Selecting an exposure lag period - is the model with the largest exposure effect the best model.
Salvan-A; Stayner-LT; Steenland-K
Am J Epidemiol 1993 Oct; 138(8):596-597
In epidemiology, there is the inclination to consider more credible the larger estimates of exposure effect. For example, higher relative risks or rate ratios are, often emphasized as a criterion for choosing among various hypothesized induction-latency parameters, for example exposure-lag values. Whereas this "higher estimate" approach may often work in practice, the validity of such a criterion is not demonstrated. The purpose of this paper is to use examples to compare exposure-lag choices based on the highest estimate approach versus those based on a statistical goodness of fit criterion (likelihood ratio test). It seems likely that most epidemiologists believe that the two criteria are equivalent, that is, they would lead to the same choice of exposure-lag parameter estimates, or they may at the most show trivial differences. The examples shown are based on both published and artificial data, in which an exposure-lag parameter is estimated by trial and error fitting: the behavior of the goodness-of-fit statistic obtained over the assigned values of the parameter is compared with that of the relative risk. The examples show that there can be conspicuous inconsistencies between the highest-estimate and likelihood-based goodness-of-fit criteria. In addition, the highest-estimate criterion is shown to lack general validity. It is therefore recommended that, in the absence of a priori biologic knowledge, the selection of exposure-lag values be based on goodness-of-fit criteria.
Epidemiology; Exposure-assessment; Exposure-levels; Exposure-limits; Exposure-methods; Models; Occupational-exposure; Time-weighted-average-exposure; Work-environment; Worker-health; Workplace-studies; Work-practices; Biological-effects; Biological-factors; Biological-monitoring; Exposure-levels; Sample-preparation; Sampling; Standards; Statistical-analysis
American Journal of Epidemiology; Proceedings of The 26th Annual Meeting Of The Society For Epidemiologic Research, Keystone, Colorado, June 16-18, 1993