Psychometric evaluation of the Chinese version of the Breastfeeding Attrition Prediction Tool (BAPT-C)

Identifying women at risk of premature weaning is key for nurses to provide breastfeeding guidance and support. Currently, there is no valid instrument to predict breastfeeding attrition in China. Herein, we introduced the Breastfeeding Attrition Prediction Tool (BAPT) to fill this gap. Brislin’s translation model was utilized, and the original BAPT was translated into Chinese. This prospective cohort study evaluated the BAPT on 500 healthy Chinese women to assess internal consistency, reliability, and validity. The Cronbach α coefficient of the subscales of Chinese BAPT (BAPT-C) ranged from 0.88 to 0.93, and the test-retest reliabilities ranged from 0.53 to 0.83. Two items were deleted during the Delphi expert consultation because of low item-content validity index (I-CVI). Next, we revised the BAPT-C from a 6-point Likert scale to a 5-point Likert scale. The exploratory factor analysis was used to determine four factors of BAPT-C, which together explained 50.10% variance, and confirmatory factor analysis provided an acceptable fit to the four-factor model (χ 2 /df = 2.196, RMSEA = 0.049, TLI = 0.902, CFI = 0.911, PGFI = 0.730). The difference in the Social and Professional Support and Breastfeeding Control scores between women with and without exclusive breastfeeding experience was significant ( p < 0.05). The exclusive formula-feeding women at 6 weeks postpartum had significantly lower scores than exclusively breastfeeding and mixed feeding women on the control subscales ( p < 0.05). A stepwise discriminant function analysis demonstrated that the Breastfeeding Control subscale accurately predicts the breastfeeding method at 6 weeks in 66.0% of the cases. The 48-item BAPT-C has good reliability and validity and shows modest predictive ability for the feeding method at 6 weeks postpartum. It is a useful instrument for the clinical nurse to identify women at high risk of premature weaning.