Recorded presentation

Elizabeth Gerken Hooten, MSPH ScD, Center for Health Promotion and Disease Prevention, University of North Carolina at Chapel Hill, 1700 Airport Road Suite 248, CB # 8140, Chapel Hill, NC 27599-8140, 919-966-6040, hooten@unc.edu, Susan P. Baker, MPH, Center for Injury Research and Policy, Johns Hopkins University School of Hygiene and Public Health, 624 North Broadway, 5th Floor, Baltimore, MD 21205, and Shrikant Bangdiwala, PhD, Biostatistics Department-CSCC, School of Public Health, University of North Carolina at Chapel Hill, CB# 8030, Suite 203, Nations Bank Plaza, 137 E. Franklin Street, Chapel Hill, NC 27514-4145.

Fatal injury at work seems easy to define and therefore, to enumerate its incidence. Imprecise definitions that vary by data source combined with limited information on the nature of work activity at the time of injury influence characterization of an injury as on-the-job or not. Given this variation, assessment of major data sources using analytic techniques is a logical extension of general surveillance system evaluation. This research employed capture-recapture (CR) methods to estimate the incidence of fatal occupational injury in North Carolina for a two-year period (1994-1995). The NC Census of Fatal Occupational Injuries (CFOI) reported 413 fatal occupational injuries. Four hundred and seventy-one cases were identified using combined data from the Medical Examiner (ME), Death Certificate (DC), Occupational Safety and Health, Workers’ Compensation Claims, Military and the National Transportation Safety Board data sources. Matching indicated about 12% of the total (60 deaths) for the two year period was missed by CFOI. This was confirmed using a two-sample CR analysis using ME and DC data. Attempting to add more than DC and ME data to the models resulted in unstable parameter estimates with very wide confidence intervals. CR methods are best employed with sources where confidence in the information affording case matches is present and where there is variance among sources to allow for point estimation with confidence intervals. Where concordance between data sources is high, little is gained in estimating underascertainment because discordance is the very thing that allows one to place parameters on what is missed.

Learning Objectives: At the conclusion of this presentation, the participant in this session will be able to

• Articulate the strengths and limitations of using capture-recapture methods in fatal occupational injury research.
• Describe the need for rethinking the definitional scheme used for classifying injury-at-work fatalities.

Keywords: Injury Control, Surveillance

Presenting author's disclosure statement:
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.

Recorded presentation