SELECTING AN EXPOSURE LAG PERIOD

In epidemiology, there is an inclination to consider more credible the larger estimates of exposure effect. For example, higher relative ris ks or rate ratios are often emphasized as a criterion for choosing amo ng various hypothesized exposure-lag values. An alternative criterion for this choice might be based on a goodness of-fit measure. We presen t examples, based on hypothetical data, in which an exposure lag param eter is:estimated by trial and error fitting: we compare the behavior of the likelihood-ratio goodness of-fit statistic ob rained over the a ssigned values of the parameter with that of the relative risk. We sho w that there can be inconsistencies between the highest estimate and l ikelihood based goodness-of-fit criteria. Concern about the validity o f the highest-estimate criterion prompts us to recommend that this cri terion not be used for the estimation of exposure weighting parameters , which should preferably be based on a priori biological knowledge ed ge or on goodness of-fit criteria.