E N D
INTRODUCTION Breastmilk is universally accepted as the best food for an infant. Despite all the evidence, breastfeeding rates in the United States need much improvement. Low-income African-American have been identified as a vulnerable population with the lowest breastfeeding initiation rates and the shortest duration of breastfeeding. This study was designed to help understand the factors that influence feeding patterns in this population and subsequently assist in the development of targeted promotion programs. The study used a prospective observational cohort design to determine if the Breastfeeding Attrition Prediction Tool (BAPT)* along with demographic variables can be used with a sample of African American pregnant women to understand and predict their infant feeding behavior. The BAPT is a 94 item scale developed based on the constructs of the theory of planned behavior to identify women who express an intention to breastfeed. The tool was developed and tested in a population of Caucasian, post-partum women with high income and educational levels. * Janke, J. (1992). “Prediction of breastfeeding attrition: instrument development.” App Nurs Res 5(1): 48-53.
METHODS Data Collection • A screening questionnaire was used to determine eligibility for the study as well as feeding intention. A written consent was obtained for the women who were eligible. • The BAPT was administered to a selective convenience sample of 201 African-American pregnant women, aged 20-35 years, in the third trimester of pregnancy. • The BAPT was first administered face-to-face in the prenatal clinic while the patient was waiting for her appointment. Two follow-up telephone interviews were done at 1 week and 6 weeks post-partum. • Each participant was provided incentives with a total cash value of $50. A diaper bag with a cash value of $30 was given after the first interview. Participants were given $10 for each follow-up interview. Both the product and cash incentives were provided by Medela, Inc.
Data Analysis • The BAPT was tested for appropriateness in the African-American population using cognitive methodology. Modifications were made to the instrument based on these findings. • Factor analysis was used to determine the factor structure of the instrument and compare it to the original instrument. • Logistic regression analysis was used to determine the predictors of breastfeeding behavior in this population.
RESULTS Demographics • Mean Age: 24.4 ± 4.1 years • Mean number of years of education: 12.5 ± 1.6 years • Marital Status: • 71.1% were single living alone, with a partner or with parents • 16.9% were married • Annual Family Income: • 34.8% had income < $10,000 • 26.9% had income between $10,000 & $20,000 • 23.4% had income between $20,000 & $30,000 • 14.9% had income > $30,000 • Employment Status at time of interview: • 48.3% were employed either part-time or full-time Feeding Intention • Feeding intention was determined using the question “How do you intend to feed your baby?” on the screening questionnaire. • The responses were categorized into breastmilk only, formula only, breastmilk and formula. • These responses were collapsed into any breastfeeding and formula feeding to create the dichotomous variable for the regression analysis.
Logistic Regression Analysis • Logistic regression analysis was used to determine the best predictors of feeding behavior at first and second follow-up. • Two Regression analysis were run using the dichotomous feeding behavior at first follow-up as the dependent variable. • Variables that were significant in the bi-variate analysis were used as the independent variables in the regressions Prediction of behavior at first follow-up • The common independent variables were as follows • number of years of education • annual family income • marital status • knowing a friend or family member who has breastfed • having breastfed before • having made the decision on how to feed the baby • complications during delivery • One of the regressions used total BAPT score as an additional independent variable and the other used the four sub-scale scores as additional independent variables.
Prediction of behavior at second follow-up • The common independent variables were as follows • annual family income • marital status • having breastfed before • having made the decision to how to feed the baby • complications during delivery • time interval between the first BAPT administration and second follow-up. • One of the regressions used total BAPT score as an additional independent variable and the other used the four sub-scale scores as additional independent variables.
CONCLUSIONS • The overall findings indicate that the BAPT can be used to predict breastfeeding behavior in a sample of pregnant African-American women. • Both the BAPT total score and 4 sub-scales used together were strong predictors of behavior at first and second follow-up. • The total or sub-scale scores, marital status, having made the decision on how to feed the baby and having breastfed before were the significant predictors of feeding behavior at the first follow-up interview. • Only the total BAPT score and having breastfed before along with the time to second follow-up were significant predictors of behavior. When the 4 sub-scale scores were used, time to second follow up and the scale scores for PBS, NBS, and SPS were significant predictors of behavi or. Demographic variables were not significant predictors of feeding behavior at the second follow-up. • In conclusion the study indicates that the BAPT can predict behavior in the first few weeks after delivery more accurately than it can predict behavior in the later post-partum period.