Federal monitoring system underestimates work-related injury and illness

The current national surveillance system may miss two-thirds of the total number of occupational injuries and illnesses, suggests a study in the April Journal of Occupational and Environmental Medicine, official publication of the American College of Occupational and Environmental Medicine (ACOEM).

Dr. Kenneth D. Rosenman and colleagues of Michigan State University, East Lansing, combined four data bases to identify work-related injuries and illnesses (resulting in more than 7 missed work days) occurring in Michigan from 1999 through 2001. The results were carefully matched to data from the national surveillance system for occupational injuries and illnesses, maintained by the U.S. Department of Labor's Bureau of Labor Statistics (BLS).

A total of 79,400 occupational injuries and illnesses were identified by the four combined data bases. In contrast, the BLS data base estimated that approximately 30,800 injuries and illnesses occurred during the three-year study period. Thus the BLS system failed to account for 61 percent of work-related injuries and illnesses.

An additional "capture-recapture" analysis–performed to identify cases missed by the combined data bases–suggested that the true total was 868,200 injuries and illnesses. Based on this figure, 68 percent of occupational injuries and illnesses were missed by the BLS system.

The BLS data base performed somewhat better in identifying occupational injuries, as opposed to illnesses. The accuracy of estimates varied by industry–the BLS data base captured 94 percent of injuries and illnesses for agriculture, compared with 45 percent in the transportation, communications, and electrical services industries.

The BLS is responsible for compiling accurate statistics on all "disabling, serious or significant" occupational injuries and illnesses. Previous studies have suggested that the current surveillance system–which uses a sampling strategy, rather than a census approach–misses some percentage of cases. In response to a 1987 study showing an undercount of work-related deaths, the BLS instituted a census system to gather more accurate data on occupational fatalities.

"Based on the results of our analysis we estimate that the number of work-related injuries and illnesses in Michigan is three times greater than the official estimate derived from the BLS annual survey," Dr. Rosenman and colleagues report. Whereas BLS statistics suggest that work-related injuries affect 1 in 15 Michigan workers per year, the new results suggest that the true rate is closer to 1 in 5.

Several factors likely contribute to the undercount–the BLS system excludes government workers and the self-employed, and employers and employees may perceive disincentives to reporting. A census approach, like that used to monitor work-related deaths, could improve reporting of injuries and illnesses as well. The investigators conclude, "A more comprehensive surveillance system for work-related injuries and illnesses would be useful to inform decision making on the allocation of public health resources to occupational health and safety…and to prioritize, target and evaluate both public health and enforcement activity to reduce work-related injuries and illnesses."