Modified endometriosis fertility index Is more accurate to predict the non-ART pregnancy rate following surgery: a cohort of Chinese women

Abstract

Objective: To examine whether a modified endometriosis fertility index (EFI) can better predict the rate of pregnancy without assisted reproductive technologies (ART) after laparoscopic surgery in infertile women with endometriosis. Design: Retrospective cohort study. Setting: A university medical center. Population: 564 infertile patients who received laparoscopic surgery for endometriosis. Methods: 472 patients were used to modify the EFI based on new, optimal cutoffs for its predictor variables. The predictive accuracy of the modified EFI was examined in the other 92 patients. Main outcome measures: post-surgical non-ART pregnancy. Results: Among the patients for the EFI modification, the multivariable Cox regression results showed that historical factors were more contribution in predicting non-ART pregnancy rate than surgical factors both in modified EFI (C-index: historical factors 0.617 vs surgical factors 0.558) and the original EFI (C-index: historical factors 0.600 vs surgical factors 0.549). No significant relationship between the prior pregnancy and post-operative non-ART pregnancy rates was detected by both modified EFI and original EFI (P=0.530 and 0.802, respectively). In the patients for assessing the modified EFI, the predictive accuracy of two modified EFI models (C-index: 0.627 and 0.632, respectively) for non-ART pregnancy rates were superior to that of the original EFI (C-index: 0.602). Conclusions: A modified EFI based on population-specific optimal cutoffs and weight might be more suitable for estimating the rate of non-ART pregnancy after laparoscopic surgery in infertile women with endometriosis. Funding: This study received no external funding.