An estimate of the BLS undercounts of nonfatal occupational injuries among small construction establishments

The accuracy of occupational safety and health surveillance data is the key to effective interventions. However, the data collected by the U.S. Bureau of Labor Statistics (BLS) appears to undercount nonfatal occupational injuries in construction, particularly for small establishments which comprise most of this industry.

To estimate the breadth of underreporting in the construction industry, we stratified fatal and nonfatal injuries by establishment size and Hispanic ethnicity. The hypothesis behind our models is that nonfatal injury rates are proportional to fatal injury rates across these subgroups. To control for the effects of changes in the data systems, such as changes in industrial and occupational coding systems and OSHA reporting criteria, data from 2003 through 2006 were used and pooled together for more reliable estimates. In addition to data from the Census of Fatal Occupational Injuries and the Survey of Occupational Injuries and Illnesses, several datasets from different sources were used for the data analyses, including County Business Patterns, Current Population Survey, National Health Interview Survey, and Medical Expenditure Panel Survey. Univariate and multivariate analyses were conducted using SAS and SUDAAN.

Our initial findings indicate that the BLS nonfatal data collection missed a large number of injuries in the construction industry. Among Hispanic construction workers employed in small establishments (1-10 employees), about two-thirds of nonfatal injuries were not captured in the BLS data. The results suggest that an improved nonfatal occupational injury data collection is urgently needed.