Can we use workers’ compensation data as an effective occupation health and safety tool?

A few days ago, I was at a CDC-NIOSH workshop in Washington DC on the topic of using workers’ compensation data for occupational safety and health. This event was organized around six draft white papers that really illustrate the potential, the barriers, and the pitfalls of this pursuit.


The first session highlighted some of the successes in using workers’ comp data in health care injury prevention. Several papers demonstrated the advantages of linking workers’ comp data with other data files. Duke University, for example, has 30,000 employees and the data in its various electronic systems can be linked on employee ID. The research presented, while preliminary, showed an elevated relative risk of musculoskeletal injuries for certain health care employees over what might be expected. It also showed that certain administrative changes have a big impact on injury rates.

The second of the themed white papers looked at the burden of occupational injuries, illnesses and fatalities. By burden, we mean the total impact on society not just the dollar impact of claims, employer cost, or loss to the worker or family. Each of these components is important and needs to be part of the equation. When that total burden is calculated, the benefit-to-cost ratio of prevention becomes more evident. In this session, there were a number of papers. For me, Seth Seabury’s presentation on the work RAND and others have done through the linking of earnings data was a highlight. The research shows that workers who suffer a work-related injury lose income relative to their non-injured counterparts. The research technique that yields this result has been tried in CA, MN, MI, NM, OR, WA, WI, Ontario and BC. While far from a national or full North American perspective, the similarity of findings is helping to quantify that loss and allow policy makers to assess the adequacy of compensation.

The third white paper dealt with the contingent workers, a growing segment of the workforce. By “contingent workers”, we mean those without an explicit or implicit contract for employment. Day labour on a construction site might be an example. In some cases, contingent workers are hired to do jobs that are considered of lower value but may be of higher risk. Contingent workers are often not captured in data sets such as the Survey of Occupational Injuries and Illnesses (SOII). Temp agencies, professional employee organizations (PEOs) and some labour contractors pose challenges for prevention, enforcement, and underwriting.

Our white paper on the use of workers’ compensation data for loss prevention provided a backdrop for presentations from WorkSafeBC, Ohio BWC, and the California Department of Industrial Relations, among others. It also fit in well with NIOSH Director, John Howard’s remarks on the importance of workers’ compensation and its data to primary, secondary and tertiary prevention. His remarks were consistent with many of the presentations that underscored the need for better metrics and standards for the data.

The fifth white paper focused on state-level analysis of workers’ compensation data for public health purposes. The presentation by Jennifer Wolf-Horejsh, executive director for IAIABC, highlighted the work of that organization in standardizing reporting through its EDI (electronic data interchange) standard and training. Presentations from various groups during this session also revealed the degree to which data from other sources and innovative techniques may be used to confirm workers’ compensation data particularly estimating the cases not captured by various data sets. The examination OSHA recordkeeping practices and workers’ compensation claims in Washington was particularly revealing and points to some of the sources of error (undercounting) that may well exist in reported work injuries.

The final white paper on leading indicators provided a great opportunity to reflect on how we might better use data for prevention purposes. One intriguing approach was Allard Dembe’s paper on O*Net. O*Net, the Occupational Information Network, (an outgrowth of the old Dictionary of Occupational Titles so many of us used in the past). It describes nearly a thousand occupations with 277 descriptors for each occupation. His research, while very preliminary, demonstrated how some of these factors might be used to predict workers’ compensation claim outcomes.

The two-day workshop provided the hundred or so researchers with a wonderful opportunity to focus on some leading edge research. When the white papers and presentations are finalized and proceedings published, I am sure they will prove a valuable resource for other researchers, policy makers, and practitioners