Capture-recapture estimates of nonfatal workplace injuries and illnesses

Ann Epidemiol. 2008 Jun;18(6):500-6. doi: 10.1016/j.annepidem.2007.11.003. Epub 2008 Feb 20.


Purpose: We examine reporting of nonfatal injury and illness reporting for the two most important sources of such data in the United States: workers' compensation data and the Bureau of Labor Statistics' (BLS) annual Survey of Occupational Injuries and Illnesses.

Methods: We linked individual case records from establishments reporting to the BLS with individual cases reported to workers' compensation systems in six states for 1998-2002 and used capture-recapture analysis to estimate the proportion of injuries reported. Data are for private sector workers and exclude mining, railroad and water transportation, temporary employment agencies, membership organizations, and small agricultural establishments.

Results: For injuries and illnesses eligible for income benefits, using conservative assumptions, we estimate that workers' compensation systems in the six states missed over 180,000 lost-time injuries in the sampled industries, that the BLS survey missed almost 340,000, and that about 69,000 injuries were unreported to either system.

Conclusions: Underreporting of nonfatal occupational injury and illness is substantial in both systems, but particularly in the Survey of Occupational Injuries and Illnesses. Using both sources improves coverage but falls far short of an accurate count for four of the six states. Reporting rates vary widely, so we cannot infer them for the entire United States.

Publication types

  • Research Support, N.I.H., Extramural
  • Research Support, Non-U.S. Gov't

MeSH terms

  • Accidents, Occupational / statistics & numerical data*
  • Data Interpretation, Statistical
  • Humans
  • Medical Record Linkage
  • Occupational Diseases / epidemiology*
  • Odds Ratio
  • Reproducibility of Results
  • United States / epidemiology
  • Workers' Compensation / statistics & numerical data
  • Wounds and Injuries / epidemiology*