Endometrial scratching in frozen embryo transfer: A stratified cohort study.

Abstract

Background: Endometrial scratching (ES) is widely used in assisted reproduction. However, randomized trials have shown no benefits in unselected populations, and evidence specific to frozen embryo transfer (FET) is limited. Therefore, we estimated the association between ES and the clinical pregnancy rate (CPR) separately within strata defined by the embryo transfer (ET) history.

Methods: In this retrospective cohort study at a tertiary center in Taiwan (January 2017-January 2021), nulliparous women aged <40 years who underwent FET were eligible. ES was performed with a Novak curette in the cycle preceding FET, and analyses were stratified by the ET history (none, 1-2, or ≥3). The primary outcome was the CPR at 6 to 8 weeks of gestation. Multivariate logistic regression within each stratum was adjusted for prespecified covariates. Differences between strata were not formally tested.

Results: Of the 822 women who were enrolled, 176 (21.4%) underwent ES. The overall CPR did not differ between the groups (46.6% vs 53.6%; p = 0.100). In women undergoing first FETs (n = 457), ES was associated with a lower crude CPR (39.4% vs 58.1%; odds ratio: 0.47, 95% CI, 0.27-0.80; p = 0.005) but not after adjustment (adjusted odds ratio [aOR]: 0.59, 95% CI, 0.31-1.15; p = 0.121). The CPR was numerically higher with ES in women with 1 to 2 previous ETs (n = 295; 50.0% vs 46.9%; aOR: 1.29, 95% CI, 0.62-2.66; p = 0.499) and in those with ≥3 previous ETs (n = 70; 54.2% vs 45.7%; aOR: 1.54, 95% CI, 0.19-12.40; p = 0.687); all adjusted associations were nonsignificant. Early miscarriage and ectopic pregnancy rates did not differ across the strata.

Conclusion: ES was not associated with improved clinical pregnancy in this FET cohort, and its routine use in treatment-naïve patients with good prognosis was not supported. In women with previous ET failures, the estimates favored ES but were imprecise, nonsignificant, and exploratory. Therefore, adequately powered randomized controlled trials are required.