Chronic health problems and risk of accidental injury in the workplace: a systematic literature review.

OBJECTIVES To investigate whether common important health conditions and their treatments increase risks of occupational injury. METHODS A systematic search was conducted of MEDLINE, EMBASE and PsycINFO databases from inception to November 2006 employing terms for occupational injury, medications, and a broad range of diseases and impairments. Papers related solely to driving, alcohol, or substance abuse were excluded, as were studies that did not allow analysis of injury risk. For each paper that was retrieved we abstracted standard information on the population, design, exposure(s), outcome(s), response rates, confounders and effect estimates; and rated the quality of information provided. RESULTS We found 38 relevant papers (33 study populations): 16 studies were of cross-sectional design, 13 were case-control and 4 were prospective. The overall quality was rated as excellent for only two studies. Most commonly investigated were problems of hearing (15 studies), mental health (11 studies) and vision (10 studies). For impaired hearing, neurotic illness, diabetes, epilepsy and use of sedating medication there were moderate positive associations with occupational injury (odds ratios 1.5-2.0), but there were major gaps in the evidence base. Studies covering vision did not present risks by category of eye disease; no evidence was found on psychotic illness; for diabetes, epilepsy and cardiovascular disease there were remarkably few papers; studies seldom distinguished risks by sub-category of external cause or anatomical site and nature of injury; and exposures and outcomes were mostly ascertained by self-report at a single time point, with a lack of clarity about exposure timings. CONCLUSION Improved research is needed to define the risks of occupational injury arising from common health complaints and treatments. Such research should delineate exposures and outcomes in more detail, and ensure by design that the former precede the latter.