We are developing and applying new methodologies to research in epidemiology. This year work has progressed in 6 areas: (1) we have developed analytic methods for handling missing data and for fitting nonmultiplicative models, when analyzing data from case-control studies; (2) we have developed statistical methods for extracting information on disease incidence from data on disease mortality; (3) we have published guidelines for designing efficient and informative studies of early (i.e. unrecognized, but biochemically detectable) pregnancy loss; (4) we have developed extensions of earlier models for the probability of conception as related to timing of intercourse (in relation to ovulation), models which now incorporate effects of reproductive toxins and can offer valuable insight into the mechanisms by which such toxins impair human fertility; (5) we have developed strategies for analyzing retrospective time-to-pregnancy studies when the reproductive toxin of interest may have changed in prevalence over calendar time, producing 'time trend' bias; and (6) we have begun work aimed at improving analytic approaches for case-control studies where """"""""disease' status is based, not on physiology, but on circumstantial evidence involving a history of failures (versus history of successes), such as history of repeated miscarriage.
