Outmigrant ascertainment for bias assessment in environmental epidemiology

In the summer of 1985, the McGill Epidemiology Rapid Response Unit undertook a study comparing two areas exposed to emissions of 'sour gas' refineries to an unexposed area. One operational objective of the project was the identification and survey of all the people who had lived in the study area but had since moved (outmigrants).

We estimated the number of outmigrants (people who had ever lived in the area during the period 1957 to 1985) to be 3363 by using information obtained from our cross-sectional survey and from population statistics for the area of interest. Ten different methods combined lead to the identification of approximately 87% of all the outmigrants who left the study area during that period. We used vital statistics to identify the outmigrants who had died and mailed questionnaires to obtain the necessary information from the others.

We confirmed the vital status of approximately 46% of them (1532/3363). The results from the outmigrant survey showed that they were younger than area residents, that they experienced lower rates of heart disease and hypertension and that they had moved for health reasons in only 1.3% of the cases. These findings were similar across comparison areas.

We concluded that there was no effect due to migration bias on the cross-sectional study results. This evidence considerably strengthened the conclusions regarding the effects of exposure, a benefit that largely justified the cost of identifying and surveying the outmigrants.