Design and evaluation of a mortality surveillance system for motor vehicle manufacturing workers

This investigation evaluates the validity and generalizability of a mortality surveillance system (MSS) based on Ford Motor Company and United Auto Workers (UAW) data. The Ford Work History File (WHF) is used to identify and follow up a cohort of 270,464 active and retired employees during 1978-1985. Workers who left employment without retiring (terminated employees) are withdrawn from follow-up one year after termination. Deaths (N = 12,130) are ascertained using the Ford Mortality File (FMF), the WHF and the Michigan and Ohio death files (MODF). Data from a conventional retrospective follow-up study (RFS) are used to evaluate the completeness of death ascertainment in the FMF and in the MSS. The MSS cohort is used to assess the validity of the RFS design versus the proportional mortality ratio (PMR) study design and of internal comparisons (Mantel-Haenszel mortality rate ratios (MHMRs) and Poisson regression models) versus external comparisons with the U.S. population (standardized mortality ratios (SMRs) and PMRs).

The FMF is only 90% complete, underascertainment being related to age and employment factors. MSS death ascertainment is 99.7% complete. MSS cohort mortality patterns confirm expectations about the relationship between SMRs and PMRs. Theoretical considerations show that SMRs are not less valid than PMRs in the presence of the "healthy worker effect" (HWE). SMRs and Poisson regression estimates are more valid than PMRs in evaluating mortality trends by length of employment and time since hire. Internal comparisons reduce the HWE in trend analyses and detect more associations with production and job groups than do external comparisons. SMRs have a higher predictive value than PMRs in detecting disease excesses found in MHMR analyses. RFS data show that terminated employees have higher mortality rates than active and retired employees. However, the overall MSS results are comparable to the RFS results.

An MSS based on the RFS design provides valid information on the mortality patterns of UAW-Ford workers. Analytic procedures employing both internal and external comparisons can assess widespread disease excesses or deficits, ensure powerful analyses by employment factors and reduce the HWE.