but estimates vary widely by study. In addition, conducting research on occupational injuries in physicians using established databases is challenging due to severe underreporting. For instance, in one study13 of 103 injured surgeons, a mere 19% reported their injury to their institution, despite the fact that 35% performed fewer operations due to the injury.
Per the Association of American Medical Colleges 2014 Physician Specialty Data Book,14 at-risk physicians comprise 20.4% (175 955 of 860 939 physicians) of the active physician workforce. This workforce is expected to face a shortage by 2025, with a lack of 25 200 to 33 200 surgeons alone, and disability is one contributing factor.15 Although research has been conducted on burnout,16 sharps injuries,17 and other occupational hazards in medicine,17– 19 little attention has been paid to the growing body of literature describing the work-related MSDs forcing these physicians to undergo surgery,3,20– 26 reduce productivity,25– 29 and at times lose their careers.3,6,20,28 Reliable estimates of the burden of work-related MSDs among these physicians are important for informing the urgency and scope of preventive efforts needed, particularly given the impending workforce shortage.30,31
We conducted a systematic review and meta-analysis of published studies among at-risk physicians, with several goals. These include determining (1) the prevalence of work-related MSDs, (2) the disability burden of work-related MSDs, and (3) the scope of interventions aimed at reducing the prevalence of work-related MSDs.
A systematic search was conducted in MEDLINE (Ovid), Embase (Elsevier), Web of Science, PubMed (National Center for Biotechnology Information), and 2 clinical trial registries from inception until December 2016, without language restriction. The Meta-analysis of Observational Studies in Epidemiology (MOOSE) guidelines for meta-analyses and systematic reviews of observational studies were used. Database search algorithms were designed by an expert biomedical librarian. Databases were queried for studies on at-risk physicians, work-related MSDs, and ergonomics. At-risk physicians were defined as surgeons and medical interventionalists. Complete search algorithms for each database are available in the Database Search Algorithms section of the Supplement.
Two of us (S.E. and B.N.T.) independently screened articles, extracted data, and performed the critical appraisal, with discrepancies discussed among the team.32,33 The principal investigators of ongoing trials were contacted regarding imminent publications. All included articles were published in English or Spanish and were read directly, and citations were screened for articles missed by the search.
Studies were included in the qualitative review if they (1) reported primary data on at-risk physicians and (2) were accepted for publication in peer-reviewed journals. Additional eligibility criteria for inclusion in the quantitative synthesis were as follows: the study (1) reported at least one prevalence estimate for one of the most commonly reported work-related MSDs of the neck, shoulder, back, or upper extremity; (2) reported prevalence estimates for a common interval; and (3) passed critical appraisal. In addition, studies reporting on the prevalence of pain were only included if the pain was assessed using a previously validated instrument, a de novo instrument for which the article reported acceptable validity or reliability scores, or a de novo instrument designed by occupational medicine or ergonomics experts.
Data from each study reporting prevalence estimates were extracted. These included the following: study design, medical specialty, procedural technique, geographic location, sample size, response rate, mean age of physicians, percentage of male physicians, mean caseload of physicians in hours performing procedures per week or number of procedures performed per week, mean number of years in practice of physicians, instrument used for diagnosis or screening, reported prevalence estimates with associated periods of work-related MSDs, and days of work lost or number of physicians requiring a leave of absence, practice restriction or modification, or early retirement due to work-related MSD.
The prevalence estimates for work-related MSDs of the neck, shoulder, back, and upper extremity were recorded. These areas were chosen because they were previously identified as areas of great concern.34 Tools used to evaluate diagnoses and pain were also recorded.
Data extracted from each study reporting outcomes of ergonomics assessments or interventions included medical specialty, procedural technique, and geographic location. Studies were sorted into one of the following 3 groups: baseline ergonomics assessments, ergonomics products and technology, or education and behavior modification strategies.
The critical appraisal checklist from The Joanna Briggs Institute35 Reviewers’ Manual 2014 was used to assess the quality of each study reporting prevalence estimates (eTable 1 in the Supplement). Low-quality studies were excluded. eTable 2 in the Supplement lists detailed exclusions.
Prevalence estimates were calculated by pooling study-specific estimates using the random-effects meta-analytic model by DerSimonian and Laird. Statistical heterogeneity was assessed with the Mantel-Haenszel method and I2 statistic. Weighted proportions and their 95% CIs were summarized in forest plots. Studies reporting a measure of disability burden or outcomes of ergonomics assessments and interventions underwent textual analysis and were summarized qualitatively.
Sensitivity analyses were conducted by stratifying on publication year for pooled injury data and by stratifying on assessment instrument for pooled pain data. We evaluated for differences in rates of injury and pain by specialty and technique using standard nonparametric bivariate methods. Meta-regression was not conducted given an insufficient number of studies with detailed sample data per analysis needed for meaningful results. All statistical tests were 2-sided with α set to .05. A software program (R, version 3.3.2; R Foundation for Statistical Computing) was used for all statistical analyses.
The search returned 3739 unique items. Most were discarded due to nonphysician patient population because many results involved physician management of occupational disorders in the general population (n = 3474). Others were discarded due to a lack of primary data of interest (n = 48) or the lack of a valid and reliable instrument used for evaluation (n = 91). Ultimately, 126 articles met inclusion criteria; 9 articles presented data for multiple outcomes of interest and were added more than once for analysis (Figure 1).
Included articles were published between 1974 and 2016 in 23 unique countries. The specialties contributing most to the literature were general surgery (53 of 135 [39.3%]), gynecology (19 of 135 [14.1%]), and urology (17 of 135 [12.6%]). Most articles (68 of 135 [50.4%]) focused on minimally invasive techniques (eTable 3 in the Supplement).
Included in the meta-analysis were 21 articles. These encompassed 5828 physicians (mean age, 46.0 years; 78.5% male; 12.8 years in practice; 14.4 hours performing procedures per week). Rates of injury and pain were similar among surgeons compared with interventionalists (eTable 4 in the Supplement), as well as among surgeons with a predominantly open approach compared with minimally invasive surgeons (eTable 5 in the Supplement).
Injury prevalence was described in 16 cross-sectional studies (Table). Studies used de novo questionnaires to ask physicians if they had been clinically diagnosed as having various work-related MSDs, with the most common being degenerative cervical spine disease, rotator cuff pathology, degenerative lumbar spine disease, and carpal tunnel syndrome. Given a great breadth of pathologic descriptions for spine disease, an expert in rheumatology and orthopedic surgery outcomes research (Jeffrey N. Katz, MD, MSc, written communication, January 2017) was consulted after data extraction to define degenerative spine disease as any of the following: spondylosis, spondyloarthropathy, herniated or ruptured cervical disc, or radiculopathy. Nonspecific paresthesia and neurapraxia were excluded.
Thirteen of the 16 studies were eligible for the quantitative synthesis (eTable 2 in the Supplement). This resulted in a pooled sample of 4245 physicians, most of whom were orthopedic surgeons (n = 1232), interventional cardiologists (n = 1118), and general surgeons (n = 582), located in North America (n = 3812).
The overall career prevalence estimate of degenerative lumbar spine disease was 19% (544 of 2449 physicians) (95% CI, 13%-27%) (I2 = 94.6%) (Figure 2). This appeared to increase over time in serial cross-sectional surveys of interventional cardiologists in the Society for Cardiac Angiography and Interventions: Ross et al24 reported an estimate of 6.5% in 1997, Goldstein et al41 reported an estimate of 16.8% in 2004, and Klein et al45 reported an estimate of 24.7% in 2015 (a change of 18.3% over 18 years). When stratified by period, studies published in the past decade resulted in an estimate of 22% (361 of 1598 physicians) (95% CI, 15%-30%) (I2 = 93.0%).
The overall career prevalence estimate of rotator cuff pathology was 18% (300 of 1513 physicians) (95% CI, 13%-25%) (I2 = 87.9%). When stratified by period, studies published in the past decade resulted in an estimate of 17% (212 of 1146 physicians) (95% CI, 10%-26%) (I2 = 90.4%).
The overall career prevalence estimate of degenerative lumbar spine disease was 19% (544 of 2449 physicians) (95% CI, 5%-16%) (I2 = 94.5%). This appeared to increase over time in serial cross-sectional surveys of interventional cardiologists in the Society for Cardiac Angiography and Interventions: Ross et al24 reported an estimate of 8% in 1997, Goldstein et al41 reported an estimate of 29% in 2004, and Klein et al45 reported an estimate of 35% in 2015 (a change of 27% over 18 years). When stratified by period, studies published in the past decade resulted in an estimate of 21% (391 of 1507 physicians) (95% CI, 15%-30%) (I2 = 92.1%).
The overall career prevalence estimate of carpal tunnel syndrome was 9% (256 of 2449 physicians) (95% CI, 5%-16%) (I2 = 94.5%). When stratified by period, studies published in the past decade resulted in an estimate of 12% (244 of 1867 physicians) (95% CI, 7%-18%) (I2 = 91.7%).
Twelve of the 18 studies were eligible for the quantitative synthesis (eTable 2 in the Supplement). This resulted in a pooled sample of 2815 physicians, most of whom were orthopedic surgeons (n = 1264), gynecologists (n = 495), and dermatologists (n = 371), located in North America (n = 1989).
The overall 12-month prevalence estimate of neck pain was 60% (1131 of 1921 physicians) (95% CI, 47%-72%) (I2 = 96.0%), and the overall 12-month prevalence estimate of shoulder pain was 52% (802 of 1360 physicians) (95% CI, 43%-61%) (I2 = 86.8%). These results are shown in Figure 3. The overall 12-month prevalence estimate of back pain was 49% (1233 of 2254 physicians) (95% CI, 36%-62%) (I2 = 96.8%), and the overall 12-month prevalence estimate of upper extremity pain was 35% (588 of 1343 physicians) (95% CI, 21%-52%) (I2 = 96.6%). These results are shown in Figure 4.
Heterogeneity was considerable for all crude analyses (mean I2 = 93.5%). It was lower on sensitivity analyses (mean I2 = 72.3%).
Five studies used the Nordic Musculoskeletal Questionnaire. For these, the 12-month prevalence estimate of neck pain was 65% (737 of 1058 physicians) (95% CI, 54%-76%) (I2 = 89.9%), shoulder pain was 52% (647 of 1058 physicians) (95% CI, 41%-63%) (I2 = 88.8%), back pain was 59% (715 of 1058 physicians) (95% CI, 45%-71%) (I2 = 92.0%), and upper extremity pain was 39% (527 of 1058 physicians) (95% CI, 25%-55%) (I2 = 94.9%).
Four studies used the Physical Discomfort Survey. For these, the 12-month prevalence estimate of neck pain was 38% (215 of 561 physicians) (95% CI, 34%-42%) (I2 not applicable), and back pain was 28% (189 of 671 physicians) (95% CI, 25%-32%) (I2 = 0.0%).
Three studies used a de novo questionnaire. For these, the 12-month prevalence estimate of neck pain was 59% (179 of 302 physicians) (95% CI, 54%-65%) (I2 = 0.0%), shoulder pain was 51% (155 of 302 physicians) (95% CI, 46%-57%) (I2 = 0.0%), back pain was 60% (329 of 525 physicians) (95% CI, 45%-73%) (I2 = 86.7%), and upper extremity pain was 21% (61 of 285 physicians) (95% CI, 17%-27%) (I2 not available).
A measure of disability burden was reported in 10 studies. Few studies reported similar specific measures, so a general measure of the number of affected physicians over a range of severity was calculated. Overall, 12% (277 of 2319 physicians) (95% CI, 7%-18%) (I2 = 92.3%) of physicians required a leave of absence, practice restriction or modification, or early retirement due to work-related MSD (eFigure in the Supplement).
Ergonomics assessments and interventions were described in 101 articles, about half of which (n = 50) were from the United States. Germany has published more studies (n = 4) directly related to policy and regulatory processes compared with other countries. All studies used short-term measures, such as symptoms experienced or electromyographic data collected during procedures, with minimal time points. No study design included long-term surveillance for work-related MSDs. Thirty-eight studies focused on ergonomics devices (eg, floor mats and supportive furniture) and ergonomically improved medical technology (eg, alternative video displays for microsurgery), 37 studies focused on education and behavioral modifications (eg, targeted stretching microbreaks), and the remainder consisted of baseline ergonomics assessments, details of which are beyond the scope of this systematic review and meta-analysis.
Several studies5,50– 52 state that surgeons and interventionalists lack awareness of applied ergonomics recommendations. Twelve at-risk specialties from 8 countries have published at least one article describing a need for ergonomics education during medical training. Specialties included dermatology,7 gastroenterology,50 general surgery,5,51 gynecology,52 interventional cardiology,53 interventional radiology,53 neurosurgery,54 ophthalmology,43 orthopedic surgery,13 otolaryngology,55,56 plastic surgery,4 and urology.10,57 Countries include Australia,58 China,57 Germany,59 Italy,60– 62 the Netherlands,63 Spain,64 the United Kingdom,65,66 and the United States.43,50,51 A few studies39,50,64 have found that ergonomics education during medical training appears feasible, accepted, and effective at changing behaviors and reducing symptoms.
This systematic review and meta-analysis of work-related MSDs among at-risk physicians found (1) high prevalence estimates of work-related MSDs, (2) a range of disability burden that included early retirement, and (3) a high demand for intervention, with poor supply. These findings are worrisome in the context of an impending shortage30,31 of surgeons and interventionalists and the large public investment required to train such specialists.
This study builds on recent work demonstrating high prevalence estimates of burnout,16 depression,67,68 and attrition69 among physicians during and after training. When considered together, these data suggest that some aspects of medical culture may be detrimental to the mental, emotional, and physical health and career longevity of physicians and subsequently may be detrimental to the volume and quality of patient care. These findings represent the first synthesis of the evidence on an important topic facing many physicians.
This study found that at-risk physicians bear a large burden of work-related MSDs. To illustrate, we found 12-month prevalence estimates for work-related musculoskeletal pain of the neck, shoulder, back, and upper extremity of 65%, 52%, 59%, and 39%, respectively, when the standardized Nordic Musculoskeletal Questionnaire was used for assessment (Figure 3 and Figure 4). Studies using this same questionnaire found mostly comparable 12-month prevalence estimates among workers in high-risk, labor-intensive occupations. For instance, a study70 of 996 unionized apprentice construction workers in the United States found 12-month prevalence estimates for the neck, shoulder, back, and upper extremity of 31.8%, 27.9%, 54.4%, and 42.4%, respectively. For 409 Portuguese nurses, these estimates were 50.1%, 37.8%, 63.1%, and 28.4%.71 For 217 Nigerian physical therapists, these estimates were 21.1%, 22.2%, 69.8%, and 20.6%.72 This epidemiologic burden translates to disability burden, with severe economic ramifications.
More people are disabled from working due to work-related MSDs than from any other group of diseases.73 In health care workers, work-related MSDs are the number one cause of absenteeism.74 The Bureau of Labor Statistics of the US Department of Labor estimates that 62% of all worker injuries and 32% of missed days from work are secondary to work-related MSDs: this amounts to an estimated economic impact of $13 to $20 billion every year.1 The economic burden of work-related MSDs has not yet been reliably defined for physicians.
This study found that 12% of physicians with work-related MSDs require a leave of absence, practice restriction or modification, or early retirement. A lack of standardized reporting across studies precluded a more meaningful measure of disability burden, but the primary literature is telling. For instance, Auerbach et al3 reported a career prevalence of 27.6% for cervical radiculopathy among orthopedic surgeons, of whom 10.7% required surgical intervention and 18.9% required time off work, ranging from days to forced early retirement. Among ophthalmologists, Sivak-Callcott et al20 reported a career prevalence of 31.3% for symptomatic bulging or herniated spinal discs; of the entire sample, 42.5% had to modify their practice due to pain, 7.6% required surgical treatment, and 9.2% were forced to cease practicing entirely.
The qualitative portion of the analysis discovered a demand for ergonomics education. Specifically, 12 specialties from 8 countries explicitly described a need for such applied education. This is supported by reports of low awareness levels of occupational injury, ranging from 11% to 41.3%.5,51 Furthermore, 85% of at-risk physicians are concerned by work-related MSDs and resulting disability, and more than 90% state that formal ergonomics education should be standard during training, yet only 6.9% to 17% report receiving any ergonomics education during their training, most of which is described as sporadic, informal intraoperative directives.39,50
Last, this systematic review and meta-analysis found evidence that procedural physicians do indeed appear to be at higher risk than nonprocedural physicians. For example, Healy et al42 found that endourologists reported more hand and wrist problems than psychiatrists, Mal and Costello48 demonstrated that otolaryngologists experience impingement syndrome more frequently than endocrinologists, Kim-Fine et al8 determined that vaginal surgeons reported a higher rate of work-related MSDs compared with primary care physicians, and Kitzmann et al44 reported a higher prevalence of neck, back, and upper extremity pain among ophthalmologists compared with family medicine physicians.
When interpreting the results of the quantitative synthesis, it is important to remember that heterogeneity was high for all analyses. A moderate amount of this heterogeneity is explained by differences in the sensitivity of instruments used for MSD evaluation. This is evidenced by the lower heterogeneity scores on subgroup analysis by instrument. Additional heterogeneity may be explained by differences in the mean age of physicians, workload, number of years in practice, proportion that are male, and geographic location among study samples. Although differences in these variables are grossly evident in the Table, determining whether any one variable independently and significantly accounts for heterogeneity was not possible because there was an insufficient number of studies for meta-regression. When more studies are available, this systematic review and meta-analysis may be updated and improved with such an analysis.
While the etiology of the heterogeneity should be determined, its existence does not render the present systematic review and meta-analysis inappropriate for 2 reasons. First, our strict protocol only pools study outcomes that are highly qualitatively homogeneous. Second, the magnitudes of all study estimates are high enough to be meaningful, such that the summary effect estimate is informative, irrespective of variability. Even though the 95% CI for the 12-month prevalence of neck pain is wide at 47% to 72%, the lower boundary of 47% is concerning enough to warrant further study and improved awareness efforts given that other high-risk occupations with similar or even lower prevalence rates receive training and support for preventing injury. These results are particularly meaningful because of our use of a conservative random-effects model.
Despite the important strengths of this systematic review and meta-analysis, the bias assessment of the primary literature revealed several limitations. First is selection bias. Of the 30 cross-sectional studies describing prevalence, all 30 used convenience sampling, only 10 achieved a response rate of at least 65%, and only 4 were analytic cross-sectional studies. The rest were descriptive studies without reference groups.
A second limitation is misclassification bias. All primary data were acquired through self-report instruments.
The third limitation is social desirability bias. Some studies administered surveys in-person and to very small cohorts of peers at the same institution.
Finally, we were unable to assess for publication bias. This was due to the setting of an insufficient number of studies needed for meaningful results.
Considering the high prevalence of work-related MSDs, subsequent disability burden, and clear demand for ergonomics education by numerous specialties, this systematic review and meta-analysis underscores the need for the development and validation of an evidence-based applied ergonomics program. Education on workplace safety and ergonomics has been shown to be effective at increasing awareness and reducing risk factors in other occupations,75– 78 but it is often not sufficient on its own. Therefore, future research should involve collaborating with experts in ergonomics and occupational medicine to develop and examine a broader, systems-based approach to ergonomics programs within the surgical or interventional suite.79,80
Ultimately, this work should be integrated with research on preventing surgeon burnout and attrition given shared risk factors. With these joint efforts, perhaps using innovative physician well-being programs81,82 as vehicles, we may be able to better protect the health and career longevity of our workforce and subsequently deliver superior care for our patients.
This systematic review and meta-analysis found that the prevalence of work-related MSDs among at-risk physicians is high and that no overarching intervention exists. At a time when practitioners are beginning to address burnout and other facets of medical culture contributing to attrition, suicide, and other markers of poor well-being, we must not forget the physical demands of a career in medicine. Like workers in other occupations, physicians have a right to practice their profession in a safe environment. The health and career longevity of our trainees, our colleagues, and ourselves rely on our dedication to bringing awareness and action to this issue.
Accepted for Publication: September 2, 2017.
Published Online: December 27, 2017. doi:10.1001/jamasurg.2017.4947
Study concept and design: Epstein, Sparer, Tran, Singhal, Lee.
Acquisition, analysis, or interpretation of data: Epstein, Sparer, Tran, Ruan, Dennerlein.
Drafting of the manuscript: Epstein, Tran, Singhal.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Epstein, Lee.
Administrative, technical, or material support: Epstein, Sparer, Ruan, Dennerlein, Singhal, Lee.
Study supervision: Epstein, Sparer, Singhal, Lee.
Conflict of Interest Disclosures: None reported.
Funding/Support: Dr Sparer was supported by grant 3R25CA057711 from the National Institutes of Health.
Role of the Funder/Sponsor: The funding source had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: The article contents are solely the responsibility of the authors and do not necessarily represent the official views of the National Institutes of Health.
Additional Contributions: Paul A. Bain, PhD (Countway Library of Medicine, Harvard Medical School), developed the search algorithms for this systematic review and meta-analysis. Theodore K. Courtney, MS, CSP, and Jeffrey N. Katz, MD, MSc (Harvard T. H. Chan School of Public Health), provided guidance regarding research in occupational medicine and ergonomics. Marcia A. Testa, PhD, MPH, and Linda Marc, ScD, MPH (Harvard T. H. Chan School of Public Health), contributed guidance regarding quantitative methods and scholarly writing. Mark Rechnic, MD (Division of Plastic and Reconstructive Surgery, University of California, San Diego, School of Medicine), inspired the development of this study. None received compensation.