Authors: Neera K. Goyal, Laura B. Attanasio, Katy B. Kozhimannil
Categories: Article, Baby-Friendly Hospital Initiative, breastfeeding, early-term, hospital practices, late preterm
Source: Birth (Berkeley, Calif.)
Doi: 10.1111/birt.12135
Authors: Neera K. Goyal, Laura B. Attanasio, Katy B. Kozhimannil
Compared with term infants (39–41 weeks), early-term (37–38 weeks) and late preterm (34–36 weeks) infants have increased breastfeeding difficulties. We evaluated how hospital practices affect breastfeeding by gestational age.
This Listening to Mothers III survey cohort included 1,860 mothers who delivered a 34–41-week singleton from July 2011 to June 2012. High hospital support was defined as at least seven practices consistent with the Baby-Friendly Hospital Initiative’s Ten Steps for United States hospitals. Logistic regression tested mediating effects of hospital support on the relationship between gestational age and breastfeeding at 1 week postpartum.
High hospital support was associated with increased exclusive breastfeeding (AOR 2.21 [95% CI 1.58–3.09]). Just 16.4 percent of late preterm infants experienced such support, compared with early-term (37.9%) and term (30.7%) infants (p = 0.004). Although overall breastfeeding rates among late preterm, early-term, and term infants were 87, 88, and 92 percent, respectively, (p = 0.21), late preterm versus term infants were less likely to exclusively breastfeed (39.8 vs. 62.3%, p = 0.002). Inclusion of hospital support in multivariable modeling did not attenuate the effect of late preterm gestation.
Differences in practices do not account for decreased exclusive breastfeeding among late preterm infants. Hospital supportive practices increase the likelihood of any breastfeeding.
Given known maternal and infant health benefits of breastfeeding, efforts such as The Joint Commission’s Perinatal Care Core Quality Measures and the Baby-Friendly Hospital Initiative include a strong emphasis on supportive practices during childbirth hospitalization to promote breastfeeding (1,2). For late preterm infants, delivered at 34–36 weeks’ gestation and representing 70 percent of all preterm infants, breastfeeding may be particularly important for long-term physical and developmental health (3). However, this population is at risk for poorer breastfeeding outcomes, including early breastfeeding cessation, parental feeding-related anxiety, and re-hospitalization for feeding issues compared with infants born at 37 weeks’ or greater gestation (3–6). It is not yet known whether these infants have differential access to evidence-based supportive hospital practices (7).
More recently, evidence has emerged that even among early-term infants, delivered at 37–38 weeks, feeding difficulty and other complications are more likely than for those born at 39–41 weeks (8). Like late preterm infants, early-term infants are less likely than term infants to initiate breastfeeding and less likely to continue breastfeeding after discharge (9). Additional research demonstrates higher risk of rehospitalization for feeding-related issues compared with term infants (10), suggesting a relative physiologic immaturity that may have important implications at a population level.
Prior studies have conclusively demonstrated the importance of hospital practices, such as limiting formula supplementation and rooming-in of infants with their mothers, for both short-term and long-term breastfeeding success (11,12). However, only limited evidence is there for how such aspects of hospital care contribute to early outcomes among infants born at earlier gestational ages (7). Among late preterm infants in particular, who are more likely than term and early-term infants to be cared for in neonatal intensive care unit (NICU) settings (13), variation in hospital management including feeding policies and protocols for length of stay may contribute to differential breastfeeding outcomes (14,15). At least one prior study has demonstrated that NICU admission for late preterm infants may be protective against readmission after discharge, although the extent to which the admission is due to differences in formula supplementation or other interventions is unclear (14).
The purpose of the current study is to determine whether supportive hospital practices mediate the relationship between gestational age at birth and breastfeeding outcomes at 1 week postpartum among late preterm, early-term, and term newborns, adjusting for demographic and clinical factors. As shown in the conceptual model presented in Fig. 1, we hypothesized that, compared with term newborns, early-term and late preterm newborns may have a lower likelihood of exclusive breastfeeding, but that this effect could be partially mediated by differences in access to supportive practices during hospitalization.
Data for this analysis came from the Listening to Mothers III survey, a national sample of women who gave birth to a singleton baby in a United States hospital between July 1, 2011, and June 30, 2012 (N = 2,400). Commissioned by Childbirth Connection and conducted by Harris Interactive, this survey addresses the labor and birth experience, including questions about breastfeeding plans, hospital practices, and feeding patterns. Women completed the survey at an average of 11 months postpartum (range 4–18 months). Potential respondents were drawn from the Harris Poll Online, GMI, Research Now/E-Rewards, and Offerwise Hispanic panels. Eligible participants completed the survey online using a secure server with advanced web-assisted interviewing technology. Survey responses were weighted using propensity score methods to adjust for potential biases associated with the online survey mode, and further weighted to match the demographic characteristics of the sample to those of the target population (16). The final, weighted sample approximates nationally representative estimates, and weights are used throughout the analysis to retain this feature. Further details about the survey methodology and questionnaire are available at the website www.childbirth-connection.org/listeningtomothers/.
This study was approved by the University of Minnesota Institutional Review Board.
All 2,400 respondents were asked at the time of the survey to recall how they intended to feed their infant as they approached the end of pregnancy. Hospital practices and breastfeeding outcomes were assessed among women who reported intending to breastfeed (either exclusively or in combination with formula); therefore, the sample was limited to infants whose mothers intended to breastfeed. The final study population included 126 late preterm, 355 early-term, and 1,379 term infants, totaling 1,860.
Gestational age was calculated based on maternal report of due date and the infant’s date of birth. This calculation was then categorized as late preterm (34 0/7–36 6/7 weeks), early-term (37 0/7–38 6/7 weeks), or term (39 0/7–41 6/7 weeks) (8). As per the most recent gestational age definitions from the American College of Obstetrics and Gynecology, our categorization of term infants includes both term (39 0/7–40 6/7 weeks) and late-term (41 0/7–41 6/7 weeks) infants (17).
Our two primary outcomes were breastfeeding at 1 week postpartum (either exclusively or in combination with formula) and exclusive breastfeeding at 1 week. Breastfeeding outcomes for other specified time intervals were not available in the survey. We also examined women’s experiences of hospital practices consistent with the Baby-Friendly Hospital Initiative’s Ten Steps for United States Hospitals (1,18), which have been shown to be positively associated with breastfeeding initiation and duration (11,12). The survey included measures for the following Baby-Friendly infant in mother’s arms during first postpartum hour, rooming-in, hospital staff helped initiate breastfeeding, hospital staff demonstrated infant positioning for breastfeeding, hospital encouraged breastfeeding on demand, hospital staff did not provide water or formula supplements, hospital staff provided information on community breastfeeding resources, and hospital staff did not offer a pacifier. Baby-Friendly practices not available in the survey were the availability of a written breastfeeding policy for hospital staff and training of hospital staff in policy implementation.
We also measured whether the woman reported having skin-to-skin contact the first time she held her baby. Often called “Kangaroo Care,” this practice is also associated with breastfeeding success (19) and may be particularly important for late preterm infants (20). We created a series of indicator variables for each practice, and a composite measure, which we calculated by assigning one point for each practice and adding up the total (range 0–9). Women with scores of 7–9 on the measure were categorized as having experienced a high level of hospital support for breastfeeding (21).
Sociodemographic and birth-related covariates were based on the conceptual model and available evidence on factors associated with breastfeeding (22,23). These included maternal age, race-ethnicity, payer source, education level, marital status, and parity. Obstetrical covariates included delivery mode (vaginal or cesarean) and a composite variable for pregnancy risk factors which included prepregnancy hypertension requiring medication, preexisting or gestational diabetes, and prepregnancy obesity (body mass index ≥ 30). An indicator variable for complications of labor or delivery represented when cesarean delivery was indicated for malpresentation, fetal distress, macrosomia, placentation disorders, failed induction, disproportion, or other major health concerns. Labor and delivery was also classified as complicated if labor was induced for macrosomia, premature rupture of membranes, oligohydramnios, fetal distress, or other major health concern. An indicator variable for NICU admission was also assigned. Finally, intention to exclusively breastfeed was also included as a measure of high maternal motivation level.
We first examined bivariate associations between breastfeeding outcomes and gestational age category, high hospital support, and other covariates. Using multivariable logistic regression, we then conducted a mediation analysis, first testing whether gestational age category was independently associated with differences in likelihood of receiving a high level of hospital support (Step 1). We then used a similar regression model to estimate the independent effect of gestational age category on odds of any and exclusive breastfeeding at 1 week (Step 2). Next, we estimated the independent effect of experiencing a high level of supportive hospital practices on odds of any and exclusive breastfeeding at 1 week (Step 3). Finally, both gestational age category and high hospital support were included in the multivariable models of any and exclusive breastfeeding to assess changes in the absolute value of the coefficient for gestational age category (Step 4) (24). In a sensitivity analysis, we tested multiple different model specifications and added interaction terms to these models to determine whether the relationship between supportive hospital practices and breastfeeding outcomes varied by gestational age.
Somers’ D values were used to provide valid C-statistic estimates for the weighted data (25). All analyses were conducted using Stata v.12.
Covariates, key predictors, and breastfeeding outcomes of the sample by gestational age category are reported in Table 1. Sample characteristics generally mirror those of the United States childbearing population, with late preterm infants generally experiencing greater prevalence of clinical complications of pregnancy and delivery, including cesarean delivery and admission to the NICU. Fewer mothers of late preterm infants experienced several hospital practices intended to promote breastfeeding. For example, 29.9 percent reported holding their infants during the first hour after birth, compared with 51.6 percent of term infants. Just over a third (36.7%) of late preterm infants roomed in with their mothers, versus 67.6 percent of early-term and 64.7 percent of term infants. However, more mothers of late preterm infants and early-term infants (58.6 and 60.6%, respectively) reported receiving information about community breastfeeding resources, compared with mothers of term infants (48.4%). Only 16.4 percent of mothers with late preterm infants experienced a high level of supportive breastfeeding practices (7–9 of the practices examined), compared with 37.9 percent of mothers with early-term infants and 30.7 percent of mothers with term infants.
Table 2 demonstrates Step 1 of the mediation analysis, in which we assessed the effect of gestational age on receipt of high hospital support after adjusting for NICU admission and other covariates. There was no significant difference in odds of high hospital breastfeeding support for late preterm infants compared with term infants (adjusted odds ratio [AOR] 0.64 [95% confidence interval (CI) 0.33–1.24]). There was a trend for early-term infants, however, to have a higher level of hospital support compared with term infants (AOR 1.44 [95% CI 1.00–2.04]). In this model, NICU admission was independently associated with a reduced likelihood of receiving high breastfeeding support, (AOR 0.34 [95% CI 0.20–0.62]).
Step 2 of the analysis, also shown in Table 2, demonstrated the independent effect of late preterm gestational age on breastfeeding outcomes at 1 week postpartum. Compared to term infants, late preterm infants had less than half the odds of being breastfed exclusively at 1 week of life (AOR 0.38 [95% CI 0.23–0.65]). C-statistic values for the any breastfeeding and exclusive breastfeeding models were 0.73 and 0.76, respectively.
Step 3 of the mediation analysis confirmed the association between high hospital support and breastfeeding outcomes after controlling for covariates. We observed that a high level of hospital support for breastfeeding was independently associated with increased odds of any breastfeeding at 1 week (AOR 3.58 [95% CI 1.72–7.50]) and exclusive breastfeeding at 1 week (AOR 2.32 [95% CI 1.67–3.24]).
For Step 4 of the mediation analysis, gestational age and high hospital support were included in a multivariable model of breastfeeding outcomes, adjusting for other covariates (Table 3). Including hospital practices in this model resulted in small increases in the C-statistic values for both any and exclusive breastfeeding, now 0.76 and 0.80, respectively. However, the effects of late preterm gestation were unchanged, such that late preterm infants still had less than half the odds of exclusive breastfeeding at 1 week postpartum compared with term infants (AOR 0.40 [95% CI 0.24–0.68]). Thus, results of this mediation analysis indicate that differential access to hospital supportive practices, although an important independent predictor, does not mediate the observed relationship between gestational age and breastfeeding outcomes.
Lastly, to further characterize the impact of specific Baby-Friendly practices on breastfeeding outcomes, we repeated the mediation analysis with each of the nine practices as individual variables (Table 4). Adjusting for other covariates, late preterm infants were less likely than term infants to be held in the mothers’ arms during the first hour of life and to room in, and they were more likely to be offered a pacifier (Table 4, column 1). There were also trends for late preterm infants to have decreased likelihood of formula supplementation withheld and decreased likelihood of early skin-to-skin contact; however, these did not achieve statistical significance (p > 0.05). When these individual practices were included in the final multivariable model of breastfeeding outcomes, the negative effect for late preterm status on exclusive breastfeeding persisted but was slightly attenuated (AOR 0.48 [95% CI 0.27–0.87]) (Table 4, column 3).
We conducted sensitivity analyses with high hospital support defined as scores of 8–9, and results were substantively unchanged. We also repeated the analysis using the hospital support variable as a continuous rather than binary measure, and results were substantively unchanged. In the full multivariable model including gestational age, NICU admission, and other covariates, increasing number of supportive practices was associated with higher odds of any and exclusive breastfeeding (AOR approximately 1.45 in both models, p < 0.001).
We also tested for interaction terms between gestational age category and hospital support, which were not significant, indicating that the effect of hospital support does not differ by gestational age (results not shown).
While the association between lower breastfeeding rates and late preterm birth has been previously described, the specific contribution of hospital practices to this relationship has not previously been investigated. We show that, although supportive practices are less frequent among late preterm infants, lower access to this support as a whole does not contribute to differences in exclusive breastfeeding between late preterm and term infants. Similar to prior studies, high levels of hospital support dramatically increase the likelihood of breastfeeding; our results also support that this bundle of practices is effective for infants at late preterm, early term, and term gestations.
Our analyses demonstrate that late preterm infants compared with term infants are less likely to experience several hospital practices intended to support breastfeeding, including being held in their mother’s arms during the first hour of life, rooming-in, and withholding of a pacifier. Our results demonstrate that as a bundle of practices, high hospital support is strongly predictive of any breastfeeding and of exclusive breastfeeding. However, our analysis including each of the practices as individual factors demonstrates that early skin-to-skin contact may be particularly important for both any and exclusive breastfeeding, independent of gestational age, NICU admission, and other covariates.
Results of our mediation analysis demonstrate that although supportive practices are strongly predictive of higher breastfeeding rates, the relationship between late preterm birth and early breastfeeding is not substantively driven by differential access to these supportive practices as a whole. The magnitude of the difference in rates of exclusive breastfeeding between late preterm and term infants is similar, whether or not these supports were provided. These results suggest that the physiologic challenges associated with late prematurity, that is, feeding dysfunction, hypoglycemia, and temperature instability, provide logistical barriers in this population that often supersede the benefits of high-quality hospital supportive care. Our findings also suggest a trend toward decreased likelihood among late preterm infants of any breastfeeding, consistent with a recent report using 2000–2008 Pregnancy Risk Assessment Monitoring System (PRAMS) data, in which Hwang et al report a 5 percent lower likelihood of initiating breastfeeding between late preterm and term infants (9). Of note, differences in breastfeeding outcomes between late preterm and term infants persisted after adjustment for NICU admission. Our observation that NICU admission was associated with decreased likelihood of supportive practices likely reflects the impact of higher clinical severity. When included in the full multivariable models with gestational age, supportive practices, and covariates, NICU admission itself was not associated with significant differences in either any or exclusive breastfeeding at 1 week.
We found that early-term infants were more likely to receive a high level of breastfeeding support compared with term infants. Possibly this difference reflects a perception of these infants as being more vulnerable compared with term infants, coupled with fewer logistical barriers compared with late preterm infants. Of note, although outcome differences between early-term and term infants were not statistically significant, we did observe trends toward lower likelihood of any and exclusive breastfeeding. For the outcome of any breastfeeding, this trend was even more pronounced than for the late preterm infants, perhaps attributable to shorter lengths of stay and less follow-up support after discharge (9).
Taken together, these findings highlight the impact of hospital support regardless of gestational age for mothers who intend to breastfeed in meeting their goals. Late preterm infants and their mothers may benefit from adaptation of Baby-Friendly Hospital practices, particularly in Level 2 and Level 3 NICU settings as the clinical situation permits. Future research may focus further on hospital-level factors contributing to differential outcomes for late preterm infants, including policies for NICU admission and late preterm infant feeding protocols. Further research may also focus on how hospital support linked with high-quality follow-up support of breastfeeding (i.e., home visits) contributes to long-term breastfeeding within these gestational age categories (26).
The primary limitation of this study is our reliance on maternal self-report and lack of confirmation of data through hospital records, further compounded by retrospective collection of information many months after the infant’s birth date. This limitation may contribute to recall or social desirability bias in reporting breastfeeding intention, particularly among women who were unable to successfully breastfeed. Success or lack of success in breastfeeding may also contribute to selective maternal recall bias about the availability of supportive hospital practices. Additional limitations of the study include sample size and lack of information on breastfeeding outcomes at other designated follow-up intervals (e.g., 3 months or 6 months). Moreover, analyses were limited to the outcomes reported in this study and did not contain details on expression of breastmilk, paced feeding, or whether supplementation was clinically warranted. Finally, one hospital-level factor we lacked information on, in addition to establishment of a written breastfeeding policy and hospital staff training, is the level of NICU care provided at the birth hospital, that is Level 2 versus Level 3, which would help further characterize the impact of NICU admission on aspects of breastfeeding support. Strengths of this study include a sample of women from across the United States, data that reflect recent practices and experience of care, and the inclusion of variables not otherwise available in medical records or administrative data.
Infants born late preterm have considerably higher risk of complications that must be balanced against the substantial benefits of breastfeeding. Our analysis demonstrates that these infants, independent of other factors, are less likely to experience hospital practices supportive of breastfeeding. However, reduced access to this support does not substantially account for their lower exclusive breastfeeding rates compared with term infants. High levels of hospital support dramatically increase the likelihood of breastfeeding for infants regardless of gestational age, suggesting that such practices should be routinely provided as the clinical situation allows.