Abstract
Abstract
Background:
Benefits of breastfeeding are well established. What is more, breastfeeding is associated with lower healthcare costs. More U.S. hospitals are adopting the World Health Organization's Ten Steps to Successful Breastfeeding; however, most hospitals fall short on key factors.
Objective:
To our knowledge, this project is the first of its kind to use national-level data and a complex statistical modeling approach to identify a more complete picture of the variables related to breastfeeding duration within the postpartum period.
Methods:
This secondary data analysis project used the Phase 7 Core PRAMS Research File (2012–2013) and the Standard Questions B1, B2, B3, and B4 variables in the statistical analysis. The outcome variable of interest was length of breastfeeding during the postpartum period. The postpartum period was defined as 8 weeks after delivery rather than the usual definition of 6 weeks to accommodate the way quit time is reported in the Pregnancy Risk Assessment Monitoring System (PRAMS) data. Univariate and multivariate analyses were conducted using PC SAS version 9.4.
Results:
The multivariate analysis indicates that many of the modifiable factors associated with quitting breastfeeding within the postpartum period are hospital related. This pilot study reinforces the importance of 7 of the 10 Steps.
Conclusions:
These results underscore the importance of hospitals adopting evidence-based best practices for breastfeeding. The relationship found between Women, Infants, and Children receiving supplemental nutrition benefits and breastfeeding quit times requires further exploration. The study results also highlight the need to address modifiable factors that may be overlooked in traditional breastfeeding promotion efforts, such as depression and alcohol use.
Introduction
T
What is more, breastfeeding is associated with lower healthcare costs. Exclusive breastfeeding for at least 3 months is associated with fewer physician office visits, fewer days of hospitalization, and fewer prescriptions for three common infant illnesses. 12 More recent cost analysis found similar results with an estimated $13 billion per year saved in healthcare costs if 90% of families with infants complied with the recommendations to breastfeed exclusively for 6 months. 13
According to the latest Breastfeeding Report Card from the Centers for Disease Control and Prevention (CDC), 14 in the United States 79.2% of infants are ever breastfed, 49.4% are breastfeeding at 6 months of age, only 40.7% are exclusively breastfed at 3 months, and only 18.8% are exclusively breastfed at 6 months, which is the recommended duration. 15 Healthy People 2020 Objectives include increasing the proportion of infants who are ever breastfed to 81.9%, at 6 months of age to 60.6%, exclusively through 3 months to 46.2%, and exclusively through 6 months of age to 25.5%. 16
Hospital breastfeeding practices are associated with breastfeeding duration. 17 More U.S. hospitals are adopting the World Health Organization's (WHO's) Ten Steps to Successful Breastfeeding, however, most hospitals fall short on key factors, such as limiting nonbreast milk feeds, rooming-in, limiting use of pacifiers, and providing postdischarge support. 18
Race/ethnicity, maternal education, maternal age, and poverty status have been standard variables used in breastfeeding promotion efforts. Research shows that Non-Hispanic Black women, women with lower education levels, younger women, and poorer women are less likely to breastfeed. 19 The decisions parents make related to childrearing, including infant feeding decisions, are influenced by many factors that may include, but are not limited to, demographic factors.
To our knowledge, this project is the first of its kind to use national level data and a complex statistical modeling approach to identify a more complete picture of the variables related to breastfeeding duration within the postpartum period. This article reports the results of an analysis of the CDC's Pregnancy Risk Assessment Monitoring System (PRAMS) data to identify factors associated with quitting breastfeeding in the postpartum period. Ethical approval for this research was provided by the University of Nebraska Medical Center Institutional Review Board (IRB #131-18-EX).
Materials and Methods
PRAMS collects state-specific, population-based data on maternal attitudes and experiences before, during, and shortly after pregnancy. PRAMS data are not available from other sources and currently covers about 83% of all U.S. births. 20 PRAMS uses a standardized data collection system (survey) that includes both core questions asked by all participating states and optional standard questions developed by the CDC with the goal of using the research to improve the health of mothers and infants.
This secondary data analysis project used the Phase 7 Core PRAMS Research File (2012–2013) and the Standard Questions B1, B2, B3, and B4 as variables in the statistical analysis. Variables were sorted into six categories: breastfeeding education and support, breastfeeding intent, mother's health status, pregnancy intent, stressors, and demographics (Table 1). PRAMS Standard question B3 includes factors related to 7 of the 10 hospital breastfeeding best practices (sometimes referred to as baby-friendly practices) as outlined in the WHO's Ten Steps to Successful Breastfeeding.
C, core question included by all states; S, standard question that is optional and not included by every state; PRAMS, Pregnancy Risk Assessment Monitoring System; WIC, Women, Infants, and Children.
The outcome variable of interest was length of breastfeeding during the postpartum period. The postpartum period was defined as 8 weeks after delivery rather than the usual definition of 6 weeks to accommodate the way quit time is reported in the PRAMS data.
PC SAS version 9.4 was used for all summaries and analyses. The statistical level of significance was set to 0.05 unless otherwise indicated. Four binary outcome variables based on breastfeeding quit time were created: Less than 1, 1–2, 3–4, and 5–8 weeks. For each of the breastfeeding quit time outcomes, univariate analyses were performed for education/support factors, intent/demographic/health status factors, and stress factors. The SURVEYFREQ procedure in SAS was used to test for an association of each factor with each breastfeeding quit time outcome. Factors with univariate SURVEYFREQ p-values less than 0.10 were included in a multivariate logistic regression model using the SURVEYLOGISTIC procedure in SAS. Backward elimination was used to arrive at a final model with statistically significant (p < 0.05) terms remaining. p-Values and odds ratios are presented (Tables 2–5).
Total N = 7543; quit N = 631. Total N varies for each postpartum time period because of missing data.
Total N = 5044; quit N = 765. Total N varies for each postpartum time period because of missing data.
Total N = 1045; quit N = 352. Total N varies for each postpartum time period because of missing data.
Total N = 2142; quit N = 1203. Total N varies for each postpartum time period because of missing data.
Results
The univariate analysis identified 35 variables that were significant for at least one of the quit times (<1 week = 18 variables, 1–2 weeks = 18 variables, 3–4 weeks = 28 variables, 5–8 weeks = 15 variables). Only four of the variables were significant in the univariate analysis for all postpartum time periods: breastfeeding within the first hour; exclusive breastfeeding in the hospital; Women, Infants, and Children (WIC) status, and percent of poverty level.
The most significant finding of the multivariate analysis is that hospital-related factors were the primary predictors of quitting breastfeeding during the postpartum period (Tables 2–5). The baby receiving supplemental feeding in the hospital was significantly associated with quitting breastfeeding within three of the quit times considered (<1 week p = <0.0001, 1–2 weeks p = <0.0001, 5–8 weeks p = 0.019). Receiving formula samples in hospital gift packs was significantly associated with quitting during two of the time periods considered (<1 week p = <0.01 and 3–4 weeks p = 0.005). Not breastfeeding on demand in the hospital was significantly associated with quitting with the first week postpartum (p = 0.015) as was not receiving information about follow-up support (i.e., a telephone number) (p = 0.006). Not rooming in was significantly associated with quitting 1–2 weeks postpartum (p = <0.0001). Not breastfeeding within the first hour after delivery and not breastfeeding in the hospital at all were significantly associated with quitting 3–4 weeks postpartum (p = 0.006 and p = 0.03 respectively). Not receiving a breast pump was significantly associated with quitting 5–8 weeks postpartum (p = 0.009).
Nonhospital-related factors associated with quit time in the multivariate analysis included mother's race, mother's age, receiving WIC Supplemental Nutrition benefits, depression, desire to be pregnant, and change in drinking. Race was significantly associated with quit time across with the <1 week (p = 0.01), 1–2-week (p = <0.001), and 5–8-week time periods (p = <0.001). Mother's age was significantly associated with quit time for quitting in the 1–2-week (p = <0.01) and 3–4-week (p = <0.01) time periods. Receiving WIC benefits was significantly associated with quit time in the 1–2-week (p = 0.021) and 3–4-week (p = <0.01) periods. Conversely, not receiving WIC benefits was significantly associated with quit time during the 5–8-week period (p = 0.020). A diagnosis of depression before birth was significantly associated with quit time in the <1-week (p = 0.027) and 3–4-week (p = 0.022) periods. Desire to be pregnant was significantly associated with quit time in the 1–2-week period (p = 0.014) with mothers who did not want to be pregnant, wanted to be pregnant later, or were unsure more likely to quit. Change in drinking was significantly associated with the 1–2-week period (p = 0.18) and the 5–8-week period (p = 0.03) with mothers who did not drink before pregnancy or who quit drinking during pregnancy being more likely to quit breastfeeding.
Discussion
The factors most traditionally considered when analyzing breastfeeding duration and exclusivity, including age, race/ethnicity, income or poverty level, and education level, were not significant in the multivariate analysis for all the time periods considered in the postpartum period. The implication of these findings is that focusing on demographic factors when identifying audiences for breastfeeding promotion efforts may ignore other, more pertinent factors.
Results of the multivariate analysis indicate that many of the modifiable factors associated with quitting breastfeeding within the postpartum period are hospital related. These results reinforce the importance of hospital best practices related to breastfeeding. Of the 10 best practices, 7 of them are supported by the results of this research, including baby rooming in with mother, showing mothers how to breastfeed, helping mothers initiate breastfeeding within 1 hour of birth, no supplemental feeding of baby, breastfeeding on demand, referring mothers to breastfeeding support on discharge, and not using pacifiers. 21 In addition, these results also show the negative impact of providing hospital gift packs that include formula samples.
Other potentially modifiable factors include depression, receiving WIC benefits, the timing of the pregnancy, and drinking behaviors. Mothers who had a diagnosis of depression before giving birth were more likely to quit breastfeeding within the first week or during 3–4 weeks postpartum. Mothers who reported being depressed sometimes since the baby was born were more likely to quit breastfeeding in the 5–8-week period postpartum. This finding suggests that mothers should be asked about depression during prenatal visits and at well-baby checks within the first weeks after giving birth, and provided resources and support specific to addressing depression.
Interestingly, receiving WIC benefits was significantly associated with quitting breastfeeding during the 1–2 and 3–4-week periods postpartum. This finding is at odds with the WIC practice of promoting breastfeeding and providing additional benefits to breastfeeding mothers. Conversely, not receiving WIC benefits was significantly associated with quitting breastfeeding during the 5–8-week period postpartum. These findings warrant further exploration and may have implications for the structure and administration of the WIC program overall as well as breastfeeding promotion efforts within WIC.
Timing of the pregnancy was also significantly associated with quitting breastfeeding. Mothers who wanted to be pregnant later, did not want to be pregnant, or were unsure about being pregnant were more likely to quit breastfeeding within the 1–2-week postpartum period than mothers who wanted to be pregnant then or sooner. This finding indicates that there may be an association between reproductive health, including family planning, and breastfeeding. It highlights the importance of viewing infant feeding decisions within the larger context of women's reproductive choices.
Change in drinking during pregnancy was also associated with quitting breastfeeding during the 1–2 and 5–8-week postpartum periods. Mothers who did not drink during pregnancy were more likely to quit breastfeeding in the 1–2-week period than mothers who drank during pregnancy. Mothers who did not drink or quit drinking during pregnancy were more likely to quit breastfeeding during the 5–8-week postpartum period. This finding also deserves further exploration. It may be that women who drink during pregnancy are less aware of the impact of their alcohol consumption on their breastfeeding infant and, as a result, their drinking is not a primary consideration in their infant feeding decisions.
There were several nonmodifiable related factors associated with quitting breastfeeding within the postpartum period, including mother's race and age. Black mothers were more likely than White or other races to quit breastfeeding in the <1-week period, however, White mothers were more likely than Black or other races to quit breastfeeding in the 1–2 and 5–8-week periods. This suggests that breastfeeding promotion efforts need to be tailored to different races during different postpartum time periods.
Mother's age was significantly associated with quitting during the 1–2 and 3–4-week periods for mothers in the 20–29 age group compared with mothers in the 30–39 age group. This suggests that breastfeeding promotion efforts should be targeted toward all mothers during the prenatal period, but more efforts focused toward younger mothers during the postpartum period.
Conclusion
Using a multivariate analysis approach when considering factors associated with breastfeeding duration during the postpartum period provides a more robust picture. The results of this study provide a foundation for moving away from “one size fits most” interventions and toward more targeted efforts to support breastfeeding in a way that recognizes the unique needs of mothers in different stages of the postpartum period. Some of the findings need further exploration, for example the relationship between receiving WIC benefits and breastfeeding quit times. The study results also highlight the need to address modifiable factors that may be overlooked in traditional breastfeeding promotion efforts, such as depression and alcohol use during pregnancy. Perhaps most importantly, these results underscore the importance of hospitals adopting evidence-based best practices for breastfeeding in the hospital.
Footnotes
Acknowledgments
The research team would like to acknowledge Dr. Mohammed Siahpush, Associate Dean for Research, University of Nebraska Medical Center College of Public Health, for his funding support for this project. The authors would also like to acknowledge Dr. Lynette Smith, Assistant Professor in the Department of Biostatistics, University of Nebraska Medical Center College of Public Health, for her guidance on statistical analysis. The authors would like to acknowledge the PRAMS Working Group and the Centers for Disease Control and Prevention (CDC) for providing the data used in the analysis reported in this article.
Disclosure Statement
The authors affirm that they have no competing financial interests.
