Premature cessation of breastfeeding in infants: development and evaluation of a predictive model in two Argentinian cohorts: the CLACYD study*, 1993-1999

The objective of this study was to develop a model to predict premature ces sation of breastfeeding of newborns, in order to detect at-risk groups that would benefit from special assistance programmes. The model was constructe d using 700 children with a birthweight of 2000 g or more, in 2 representat ive cohorts in 1993 and 1995 (CLACYD I sample) in Cordoba, Argentina. Data were analysed from 632 of the cases. Mothers were selected during hospital admittance for childbirth and interviewed in their homes at 1 mo and 6 mo. To evaluate: the model, an additional sample with similar characteristics w as drawn during 1998 (CLACYD II sample). A questionnaire was administered t o 347 mothers during the first 24-48 h after birth and a follow-up was comp leted at 6 mo, with weaning information on 291 cases. premature cessation o f bnastfeeding was considered when it occurred prior to 6 mo. A logistic re gression model was fitted to predict premature end of breastfeeding, and wa s applied to the CLACYD II sample. The calibration (Hosmer-Lemeshow C stati stic) and the discrimination [area under the receiver operating characteris tics (ROC) curve] of the model were evaluated. The predictive factors of pr emature end of breastfeeding were: mother breastfed for less than 6 mo [odd s ratio (OR) = 1.84, 95% confidence interval (CI) 1.26-2.70], breastfeeding of previous child fur less than 6 mo (OR = 4.01, 95% CI 2.58-6.20), the co ndition of the firstborn child (OR = 2.75, 95% CI 1.79-4.21), the first mot her-child contact occurring after 90 min of life (OR = 1.88; 95% CI 1.23-2. 91) and having an unplanned pregnancy (OR = 1.50, 95% CI 1.05-2.15). The ca libration of the model was acceptable in the CLACYD I sample (p = 0.54), as well as in the CLACYD II sample (p = 0.18). The areas under the ROC curve were 0.72 and 0.68, respectively. Conclusion: A model has been suggested that provides some insight unto back ground factors for the premature end of breast-feeding. Although some limit ations prevent its general use at a population level, it may be a useful to ol in the identification of women with a high probability of early weaning.