OBJECTIVES: To fit a model which describes the complex interaction between various work-related factors and both musculoskeletal and psychosomatic complaints of nurses simultaneously and to compare the results with those of two LISREL models, in which the health outcome variables are analyzed separately in terms of their correlation with the same set of work-related factors. MATERIALS AND METHODS: Nurses (n=718) working in four Dutch nursing homes completed a questionnaire about work-related factors and musculoskeletal and psychosomatic complaints (response 81%). LISREL was used to describe and analyse the models. RESULTS: The full model had a reasonable overall fit. The strongest (and statistically significant) correlations were found between work postures and musculoskeletal complaints and between psychosocial work demands and psychosomatic complaints. In the two single models, health outcomes showed statistically significant correlations with the same work-related factors as they did in the full model. The full model could be slightly improved when the variable somatized complaints (e.g., dizzyness, headache) was related both to musculoskeletal complaints and to psychosomatic complaints. CONCLUSIONS: Although a model that simultaneously describes more dependent variables appears to provide a better approximation of the complexity of the actual relationship between work-related factors and health effects, the models in which only one health outcome is studied seem to give a proper description of the data at hand. This is an important finding because these single health outcome models resemble the usual analysis strategy by means of standard regression analysis. The adapted version of the full model suggests that musculoskeletal complaints of nurses can partly be understood as non-specific health complaints.
OBJECTIVES: To fit a model which describes the complex interaction between various work-related factors and both musculoskeletal and psychosomatic complaints of nurses simultaneously and to compare the results with those of two LISREL models, in which the health outcome variables are analyzed separately in terms of their correlation with the same set of work-related factors. MATERIALS AND METHODS: Nurses (n=718) working in four Dutch nursing homes completed a questionnaire about work-related factors and musculoskeletal and psychosomatic complaints (response 81%). LISREL was used to describe and analyse the models. RESULTS: The full model had a reasonable overall fit. The strongest (and statistically significant) correlations were found between work postures and musculoskeletal complaints and between psychosocial work demands and psychosomatic complaints. In the two single models, health outcomes showed statistically significant correlations with the same work-related factors as they did in the full model. The full model could be slightly improved when the variable somatized complaints (e.g., dizzyness, headache) was related both to musculoskeletal complaints and to psychosomatic complaints. CONCLUSIONS: Although a model that simultaneously describes more dependent variables appears to provide a better approximation of the complexity of the actual relationship between work-related factors and health effects, the models in which only one health outcome is studied seem to give a proper description of the data at hand. This is an important finding because these single health outcome models resemble the usual analysis strategy by means of standard regression analysis. The adapted version of the full model suggests that musculoskeletal complaints of nurses can partly be understood as non-specific health complaints.
Authors: Donald C Cole; Sheilah Hogg-Johnson; Michael Manno; Selahadin Ibrahim; Richard P Wells; Sue E Ferrier Journal: Int Arch Occup Environ Health Date: 2006-05-31 Impact factor: 3.015