To evaluate the capacity of the postcoital test (PCT) to predict spontaneous pregnancy in a large cohort study of subfertile couples.Prospective study.Department of reproductive medicine of 38 hospitals in the Netherlands.Between January 2002 and February 2004, we prospectively included consecutive subfertile couples who had not been evaluated previously for subfertility.We estimated the contribution of the PCT result to the existing prediction model for spontaneous pregnancy by calculating the adjusted hazard ratio (HR) of an abnormal PCT result. We constructed a second prediction model (PCT model) based on the reference model including the PCT.Primary endpoint in this study was ongoing pregnancy. We evaluated the performance of the PCT model in comparison with the reference model by calculating goodness of fit, discrimination, calibration, and the "net reclassification improvement".We included 3,021 couples of whom 537 (18%) had a spontaneous pregnancy and 55 (1.8%) a nonsuccessful pregnancy; 1,316 (44%) started treatment within 12 months, 824 (27%) neither started treatment nor became pregnant, and 289 (10%) became lost to follow-up within 12 months. The adjusted HR for an abnormal PCT was 0.76 (95% confidence interval [CI]: 0.62 to 0.94). The adjusted HR for an abnormal PCT was 0.63 (95% CI: 0.47 to 0.84) in case of no spermatozoa, 0.81 (95% CI: 0.57 to 1.2) in case of nonmotile spermatozoa, and 1.2 (95% CI: 0.8 to 1.8) in case of motile, nonprogressive spermatozoa. Adding the PCT result to the reference model did not improve goodness of fit. Discrimination was equally poor for the PCT model and the reference model. The calibration plots of both models showed comparably good calibration. The net reclassification improvement of the predictions of the PCT model compared with the reference model was -1.1%.This study demonstrated that the PCT has prognostic value but does not add substantially to a prognostic model for spontaneous pregnancy. American Society for Reproductive Medicine.
- Postcoital test
- prediction model
- hazard ratio
- net reclassification improvement