Abstract
Breastfeeding is strongly endorsed in the Healthy People 2020 goals; however, there remain many disparities in breastfeeding prevalence. The purpose of this study was to examine the association between breastfeeding and the Federal Poverty Level in the United States. Data from 5,397 women in the National Survey of Family Growth 2011–2013 survey were included in this study. The data were analyzed for descriptive features and logistic regressions of the Federal Poverty Level on breastfeeding. There were 64.1% of women who reported breastfeeding. Over one-third (35.2%) of women reported having a household income of 0–99% of the Federal Poverty Level. There were 15.2% of women who reported an income of 400% and above the Federal Poverty Level. With statistical adjustment for maternal age, race/ethnicity, education, marital status, parity, preterm birth, birth weight, insurance, and dwelling, the Federal Poverty Level was not significantly associated with breastfeeding. In this recent survey of mothers, Federal Poverty Level was not shown to be a significant factor in breastfeeding.
1. Introduction
The American Academy of Pediatrics recommends the exclusive breastfeeding of infants to age six months, with continued breastfeeding (complemented by solid foods) for one year or longer [1]. The United States (US) Department of Health and Human Services recognizes the public health benefits of breastfeeding and has nine breastfeeding-related objectives for Healthy People 2020 goals [2]. These objectives include increasing the number of infants having ever been breastfed from the baseline of 74.0% to 81.9%; increasing the number of infants who are breastfed to age 6 months from the baseline of 43.5% to 60.6%; and increasing the number of facilities that provide recommended care for lactating mothers and newborns from a baseline of 2.9% to 8.1% [2].
There are many barriers to breastfeeding that have been reported in earlier studies including lack of support [3, 4], public beliefs [3], difficulty with the breast pump [5], young age of mother, less education, unmarried status, fear of embarrassment, fear of being fired, privacy, sexualization of the breast, change in appearance of the breast, pain, bleeding, difficulty latching-on, insufficient milk, race/ethnicity, and low income [6]. In a population-based study examining the influence of poverty and participation in the federal Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) in South Carolina, researchers found that WIC participation was the strongest predictor of lack of breastfeeding initiation in that state [7].
Women who participated in WIC programs faced additional barriers to breastfeeding [8]. One of the themes that emerged in a qualitative study of WIC counselors serving primarily African American families was that formula use was seen as a sign of wealth [9]. Prior to the recently revised WIC breastfeeding incentive program of augmented food packages for breastfeeding women, WIC participation had been associated with lower breastfeeding initiation and duration rates [10]. WIC credits can be used for supplemental formula, and many clients viewed the supplemental formula as more valuable than the offset of expanded food packages [8].
With goals in place and concerted efforts to increase breastfeeding rates, research results have been inconsistent regarding the association between family income and breastfeeding; some researchers indicate no association [11, 12], others support an association [13, 14], and others report equivocal results [15]. The aim of this study was to determine if there was an association between breastfeeding and the Federal Poverty Level (FPL) using data from the National Survey of Family Growth 2011–2013.
2. Methods and Materials
Data from the National Survey of Family Growth (NSFG) 2011–2013 data were used to conduct a cross-sectional secondary data analysis of the association of FPL and breastfeeding. The 2011–2013 survey is the NSFG’s 8th data file release since 1973 (National Survey of Family Growth 2015) [16]. The NSFG survey was specifically designed to determine family trends as well as differences among groups in family sizes, family structure, use of contraception, sexual activity, and infertility for use in designing health services and educational programs [16]. The sampling was a multistage probability-based national representative of US households [16]. Details of the survey are provided at the NSFG website, http://www.cdc.gov/nchs/nsfg/nsfg_2011_2013_puf.htm.
This study received the West Virginia University Institutional Review Board study acknowledgement (protocol number 1502572781). The research was conducted in accordance with prevailing ethical principles.
2.1. Study Population
Participants in the current study of the association between breastfeeding and FPL were women of childbearing age (14 to 44 years) who had completed NSFG 2011–2013 data for the following variables: breastfeeding; FPL status; and race/ethnicity. The sample size was 5,397 women.
2.2. Variable Definitions
The dependent variable was breastfeeding, defined as breastfeeding one week or more. (The definition did not include intent to breastfeed.) The variable was a dichotomized “yes” or “no” variable.
The independent variable was the FPL of the mother. FPL was provided by the NSFG as a recoded variable with five levels: 0–99% FPL, 100–199% FPL, 200–299% FPL, 300–399% FPL, and 400% and above FPL. Other sociodemographic and health variables included in the analyses were maternal age (14 to less than 20 years; 20 to less than 25 years; 25 to less than 30 years; or 30 to 44 years); maternal race/ethnicity (non-Hispanic white; non-Hispanic black; or Hispanic); maternal education (less than high school; high school graduate; some higher education; or Associate degree and above); marital status (married or single/divorced/separated/widowed); parity (first infant or 2 or more); preterm birth (yes, less than 37 weeks, or no, 37 weeks and above); low birth weight (yes, less than 2,500 grams; no, 2,500 grams and above); insurance (private or MediGap; Medicaid, CHIP, or a state-sponsored plan; Medicare, military, or other government plan; or single service, Indian Health Service, or not covered); and dwelling (urban, principal city; urban, other than principal city; or rural).
2.3. Analyses
Due to the complex sampling design of the NSFG 2011–2013, analyses were conducted to account for the computational units, strata, and final weights. SAS version 9.3® (Cary, NC) software was used to determine the descriptive characteristics of the sample, and the relationship between breastfeeding and FPL in logistic regression. The model-based imputations in the NSFG 2011–2013 data for the variables chosen had a maximum imputation occurring with FPL (5.94%). An adjusted regression model was built which included maternal age, maternal race/ethnicity; maternal education; marital status; parity; preterm birth, birth weight; insurance; and dwelling.
3. Results
The study population was derived from the National Survey of Family Growth 2011–2013 data. There were 5,397 participants in this study. There were 3,302 (64.1%) participants who breastfed and 2,095 (35.9%) who did not breastfeed. There were 2,406 (35.3%) at 0–99% of the FPL; 1,324 (23.3%) at the 100–199% FPL; 701 (15.9%) at the 200–299% FPL; 414 (10.4%) at the 300–399% FPL; and 552 (15.1%) at or above the 400% FPL. The sample description is presented in Table 1.
In unadjusted logistic analysis, there were significant relationships between breastfeeding and income for all of the participants. In stratified analyses, there were significant associations between breastfeeding and income with maternal age in the 25–30 year category; maternal age in the 30 year and older category; maternal race/ethnicity; maternal education in the Associate degree and above category; parity in the second baby or above category; preterm birth; insurance in all categories except the Medicaid, Chip, State sponsored plan category; and urban dwelling. The results of the unadjusted logistic regressions are presented in Table 2.
The results of two logistic regression models on breastfeeding are presented in Table 3. The first model demonstrates the combined effect of entering maternal age, maternal education, marital status, and dwelling on the relationship between breastfeeding and FPL. FPL is attenuated and no longer significant in this parsimonious model. The complete model further attenuated the relationship.
There were no interactions with FPL and marital status, preterm birth, and dwelling when each variable was entered individually in the logistic regressions. There was a slight attenuation of the association of FPL and breastfeeding when maternal age and maternal education were entered individually in logistic regressions. It was the combined effects of the variables which altered the FPL-breastfeeding relationship. The results of the adjusted logistic regression analyses stratified for each category of maternal race/ethnicity, marital status, and preterm birth are presented in Table 4.
4. Discussion
The odds ratio for the association between FPL and breastfeeding failed to reach significance in an adjusted logistic regression with the covariates of maternal age, maternal race/ethnicity, maternal education, marital status, parity, preterm birth, low birth weight, insurance, and dwelling. Although studies exist which address only maternal low income and breastfeeding, there is a paucity of information concerning income disparity (concerning a more comprehensive and inclusive approach to income) and breastfeeding in the US. Nevertheless, similar results to support this study were found in a study of 10,519 women in California who gave birth between 1999 and 2001 in which Heck and colleagues [11] found that family income was not associated with breastfeeding. Lutter and Morrow [12] reported a trend in Africa, Asia, Latin America, and the Middle East over the previous two decades in which annual increases in breastfeeding were not associated with the gross national income of the participants’ respective countries.
Researchers conducting a study of three hospitals in Canada found conflicting results regarding breastfeeding initiation and maternal income. Overall breastfeeding initiation increased over time, although for one hospital the rate difference between maternal high income and low income decreased, for another hospital the rate difference remained the same, and for the third hospital the rate difference increased [15].
Supporting the association between breastfeeding rates and income, researchers using the 1999–2006 National Health and Nutrition Examination Surveys suggested that women with higher incomes were significantly more likely to breastfeed than women with low incomes [13]. In a large population-based ecological study in Ontario, Canada, women with higher incomes were more likely to breastfeed their infants than women with lower incomes [14].
Future studies are needed to further clarify the role of FPL on breastfeeding and to examine the role of supportive services, such as WIC, in encouraging breastfeeding. This study’s main limitation is that the data collected were self-reported. Self-reported measures are subject to social desirability bias, which occur when a participant responds in a way to appear more positive. Second, a causal interpretation cannot be applied as the cross-sectional design of the study does not indicate temporality. However, the study has several strengths. It is a large study using national data and the sampling used a multistage probability-based and representative design.
5. Conclusion
While FPL was significantly associated with lack of breastfeeding in unadjusted analyses, the role of FPL failed to reach significance in the adjusted regression analyses of the study, suggesting a need for all women to have breastfeeding initiation support. Women of childbearing age should be educated about the benefits of breastfeeding for themselves and their infants. The workers in supportive programs, such as WIC, are important in encouraging breastfeeding in their clients. Healthcare providers should continue to promote breastfeeding and to help meet Healthy People 2020 goals.
Disclaimer
The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Conflict of Interests
The authors declare that there is no conflict of interests regarding the publication of this paper.
Acknowledgment
Research in this publication was supported by the National Institute of General Medical Sciences of the National Institutes of Health under Award nos. U54GM104942; WVCTSI; R01 DE014899.