Validity of self-reported physical work load in epidemiologic studies on musculoskeletal disorders

OBJECTIVES: This study assessed the validity of self-reported physical work load by questionnaire and logbook against task analysis and observation. It also investigated factors (job type and low-back or neck pain) affecting the self-assessment of physical work load and compared the assessments between the questionnaire and the logbook. METHODS: A self-administered questionnaire including 10 questions (ordinal scales) on physical work load and musculoskeletal symptoms was filled out by 2756 men in the forest industry. From this population, 36 men were selected for task analysis and observation. Logbooks including 10 continuous variables were analyzed for 386 men. RESULTS: The Spearman rank correlation coefficients between the self-assessments and observations for the frequency of manual handling, duration of trunk flexion, neck rotation, hand above shoulder level, and squatting or kneeling ranged between 0.42 and 0.55. The correlation coefficients for the questionnaire items were higher in general, and the accuracy better, for those with no low-back pain than for those with pain. The duration of trunk flexion, neck flexion and hand above shoulder level was overestimated in the questionnaires and less so in the logbooks. CONCLUSIONS: Self-administered questionnaires may help to classify groups with heterogeneous occupational tasks according to some work-load factors. The accuracy of the assessments is not good for studying quantitative exposure-effect relationships, however. The logbook method might give more valid information. The perception of musculoskeletal pain may bias the self-assessment of work load.