Using Workers’ Compensation Data for Occupational Safety and Health: A Work in Progress

By Anthony Robbins

On 19 June, the National Institute for Occupational Safety and Health and other Federal agencies and private sector groups concerned with worker health convened a two-day workshop at the Department of Labor’s Frances Perkins Building in Washington.  About 100 researchers gathered to discuss how workers compensation data could be analyzed and used to study worker safety and health. NIOSH asked me to moderate the first session. I was flattered, but for all my roles in public health since leaving NIOSH’s directorship in 1981, I was surely not an expert in workers’ compensation.  For the last few years, as a Professor at Tufts, I had settled into editing the Journal of Public Health Policy.

I diligently read the papers, as I had a lot of catch-up to do.  I remembered that in 1979 I had asked my colleagues at NIOSH about using workers’ comp data to study injuries and illnesses, and I think I was told that NIOSH was not doing that. As I read the workshop papers, I realized that today, a sizeable group of researchers, including several from NIOSH, was mining workers’ comp data.  What could I add to their excellent work?

The workshop papers deserve more than a casual look. They will be cleaned-up and published (as were the presentations and posters from the first workshop).  They shed light on key issues:

1.    Preventing injuries in the health care industry: Lisa Pompeii and her coauthors, including my colleague John Dement from NIOSH days, explained how hospitals, already attuned to illness and injuries, are particularly quick to make use of workers compensation data.

2.    Costs of injuries and illnesses: In a session moderated by Les Boden from Boston University, Seth Seabury of RAND Corporation described progress in linking compensation data to earnings to study the economic impact of disabling injuries.

3.    Special problems facing contingent workers and those who try to protect them: The SHARP program in Washington’s Department of Labor and Industries used the state’s compensation claims database to study occupational safety and health in the “temporary help supply industry.” For nearly every risk class, claims rates are higher for temporary than permanent workers.

4.    How some programs have used workers comp data to help protect workers: There were presentations from British Columbia, California, and Ohio, but one study garnered particular attention. A study by Michael Toffel from Harvard Business School suggested that random state safety inspections reduced injuries without causing any detectable job loss.

5.    How state and provincial workers compensation data compared with other information: These presentations demonstrated, beside many particular findings, the virtues of universal medical insurance in Canadian provinces for generating reliable data.

6.    How leading indicators can be found in state-level data and other methodological issues:  These presentations further explored why different data sources give different answers.  What are the sources of undercounts or over counts of occupational injuries?

Reflecting the growing interest in this field, the organizers created posters sessions to accommodate more presentations.  I was particularly intrigued by Frank Neuhauser’s research that suggested that “after controlling for risk and exposure, injury risk declines substantially with age for men, but remains constant or increases for women.”

I struggled with what to say in my short introduction before the timer’s light would go from green to yellow to red. With my editor’s cap on, I tried to be helpful. I told the audience of researchers that I detected a problem I had seen before.  Insufficient effort had gone into making the research results understandable and useable by policymakers and program managers–non-experts who might be able to improve worker protection.  I acknowledged that to do so, may not be easy because the research is often sophisticated and methodologically complex.  But if researchers expect their results to be used, they must be sure that there are users prepared to grasp the findings and the implications for policy and programs.  Targeted users may need education about workers comp data and studies that use them.

It would be worth revisiting the experience at the National Center for Health Services Research in the 1970s, when the Center used some of its limited appropriated funds to convene legislators and health program managers from around the country to teach them how to use NCHSR’s research and health services research, in general.  As these potential users learned the value of health services research, the Center was rewarded by a rapid expansion of its budget.

Standing on the outside looking into workers’ compensation research, I remain concerned that workers’ comp data will have limited value until the same Federal agencies and outside groups that organized the workshop invest in educating potential users about how to use these data.

Anthony Robbins, MD, MPA is co-Editor of the Journal of Public Health Policy.