Breastfeeding and use of the health care system in Bahia State, Brazil: three multivariate analyses

Three multivariate methods--deviation from total proportion breastfeeding, polynomial regression, and proportional hazards--are used on current status breastfeeding data to explain an inverse relationship between use of the health care system and breastfeeding in Bahia State, Brazil. Among the intervening variables analyzed--urban/rural residence, educational attainment, age of woman, and place of last live birth--education and place of last live birth are found to have a net effect on breastfeeding by all three methods of analysis. The advantages and disadvantages of each method are described.