Construction and validation of a predictive model for exclusive breastfeeding at discharge based on the information-motivation-behavioral skills theory

Despite the well-established health benefits of exclusive breastfeeding for both mothers and infants, breastfeeding rates in China remain suboptimal. This study, guided by the Information-Motivation-Behavioral Skills (IMB) model, aimed to develop and validate a predictive model for exclusive breastfeeding at discharge to facilitate the early identification of high-risk mothers and enable timely clinical interventions.

In this prospective observational study conducted from February to June 2025, a total of 623 postpartum women were recruited, with 592 meeting the inclusion criteria. Of these, 448 were allocated to the model development group, while 144 from a different hospital formed the external validation group. Demographic and breastfeeding-related variables were collected via questionnaires and electronic medical records. Logistic regression was employed to identify significant predictors and construct a nomogram. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), calibration plots, the Hosmer-Lemeshow goodness-of-fit test, and decision curve analysis (DCA), and externally validated using an independent cohort.

Both univariate and multivariate logistic regression analyses identified newborn sex, early skin-to-skin contact, breastfeeding attitude, breastfeeding self-efficacy, and LATCH score as significant predictors of exclusive breastfeeding at discharge. The nomogram exhibited good discriminatory ability, with an AUC of 0.76 (95% CI, 0.70-0.81) in the development group and 0.66 (95% CI, 0.56-0.75) in the validation group. The Hosmer-Lemeshow test indicated good model calibration (p > 0.05), and decision curve analysis demonstrated favorable clinical applicability.

This study successfully constructed and preliminarily validated a pure breastfeeding prediction model based on the IMB theory. The model demonstrates good calibration and moderate discriminatory ability, enabling clinicians to identify mothers at higher risk for exclusive breastfeeding failure early before discharge. Although its external validation performance suggests that its generalizability requires further validation in larger samples and more centers, its robust theoretical foundation positions it as a valuable risk assessment and screening tool. This provides a meaningful reference framework and methodological starting point for developing precise, efficient, and targeted nursing interventions in the future.