Work-Related Musculoskeletal Disorders in Physical Therapists: A Prospective Cohort Study With 1-Year Follow-up

Abstract

Background: Work-related musculoskeletal disorders (WMSDs) have a significant impact on physical therapists, but few studies have addressed the issue. Research is needed to determine the scope of the problem and the effects of specific risk factors.

Objectives: The objectives of this study were: (1) to determine the 1-year incidence rate of WMSDs in physical therapists and (2) to determine the effects of specific risk factors.

Design: This was a prospective cohort study with 1-year follow-up.

Methods: Subjects were randomly selected American Physical Therapy Association members (N=882). Exposure assessment included demographic data, physical risk factors, job strain, and specific physical therapy tasks. The primary outcome was WMSDs, with a severity rating of at least 4/10 and present at least once a month or lasting longer than a week.

Results: The response rate to the baseline questionnaire was 67%. Ninety-three percent of the subjects who responded to the baseline questionnaire responded to the follow-up questionnaire. The 1-year incidence rate of WMSDs was 20.7%. Factors that increased the risk for WMSDs included patient transfers, patient repositioning, bent or twisted postures, joint mobilization, soft tissue work, and job strain.

Limitations: The primary limitation of this study was the number of therapists who had a change in their job situation during the follow-up year.

Conclusions: Work-related musculoskeletal disorders are prevalent in physical therapists. Physical therapy exposures, patient handling, and manual therapy, in particular, increase the risk for WMSDs.

There is evidence that work-related musculoskeletal disorders (WMSDs) have a significant impact on physical therapists. In prior studies,^1–6 physical therapists reported taking sick time, changing practice habits, changing work settings, or leaving the profession due to WMSDs. Cromie et al² reported that 1 in 6 physical therapists changed settings or left the profession due to WMSDs. Glover et al³ reported that 32% of physical therapists with WMSDs lost work time. Molumphy et al⁶ reported that 18% of physical therapists with WMSDs of the low back changed their work setting and that 12% of the physical therapists reduced their patient care hours.

Exposure to risk factors for WMSDs is likely to result from patient care activities that include lifting patients, transferring patients, and the performance of manual therapy. Each activity involves the application of relatively high levels of force, and each activity may have to be performed in hazardous postures. Patient handling has been consistently associated with WMSDs in nurses,^7–11 and biomechanical studies^12,13 have demonstrated very high associated loads. Nurses have one of the highest rates of nonfatal occupational musculoskeletal injuries.^14,15

Studies to date on physical therapists provide some indication of which job tasks may increase the risk for WMSDs. However, the exposure assessments were inadequate for determining the effect of specific risk factors. New research is needed to define the scope of the problem and to demonstrate the association between physical therapy work and WMSDs. The objectives of this study were: (1) to determine the 1-year incidence and prevalence rates of WMSDs in physical therapists and (2) to determine the effects of specific risk factors.

Method

Study Design

The study was a nonexperimental, prospective cohort study with a 1-year follow-up. Data were gathered using 4-page, self-report, mailed questionnaires at baseline and at follow-up. The questionnaires and protocols of this project were approved by the University Committee on Activities Involving Human Subjects at the Graduate School of Arts and Science, New York University.

Subjects

The American Physical Therapy Association (APTA) randomly selected 1,500 members using a Microsoft Access (version 2003)^* query. Of the 1,500 names transmitted, 1,486 names were determined to be eligible. Some of the members did not reside in the United States (n=9) or had invalid addresses (n=4). One member was on the doctoral dissertation committee of the primary investigator (MC) and was excluded. Questionnaires were sent to the 1,486 members who were potentially eligible. One year later, the follow-up questionnaire was mailed to every subject who responded to the baseline questionnaire.

Inclusion and Exclusion Criteria

The project included licensed physical therapists who were APTA members and involved in direct patient care at least 1 hour per week at their primary position. The therapists had to return both questionnaires, and they had to reside in the United States. There were no exclusion criteria.

Background/Demographics

A series of background and demographic items including age, sex, experience, hours worked per week, practice setting, hours of patient care per week, and holding a second job were included in the baseline questionnaire. Patient care hours were determined by the question “Approximately how many hours per week do you provide direct patient care at your primary job?”

Exposure Assessment

Exposure was assessed at baseline. Exposure assessment included specific physical therapy tasks, general physical risk factors, and psychosocial factors. Specific physical therapy tasks included patient transfers and patient repositioning (almost never or not at all, 1–5 times per day, 6–10 times per day, 11–15 times per day, or more than 15 times per day) and passive range of motion (PROM), joint mobilization, and soft tissue work (performed on 0 patients per day, 1–5 patients per day, 6–10 patients per day, 11–15 patients per day, or more than 15 patients per day). General physical risk factors included kneeling or squatting, working with the trunk bent or twisted, awkward postures, static postures, and repetition (almost never or not at all, about 10% of the workday, about 25% of the workday, half of the workday or more).

Selection of risk factors and scaling of questions were based on reviews of the physical therapy, nursing, and industrial literature.^{2–4,10,16–21} A panel discussion with 8 physical therapists and informal conversations with physical therapists also were used to select variables of interest and to scale the questions appropriately.

Psychosocial risk factors were assessed with scales from the Job Content Questionnaire (JCQ).²² The JCQ has been translated into several languages and has been shown to yield reliable and valid data.²³ Used initially to study heart disease and cardiovascular outcomes, it also has been applied in studies with musculoskeletal outcomes. Scales for work demands (5 questions) and job control (8 questions) were chosen because these dimensions have been associated with WMSDs.^24–28 Scales were totaled using specific formulas developed for the JCQ and then dichotomized based on the scale averages as recommended by the JCQ center. Physical therapists with high demands and low control were classified as having job strain.

Three questions were included in the follow-up questionnaire to determine whether respondents had changed jobs, rotated positions, or left the profession during the follow-up year.

Outcome Variables

Outcomes were assessed at baseline to establish prior cases and again at follow-up to establish the 1-year incidence rate. Outcomes were assessed with a series of questions adapted from Hagberg et al²⁹ and the Nordic Musculoskeletal Questionnaire (NMQ).³⁰ The NMQ is a widely used, valid, and reliable musculoskeletal surveillance and exposure assessment tool.^31–33

The primary outcome variable was the 1-year incidence of WMSDs. A case definition was developed that was broad enough to capture WMSDs that were sufficiently serious to cause problems at work but restrictive enough to avoid WMSDs that were minor complaints. A case was defined as a report of WMSD rated at least 4/10 on a visual analog pain scale from 0 to 10 and lasting more than 1 week or present at least once a month. Similar definitions have been used in other studies and have been found to be more restrictive.^26,34 An incident case was considered to be when a subject met the case definition during the follow-up period but was free of the disorder for 8 weeks prior to baseline. Secondary outcomes included visiting a physician, changing settings, and leaving the profession due to WMSDs.

A complaint of WMSD was established with the question “Have you had any musculoskeletal pain during the past 12 months that you believe to be related to your work?” The question “Have you had this pain or discomfort during the past 8 weeks?” was used to establish who was free of a particular WMSD for 8 weeks prior to baseline. Duration of WMSD was assessed by the question “How long does the pain or discomfort usually last (24 hours or less, 24 hours to 1 week, >1 week to 1 month, >1 month to 6 months, or >6 months)?” Frequency of WMSD was assessed by the question “How many times have you had the pain or discomfort (once every 6 months or less, once every 2–3 months, once a month, once a week, or more often than once a week)?” Respondents were directed to answer each question separately for each body region.

Pilot Testing

The questionnaire was given to physical therapists and physical therapy educators for informal feedback. The questionnaire was formally pilot tested in 2 phases. The first phase was a panel discussion with 8 physical therapists. Panel participants were selected by judgment sampling so that they included both expert and novice clinicians. The mean age of panel participants was 38.7 years, and the mean experience was 11.4 years. The panel included 3 board-certified clinical specialists.

In response to comments by panel participants, the scaling of the highest exposure category was changed from a frequency of 20 to 15. Both during the panel discussion and during informal feedback from physical therapists, a pain level of 4/10 was determined to be a reasonable level to differentiate minor complaints from more serious WMSDs.

The second phase was a test-retest reliability study with a 1-month follow-up period. Intraclass correlation coefficients (2-way, mixed model for absolute agreement) for test-retest stability of questions related to specific physical therapy tasks ranged from .80 to .91.³⁵ Questions related to general physical risk factors demonstrated moderate to good reliability. Intraclass correlation coefficients for those items ranged from .49 to .69. Test-retest stability of the JCQ scales was good for decision authority (ICC=.71) and psychological demands (ICC=.69) but fair for skill discretion (ICC= .35). The sample size of the pilot study was small (n=19), and some of the scales and questions were highly influenced by 1 or 2 observations. Content validity, face validity, and test-retest reliability of the questionnaires were determined to be adequate for use.

Data Entry

The questionnaires were entered and coded using SPSS Data Entry Builder (version 4.0.1).^† The questionnaires were entered twice into separate files and then were checked for errors by comparing the 2 versions. Discrepancies between the 2 files were checked against the questionnaires and then corrected. Twenty questionnaires were randomly selected for final error verification by comparing the entered data with the original questionnaires.

Data Analysis

Data were analyzed using SPSS for Windows (Graduate Pack, version 14.1)^† and Intercooled Stata (version 9.2 for Windows)^‡ computer programs. Exposure and background data were taken from the baseline questionnaire. Outcome data were taken from the follow-up questionnaire.

Descriptive statistics were produced for all background factors, exposure factors, and outcomes. The effect of background factors on WMSDs was analyzed using independent-sample t tests for continuous variables and the chi-square test of association for categorical variables. Prevalence (for each body region) was calculated by taking the number of cases in that body region and dividing it by the total number of therapists who responded to the follow-up questionnaire. Incidence (for each body region) was calculated by taking the number of cases in that body region and dividing it by the number of therapists who did not have a WMSD in the same body region for at least 8 weeks prior to baseline.

Associations among primary risk factors.

Associations among primary risk factors were assessed using the Kendall tau B statistic. The Kendall tau B is a nonparametric correlation coefficient based on the ranks of the data. It is appropriate for ordinal data with equal or near-equal numbers of categories.^36,37 Multicollinearity among background factors and primary risk factors was evaluated by calculating the variance inflation factors, variance proportions, and eigenvalues.³⁸ Where 2 background factors demonstrated multicollinearity, only 1 background factor was used for adjustment during logistic regression.

Associations between primary risk factors and WMSDs.

The effects of specific physical therapy tasks, general physical risk factors, and psychosocial risk factors on WMSDs in specific body regions (the low back and the wrist and hand, the 2 most prevalent outcomes) were analyzed using unconditional logistic regression. Subjects who were cases at baseline for the body region in question were eliminated from the data file for that analysis. Primary, ordinal predictor terms were coded as n−1 dummy variables. For each factor, odds ratios (ORs) adjusted for age, sex, hours worked per week at primary job, and holding a second job were calculated. Ninety-five percent confidence intervals (CIs) also were calculated. Hours worked per week and age were dichotomized based on the median values. Prior to logistic regression, missing values for age (25 missing) and hours per week (5 missing) were replaced with the sample median. According to Tabachnik and Fidell,³⁸ different missing value strategies are likely to yield similar results when the total number of missing values is less than 5% of potential responses.

Trends were evaluated by treating ordinal predictors as continuous terms and reporting the P values of the Wald statistic. The highest level of exposure for the specific physical therapy tasks (>15 per day) was collapsed with the next highest category due to the low number of respondents at the highest level of exposure. Where there were zero cells at any level of exposure, the variable was collapsed further.

Multivariate models were not reported because there were not enough outcomes for adequate power and because there was multicollinearity. Although controversial, a general recommendation of at least 10 cases for each predictor has been proposed.³⁸ Multivariate models for the low back and the wrist and hand included several primary predictors, each coded as n−1 dummy variables, as well as several potential confounders. This substantially exceeded reasonable power limitations.

Effects of changing jobs.

To evaluate the stability of the OR estimates when considering that physical therapists change jobs, the risk factor analysis was completed on a reduced cohort of physical therapists who had no change in job situation.

General Protocol

The baseline questionnaire, a cover letter, a return envelope, and an institutional review board statement were mailed in July 2005. A reminder postcard was sent 2 weeks after the initial mailing.

The follow-up mailing occurred in July 2006. A reminder postcard was sent 2 weeks later. A replacement questionnaire, a cover letter, and a return envelope were sent at 5 weeks to subjects who had not responded. A final replacement questionnaire, a cover letter, and a return envelope were sent at 9 weeks. Any mailing that came back with a change-of-address label was resent to the current address. At 5 weeks, the addresses of nonrespondents were checked against the US Postal Service NCOA^Link database³⁹ to check for change of address status. If a change of address was found, a new packet was sent to the current address.

To increase the response rates, stamps were used on both the outgoing and return envelopes. All cover letters were personally addressed and hand signed. Each packet also contained a $3 Starbucks gift card with a “Thank You!” note hand written on each card. All correspondence was printed on customized letterhead. One subject from both the baseline and follow–up phases was randomly chosen to receive a $500 APTA gift card. The primary investigator's name was hand written above the return address on each outgoing envelope. Responses were accepted through October 2006.

Role of the Funding Source

This project was supported by 2 National Institutes of Science EARDA grants (HD 035965), a National Institute for Occupational Safety and Health ERC Pilot Project Award (T42 OH008422), and a Smart Family Foundation Grant. The grants were used to cover direct and indirect expenses related to the pilot studies and the full project. Dr Koenig's work was supported, in part, by a Center grant from the National Institute of Environmental Health Sciences (ES00260).

Results

Response

One thousand twenty-nine physical therapists responded to the baseline questionnaire. Seventy-seven respondents were not included in the study because they did not have any clinical hours (n=21), were retired (n=18), did not have any patient care responsibilities (n=35), planned on retiring before follow-up (n=1), were students (n=1), or declined to complete the questionnaire (n=1). The follow-up questionnaire was mailed to each of the remaining 952 subjects who responded to the baseline questionnaire. Responses to the follow-up questionnaire mailing were received from 882 physical therapists (93%) before the deadline. The mean time between responses was 369 days (SD=20).

Missing Values and Accuracy

Exposure and background data were taken from the baseline data file. Outcomes were taken from the follow-up data file. None of the questions had more than 5% missing values. The majority of questions had 5 or fewer missing values. An accuracy check of 20 randomly selected questionnaires from both the baseline and the follow-up phases yielded no errors after data checking and cleaning procedures.

Background and Demographic Factors

Background and demographic information for subjects who responded to both questionnaires was compared with APTA's membership profile for the same year (2005). Subjects were a little younger (mean age of respondents=40.3 years, mean age of APTA membership=41.6 years) and less experienced (mean number of years of experience of respondents=14.3 years, mean number of years of experience of APTA membership=16.3 years). There also was a higher proportion of female therapists compared with the APTA membership (proportion of respondents=0.71, proportion of APTA membership=0.68).

Practice setting distributions were generally similar to the APTA demographic profile. There were significantly fewer members from the academic setting in the study sample. More sample therapists reported their work setting as “other,” and more were employed in private outpatient practices.

Changes in Job Situation

About 1 in 4 respondents (23.4%) reported some type of job change during the follow-up period. Respondents reported rotating to another unit within their facility (6.9%), changing jobs (15.5%), and leaving their jobs and not returning to work at time of follow-up (2.4%).

Work-Related Musculoskeletal Disorders

In the follow-up year, 507 physical therapists (57.5%) reported a work-related ache, pain, or discomfort in 678 body regions. The 1-year prevalence rate for WMSDs in any body region was 28.0%. The 1-year incidence rate was 20.7%. In total, 183 physical therapists reported 243 incident cases. Table 1 lists the proportions of incident cases for the follow-up year. The greatest proportion of incident cases was seen in the low back, followed by the wrist and hand, neck, and shoulder.

Table 1.

1-Year Incidence of Work-Related Musculoskeletal Disorders

Physical therapists with incident cases had to be pain-free for 8 weeks prior to baseline. Some therapists could have had incident cases during the baseline year but were pain-free at baseline. This was not common. For the wrist and hand, 5 therapists had prevalent cases during the baseline year but were pain-free at baseline. For the low back, 3 therapists had prevalent cases during the baseline year but were pain-free at baseline. Incident cases were primarily new cases rather than recurrences of cases from the baseline year.

Many of the therapists who reported having incident cases during the follow-up year, however, had reported complaints that did not reach case status during the baseline year. For the low back, wrist and hands, upper back, and neck, at least one half of the therapists who developed new WMSDs during the follow-up year had reported work-related aches, pain, or discomfort during the baseline year that did not qualify as cases of WMSDs.

Impact of WMSDs

Table 2 lists the numbers and percentages of physical therapists who reported consulting a physician, losing work time, changing settings, or leaving the profession due to WMSDs. One hundred fifteen physical therapists (13%) reported visiting a physician due to WMSDs. Sixty-two therapists (7%) lost work time due to WMSDs. Eighteen therapists (2.0%) reported changing settings due to WMSDs. Four physical therapists (0.5%) reported leaving the profession due to WMSDs.

Table 2.

Impact of Work-Related Musculoskeletal Disorders (WMSDs)

^aThe total reflects percentage of subjects who reported the impact in any body region. Impacts of WMSDs in multiple body regions were counted only once.

Associations Between Background Factors and WMSDs

Associations between background factors and WMSDs are summarized in Table 3. Female therapists, older therapists, and more experienced therapists were more likely to develop WMSDs. The mean difference between therapists with and without WMSDs was marginally significant for age (mean difference=1.8 years, P=.05) but not for experience (mean difference=1.2 years, P=.18). The difference in proportions of WMSDs among male and female physical therapists was not significant. Therapists with WMSDs were similar to therapists without WMSDs with regard to hours worked per week, hours of patient care, and holding a second job.

Table 3.

Associations Between Background Factors and Work-Related Musculoskeletal Disorders (WMSDs)

^aIndependent samples, 2-tailed t test.

^bPearson chi-square test of association.

Settings

The proportion of cases varied across settings. The highest proportions of WMSDs (in settings with more than 10 respondents) were seen in school system (31.3%), “other” (24.5%), private outpatient physical therapy (22.3%), home care (21.0%), and acute care (17.3%) settings.

Associations Among Risk Factors

Significant correlations were noted between patient positioning and patient transfers (Kendall tau B=.77, P<.01). Manual techniques also were highly correlated with each other. Soft tissue work was highly correlated with joint mobilization (Kendall tau B=.76, P<.01) and moderately correlated with PROM (Kendall tau B=.51, P<.01). Joint mobilization and PROM also were moderately correlated (Kendall tau B=.55, P<.01). Among the general physical risk factors, a moderate correlation was noted between awkward postures and bent or twisted postures (Kendall tau B=.51, P<.01). Multicollinearity was noted between age and experience and between hours per week and patient care hours. Only age and hours per week were retained for the logistic regression analysis.

Risk Factors for WMSDs of the Low Back

Associations between risk factors and WMSDs of the low back are summarized in Table 4 (only risk factors with significant or near-significant ORs or trends are listed). Patient transfers, patient repositioning, bent or twisted postures, and job strain increased the risk for low back WMSDs. Therapists who transferred patients 6 to 10 times per day had odds of developing low back WMSDs that were 2.4 times higher than those of therapists who did not perform any transfers (OR=2.40, 95% CI=1.03–5.62). The OR for therapists who transferred patients 10 or more times per day was higher but not significant. A trend of increased incidence of low back WMSDs across increasing levels of patient transfers (P=.02) was noted. The ORs and test for trend for positioning approached significance. Therapists who repositioned patients more than 10 times per day had odds of developing low back WMSDs that were 2.61 times higher than those of therapists who did not reposition patients (OR=2.61, 95% CI=0.96–7.08).

Table 4.

Univariate Associations Between Risk Factors and Work-Related Musculoskeletal Disorders (WMSDs) of the Low Back^a

^aOR=odds ratio, CI=confidence interval.

^bAdjusted for age, sex, hours per week at primary job, and holding a second job.

Bent or twisted postures increased the risk for low back WMSDs. Therapists who reported exposure to bent or twisted postures had 5.74 times higher odds of developing WMSDs than therapists who reported no exposure to bent postures (OR=5.74, 95% CI=1.60–20.65). A trend of increasing low back WMSDs across increasing levels of bent postures also was noted (P=.02). Therapists who reported job strain had 2.52 times higher odds of developing a low back WMSDs during the follow-up year (OR=2.52, 95% CI=1.38–4.61).

Risk Factors for WMSDs of the Wrist and Hand

Associations between risk factors and WMSDs of the wrist and hand are summarized in Table 5 (only risk factors with significant or near-significant ORs or trends are listed). All 3 manual techniques increased the risk for WMSDs of the wrist and hand. Of the 3 techniques, the most substantial effect was seen with higher levels of soft tissue work. Therapists who performed soft tissue work on more than 10 patients per day had odds of developing WMSDs that were 13.61 times higher than those therapists who performed no soft tissue work (OR= 13.61, 95% CI=2.91–63.78). Therapists who performed joint mobilization on more than 10 patients per day had odds of developing wrist WMSDs that were 7.95 times higher than those of therapists who did not perform joint mobilization (OR=7.95, 95% CI=2.18–29.04). Although the individual ORs for PROM were not significant, a trend of increasing odds across increasing levels of exposure for all 3 manual techniques was noted. Therapists who performed more than 5 transfers (OR=0.20, 95% CI=0.06–0.69) or who repositioned patients more than 5 times (OR=0.21, 95% CI=0.06–0.74) had lower odds of developing wrist and hand WMSDs.

Table 5.

Univariate Associations Between Risk Factors and Work-Related Musculoskeletal Disorders (WMSDs) of the Wrist and Hand^a

^aOR=odds ratio, CI=confidence interval, PROM=passive range of motion.

^bAdjusted for age, sex, hours per week at primary job, and holding a second job.

Comparison of Risk Factors Between Full Cohort and Partial Cohort Without Therapists Who Changed Jobs

Odds ratios for significant risk factors in the full cohort analysis (n=882) were compared with ORs for a partial cohort that did not include therapists who had a change in job situation during the follow-up year (n=676). The results for variables with significant or near-significant associations are summarized in Table 6. The point estimates of the ORs, in general, were similar for low back WMSDs and reduced for wrist WMSDs in the smaller cohort. The CIs were broader in the smaller cohort.

Table 6.

Comparison of Risk Factors Among Physical Therapists Who Did and Did Not Change Jobs^a

^aOR=odds ratio, CI=confidence interval, PROM=passive range of motion.

^bAdjusted for age, sex, hours per week at primary job. and holding a second job.

Discussion

A substantial number of respondents reported WMSDs. The prevalence of work-related musculoskeletal complaints (57.5%) was similar to the rates found in prior studies^1,3,4 and similar to the rate reported at baseline (60.8%) but lower than the rate found by Cromie et al² (80%). The 1-year incidence rate of 20.7% represented a substantial proportion of therapists who developed a new WMSD based on a fairly stringent case definition. Incidence for some therapists represented a completely new WMSD. For other therapists, it represented a transition from complaint to case status. Intervention strategies, therefore, should be focused on secondary prevention as well as primary prevention.

Comparison of this incidence rate with those of other studies in musculoskeletal epidemiology is not possible. The vast majority of studies do not evaluate every body region and among those that do, a standardized case definition is rarely used despite being recommended.²⁹ Additionally, very few studies specify that pain be work-related.

Physical therapists did not change settings due to WMSDs or leave the profession due to WMSDs in substantial numbers. This finding contrasts with the study by Cromie et al,² where 1 in 6 physical therapists changed settings or left the profession due to WMSDs. Few therapists saw a physician because of their WMSDs. There is evidence that therapists prefer to treat themselves or seek informal treatment from colleagues as opposed to seeking formal evaluation.³ Very few therapists in this sample lost work time due to WMSDs (7%).

The physical therapists did not appear to seek treatment, take time off from work, or seek evaluation in response to WMSDs. Cultural factors may be responsible, in part, for this behavior.^40,41 Potential solutions should include a discussion of the professional culture where dedication and good health are valued to the point where it may be too embarrassing for a physical therapist to discuss injuries, seek help, or take time off from work.

Work-related musculoskeletal disorders were associated with increasing age. As physical therapists become older, strategies must be developed that will allow them to continue clinical practice.

Patient handling (transfers and repositioning) increased the risk for WMSDs of the low back. This finding was consistent with a cross-sectional analysis of baseline data from this project and with prior research. Patient handling has been associated with WMSDs in physical therapists^1,2 and nurses.^7,10,42,43 Differences between the 2 tasks in terms of risk are difficult to determine because they are usually performed in the same settings. It is likely that each task imparts substantial risk.

Protective measures should be considered for lifting and handling patients whenever possible. Use of equipment such as sliding boards, sit-to-stand devices, sliding sheets, lifting equipment (free-standing and ceiling mounted), and height-adjustable beds can reduce the risk of WMSDs.^44–46 Lifting policies also should be considered on the facility level. Minimal lift policies and lifting teams have been proven to reduce caregiver injury rates.^47,48

All 3 manual techniques increased the risk for wrist WMSDs. This finding was consistent with the results of prior studies^1–4 and the baseline analysis. Soft tissue work emerged as the most substantial risk factor, with the ORs for the higher exposure categories representing a severe increase in risk.

All 3 techniques impart very high loads to the wrists and hands and could increase the risk for wrist and hand WMSDs. However, there appears to be something especially hazardous about soft tissue work. Soft tissue techniques usually are performed for longer durations. That may partially explain the results. Manual therapy has been studied as a risk factor for wrist and hand injuries,^49,50 but more focus on soft tissue work is urgently needed.

A variety of protective measures for manual therapy such as thumb splints, mobilization wedges, and soft tissue devices are available. These devices have not been studied in depth, but they may provide some measure of protection. Physical therapists should be introduced to this equipment as they learn to perform manual techniques. Clinicians also may benefit from judicious application of certain techniques, considering them only for patients who truly need them.

Job strain was associated with WMSDs of the low back in both the baseline and longitudinal analyses. This association was consistent with findings from studies in other populations. Work-related musculoskeletal disorders have been associated with job strain,^24,51 increased job demands,^52–55 and reduced job control.^53,56–58 Physical therapists may believe that WMSDs have a mostly mechanical etiology and can be prevented with attention to mechanical factors and body mechanics.⁴⁰ The current study suggests that psychosocial risk factors play an important role in the etiology and perpetuation of WMSDs.

The main limitation of this study was the instability of the exposure data from baseline. About 23% of the subjects had a significant change in their work status over the follow-up year. It was possible that many of the therapists who changed jobs went to positions in similar settings. It also was possible that they developed their WMSDs prior to changing positions. Comparison of the cohorts with and without job changes revealed different magnitudes of association and varying levels of significance, but very similar risks overall. The findings also were generally similar to an analysis of the cross-sectional, baseline data.

This study included APTA members only. The results, therefore, may be generalizable to the population of APTA members but not necessarily to physical therapists in general. Based on a survey by APTA in 2005,⁵⁹ employment information for nonmembers is not very different from that of members. However, it is unknown whether any differences would affect their responses to this questionnaire. Nonmembers were more likely to be female, but sex was not associated with WMSDs in this analysis. A smaller proportion of nonmembers worked full-time. Work hours, however, were not associated with WMSDs in this analysis. Further research should include both APTA members and nonmembers to account for the differences between the 2 groups.

Power for evaluating the specific effects of certain exposure levels was limited. There are no generally accepted guidelines for power in logistic regression. Tabachnik and Fidell³⁸ have suggested 10 cases for each predictor, but they qualify that there is very little substantiation for this number. In the current study, for some predictors, low numbers of therapists with cases in some of the exposure levels may have served to either inflate or reduce the ORs and could have caused some instability. Very high ORs should be interpreted with caution.

Confounding (in both measured and unexplored variables) was a consideration. A bent posture, for example, could have been considered an awkward posture. However, therapists in the panel discussion and therapists who gave informal feedback were confident that the specific physical therapy techniques were easily distinguishable. Therefore, the high correlations among specific physical therapy tasks were likely to reflect the patient populations and tasks required in certain settings. Multivariate models can control for confounding in measured variables. However, the high number of dummy coded predictors, high correlations among some predictors, and low numbers of cases at certain levels of exposure precluded the use of multivariate models. Future studies should include larger sample sizes and exposure assessments that avoid multicollinearity. Future studies also could use a less restrictive case definition that would result in additional cases.

The seemingly protective effect of transfers and repositioning for wrist WMSDs requires explanation. It is likely that wrist and hand injuries were less common in settings where physical therapists have to transfer and reposition patients a lot and, therefore, do not perform manual therapy frequently. Thus, it could have been the lack of manual techniques that reduced the risk for wrist and hand WMSDs as opposed to the protective effect of patient handling.

In conclusion, there were 2 major findings to consider from these data. The first major finding was that about 57% of the respondents reported a complaint of a work-related ache, pain, or discomfort. About 1 in 5 respondents (20.7%) reported a WMSD that met a stringent incident case definition. The second major finding was a clear association between physical therapy work and WMSDs. Patient transfers and repositioning were associated with WMSDs in the low back, and soft tissue work and joint mobilization were associated with WMSDs in the wrist and hand. These associations were mostly independent of hours worked, age, sex, holding a second job, and other primary risk factors. These associations also are consistent with criteria required for causal inference, including theoretical plausibility, statistical association, evidence of dose-effect relationships, and consistency with prior studies.⁶⁰

To reduce the rate of WMSDs in physical therapists, 2 recommendations are proposed. The first recommendation is wider consideration of safe patient-handling and movement policies within the profession. The American Nurses Association has established a campaign to advance safe patient-handling practices.⁶¹ The APTA is collaborating in these efforts.⁶² These efforts, however, are primarily focused on the movement of dependent or partially dependent patients. The concept of patient handling should be expanded to include manual therapy, which also poses considerable risk.

Part of the effort to advance safe patient handling would include changes in professional (entry-level) physical therapist education. These changes would include expanded content on WMSDs in physical therapists as well as the use of equipment and protective devices. This also should include a discussion of the role that the professional culture may play in the development of WMSDs.

The second recommendation is an expansion of the research agenda related to WMSDs in physical therapists and other health care workers. Studies with large sample sizes and multivariate models are needed to further examine the link between physical therapy exposures and WMSDs. Because the majority of therapists who have WMSDs do not take time off from work, studies that look at the impact of WMSDs, while at work as well as over the long term, also are needed.