2004; McNutt et al. 2003; Skov et al. 1998). Based on prior knowledge (scientific and clinical), age (dichotomised into groups ≤45 or >45), gender and physical activity levels (Saltin 1968) were evaluated as possible confounders following the criteria for a confounding factor by Rothman et al. (2008). Finally, potential confounders were included in the model if the change between adjusted and crude RR for the exposure variables was at least 10 % (Hosmer 2000; Rothman et al. 2008). Only the final models are shown in the results. Results Women accounted for four out of five participants, which well mirrors the situation in Swedish health care (Table 1). Twenty-six percent (n = 197) reported frequent musculoskeletal

pain, and 21 % (n = 154) had experienced long-lasting stress at baseline. Decreased work performance at follow-up was reported HSP inhibitor by 9 % (n = 66) and reduced work ability by 34 % (n = 246) among those who at baseline reported good work ability and no decrease in work performance. Table 1 Characteristics of the study population at baseline Characteristics Distribution  % (n) Gender

   Men 20 (151)  Women 80 (595) Age    −44 38 (283)  45+ 62 (463) Physical activity    Sedentary 8 (60)  LPA 51 (381)  MVPA 41 (305) Stress    No 79 (589)  Yes 21 (157) Pain    GSK1904529A price No-infrequent 74 (549)  Frequent 26 (197) Stress/pain    No/no-infrequent 61 (452)  No/frequent 18 (137)  Yes/no-infrequent 13 (97)  Yes/frequent 8 (60) Distribution between categories in percent selleck kinase inhibitor (%) and numbers (n) Participants with complete data for the analyses of work performance (N = 746) LPA light physical activity, MVPA moderate to vigorous physical activity Workers who at baseline were categorized as having frequent pain had a higher risk for reporting reduced work ability at follow-up compared to workers without such pain (Table 2). The result was similar to the outcome work performance. Stress was not clearly related to any of

the outcomes, although the increased risk estimate for reduced work ability showed a trend towards an association (95 % CI 1.00–1.58). Age was included as a possible confounder in the models for decreased work performance, but not in the models Lenvatinib nmr for work ability since it did not change the risk estimates for neither pain nor stress. Gender and physical activity were not associated with either outcome and therefore omitted from the final analyses. Table 2 Percentages, frequencies (n) and risk ratios (RR) with 95 % confidence intervals (CI) for stress and musculoskeletal pain in relation to reduced work ability (WAI) and decreased work performance (DWP)   WAI DWP % (n) RR (95 % CI) % (n) RRa (95 % CI) Stress          No 32 (184) 1 9 (51) 1  Yes 40 (62) 1.3 (1.00; 1.58) 10 (15) 1.1 (0.63; 1.89) Pain          No-infrequent 30 (159) 1 7 (40) 1  Frequent 44 (87) 1.5 (1.21; 1.81) 13 (26) 1.5 (1.22; 1.85) Stress/pain          No/no-infrequent 29 (126) 1 8 (34) 1  No/frequent 42 (58) 1.5 (1.14; 1.86) 12 (17) 1.5 (1.15; 1.89)  Yes/no-infrequent 35 (33) 1.