MEASUREMENT OF PERSONAL RISK BEHAVI-OR IN OCCUPATIONAL RISK STUDIES AMONG HEALTH CARE WORKERS.

Risky behaviours, particularly illegal and heavily stigmatized behaviours, are difficult to measure through self-report in both high risk groups and the general population. Underreporting can result in substantially biased estimates of non-injection drug use (IDU) risk of hepatitis C virus (HCV) infection. We hypothesized that asking about the existence of social networks injection drug use may be a useful marker of IDU. A cross-sectional survey of physicians and nurses was conducted in seven hospitals in Georgia. Based on survey responses participants were categorized into three IDU risk groups: ever used injecting drugs (Self IDU), reported a friend, family member or colleague used injecting drugs (Associate IDU), or reported neither (No IDU). Testing on anti-HCV was done using third generation ELISA methods. Both unadjusted and adjusted prevalence ratios between IDU risk groups and HCV prevalence were estimated. Of the 1312 (82.2%) participants, 10 (0.8%), 75 (5.7%), and 1227 (93.5%) were categorized as Self IDU, Associate IDU and No IDU, respectively; with HCV prevalence of 20%, 9.3% and 4.6%, respectively (p=0.016). The association was due primarily to women's reports. Those who reported some IDU risk were more likely to report other personal risk behaviors (e.g., multiple sex partners) and occupational risk behaviors (e.g., frequent exposure to blood and body fluids). This study represents a start of measurement development by assessing the potential usefulness of a marker to measure of IDU. Improved measurement of stigmatized behaviors is needed for confounding adjustment to improve estimates of occupational risks of blood-borne infections.