Abstract: Objective To systematically search and evaluate the risk prediction models of postpartum breastfeeding behavior interruption, and to provide reference for clinical development, optimization and application of appropriate models. Methods The literature on risk prediction models of breastfeeding behavior interruption were searched in Web of Science, Cochrane Library, PubMed, Embase, CINAHL, CNKI, CBM, VIP and Wanfang Data. The retrieval time spanned from the inception to April 17, 2024. Two researchers independently screened the literature and extracted data, and the included studies were evaluated using the prediction model risk of bias assessment tool (PROBAST). Results A total of 13 studies were included, involving 18 models. Differentiation was reported in 11 studies; calibration methods were reported in 7 studies; internal validation was performed in 6 studies; external validation was performed in 3 studies, and a combination of internal and external validation was used in 1 study. Maternal age, previous feeding experience, breastfeeding self-efficacy, and multiple pregnancies were the most common predictors reported. Six studies had good applicability but a higher risk of overall bias. Conclusion The research on risk prediction models of breastfeeding behavior interruption is still in the development stage. The risk of bias of the models is generally high, which is mainly reflected in data analysis. In the future, the reporting process and statistical analysis should be further standardized, and the effectiveness and feasibility of the model in clinical practice should be evaluated through external verification.

Key words: breastfeeding interruption, predictive model, systematic review, evidence-based nursing