Objective: The ability of workers to accurately recall exposures that occur on the day of their injury is considered a potential limitation of case-crossover studies. This study assessed validity of occupational exposures reported by uninjured workers at a Midwestern meatpacking plant.
Methods: One hundred thirty-six workers were observed for 60 min while working and then interviewed within 8 days (median 3 days) about exposures during the observation period. The level of agreement between self-reports and direct observations was assessed using kappas and intraclass correlation coefficients.
Results: Excellent agreement was found between observed and reported work location (kappa = 0.97, 95% CI: 0.92-1.0), task (kappa = 0.83, 95% CI: 0.76-0.91) and tools used (kappa = 0.88, 95% CI: 0.81-0.95). Personal protective equipment varied by work type and location, and agreement between observed and reported usage varied from excellent to poor for various items. Excellent agreement was found for tool sharpening (kappa = 0.89, 95% CI: 0.82-0.97); good agreement for occurrence of break during the observation period (kappa = 0.60, 95% CI: 0.45-0.74); and poor agreement for equipment malfunction, line stoppages, being tired, unusual task, unusual work method, being distracted, rushing, slipping, or falling.
Conclusions: Agreement between observed and reported occupational exposures varied widely. Self-reported exposures are utilized in many occupational studies, and future exposure validity assessment studies should continue to improve retrospective study methods. Valid exposures will allow researchers to better understand injury etiology and ultimately prevent injuries from occurring.
(c) 2009 Wiley-Liss, Inc.