Reply: Pregnancy is predictable: a large-scale prospective external validation of the prediction of spontaneous pregnancy in subfertile couples

Abstract

Sir, We thank Drs Smeenk and colleagues for their interest in our work. The authors address several issues. First, they state that a prognostic model is only useful if it changes the prior probability for pregnancy in an accurate way, and suggest that the Hunault model does not change the prior probability of 32% that most couples would have for spontaneous conception within 12 months. We disagree with the interpretation of Smeenk et al. Figure 3 of our article shows clearly that the 10% of the couples with the best prognosis have a mean predicted probability of over 50%, and indeed 50% of these couples conceive. Even so, of the 20% of couples that have a predicted probability of around 20%, indeed < 20% conceive. We feel that such information is relevant for clinical practice. Secondly, Smeenk et al. remark that in 1104 couples (36%) the probability of a spontaneous pregnancy was over 40%, whereas only two deciles corresponding with 611 couples with a probability of over 40% were described in the calibration plot. The authors are correct to point this out, which provides us with the opportunity to clarify this apparent discrepancy. The number of 1104 couples was based on predictions with a model with the Post Coital Test (PCT), whereas the 611 couples were based on the model without the PCT.