Meeting breastfeeding intentions differ by race/ethnicity, Infant and Toddler Feeding Practices Study‐2

Abstract Prenatal breastfeeding intentions impact breastfeeding practices. Racial/ethnic disparities exist in breastfeeding rates; it is unknown if prenatal intentions and meeting intentions differ by race/ethnicity. A longitudinal cohort of USDA's Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) which enrolled participants beginning in 2013 were used to estimate prenatal intentions for breastfeeding initiation, exclusive breast milk feeds at 1 and 3 months by race/ethnicity (n = 2070). Meeting intentions were determined by reported breast milk consumption at birth, 1 month and 3 months. Multivariable logistic regression was used to determine the association of race/ethnicity with meeting intentions. There were no differences in prenatal breastfeeding intentions between non‐Hispanic White and non‐Hispanic Black women (initiation: 86.9% and 87.2%; Month 1: 52.3% and 48.3%; Month 3: 43.8% and 40.9%; respectively), but a higher percentage of Hispanic women intended to breastfeed at all time points (95.5%, 68.3% and 56.4%; respectively, P < 0.05). Among women who intended to breastfeed at Month 1, non‐Hispanic Black and Hispanic women had significantly lower odds of meeting intentions compared with non‐Hispanic White women after adjusting for covariates (aORs: 0.63 [95% CI: 0.41, 0.98]; 0.64 [95% CI: 0.44, 0.92], respectively). Similar findings were seen for Month 3. Despite no differences in breastfeeding intentions, non‐Hispanic Black women were less likely to meet their breastfeeding intentions than non‐Hispanic White women. Hispanic women were more likely to intend to breastfeed yet were less likely to meet their intentions. This suggests that non‐Hispanic Black and Hispanic women face challenges to meeting their longer breastfeeding intentions. Understanding how racism, bias and discrimination contribute to women not meeting their breastfeeding intentions may help efforts to reduce breastfeeding disparities.


| INTRODUCTION
Breast milk is the optimal source of nutrition for most infants (Section on Breastfeeding, 2012). The American Academy of Pediatrics recommends that infants receive only breast milk for about the first 6 months of life, followed by complementary food introduction with continued breastfeeding for as long as mother and baby desire (Section on Breastfeeding, 2012). Although breastfeeding rates have increased overall in the United States for the last several decades, There are many factors the can impact breastfeeding intentions, initiation and duration and these factors may differ by race/ethnicity (Asiodu, Waters, Dailey, & Lyndon, 2017;Hinson, Skinner, Lich, & Spatz, 2018;Spencer & Grassley, 2013). Historically, advice and/or support for breastfeeding from the US Department of Agriculture's (USDA's) Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) programme has differed between Black and White women with Black women being more likely to receive support for bottle feeding and less likely to receive advice on breastfeeding (Beal, Kuhlthau, & Perrin, 2003). More recently, African American women who breastfed reported that WIC peer counsellors or participation in a WIC peer breastfeeding support group played a positive role in helping them breastfeed (Gross, Davis, Anderson, Hall, & Hilyard, 2017;Mickens, Modeste, Montgomery, & Taylor, 2009;Robinson, VandeVusse, & Foster, 2016). The decision to breastfeed is often made prenatally (Gurka et al., 2014) and has been found to be a strong predictor of actual breastfeeding practice (Donath, Amir, & The ALSPAC Study Team, 2003). Limited data exist on whether prenatal breastfeeding intentions differ by race/ethnicity and whether or not women are able to achieve these intentions. The objective of our study is to describe women's prenatal breastfeeding intentions by race/ethnicity and assess the impact of race/ethnicity in meeting these intentions.

| METHODS
USDA's WIC programme conducted a longitudinal cohort study of women and their children enrolled in WIC entitled the WIC Infant and Toddler Feeding Practices Study 2 (ITFPS-2). This longitudinal cohort study is designed to provide information on feeding practices and nutrition outcomes among women and children enrolled in WIC (May et al., 2017). Sampling for the study was done using a two stage stratified approach as described in the WIC ITFPS-2: Infant Report (May et al., 2017). Briefly, the study enrolled participants in 2013 from 80 WIC sites across 27 states. WIC sites that were projected to enrol ≥30 participants per month were eligible to participate. These sites represented 37% of WIC sites and 87% of WIC participants (May et al., 2017). Women were invited to participate if they were ≥16 years of age, spoke English or Spanish and were either pregnant or their infant was less than two and half months old (May et al., 2017). Women completed an in-person screener to determine eligibility. Overall, a total of 4489 women were screened and eligible to complete the study and 4367 enrolled (97.3%). Of these, 2649 enrolled prenatally and completed a prenatal interview; an additional 1400 women enrolled after birth and completed the 1-or 3-month interview; and 318 did not complete either a prenatal, Month 1 or Month 3 interview after enrollment. Prenatal interviews and all follow-up surveys were completed via telephone in English or Spanish. A total of 16 follow-up surveys were completed from birth to 60 months at approximately 2-to 6-month intervals (i.e., Months 1,3,5,7,11,13,15,18,24,30,36,42,48,54,60 and 72). Prenatal interviews were completed within 60 days of enrollment and follow-up surveys were open for 4 weeks (i.e., starting 14 days before through 14 days after the participant reached the age of the monthly survey; meaning this could be as early as 2 weeks of age or as late as 6 weeks of age for the Month 1 survey). Women provided written informed consent and were provided incentives for enrolling and for each survey completed (May et al., 2017). We use three survey time points for this analysis: the prenatal, Month 1 and Month 3 interviews. For this analysis, no infants enrolled after birth were included because data were needed on prenatal intentions. Response rates for these interviews ranged from 87.7% to 92.3% (May et al., 2017). Detailed study methodology has been published (May et al., 2017).

| Breastfeeding intentions and meeting intentions
Prenatal breastfeeding intentions were assessed using a 5-point Likert scale to three questions: (1) 'I am planning to breastfeed my baby or

Key messages
• Among a longitudinal cohort of WIC participants, no differences in prenatal breastfeeding intentions existed between non-Hispanic Black and non-Hispanic White women.
• Meeting breastfeeding intentions differed by race/ethnicity.
• Hispanic and non-Hispanic Black women were less likely to meet their longer breastfeeding intentions than non-Hispanic White women.
• Understanding how racism, bias and discrimination contribute to women not meeting their breastfeeding intentions may help efforts to reduce breastfeeding disparities.
at least try', (2) 'When my baby is one month old, I will be breastfeeding without using any formula or other milk' , and (3) 'When my baby is three months old, I will be breastfeeding without using any formula or other milk' . Women who reported strongly agree/agree to the questions were categorized as intending to initiate breastfeeding (question one), breastfeed at Month 1 (Question 2) or breastfeed at Month 3 (Question 3).
Meeting breastfeeding intentions were determined only among women who intended to initiate breastfeeding, breastfeed at Month 1 or breastfeed at Month 3. Thus, meeting each of the three breastfeeding intention outcomes were examined independently and had a different denominator (those who disagreed/strongly disagreed or were neutral were not included in the denominator at each intention time point). Meeting breastfeeding initiation intention was defined as reporting that they initiated breastfeeding (May et al., 2017). Breastfeeding initiation was based on indication of breast milk consumption in any of a series of questions 1 assessing the first thing fed, feeding in the hospital, current feeding practices and feeding from a bottle. If mothers did not indicate breast milk consumption, they were classified as not initiating breastfeeding. Meeting intentions to breastfeed at Month 1 or at Month 3 were defined as the infant currently only receiving breast milk (no formula) at the Month 1 interview or at the Month 3 interview, respectively. Not meeting intentions to breastfeed at Month 1 or 3 included infants who were currently (1) receiving both breast milk and infant formula or (2) receiving only infant formula. These were the only three response options available and respondents may not have considered any other food or beverage a child was consuming when answering. Therefore, this categorization cannot be interpreted as a true reflection of exclusive breastfeeding because they do not account for any beverage or food other than breast milk or infant formula that was additionally reported.

| Analytic sample
We limited our analyses to women who completed at least 50% of the prenatal interview in either English or Spanish (n = 2649). Women were excluded if they did not complete the Month 1 interview (n = 414) and the Month 3 interview (n = 164) or if they were missing data on the intentions questions (n = 1). This left a final sample size of 2070, representing 78% of those who enrolled prenatally.

| Statistical analyses
We estimated the percentage of women's prenatal breastfeeding intentions and the percentage meeting those intentions for initiation and at Month 1 and Month 3, overall and by race/ethnicity. Differences in proportions by race/ethnicity were assessed using t-tests (P < 0.05). Multivariable logistic regression was done to assess the association of race/ethnicity with meeting prenatal breastfeeding intentions for (1) initiating breastfeeding, (2)  was used in all analyses to account for the complex sample design.
Analyses were weighted using the prenatal interview core weight which adjusts for differential probability of selection and nonresponse. Analyses are therefore representative of the WIC population who were enrolled in WIC sites with ≥30 participants/month. Defined as a mother indicating any breast milk consumption in a series of questions that assessed the first thing fed, feeding in the hospital, current feeding practices and feeding from a bottle.

breastfeeding goals. A population-based cohort study in the United
Kingdom found that 96% of women who intended to breastfeed for at least 4 months initiated breastfeeding (Donath et al., 2003). This may be comparable with our findings that 84-95% of women who intended to at least try breastfeeding initiated breastfeeding.
However, we also saw that one in 10 women did not meet their intention to initiate breastfeeding. This may be concerning given that meeting this intention could have been accomplished by putting the baby to the breast only one time. Understanding the reasons behind this estimate are warranted but were beyond the scope of this analysis.

Research assessing longer breastfeeding duration intentions has
found that not all women can breastfeed as long as they desire (Odom, Li, Scanlon, Perrine, & Grummer-Strawn, 2013 Fien, 2008). In contrast, we assessed meeting breastfeeding intentions at specific time points up to Month 3 and found high proportions of women who did not meet their intentions (e.g., 55% at Month 1). The two sample populations differ substantially and may be one of the reasons for the differences in findings. Our analyses were conducted among a low-income population who were participating in WIC, compared with IFPS II which was conducted among a sample drawn from a consumer opinion mail panel survey in which women tended to be more educated, employed and were more likely to be White than the national population (Fein et al., 2008).
Research has consistently found that breastfeeding intentions are a strong predictor of breastfeeding behaviours (DiGirolamo, Thompson, Martorell, Fein, & Grummer-Strawn, 2005;Donath et al., 2003;Langellier et al., 2014). However, our data suggest that women are experiencing challenges to breastfeeding in the first few weeks or months of their infant's life and that these may be more pronounced, Understanding the underlying factors that contribute to the differential experience of these factors for non-Hispanic Black women and Hispanic women is an important step in helping women meet their breastfeeding intentions and reduce breastfeeding disparities.
Research suggests that non-Hispanic Black women and Hispanic women face differential treatment and expectations when it comes to breastfeeding. For example, historically, Beal et al. (2003) found that African American women were more likely to receive support for bottle feeding and less likely to receive advice on breastfeeding than White women from the WIC programme. More recently, however, African American women who breastfed reported that WIC peer counsellors or participation in a WIC peer breastfeeding support group played a positive role in helping them breastfeed (Gross et al., 2017;Mickens et al., 2009;Robinson et al., 2016). A recent review by Robinson, Fial, and Hanson (2019) recommended a more focused attention on the impacts of racism, bias and discrimination within breastfeeding research which could provide important contextual information on ways to improve breastfeeding support programs and potentially aid in the reduction of breastfeeding disparities.
There are at least two additional factors that may help explain our findings for Hispanic women. Acculturation, the process of entering a new culture in which individuals may change their attitudes and/or Among women who strongly agreed/agreed with statement on breastfeeding intentions during pregnancy, 'When my baby is one month old, I will be breastfeeding without using any formula or other milk' . Meeting intention was defined as the infant only consuming breast milk at Month 1. d Among women who strongly agreed/agreed with statement on breastfeeding intentions during pregnancy, 'When my baby is three months old, I will be breastfeeding without using any formula or other milk' . Meeting intention was defined as the infant only consuming breast milk at Month 3. e Odds ratios are adjusted for maternal age at birth, education status, breastfeeding history, type of delivery, preterm status, presence of mother and father in same household and 2013 poverty guidelines.
f Models do not include women who responded Do not Know for general breastfeeding intention (n = 1), for intention at 1 month (n = 26), and for intention at 3 months (n = 32).
behaviours, has been inversely associated with breastfeeding outcomes (Ayala, Baquero, & Klinger, 2008;Lara, Gamboa, Kahramanian, Morales, & Bautista, 2005). A validated scale of acculturation was not collected in this study, limiting our ability to determine if the analytic sample was made up of more, or less, acculturated women which could have impacted our findings. Second, the concept of 'los dos' ('best of both') is the belief that providing both breast milk and infant formula provides twice the benefits for the baby (Waldrop, 2013). We found a high proportion of Hispanic women intending to breastfeed, which is consistent with other literature (Linares, Rayens, Gomez, Gokum, & Dignan, 2015;McKinney et al., 2012), but we defined meeting intentions as only providing breast milk which is in direct contrast to 'los dos' . We found a slightly higher proportion of Hispanic women reported consuming both breast milk and infant formula than observed among non-Hispanic White women which could suggest that 'los dos' is playing a role within this sample. Although not all WIC participants may receive the full breadth of breastfeeding supports offered (Rasmussen et al., 2017), these services are not available to the general US population. Although we still found significant racial/ethnic differences in the ability to meet breastfeeding intentions, these data do not allow us to compare what the differences in intention or meeting intentions may have been without the support of the WIC program. Recent findings have indicated that the WIC program plays a positive role in breastfeeding initiation and duration among African American women (Gross et al., 2017;Mickens et al., 2009;Robinson et al., 2016).
This study has several strengths and limitations. First, the study design allowed us to assess prenatal breastfeeding intentions and determine whether or not women met these intentions in a prospective manner, reducing recall bias. Second, this sample consisted of low-income women and children enrolled in WIC and had adequate sample sizes to allow for stratification by race/ethnicity, providing a unique opportunity to assess practices within a high-risk population.
However, we were unable to account for any within race/ethnicity group differences which may have impacted breastfeeding intentions and practices. Third, although our analytic sample represented 78% of women who were enrolled prenatally and therefore had responses to prenatal breastfeeding intentions, we still had missing data. Approximately 18% of those excluded (n = 106) had experienced a pregnancy loss and did not complete follow-up surveys. A sensitivity analysis was conducted to assess the sociodemographic differences between those included in the analysis compared with those excluded. Results indicated that excluded women tended to be younger, have lower levels of education and were less likely to have previous breastfeeding experience. Excluding these women may have impacted our findings, but it is not clear if this would have resulted in an overestimation or an underestimation. We did not address environmental breastfeeding supports or challenges that may impact the ability to meet breastfeeding intentions. This was beyond the scope of the current analysis but is an important area for future study. Lastly, we did not have information on whether or not women changed their breastfeeding intentions after birth or if they were satisfied with the length of time they breastfed.

| CONCLUSION
Among a sample of women and children enrolled in USDA's ITFPS-2, we found no differences in prenatal intentions to breastfeed at any time point between non-Hispanic White and non-Hispanic Black women, but a significantly higher percentage of Hispanic women intended to breastfeed at all time points. However, non-Hispanic Black and Hispanic women were significantly less likely than non-Hispanic White women to meet their breastfeeding intentions at either Month 1 or 3. Based on our review of the literature and consistent with our findings, non-Hispanic Black and Hispanic women may face additional barriers and challenges to meeting their longer breastfeeding intentions post-partum. Efforts to identify and address these challenges may require a more thorough assessment of the underlying factors that contribute to the differential experiences non-Hispanic Black women and Hispanic women face. Attention and inclusion of issues related to racism, bias and discrimination within breastfeeding research may be one step towards providing important contextual information to improve breastfeeding support programs and potentially aid in the reduction of breastfeeding disparities.

ACKNOWLEDGMENTS
We would like to thank the mothers and children who participated in the WIC ITFPS-2 study. In addition, we would like to thank Drs. Kelley Scanlon, Danielle Berman and Courtney Paolicelli for their guidance on describing the study methodology and determining the best analytic approach.

CONFLICTS OF INTEREST
The authors declare that they have no conflicts of interest. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.

CONTRIBUTIONS
HCH designed the study, conducted the analysis and drafted the initial manuscript. JLB, RL, JMN and CGP provided substantial