Abstract
BACKGROUND: Recurrent miscarriage is a debilitating disorder associated with considerable physical and psychological morbidity. An estimated 50% of first trimester miscarriages remain unexplained. The aim of this study was to provide a personalised framework to guide the expectations of women experiencing recurrent miscarriage, with the ultimate goal of transforming clinical practice. METHODS: We used real-world data from a UK longitudinal study of 1201 couples attending National Health Service (NHS) miscarriage clinics, with a history of previous miscarriages, comprising medical and obstetric history, results of investigations and pregnancy and neonatal outcome. We developed, parametrised, and validated predictive models for the probability that the next pregnancy is viable and for the time to next pregnancy. Time to next pregnancy separates couples into two groups, a group with subfertility, i.e., delay in conception, and a group with no significant delay in conception. We used Bayesian inference for the latter model.Trial registration number: The prospective data collections were pre-registered ISRCTN17732518; https://doi.org/10.1186/ISRCTN17732518. FINDINGS: Predictive models of the time to pregnancy, the probability of the couple being subfertility and the probability of having a viable pregnancy can be parametrised from longitudinal study data. We identified several predictors for such models. In the viable pregnancy model, increased maternal age, higher Body Mass Index (BMI), having Polycystic Ovaries Syndrome (PCOS) and the number of previous miscarriages were associated with reduced odds of viable pregnancy. In contrast, having had previous live births increased the odds of a viable pregnancy. Model validation against a second external dataset gave an Area Under Curve (AUC) of 0.65 (95% Confidence Interval (CI): 0.55, 0.76). Of the 942 women referred to our recurrent miscarriage clinics and followed up over a period of 3 years, 10.7% (101) did not conceive during this time, indicating a potential subfertility problem. In the time to pregnancy model, increased maternal age, higher BMI, and smoking were associated with reduced likelihood of conception. Conversely, taking folic acid supplements and having a history of previous conceptions were associated with increased fertility. In our cohort, 53.4% (577 out of 1080 women) reported a pregnancy within 12 months. Additionally, 22.8% (277 out of 996 women who were followed up over a 2-year period) experienced a first pregnancy event in the second year. The area under the curve (AUC) for predicting pregnancy within 12 months was 0.60 (95% CI: 0.50-0.70) in an external validation using a second dataset. INTERPRETATION: The pregnancy journey can be predicted on a personalised basis by integrating the validated models. We provide a framework for evidence-based management of women with miscarriage, comprising informed decision-making, including optimal referral to fertility services, and a tailored insight into fertility outcomes, thereby guiding expectations and providing psychological support. FUNDING: Tommy's National Centre for Miscarriage Research.