A multivariable risk prediction model for pregnancies of uncertain viability without a fetal pole: a prospective...

Pregnancy of uncertain viability affects up to 30% of early pregnancies. Among women presenting with per-vaginal bleeding and absence of a fetal pole, prognosis remains indeterminate. The lack of validated prediction tools contributes to clinical uncertainty and maternal anxiety.

This prospective cohort study was conducted at KK Women's and Children's Hospital, Singapore, from October 2017 to May 2023. We recruited 446 women with single intrauterine pregnancies between 5 and 12 weeks' gestation by last menstrual period (LMP), presenting with either per-vaginal bleeding or abdominal pain and pregnancy of uncertain viability without a fetal pole. Predictor variables included maternal age, ethnicity, nausea, history of miscarriage, serum progesterone, and gestational age by LMP. The primary outcome was spontaneous miscarriage by 16 weeks. Multivariable logistic regression models were evaluated using the area under the receiver operating characteristic curve (AUROC) and Akaike information criterion. Risk scores derived from β-coefficients were categorized into low (≤1), medium (1-4), and high (4-7) risk groups.

Among 1,938 women screened, 446 were eligible, and 187 (41.9%) had early pregnancy loss. Low serum progesterone (<35 nmol/L; OR 22.0, 95% CI 13.2-36.5), absence of nausea (2.34, 1.34-4.08), greater gestational age (2.26, 1.87-2.72), and older maternal age (1.08, 1.04-1.12) independently predicted early pregnancy loss. The four-factor model achieved an AUROC of 0.88; the corresponding risk-score model achieved 0.86, with sensitivity 0.74 and specificity 0.88. Early pregnancy loss rates were 11.5%, 33.8%, and 88.2% across risk groups.

A simple four-factor model accurately predicts early pregnancy loss in women with pregnancy of uncertain viability without a fetal pole, supporting early risk stratification and individualized counseling.