Assessing the accuracy of OSHA's projections of the benefits of new safety standards.
Am J Ind Med 2004 Apr; 45(4):313-328
BACKGROUND: In the preambles to the safety and health standards that it has issued since 1987, the Occupational Safety and Health Administration (OSHA) projected that new safety standards would prevent over 2,600 death each year. For six safety standards issued since 1990, we compare OSHA's projections of the impact of full compliance on fatalities with actual fatality changes and examine the reasons for the differences. METHODS: We reviewed the preambles to OSHA standards and the Regulatory Impact Analyses (RIAs) prepared for them to identify the baseline and the prevention factor that the agency used to project the number of deaths that would be prevented. We used three data sources to track the relevant categories of fatalities: the Census of Fatal Occupational Injuries (CFOI), the National Traumatic Occupational Fatality program, and OSHAs Fatality/Catastrophe investigations. RESULTS: For all six standards, OSHA appeared to overestimate the number of deaths prevented. The availability of CFOI led to better estimates of the fatality baseline, but the prevention factor was always overestimated, especially for standards which emphasized training. CONCLUSIONS: OSHA needs to develop better methods for projecting injury impacts. Research is needed to help OSHA predict the effects of behavioral requirements (e.g., training) on actual work practices and injury outcomes. For non-fatal injuries, new methods of data collection will be required.
Workplace-studies; Injuries; Health-protection; Health-hazards; Construction-industry; Construction-workers; Injury-prevention; Accident-prevention; Statistical-analysis; Behavior-patterns
John Mendeloff, Graduate School of Public Health and Graduate School of Public and International Affairs, University of Pittsburgh, 3E34 Posvar Hall, Pittsburgh, PA 15260
Research Tools and Approaches; Intervention Effectiveness Research
American Journal of Industrial Medicine
University of Pittsburgh, Pittsburgh, PA 15260