Parental Cohabitation and Breastfeeding Outcomes Among United States Adolescent Mothers

Introduction

Numerous risks for infants born to adolescent mothers, including infant mortality and nutritional deficiencies,^1–5 can be mitigated by breastfeeding or the receipt of breast milk. ⁶ Among the 180,000 births to adolescent mothers in the United States, ⁷ the maternal and child health benefits that breastfeeding affords would have lasting health and development effects on children. However, nearly all (83%) adolescent mothers do not comply with international recommendations to breastfeed for 6 months.^8,9

Adolescence is a time when exploration of self-identity occurs. Mothering while in the adolescent phase of life may promote self-identity and help build self-esteem. ¹⁰ While previous theories speculated that adolescent motherhood restricts ones self-identify, current research states that adolescent mothers have an increase in aspirations and motivations.^11,12 Yet, adolescent mothers often report low self-esteem^13,14—which is a strong and well-known predictor for breastfeeding duration^14,15 and could explain the low rates of breastfeeding among these mothers.

Compounded with social stigma and isolation from their peers, ¹⁶ adolescent mothers often lack a critical support network that is shown to help breastfeeding mothers. ¹⁷ In fact, adolescent mothers who wanted to breastfeed were more successful if they had support from their social network or health care providers. ¹⁸ Yet, stigmatization from health care providers is pervasive, and often appears as widespread negative stereotyping. ¹⁹ Further, the Centers for Disease Control and Prevention reported that up to 71% of adolescent mothers did not experience any of the Baby-Friendly Hospital Initiative maternal practices—a policy that directly supports mothers with breastfeeding. ²⁰

Our understanding of breastfeeding experiences for adolescent mothers is often speculative, ²¹ reducing confidence in current breastfeeding recommendations for adolescents. One factor with potential positive effects on adolescent breastfeeding behaviors is paternal support. Evidence suggests that support from the infant's father is one of the most influential factors for breastfeeding outcomes.^22,23 However, difficulties of measuring paternal support have lead researchers to identify proxies that can be easily measured and utilized in breastfeeding programs. In older mothers, marital status is the most common proxy for paternal support.^24,25 Research has repeatedly found that married women are more likely to initiate and breastfeed for a longer duration.^4,26–28 Although very few adolescent mothers are married to the infant's father at birth, almost half of unmarried adolescent mothers (42%) cohabit with a partner during their pregnancy or at the birth of their infant. ²⁹ Thus, for adolescent mothers, cohabitation may be a more appropriate measure for paternal support than marital status.

To our knowledge, only two articles investigate parental cohabitation and breastfeeding outcomes in mothers of all ages, both of which were conducted outside the United States and did not focus specifically on adolescents.^30,31 Nevertheless, evidence was found between cohabitation and breastfeeding. Kiernan and Pickett ³⁰ explored the relationship between parental connectedness, measured by varying relationship forms (married, cohabiting, single-closely involved with the infant's father, and single-not closely involved with infant's father), and breastfeeding initiation and duration. Evidence supported a dose-response relationship with suspected levels of partner support, with the lowest risk of never breastfeeding among cohabiting mothers and the highest risk among mothers not closely involved with the infant's father. ³⁰ Levinienė et al. reported that married women breastfed longer than women living with a partner or single. ³¹

With nearly half of US adolescent mothers cohabiting with their baby's father, it is important to investigate its relationship with breastfeeding initiation and duration. Our study investigates the relationship between parental cohabitation and breastfeeding initiation and duration among US adolescent mothers. We hypothesize that parental cohabitation is associated with greater odds of breastfeeding initiation and a longer breastfeeding.

Materials and Methods

Our cross-sectional study used female respondent and pregnancy data from the 2011–2017 National Survey of Family Growth (NSFG). NSFG is a weighted, nationally representative survey of people aged 15–49 years living in the United States that oversamples those identifying as non-Hispanic black or Hispanic, and adolescents aged 15–19 years.^32,33 In total, 16,854 women were interviewed about sexual and reproductive history, cohabitation, marriage, divorce, and attitudes on sex, family life, and gender. Trained female interviewers collected data in-person using computer-assisted personal interviewing (CAPI) software. More detailed information about NSFG can be found in an alternative source. ³³ This study was approved by the Institutional Review Board of the University of California Berkeley.

Our study sample included all women who reported having a singleton birth during adolescence. Adolescence was defined as age 15–19 years to enable comparisons to previous literature and nationally reported US teen pregnancy rates.^34–36 We excluded women whose child was older than 18 years at interview in 2011–2017 (n = 1,039) or whose child's age was missing (n = 2) due to the inability to collect informed consent from these now-adult children. After excluding respondents with missing cohabitation (n = 42) and breastfeeding (n = 95) data, our final sample size was 1,867 women.

Parental cohabitation status at birth (yes; no) was defined using the following survey question: “When [baby name] was born, were you either married to or living with his/her father?” Respondents who refused to answer or did not know were categorized as missing.

The two outcomes, breastfeeding initiation and duration, were analyzed separately. Breastfeeding initiation (yes; no), was defined using the following survey question: “When [baby name] was an infant, did you breastfeed him/her at all?” Among those who initiated, breastfeeding duration was analyzed as weeks the infant was breastfed using the following survey question: “How old was [she/he] when you stopped breastfeeding [her/him] altogether?” Due to data limitations, breastfeeding exclusivity could not be assessed.

Potential confounding factors selected a priori based on previous literature and availability of data^26,29,37,38 include maternal educational attainment, race/ethnicity, nativity, religion raised, maternal age at first birth, paternal age at birth, parental pregnancy intention, and year of infant's birth. Table 1 shows categorization schemes for all confounders. It is important to note that NSFG does not release data on further race/ethnicity distinctions for “non-Hispanic other.” We also included infant's year of birth to account for changes in breastfeeding and cohabitation norms over the 24-year range of births.

We examined the association of potential confounders with cohabitation and breastfeeding initiation using Rao Scott second order corrections to approximate an F statistic. We also examined the association between breastfeeding duration and potential confounders by identifying differences on the log scale using a Wald test. All reported estimates are weighted proportions and are representative of US adolescent women who gave birth to a singleton infant.

Breastfeeding initiation was analyzed using multivariable logistic regression. Among those who initiated, breastfeeding duration was analyzed using multivariate Cox proportional hazards model. In the Cox model, we censored women who reported still breastfeeding at the infant's age at interview (n = 12). Kaplan–Meier curves and the Schoenfeld residual test were used to determine if the proportionality assumption of the Cox model was met (p = 0.69). All models were fully adjusted for the a priori confounders. We also tested race/ethnicity and parental pregnancy intention for effect modification using cross product terms in both the multivariate logistic regression and multivariate Cox model with a conservative alpha of 0.15. All analyses were performed using Stata 15.1 statistical software.

Results

Our study population is ethnically diverse, with 31% identifying as Hispanic, 42% as non-Hispanic white, 21.3% as non-Hispanic black, and 5.7% as other. The majority of adolescent, primipara women were born in the United States, raised Protestant or Catholic, and had a mother (i.e., the infant's grandmother) with a high school diploma/GED or less than high school. Approximately 8 in 10 fathers were less than 25 years old at the infants birth and more than half of pregnancies were unintended by both parents. Over half of adolescent mothers reported living with the infant's father at time of delivery. Cohabiting adolescent mothers were more likely to be 18 and 19 years old, Hispanic or non-Hispanic white, and born outside of the United States. In addition, the infants' father was more likely to be 20–29 years and the pregnancy to be intended by both parents (Table 1).

Table 2 displays characteristics by breastfeeding initiation and duration. The majority (55.2%) of adolescent mothers initiated breastfeeding. Breastfeeding initiation was highest among Hispanic or non-Hispanic whites, mothers raised Catholic or other religion, and mothers born outside the United States. As infant's birth year became more recent, breastfeeding initiation increased steadily; however, breastfeeding duration did not change substantially. Breastfeeding duration was longest among adolescent mothers identifying as Hispanic or other race/ethnicity, born outside the United States, and raised as other religion (Table 2).

After adjusting for all a priori identified confounders, adolescent mothers cohabiting with the infant's father at birth had 1.5 times the odds (95% confidence interval: 1.08–2.16) of initiating breastfeeding than adolescent mothers not cohabiting with the infant's father. Race/ethnicity was found to be a significant effect modifier in the logistic regression model using Wald tests (p-value = 0.12). After stratifying by race/ethnicity, adolescent mothers identifying as Hispanic or non-Hispanic white were more likely to initiate breastfeeding if cohabiting with the infant's father compared to those not cohabiting. We found no evidence of an association among adolescent mothers identifying as non-Hispanic black or non-Hispanic other (Table 3). Parental pregnancy intention was not an effect modifier in our model.

Table 4 displays hazard ratios (HR) for parental cohabitation and breastfeeding cessation. We found no significant association between parental cohabitation and breastfeeding duration in the Cox model (HR_adjusted: 1.18; 95% confidence limit: 0.94, 1.48; Table 4). The Kaplan–Meier curve showed a similar proportion of mothers breastfeeding overtime for both cohabiting and noncohabiting adolescent mothers. We did not find evidence of effect modification by race/ethnicity for breastfeeding duration or parental pregnancy intention.

Discussion

Our results suggest that US adolescent mothers cohabiting with the infant's father at birth are more likely to initiate breastfeeding than mothers not cohabiting with the infant's father. However, this relationship varies by race/ethnicity. Further, despite strong evidence of the importance of paternal support and involvement for breastfeeding duration, ³⁹ we found no association between parental cohabitation and breastfeeding duration. To our knowledge, there are no comparable studies investigating cohabitation and breastfeeding outcomes among US adolescent mothers. Nevertheless, our results are supported by international research among mothers of all ages, including Kiernan and Pickett's study, which found that cohabiting women are at lower risk of not initiating breastfeeding. ³⁰ Consistent with our findings, they also reported no association between cohabitation and breastfeeding duration. ³⁰

Contradictory to our hypothesis, we found no association between parental cohabitation and breastfeeding duration among adolescent mothers. This may be explained by the varying support systems adolescent mothers may have, including health care providers, parents, friends, and partners.^{21,26,40–44} In fact, due to the varying support networks, adolescent mothers may utilize, paternal support may not be as influential on breastfeeding practices as it is among older mothers. However, this theory has yet been tested since the majority of research focuses on older mothers' support networks, rather than adolescents’. Another potential explanation is the significant hardships adolescent mothers face when transitioning quickly from adolescence to adulthood. The early start to parenthood is not structurally supported by societal social norms, increasing adolescent mothers' stress.^45,46 Yet, some research has shown that some adolescent mothers may utilize the transitional period for growth and adaption, accepting their new social role into motherhood.^45,46

Our findings suggest a differing relationship between cohabitation and breastfeeding initiation based on race/ethnicity. Specifically, parental cohabitation was associated with higher odds of breastfeeding initiation in Hispanic and non-Hispanic white mothers, but not significantly associated for non-Hispanic black and other race/ethnicities. Familial history around breastfeeding may help explain these differences. In the United States, infant feeding norms vary substantially by race/ethnicity. Both Hispanic and non-Hispanic white mothers are more likely to breastfeed their infants^40,47 compared to non-Hispanic black mothers. Current trends in breastfeeding practice may directly influence partners breastfeeding support and encouragement, which is often aligned with social norms and their family's previous breastfeeding behaviors.^40,48

Limitations to this study include our inability to assess breastfeeding exclusivity and adjust for other potential confounders, such as school/work status at birth and other people living with the mother at birth. We were also unable to control for changes in policies that may directly impact experiences of adolescent mothers, including Temporary Assistance for Needy Families and Title X, or societal changes and norms surrounding breastfeeding. Although NSFG's data collection strategies attempt to reduce bias, social desirability bias in breastfeeding reports and recall bias in both cohabitation and breastfeeding reporting may still be present. ³³ However, literature shows that breastfeeding recall is reliable long after giving birth. ⁴⁹ Finally, NSFG may not accurately reflect contemporary adolescent mothers experiences due to the need to combine multiple years of data to appropriately estimate relationships.”

Conclusions

Breastfeeding may change the life-course for adolescent mothers and their infants by improving economic and health outcomes. ⁶ Our study sheds light on the role of cohabitation in breastfeeding initiation and duration for adolescent mothers. Implementing adolescent-friendly health services and adapting current systems, including the Baby-Friendly Hospital Initiative, could improve prenatal, delivery, and postpartum experiences for adolescent mothers. Before birth, practitioners should also consider whom the pregnant adolescent is living with and what beliefs they hold around breastfeeding, while also taking into account social norms, family breastfeeding history, and available resources. More research is needed on proxies for paternal support and how different sources of support are associated with breastfeeding among adolescent mothers.