Health surveillance using an occupational medical database

This pilot study sought associations between liver function tests (LFTs) and membership in homogeneous exposure groups (HEGs) at a target plant as pre-clinical indications of possible future occupational health problems. A large company database yielded linear models for each of six LFTs (total bilirubin, alkaline phosphatase, gammaglutamyl transferase, lactate dehydrogenase, aspartate transaminase, and alanine transaminase) in terms of sex, body mass index, age, race (white/non-white), alcohol and cigarette consumption, and production/non-production (P/NP) job, permitting control for these in analyses of LFTs vs HEGs at the plant. These analyses, with HEG substituted for P/NP in the large group model, resulted in loosely "suspect" associations significant at P < 0.10. Collapsed HEG variable (containing "suspects" separately and all other non-significant HEG levels pooled) yielded "confirmed suspects" at P < 0.05 in the analysis of an independent LFT set taken at the plant approximately one year later.