Heterogeneity in patient populations explains differences in in vitro fertilization programs.

OBJECTIVE: To evaluate the efficacy of in vitro fertilization and embryo transfer (IVF-ET). DESIGN: Two different models for pregnancy rates in IVF-ET have been developed: a beta-geometric model and a split population model. SETTINGS: All IVF cycles were performed at the Yale University School of Medicine in New Haven, Connecticut. PATIENTS, PARTICIPANTS: All patients undergoing 1,257 IVF-ET cycles performed between the years 1983 through 1987. INTERVENTIONS: None. MAIN OUTCOME MEASURE: The probability of achieving a viable pregnancy per cycle. RESULTS: Because both models provide excellent explanations for our data distribution, the decline in the conditional probability of achieving pregnancy after a given number of unsuccessful cycles may be inferred. CONCLUSIONS: These findings question the justification of continuing IVF-ET treatment beyond some threshold number of cycles.