Prospective validation of two models predicting pregnancy leading to live birth among untreated subfertile couples

BACKGROUND: Models predicting clinical outcome need external validation before they can be applied safely in daily practice. This study aimed to validate two models for the prediction of the chance of treatment-independent pregnancy leading to live birth among subfertile couples.

METHODS: The first model uses the woman's age, duration and type of subfertility, percentage of progressive sperm motility and referral status. The second model in addition uses the result of the post-coital test (PCT). For validation, these characteristics were collected prospectively in two University hospitals for 302 couples consulting for subfertility. The models' ability to distinguish between women who became pregnant and women who did not (discrimination) and the agreement between predicted and observed probabilities of treatment-independent pregnancy (calibration) were assessed.

RESULTS: The discrimination of both models was slightly lower in the validation sample than in the original sample which provided the model. Calibration was good: the observed and predicted probabilities of treatment-independent pregnancy leading to live birth did not differ for both models.

CONCLUSIONS: The chance of pregnancy leading to live birth was reliably estimated in the validation sample by both models. The use of PCT improved the discrimination of the models. These models can be useful in counselling subfertile couples.