Musculoskeletal disorders (MSDs) continue to represent a major proportion of injury/ illness incidence and cost in the U.S. Manufacturing (MNF) sector. In 2015, 34% of non-fatal injuries and illnesses involving lost time (LT) in the MNF sector involved MSDs and the MNF sector had one of the highest rates of MSD LT cases (1). Specific shoulder disorders and shoulder pain are prevalent in the general population and in the workplace, and have been estimated to represent approximately 20% of all disability payments for MSDs (2). In 2015 shoulder injuries accounted for 72,270 cases with lost work days (1). Approximately 15% (10,840) of these were in the Manufacturing sector and 9% of Manufacturing sector injuries with LT were shoulder injuries.

Despite of many engineering controls and workplace tools and equipment to reduce MSD risk factors, automotive manufacturing is characterized by repetitive manual work with short cycle times and sustained musculoskeletal loading. The prevalence of MSDs in this industry continues to be a concern. The motor vehicle manufacturing sub-sector (NAICS 3361) has been among the highest of MNF sub sectors, with MSD LT rates that were on average 96% higher than the general manufacturing MSD LT rates from 2003-2007. Assembly processes in this industry involve overhead conveyance of the vehicle chassis and overhead use of tools with elevated arm postures and musculoskeletal disorders of the shoulder are of particular concern. Shoulder MSDs account for upwards of 50% of all injuries in some automotive manufacturing departments. At the manufacturing facility where this study took place, shoulder disorders accounted for 17% of all injury cases. At this facility, shoulder injuries had been the highest cost class of injury for six consecutive years, and for eight of the prior nine years.

Prolonged overhead working postures are believed to be associated with symptoms of upper limb discomfort, fatigue, and impingement syndromes (3) and may contribute to shoulder disorders independent of high forces or load exerted with the hands. Fatigue of the shoulder muscles may result in changes in scapulo-thoracic and glenohumeral kinematics that affect risk for shoulder impingement syndromes (4,5). A number of studies and subsequent reviews suggest that there is limited evidence for exercise having benefit on preventing complaints of the arm, neck, shoulder (6,7,8)

A proactive individualized strategy to improve shoulder function of employees led to development and pilot testing of a shoulder conditioning program at this manufacturing facility, and anecdotal benefits had been reported by pilot participants. In a small group of pilot participants the reported percentage of discomfort free days increased over four weeks of exercise program participation. The program was developed by a physical therapist to condition the musculature of the shoulder complex to increase resistance to fatigue and improve musculoskeletal responses to, and tolerance of, overhead work. The general strategies include: (i) strengthening of serratus anterior and the rotator cuff musculature; (ii) enhancing normal patterns of scapulo-humeral motion that reduce activation of upper trapezius; and (iii) stretching to reduce “tightness” in pectoralis minor and the posterior capsule of the shoulder. The exercise routine is similar to a number of shoulder conditioning protocols that have been designed to improve motion and muscle activation and are believed to be beneficial to those whose occupation requires significant exposure to humeral elevation and overhead work (9,10,11). In a previous review of workplace exercise programs for the neck/shoulder few evaluative studies were found in Manufacturing industries that included a non-exercise control group and that followed outcomes for more than 12 weeks (8). Studies have specifically addressed construction work (10, 11), however, previous prospective studies do not appear to have specifically addressed Manufacturing industry employees with significant exposure to overhead work.

The objective of the present study was to prospectively evaluate the longer term effect of a workplace shoulder exercise program on shoulder and upper limb symptoms among Manufacturing industry workers with jobs involving significant overhead work. All of these individuals performed overhead automotive assembly work. Although the exercise program was ultimately followed for six months, the duration of the prospective study period was defined to be at least four months.

Interpretation of the differences in SRQ between the Exercise (Exercise) and Control groups over the months of the exercise program should consider multiple factors including: a possible regressing to baseline levels beginning three months after the exercise program began, and the absence of any relationship between exercise dose and SRQ score among exercisers. A slight improvement in the SRQ was observed in the Exercise group in the first five months after the exercise program began. However, more of the difference between groups in the exercise months was due to a decline in SRQ score in the control group; the decline was most marked in the month after the exercise program began. The early differences between the groups observed during the exercise period may reflect some preventive effect of exercise in reducing a decline in SRQ. The absence of any relationship between exercise dose (sessions attended) and SRQ among the exercise group suggests minimal improvement in SRQ score among the exercise group. Three months after the exercise program commenced SRQ scores in the control group improved steadily, regressing towards the baseline level. The reason for this is not clear, but the inconsistency of any trends calls into question the long term effectiveness of exercise on controlling shoulder symptoms.

The Nordic Questionnaire confirmed that the shoulder was the body region exhibiting the highest percentage of employees reporting problems and that this was the case in the seven months preceding, and six months during, the exercise program. A general decreasing trend in the percentage of responses reporting “trouble at any time during the last 7 days” is apparent in both Exercise group and the Controls for many body regions, including the shoulders. This finding is inconsistent with that of the SRQ score, which suggests a worsening of symptoms in the controls, and minimal improvement in symptoms among exercisers, during months 8-11.

Adherence to the questionnaire administration protocol declined over the course of the study period, resulting in lower questionnaire response rates and smaller sample sizes in the later months. The adherence to questionnaire completion was unrelated to exercise session participation. Exercise participation (compliance) was fairly consistent across months - between 28%-43% of sessions attended.

The differences in mean SRQ scores observed between Exercise and Control groups during the December '13 to May '14 period was similar to the 12 unit change in SRQ in a symptomatic group over an 8-12 week exercise program reported by Ludewig and Bordstad (11), however, in that study the SRQ improved among symptomatic workers who exercised, while in the present study the difference was observed as a decrease in SRQ among the Control group participants. SRQ scores, prior to, and following, the 6 month exercise program were lower in the present study than those reported in asymptomatic shoulders of construction workers (11). Ludewig and Borstad (11) reported a mean SRQ of 93.9 in an asymptomatic population of Construction trade journeymen, which is 10-12 points greater than the average pre-Exercise scores in the present study. Construction journeymen presenting with positive shoulder impingement tests and pain reproduction symptoms in specific clinical tests (classified as symptomatic) reported an SRQ in the range of 65.9-72.5. In the present study, the three months of lowest SRQ for the control group dropped to that level, but this appeared to be an inconsistent trend. The assembly line employees in the present study could be considered moderately symptomatic, with SRQ in the 70 – 90 range, at baseline and generally remained so throughout the exercise program.

A potential study limitation is the group randomization that resulted in the exercise program being assigned to only second shift employees. This confounding of the exercise treatment with work shift was due to a study design strategy that was anticipating an ergonomic tooling intervention that was ultimately not successfully integrated during the study period. A consequence of this study design was that employees to whom the exercise condition was assigned were slightly younger, by four years on average, on the second shift. Otherwise, the study groups were homogenous in terms of workplace exposures to risk factors and similar work environment factors at the same manufacturing facility. Another study limitation is that the duration of the pre-intervention and intervention periods resulted in missing follow-through observations (“drop out”) due to factors such as employee transfer out of the specific work groups or merely declining compliance with monthly symptom questionnaire completion. Issues of employee turnover and continuation through the study period are a challenge to any prospective workplace intervention evaluation study, and compliance with a voluntary exercise program added another layer of challenge in this study. It is not surprising that most studies of workplace exercise reporting positive effects on shoulder symptoms have been of less than 16 weeks duration (8).

In a review of 38 studies of workplace exercise for control of neck/shoulder disorders conducted between 1997-2014 half of the study populations were employees in office environments with computer work exposures (8). This may be due to the ubiquity of office work and office/computer workstation exposures making this environment efficient for recruiting larger homogenous samples. Or, office work may afford greater schedule flexibility relative to other industries for exercise session participation during the workday. Regardless, that review identified only five studies since 1997 that evaluated exercise benefits specific to shoulder musculoskeletal symptoms of manufacturing industry employees (20-24). Two of these four studies were cohort designs: a 12-week pre-test/post-test study in beverage and tin mill industries reporting a positive effect of a daily stretching session (20), and an 8-week study of assembly line workers reporting a positive effect of two weekly sessions of strengthening and stretching (21). A third study, describing a group of industrial workers including those assembling printed circuit boards, found no difference between a strength and an endurance training program on pain and perceived exertion (22). Two of the five studies included a non-exercising control group. Maher et al. (23) evaluated a 6-month resistance band strengthening program among n=4 exercisers compared to n=5 controls at a diesel engine assembly facility. That study reported no significant difference between exercisers and controls in 15 of the 16 disability index items queried and only one of the four exercise participants completed exercise logs to track compliance. Sundstrup et al (24) evaluated effects on a 7-item Workability Index outcome resulting from a 10-week resistance training program administered to slaughterhouse facility workers. Workability Index declined in a non-exercising control group relative to the exercisers. Compliance with the supervised exercise program averaged 2.4 (of 3 recommended) sessions per week attended. Compliance in the present study was weaker, with an average of 1.8 sessions per week attended. Other than Maher et al (23) we are not aware of other studies of workplace exercise longer than 10 weeks in duration that addressed shoulder disorders among manufacturing employees and that contrasted the exercise group to comparable non-exercising controls. In only one of the Manufacturing industry studies (24) was exercise compliance reported sufficiently for the determination and comparison of exercise dose.

A strength of the present study was the supervision and monitoring of the exercise and the reliable measure of exercise compliance and intensity progression by the participants. Many occupational studies in which the exercise was done without direct supervision at the workplace have less reliable measures of actual exercise compliance of participants. For example, Ludewig and Bordstad (11) used self-reported compliance through daily exercise logs and 39% of participants did not return the logs and among the 61% who did only 9 of 34 participants reported 75% or greater compliance with the program. Compliance among the present study participants is comparable to that figure - confirming 10 of 39 individuals participating in three quarters of the sessions. Other studies have reported program compliance based on meeting a low threshold for participation, e.g. attending as few as one exercise session per week, so that a high percentage of participants achieve the minimum compliance, when in fact, participants experienced a low exercise dose. Conversely, an “acceptable” exercise compliance is difficult to establish as no consensus exists regarding minimum effective dose for exercise. A more important indicator than simply attendance of sessions might be the degree of progression in increasing the training stimulus. Of 26 participants who participated in any exercise sessions 15 were documented with progression through one or more increases in band resistance during the program. The progression to higher resistance bands indicates that training stimulus was increased for these individuals, and this is suggestive of increased performance and sincerity of training effort. However, band resistance progression was not predictive of reported musculoskeletal symptoms among the exercisers over time.

The present study evaluated a voluntary workplace exercise program, in the automotive manufacturing industry, that can be implemented as a countermeasure to control the burden of shoulder disorders. When employees were incentivized some participated regularly by attending two or three sessions per week, which was believed to be a minimum target. Several individual comments were in praise of the program. One employee commented that he “…felt better able to perform at the beginning of shift.” Another reported feeling “…’loose’ at the beginning of the shift.” An employee with one of the highest levels of participation reported that he “…felt as though it did prevent him from getting hurt.” Conversely, a negative response to the program was that “…they (employees) work hard enough on-line and more exercise would tire them out too much.”

Automotive assembly employees in this study spent approximately 388 hours (1,550 exercise sessions attended) participating in exercise with 39 employees having the opportunity to participate in daily pre-shift sessions. We based cost estimates of the exercise program on representative U.S. national wage averages and the participation time observed in the present study. The U.S. Bureau of Labor Statistics (BLS) average hourly wage for Team Assemblers (OES code 51-2092) in Motor Vehicle Manufacturing is $23.34 (25). We assumed an hourly wage estimate for NATA certified trainers (exercise session leaders) of $23.77 in the Industrial/Occupational/Corporate setting (26). These result in a cost estimate for this exercise program over the six months as $10,209 ($23.34 × 388 for employee exercise time plus the exercise group leader time of $23.77 × 48.5 hours). Equipment costs associated with this program included only elastic resistance bands which are available in economical bulk rolls. These costs were trivial and thus disregarded. We estimated the average cost of a compensable shoulder injury claim to be $31,000 in the automotive manufacturing industry, based on an unpublished analysis of 106 compensable claims for shoulder injuries in that industry (NAICS 3361) in the state of Ohio. Since additional indirect costs would be incurred to train a replacement employee, true cost avoidance (benefit) would exceed a direct injury cost of $31,000. Thus, through simplified cost-benefit considerations, the program break-even point could be considered to be less than one-third of a single compensable injury avoidance.

The workplace administered shoulder exercise program may have lessened a decline in SRQ score, an instrument specific to shoulder symptoms, for a few months. The program had no effect on DASH score, which more broadly queries all upper limb symptoms. Observed trends in the SRQ scores among exercisers and controls in the first three months after the Exercise intervention did not appear to stabilize and cast doubt on whether shorter term effects can be predictive of longer term outcomes. The findings are also difficult to interpret in the face of low employee participation, which may be common to voluntary workplace exercise programs. Only 19 of the 41 employees who enrolled in the exercise condition attended an average of two or more exercise sessions per week over the study period. The present study represents one of few prospective studies of workplace administered exercise program interventions with employees in a manufacturing industry. It may be the only such study to include an occupational group with significant workplace exposure to overhead work.