Introduction 1 2 3 4 5 Not all musculoskeletal symptoms involve sickness absence or productivity loss. It would be interesting to know more about factors that might influence productivity loss in symptomatic workers. We are inclined to think that these factors might be similar to risk factors for the occurrence of symptoms or for sickness absence due to musculoskeletal symptoms. However, this is not necessarily true and these factors could easily diverge. Knowledge on both types of risk factors is important for primary, secondary and tertiary prevention. While knowledge on risk factors for the occurrence of symptoms is required to prevent them, knowledge on factors associated with productivity loss is needed in the process of retention, reintegration and rehabilitation of workers with symptoms. 6 9 10 14 The objective of the present study is to describe the extent of productivity loss among computer workers with neck/shoulder symptoms and hand/arm symptoms, and to examine associations between pain intensity, various physical and psychosocial factors and productivity loss in computer workers with neck/shoulder and hand/arm symptoms. Methods Study Population 15 The study included workers from five different companies. The five participating companies comprised an insurance company, a department of a university, a public transport company, a brewery, and a financial consultancy firm. Employees from these companies had administrative, professional or management jobs. Altogether approximately 9,000 employees were working in these companies. All employees were invited to participate in the study. To maximize the participation rate, various activities were arranged, varying from the dissemination of brochures to visits at the worksite. Almost 2,500 employees subscribed to the project and signed informed consent. They were requested by e-mail to fill out an electronic questionnaire, accessible via the Internet. Participants who did not want to fill out an electronic questionnaire could fill out a paper version. Out of these employees, 1,951 filled out the questionnaire at baseline, resulting in a response rate of 79% of the subscribed employees. In the PROMO-study, productivity loss due to neck/shoulder and hand/arm symptoms was self-reported and only assessed in workers reporting symptoms. Therefore, analyses concerning associations between various factors and productivity were limited to respondents reporting work-related neck/shoulder symptoms or hand/arm symptoms during the past three months. This selection contained 654 computer workers. Productivity Loss 16 Neck/Shoulder Symptoms and Arm/Hand Symptoms 17 Subjects were asked to rate the occurrence of pain or discomfort in the neck/shoulder region as well as in the hand/arm region in the previous 3 months on a four-point scale: ‘no, never’; ‘yes, sometimes’; ‘yes, regularly’; ‘yes, prolonged.’ Subsequently, subjects were asked to estimate whether these symptoms were related to their work, which they could answer with ‘yes, completely’; ‘yes, partly’; ‘possibly’ or ‘no.’ Also, 8 possible specific causes of these symptoms were summed: sport injuries, accidents, skin diseases, a twist or sprain, a cut or burn, a congenital defect, rheumatic disorders and a slipped disc. Neck/shoulder symptoms and arm/hand symptoms were defined as regular or prolonged pain, completely, partly or possibly related to work, and not caused by any listed specific cause. Independent Variables To examine which factors are associated with productivity loss in computer workers with neck/shoulder and hand/arm symptoms, analyses were carried out with the following variables: pain intensity, physical activity in leisure time (with BMI as a possible confounder), working hours, mouse position, psychosocial load and overcommitment. Intensity of Symptoms 18 Physical Activity and BMI 19 20 21 22 2 Working Hours Respondents were asked how many hours per week they worked according to their contract. A dichotomous variable was constructed that distinguished between full-time workers working 37–40 h and part-time workers working 4–36 h per week. Mouse Position 16 Psychosocial Load 23 24 25 26 27 To assess job satisfaction respondents were asked how they enjoyed their work. To answer this question four categories were presented: ‘never,’ ‘sometimes,’ ‘often,’ and ‘always.’ This variable was dichotomized, resulting in a positive score on job satisfaction containing the responses ‘often’ or ‘always,’ and a negative score containing the responses ‘never’ or ‘sometimes.’ Personal Factor—Overcommitment 25 23 25 Analysis First, descriptive statistics were used to examine how often symptoms resulted in productivity loss, and to describe the components of productivity loss in terms of sickness absence and decreased performance at work. The association between productivity loss and several determinants was examined with logistic regression analyses using productivity loss as the dichotomous outcome variable. Univariate and multivariate analyses were carried out. In all multivariate analyses, age, gender, level of education, and intensity of symptoms were included as covariates. For the analysis with physical activity, BMI was checked for confounding and for the analysis with the psychosocial work characteristics, the other psychosocial work characteristics were checked for confounding. If their inclusion in the model did not result in a change of more than 10% in the effect estimate, these covariates were not considered as a confounder and not included in the final model. Results Productivity Loss 1 Fig. 1 n  2 Fig. 2 n  Associations with Productivity Loss 1 Table 1 n  n Crude OR (95%CI) Adjusted* OR (95% CI) R 2 N  Full model**    .140 Pain intensity     Continuous measure (0–10)  1.24 (1.13–1.37)  1.26 (1.12–1.41)  .104 Physical activity in leisure time     No significant physical activity 63 (408) 1.00 1.00 .131     Moderate intensity (5 × 30 min/week) 22 (141) 0.73 (0.46–1.17) 0.74 (0.43–1.27)      Vigorous intensity (3 × 20 min/week) 16 (99) 1.48 (0.92–2.39) 1.27 (0.70–2.27)  Working hours     Part-time (4–36 h) 61 (399) 1.00 1.00 .135     Full-time (37–40 h) 39 (255) 1.37 (0.96–1.96)  1.36 (0.87–2.11)   Mouse position     Close to the keyboard 35 (229) 1.00 1.00 .134     Other position 65 (425) 0.93 (0.64–1.34)  0.70 (0.45–1.08)   Psychosocial load Effort-Reward Imbalance     No high effort, no low reward 40 (258) 1.00 1.00 .115     No high effort, low reward 20 (130) 1.81 (1.10–2.99) 1.43 (0.76–2.67)      High effort, no low reward 18 (119) 1.89 (1.14–3.16) 2.26 (1.24–4.12)      High effort, low reward 21 (138) 2.69 (1.68–4.32) 1.95 (1.09–3.50)  Job satisfaction/task enjoyment     Always 21 (138) 1.00 1.00 .110     Often 65 (423) 1.29 (0.80–2.08) 1.62 (0.89–2.92)      Never/sometimes 14 (93) 3.10 (1.72–5.58) 3.10 (1.44–6.67)  Overcommitment     Continuous measure (0–6)  1.06 (0.96–1.16)  1.09 (0.97–1.22)  .140 R 2 N R 2 * Adjusted for gender, age, level of education and intensity of symptoms; the analyses with effort-reward imbalance were additionally adjusted for job satisfaction and the analyses with job satisfaction were additionally adjusted for effort-reward imbalance ** The full model contained gender, age, level of education and all variables mentioned in the table Physical activity in leisure time, working hours, mouse position and overcommitment were not associated with productivity loss in computer workers with neck/shoulder symptoms or arm/hand symptoms. Additional adjustment for BMI in the analyses with physical activity did not result in a change in odds ratio for more than 10%. Discussion The purpose of this study was to describe the extent of productivity loss among computer workers with neck/shoulder symptoms and hand/arm symptoms and to examine associations between various physical, psychosocial and personal factors and productivity. The results show that in 26% of all cases reporting regular or prolonged symptoms in the past three months, productivity loss was involved. Most productivity loss was found in workers reporting both neck/shoulder symptoms and hand/arm symptoms. Overall, about 32% of the productivity loss was coming from sickness absence. Sickness absence occurred more frequently in workers reporting both symptoms (43%) and considerably less frequent in workers reporting only hand/arm symptoms (11%). Symptomatic workers reporting unfavorable psychosocial work characteristics reported more productivity loss. Comparison with Previous Research The study population of the present study consisted of office workers with neck/shoulder or hand/arm symptoms. So far, almost all studies using productivity loss as an outcome measure were studies in a mixed population, containing subjects with and without musculoskeletal symptoms. Moreover, most previous research on productivity loss concerned sickness absence, while in the present study a decreased productivity while working was also included. Therefore, the results of previous studies are hard to compare with the present study. In a mixed population, associations between productivity loss and potential risk factors partly reflect the risks for symptom occurrence. In the present study we wanted to examine which factors are related to decreased productivity once symptoms have occurred. These factors do not have to be similar. Nevertheless, we will mention the results of previous research in this section. 28 29 30 31 32 33 16 10 17 However, we did examine the association with part-time versus full-time work, because this variable is probably not biased. Respondents were asked how many hours they worked according to their contract. The distinction might indicate a form of physical load, assuming that full-time workers are exposed to a higher physical load. Nevertheless, no relation was found between productivity loss and part-time/full-time work. No previous studies were found examining this relationship. Psychosocial load, in this study defined as effort-reward imbalance and job satisfaction, was strongly associated with productivity loss. No studies were found that examined these psychosocial work characteristics in relation to productivity loss or sickness absence due to musculoskeletal symptoms. Also, no studies were found that examined the relation between psychosocial work characteristics in general and productivity loss, other than sickness absence. Finally, no studies were found that examined these relations in a symptomatic population. 34 36 37 38 Limitations of the Study 39 16 39 Although in the PROMO-study follow-up measurements were available, we choose to use a cross-sectional design. The main reason is that data on productivity loss were only available for workers with symptoms. Since symptoms generally have an episodic nature the selection of workers with symptoms at baseline is not identical to the selection of workers with symptoms at follow-up. To select workers with data on productivity at baseline as well as at one or more of the follow-up measurements would mean a selection of workers with symptoms at all these measurements. Such a selection would result in a small study population of workers with chronic symptoms. If we assume that the associations in this study are causal, still the direction of causality cannot be established due to the cross-sectional design. It could be that either psychosocial working conditions caused productivity loss, or productivity loss caused the reporting of adverse psychosocial working conditions, or both. Intervention studies focusing on the prevention of productivity loss among symptomatic workers might shed more light on the direction(s) of causality. Implications for Practice The results of this study show that employers should be aware that the consequences of neck/shoulder and hand/arm symptoms are more extensive than the visible sickness absence due to these symptoms. Only one third of workers who experience productivity loss due to their symptoms actually take sick leave. For the other workers productivity loss expresses itself in decreased performance at work. The results concerning factors associated with productivity loss are more difficult to interpret, since there is still a lack of knowledge on how they relate. It seems that symptomatic workers perform better in favorable psychosocial working conditions. An advantageous psychosocial climate might prevent productivity loss in symptomatic workers. In conclusion, most workers with neck/shoulder symptoms or hand/arm symptoms experience productivity loss from a decreased performance at work and not from sickness absence. Favorable psychosocial work characteristics might prevent productivity loss in symptomatic workers.