Management of early pregnancy loss with mifepristone and misoprostol: clinical predictors of treatment success from a randomized trial

Background: Early pregnancy loss is a common event in the first trimester, occurring in 15%–20% of confirmed pregnancies. A common evidence-based medical regimen for early pregnancy loss uses misoprostol, a prostaglandin E1 analog, with a dosage of 800 μg, self-administered vaginally. The clinical utility of this regimen is limited by suboptimal effectiveness in patients with a closed cervical os, with 29% of patients experiencing early pregnancy loss requiring a second dose after 3 days and 16% of patients eventually requiring a uterine aspiration procedure. Objective: This study aimed to evaluate clinical predictors associated with treatment success in patients receiving medical management with mifepristone–misoprostol or misoprostol alone for early pregnancy loss. Study Design: We performed a planned secondary analysis of a randomized trial comparing mifepristone–misoprostol with misoprostol alone for management of early pregnancy loss. The published prediction model for treatment success of single-dose misoprostol administered vaginally included the following variables: active bleeding, type of early pregnancy loss (anembryonic pregnancy or embryonic and/or fetal demise), parity, gestational age, and treatment site; previous significant predictors were vaginal bleeding within the past 24 hours and parity of 0 or 1 vs >1. To determine if these characteristics predicted differential proportions of patients with treatment success or failure, we performed bivariate analyses; given the small proportion of treatment failures in the combined treatment arm, both arms were combined for analysis. Thereafter, we performed a logistic regression analysis to assess the effect of these predictors collectively in each of the 2 treatment groups separately as well as in the full cohort as a proxy for the combined treatment arm. Finally, by using receiver operating characteristic curves, we tested the ability of these predictors in association with misoprostol treatment success to discriminate between treatment success and treatment failure. To quantify the ability of the score to discriminate between treatment success and treatment failure in each treatment arm as well as in the entire cohort, we calculated the area under the curve. Using multivariable logistic regression, we then assessed our study population for other predictors of treatment success in both treatment groups, with and without mifepristone pretreatment. Results: Overall, 297 evaluable participants were included in the primary study, with 148 in the mifepristone–misoprostol combined treatment group and 149 in the misoprostol-alone treatment group. Among patients who had vaginal bleeding at the time of treatment, 15 of 17 (88%) in the mifepristone–misoprostol combined treatment group and 12 of 17 (71%) in the misoprostol-alone treatment group experienced expulsion of pregnancy tissue. Among patients with a parity of 0 or 1, 94 of 108 (87%) in the mifepristone–misoprostol treatment group and 66 of 95 (69%) in the misoprostol-alone treatment group experienced expulsion of pregnancy tissue. These clinical characteristics did not predict treatment success in the combined cohort alone (area under the curve=0.56; 95% confidence interval, 0.48–0.64). No other baseline clinical factors predicted treatment success in the misoprostol-alone treatment arm or mifepristone pretreatment arm. In the full cohort, the significant predictors of treatment success were pretreatment with mifepristone (adjusted odds ratio=2.51; 95% confidence interval, 1.43–4.43) and smoking (adjusted odds ratio=2.15; 95% confidence interval, 1.03–4.49). Conclusion: No baseline clinical factors predicted treatment success in women receiving medical management with misoprostol for early pregnancy loss. Adding mifepristone to the medical management regimen of early pregnancy loss improved treatment success; thus, mifepristone treatment should be considered for management of early pregnancy loss regardless of baseline clinical factors.