Validity of a mortality study based on a corporate health surveillance system

We evaluated the Shell Health Surveillance System (HSS) by conducting two mortality studies at an oil refinery. Study A used the HSS to measure the mortality of active and retired workers during 1973 to 1982. Study B used additional information sources and followed up terminated employees. For subjects included in both studies, results were very similar. However, the mortality experience of terminees before 1973 (included only in study B) was different from that of study A subjects, reflecting differences in length of employment and time since hire. HSS-based studies provide valid measures of long-term effects of past exposures among retirees and of short-term effects of recent exposures among active employees. However, they cannot detect short-term effects of past exposures, and they have limited power for evaluating dose-response relationships.