Risk factors and prediction model for early pregnancy loss in polycystic ovary syndrome - development and validation:...

Polycystic ovary syndrome (PCOS) is a common endocrine disorder affecting reproductive-aged women and is associated with an increased risk of early pregnancy loss (EPL). The underlying mechanisms remain multifactorial and poorly understood. This study aimed to identify key risk factors for EPL in PCOS patients and develop a corresponding predictive model. A retrospective and observational cohort study was conducted on 134 PCOS patients between January 2023 and June 2025. Patients were randomly assigned to training (60%, n = 80) and validation (40%, n = 54) cohorts using an internal validation method. Clinical data, including patients' demographic and clinical factors (age, body mass index, history of adverse pregnancy outcomes), and laboratory test results (fasting blood glucose, fasting insulin(FINS), impaired glucose tolerance, triglyceride(TG), total cholesterol, high-density lipoprotein cholesterol(HDL-C), thyroid-stimulating hormone, anti-müllerian hormone(AMH), vitamin D(VD), estradiol, progesterone, luteinizing hormone (LH), follicle-stimulating hormone, prolactin and testosterone), were collected and analyzed. Univariable and multivariable logistic regression identified independent risk factors, which were incorporated into a nomogram prediction model. Model performance was assessed using the C-index, receiver operating characteristic curves, calibration plots, and decision curve analysis. Univariable logistic analysis revealed statistical differences between the EPL group and non-EPL group in body mass index, FINS, LH, LH/FSH, PRL, Testo, AMH, and VD (P <.05). Multivariable logistic analysis identified FINS, LH, Testo, and AMH as independent risk factors for EPL (P < .05). A nomogram prediction model incorporating these factors demonstrated strong performance, with C-indices of 0.869 (training cohort) and 0.860 (validation cohort), and AUC values of 0.869 and 0.860, respectively. Calibration curves showed good agreement between predicted and observed probabilities, and decision curve analysis confirmed the clinical utility and net benefit of the model. This study developed and validated a nomogram based on FINS, LH, Testo, and AMH for predicting EPL risk in PCOS patients. The model exhibits reasonable predictive accuracy and certain clinical applicability, offering a potential tool for early identification and personalized management of high-risk pregnancies. However, due to the limitations of a single-center study design and a relatively small sample size, the conclusions of this study still need to be validated through larger-scale clinical research before they can be applied in clinical practice.