Results of studies that have assessed the accuracy of the Endometriosis Fertility Index (EFI) for predicting non-assisted reproductive technology (ART) pregnancy are inconsistent.
We intended to evaluate the accuracy of EFI for the prediction of non-ART pregnancy.
Embase, Medline, Scopus, and Web of Science were searched up to 5 Oct 2019.
We included studies that used EFI to predict non-ART pregnancy in patients with surgically documented endometriosis.
A total of 5547 studies were identified, from which we included 17 studies on 4598 women in the meta-analysis. Eight studies were classified as good quality, and the rest were considered to be of fair quality. Only five (29.41%) studies used appropriate approaches to account for potential confounders. Pooled effect sizes with corresponding 95% confidence intervals (CI) were calculated using random-effects models.
The cumulative non-ART pregnancy at 36 months was 10% (95%CI: 3, 16; P< 0.001) for women with an EFI of 0-2, which significantly increased to 69% (95%CI: 58, 79; P< 0.001) for women with an EFI of 9-10. Compared to women with with an EFI of 3-4 (18%, 95%CI: 12, 24; P< 0.001), the combined cumulative non-ART pregnancies were 44% (95%CI: 26, 63; P< 0.001) for women with an EFT of 5-6 and 55% (95%CI: 47, 64; P< 0.001) for women with an EFI of 7-8. Paired comparison by the chi-square test showed a significant difference between all categories (P< 0.001). The odds ratio (OR) for EFI was 1.33 (95% CI: 1.17 to 1.49, P< 0.001) and the summary area under the curve (AUC) was 72% (95% CI: 65 to 80, P< 0.001).
The current findings highlighted the good performance of the EFI score in predicting the non-ART pregnancy rate. However, these findings should be considered with caution due to the substantial heterogeneity between studies.

© 2020 Royal College of Obstetricians and Gynaecologists.