A case study of the applicability of a prediction model for the selection of patients undergoing in vitro fertilization for single embryo transfer in another center

Objective: To evaluate the application in a different fertility clinic of a prediction model for selecting IVF patients for elective single embryo transfer. Design: Retrospective analysis of a large database obtained from a tertiary infertility center. Setting: University medical center.
Patient(s): The model, derived at the "development center" was applied in 494 consecutive first IVF cycles carried out at the "application center".
Intervention(s): After adjustment of embryo scoring system to be compatible with that used by the prediction model, it was applied to the development center data. A score chart for predicting the probability of singleton or twin pregnancy was constructed. Main Outcome
Measure(s): The area under the receiver operator curve (ROC) was determined to measure the ability of the model to discriminate between ongoing pregnancy and twin pregnancy. Calibration plots were made to assess agreement between predicted and observed pregnancy rates (PR).
Results: The areas under the ROC for predicting ongoing pregnancy and twin pregnancy were 0.63 and 0.66, respectively. Insertion of a correction factor equivalent to the difference in odds ratios for ongoing PR between the two centers was required to improve the calibration of the model.
Conclusion(s): After adaptation, the model performed well in the application center.