L26/P-556 associations between luteal phase duration and early pregnancy loss: evidence from a digital fertility cohort

Abstract Study question Is there an association between luteal phase duration and the probability of first trimester (≤12 weeks’ gestation) miscarriage? Summary answer Shorter luteal phases were associated with a higher predicted probability of first-trimester miscarriage, particularly before 7 weeks’ gestation, although the findings lacked statistical certainty. What is known already Luteal phase deficiency (LPD) is broadly defined as a short luteal phase of 9-11 days. The corpus luteum is the principal source of progesterone during the first seven weeks of gestation, and shortened luteal phases and lower luteal progesterone levels have been associated with infertility and miscarriage. Digital Fertility applications use temperature and menstrual cycle data to predict fertile and non-fertile days. These platforms generate large datasets containing pregnancy outcomes and relevant patient covariates, providing an opportunity to examine the association between luteal phase duration and first-trimester miscarriage (FTM) in a large cohort, where evidence from large studies remains limited. Study design, size, duration Complete menstrual cycle data recorded in a digital fertility application between 01 May 2022 and 14 July 2025 was retrospectively analysed. Generalised linear models with binomial logistic regression were used to estimate the association between mean luteal phase length and the probability of overall FTM, with separate models for pre and post placentation: early FTM (≤7 weeks’ gestation), and late FTM (7≤12 weeks’ gestation). Models were adjusted for age and body mass index (BMI). Participants/materials, setting, methods We included users aged 18-35 who recorded at least two ovulations and three complete cycles after app registration, had a BMI <35 kg/m2, a mean luteal phase length ≤20 days, and no known reproductive disorders. The latest cycle resulting in a pregnancy was included if the pregnancy was not a termination or molar pregnancy. A total of 9,387 menstrual cycles met the inclusion criteria, from an original dataset of 152,031 menstrual cycles (42,404 unique users). Main results and the role of chance Of the 9,387 included pregnancies, 42.7% (4,011/9,387) experienced a first-trimester miscarriage, and 26.0% (2,441/9,387) experienced a FTM. Each additional day of luteal phase length was associated with a 1.9% relative and 0.38% absolute reduction in the risk of FTM (OR 0.98; 95% CI 0.95-1.01; p = 0.06). In the pre-placentation model, each additional day of luteal phase length was associated with a 2.6% relative and 0.52% absolute decrease in the probability of miscarriage before seven weeks, although this finding lacked statistical certainty (OR 0.97; 95% CI 0.94-1.01; p = 0.1779 ). By contrast, the post-placentation model showed that each additional day of luteal phase length was associated with an 0.8% relative and 0.12% absolute reduction in miscarriage risk during this period (OR 0.99; 95% CI 0.95-1.03; p = 0.06921). Limitations, reasons for caution The final dataset had high levels of missing data for demographic variables which may be relevant to calculating miscarriage risk, such as smoking, parity, and ethnicity. This may be a unique challenge for digital fertility applications, whereby inputting such data is optional and at the discretion of the app user. Wider implications of the findings To our knowledge, this is the largest study investigating the association between luteal phase duration and FTM. The effect of luteal phase duration on miscarriage appeared to be largely confined to pre-placentation losses, illustrating the utility of digital fertility datasets to capture otherwise inaccessible early gestational events. Trial registration number No