Medium- and long-term reproducibility of self-reported exposure to physical ergonomics factors at work

The literature is sparse on reproducibility of self-reported exposure to physical ergonomics risk factors for musculoskeletal disorders (MSDs). Aims of this study were to evaluate, in a cohort of workers interviewed up to three times: 1-year test-retest reliability; and 5- and 6-year recall of physical exposures. We also examined whether reproducibility was influenced by the presence of UE MSD or by technological changes introduced between the last two surveys.

A cohort of automobile manufacturing employees was interviewed at baseline, one and six years later about work history, physical and psychosocial exposures at work, upper limb symptoms, injury and medical history, and demographics. Agreement between interviews was evaluated by intraclass correlation and Spearman coefficients. Differences in exposure between 1- and 6-year follow-up were analyzed by Wilcoxon matched-pairs signed-ranks test.

Large and significant decreases in work pace and physical effort were observed from baseline, although an upper extremity composite index was quite stable in the total population. One-year test-retest reliability was fair to good for the composite exposure index (ICC=0.58), whole-body vibration, handling parts, and tool use, but poor for the other variables considered. Long-term reproducibility, from baseline or 1-year follow-up to 6-year follow-up, was poor for the composite index and almost all single items. UE MSD case status influenced 1-year test-retest reliability, with subjects who changed case status from baseline displaying higher reliability, but not reproducibility of recalled exposures. A strong regression to the mean effect was observed on exposures reported at follow-up surveys.

Recalled ergonomics exposures could be employed in retrospective cohort studies as a somewhat reliable and unbiased estimate of the self-reported exposures that would have been obtained up to one year earlier, but not over a longer period (5-6 years). These longer-term results may have been limited by difficulty in matching jobs between interviews; also the regression to the mean effect likely contributed to reduce agreement. Changes in production technology and work organization produced a decrease in physical workload intensity and job pace, but did not have a substantial impact on an exposure index for the upper limb.