|
Abstract:
|
Monitoring human fecundity is an important public-health issue; a common procedure being to assess time to pregnancy, the duration from the 'initiation' time when a couple starts trying to become pregnant until they succeed. Statistical tools for designing and analysing time to pregnancy studies belong to the general area of survival analysis, but several features require special attention. Recruiting at initiation is however difficult to carry out, so that prospective follow-up is more realistically achieved in a prevalent cohort design. The current duration design starts from a cross-sectional sample of couples currently trying to become pregnant, using the backward recurrence time as basis for the estimation of time to pregnancy or time to the end of the pregnancy attempt. On the basis of a simulation study, this paper studies the effect of unmeasured population heterogeneity on estimates of the distribution of time to pregnancy and of the effect of risk factors. An important result is that prevalent cohort studies are much more sensitive to unobserved heterogeneity than ordinary cohort (follow-up) studies.
|
