Objective: To evaluate whether basal FSH (bFSH; measured on menstrual day 1-4) adds relevant clinical information to the prediction of ongoing pregnancy rates (OPRs) after IVF, once age and diagnostic characteristics have been taken into account.
Design: Retrospective.
Setting: Academic fertility center.
Patient(s): 435 women undergoing their first IVF cycle.
Intervention(s): None.
Main outcome measure(s): Ongoing pregnancy rate.
Result(s): The likelihood ratio of bFSH as a single prognosticator for treatment failure at a cutoff level of 15 IU/L was 3.87. The proportion of patients with such a bFSH level was 5%. Multivariate logistic regression analysis selected age, bFSH level, and infertility diagnosis as relevant predictors of ongoing pregnancy. When compared to a predictive model for OPRs based on age and infertility diagnosis, the inclusion of bFSH into this model helped to identify more patients (22 vs. 1) whose predicted OPR decreased from a low level (5%-12%) towards an extremely low level (<5%).
Conclusion(s): An acceptable performance of bFSH as a single test to predict treatment failure is only obtained above a high cutoff level. Thus, the number of patients for whom bFSH provides relevant information is small. The predictive model including bFSH identified significantly more patients with an extremely poor prognosis than did the predictive model without bFSH. However, predictions based solely on age and infertility diagnosis usually were already poor in these patients. Measurement of bFSH adds little in only a few patients and is, therefore, debatable.