Joint modeling of success and treatment discontinuation in in vitro fertilization programs: a retrospective cohort study.

Abstract : ABSTRACT: BACKGROUND: As discontinuation in in vitro fertilization (IVF) programs has been associated with a poor prognosis, one hypothesis is that some couple-specific predictive factors in IVF may be shared with opposite effect by both successful (i.e. live birth) and treatment discontinuation processes. Our objective was to perform a joint analysis of these two processes to examine the hypothesis of a link between the two processes. METHODS: Analyses were conducted on a retrospective cohort of 3,002 women who began IVF between 1998 and 2002 in two French IVF centers: a Parisian center and a center in a medium-sized city in central France. A shared random effects model based on a joint modelization of IVF treatment success and discontinuation was used to study the link between the two processes. RESULTS: Success and discontinuation processes were significantly linked in the medium-sized city center, whereas they were not linked in the Parisian center. The center influenced risk of treatment discontinuation but not chance of success. The well-known inverse-J relation between the woman's age and chance of success was observed, as expected. Risk of discontinuation globally increased as the woman's age increased. CONCLUSIONS: The link between success and discontinuation processes could depend on the fertility center. In particular, the woman's decision to pursue or to discontinue IVF in a particular center could depend on the presence of other IVF centers in the surrounding area.