Citation

Pollack KM, Agnew J, Slade MD, Cantley L, Taiwo O, Vegso S, Sircar K, Cullen MR. Am. J. Ind. Med. 2007; 50(9): 676-686.

Affiliation

Department of Health Policy and Management, Center for Injury Research and Policy, The Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland 21218, USA. kpollack@jhsph.edu

Copyright

(Copyright © 2007, John Wiley and Sons)

DOI

10.1002/ajim.20493

PMID

17676586

Abstract

BACKGROUND: Employer administrative files are an underutilized source of data in epidemiologic studies of occupational injuries. METHODS: Personnel files, occupational health surveillance data, industrial hygiene data, and a real-time incident and injury management system from a large multi-site aluminum manufacturer were linked deterministically. An ecological-level measure of physical job demand was also linked. This method successfully created a database containing over 100 variables for 9,101 hourly employees from eight geographically dispersed U.S. plants. RESULTS: Between 2002 and 2004, there were 3,563 traumatic injuries to 2,495 employees. The most common injuries were sprain/strains (32%), contusions (24%), and lacerations (14%). A multivariable logistic regression model revealed that physical job demand was the strongest predictor of injury risk, in a dose dependent fashion. Other strong predictors of injury included female gender, young age, short company tenure and short time on current job. CONCLUSIONS: Employer administrative files are a useful source of data, as they permit the exploration of risk factors and potential confounders that are not included in many population-based surveys. The ability to link employer administrative files with injury surveillance data is a valuable analysis strategy for comprehensively studying workplace injuries, identifying salient risk factors, and targeting workforce populations disproportionately affected.

Language: en