NYC Injury Headlines

Occupational Injury Costs Per Employee Pinpointing the Risks

Objectives: 1. To estimate the medical and work loss costs of lost-workday occupational injuries reported to the Bureau of Labor Statistics (BLS). 2. To determine which occupations, industries, sources, events, age groups, and gender are associated with the highest costs per employee. Methods: Work Loss Costs. BLS annual survey data show work days lost through a fixed date. Therefore, durations are censored for some cases. By major injury grouping, we built and applied non-linear regression models to estimate the full duration for censored cases in the 1993 annual survey. This was a massive modelling effort. It corrected for heterogeneity in the data and accounted for the existence of permanently and totally disabling injuries. Once work-loss durations were available for all cases, we developed algorithms to compute lost wages. One approach used wages by occupation, industry, sex, and age group from the Current Population Survey. A problem with this approach is that an executive's injury can be weighted much more heavily than a production worker's, obscuring where the injury problems lie. A second, more egalitarian approach used an average daily wage loss. The second approach facilitates injury risk comparisons between groups, but does not accurately depict employer or societal costs. The work loss costs were supplemented by fringe benefit costs and by household work loss costs. Household work loss was estimated from work loss duration, data showing workers typically return to housework 10% sooner than wage work (but possibly trading for less demanding tasks), and studies of the value of housework.

Medical Costs. Medical costs by diagnosis were derived separately by injury diagnosis for hospital-admitted and non-admitted cases. Diagnosis-specific costs for admitted cases came from national average lengths of stay for cases covered by Workers Compensation (from 1987-1992 National Hospital Discharge Survey data) and costs per hospital day from states where cost control regulatory agencies force hospitals to accurately report these costs. Post-discharge costs in the acute care phase came from 1987 National Medical Expenditure Survey (NMES) data. Longer-term costs came National Council on Compensation Insurance Detailed Claims Information (DCI) data. For non-admitted cases, costs per visit came from Civilian Health and Medical Program of the Uniformed Services (CHAMPUS) data. NMES described short-term visits per case and DCI described long-term costs. 1987-1992 National Health Interview Survey data were used to compute the number of medically treated cases without work loss from the BLS lost workday case counts. Data scarcity forced us to perform these computations for broad diagnosis groupings.

Results. The analysis is in process. The data set lets us rank costs per employee by occupation, industry, source (e.g., a machine), event (e.g., a fall), body region, diagnosis, age group, and gender.