Development, validation and clinical utility of a risk prediction model for maternal and infant adverse outcomes in women with hypothyroidism

Abstract

Abstract Purpose. The ability to calculate the absolute risk of adverse maternal and infant outcomes for an individual woman with hypothyroidism would allow preventative and therapeutic interventions to be delivered to women and infant at high-risk, sparing women and infant at low-risk from unnecessary care. We aimed to develop, validate and evaluate the clinical utility of a prediction model for adverse maternal and infant adverse outcomes in women with hypothyroidism. Methods. A prediction model development and validation study was conducted on data from a retrospective cohort. Participants included all women with hypothyroidism from a tertiary hospital in Wuxi, Jiangsu, China. The development and validation cohort comprised those who delivered between 1 October 2020 to 31 December 2022.The main outcome was a composite of critically important maternal and neonatal complications. Logistic regression was used to develop prediction models. Model performance was measured in terms of discrimination, calibration, and clinical utility. Results. Nine variables were selected to establish the prediction model of adverse maternal and infant outcomes in pregnancy with hypothyroidism. AUC indicated that the discriminant power of the nomogram was satisfactory. In the model for predicting adverse maternal outcomes, the training set AUC was 0.845 and the validation set AUC was 0.779. In the model for predicting adverse neonatal outcomes, the training set AUC was 0.685 and the validation set AUC was 0.787. The calibration plots show good agreement between the predictions of the nomograms and the actual observations in both the training and validation cohorts. The established nomograms (partial factors) performed significantly better than the nomograms constructed with all factors. In addition, DCA suggests that nomograms are clinically useful and have better discriminative power to identify high-risk mother-infant patients. Conclusion Predictive models were developed and validated to help clinicians assess maternal and infant outcomes in pregnancy with hypothyroidism and to aid in decision-making on treatment.