Smoking and Race Associated with Decreased Breastfeeding Initiation and Duration Among a Low Income Population

Abstract

Objectives:

To assess associations between smoking behaviors during pregnancy and postpartum and breastfeeding initiation and duration, among DC Women, Infants, and Children (WIC) recipients, the majority of whom are non-White, controlling for sociodemographic factors, low birth weight (LBW), and delivery in a Baby-Friendly hospital (BFH).

Materials and Methods:

A series of ordinary least squares and logistic regressions estimated the association between smoking during pregnancy and postpartum on breastfeeding initiation and duration, respectively. A multilevel modeling approach accounted for WIC site effects.

Results:

Overall, 8.8% of DC WIC recipients smoked during pregnancy. Smoking during pregnancy and postpartum was negatively associated with breastfeeding initiation (adjusted odds ratio [aOR] = 0.47, 95% confidence interval [CI]: 0.36, 0.61) and duration (adjusted coefficient [B] = −37.96, 95% CI: −62.92, −13.00) in adjusted models, respectively. Maternal age, race, marital status, LBW, and delivery in a BFH were also significantly associated with initiation, while age, race, and marital status were significantly associated with duration. Likewise, the number of cigarettes smoked per day was negatively associated with initiation (aOR = 0.79, 95% CI: 0.72, 0.87) and duration (B = −8.98, 95% CI: −15.55, −2.41) in adjusted models.

Conclusions:

Both smoking and number of cigarettes smoked during pregnancy and postpartum are significant factors associated with less breastfeeding in the DC WIC population. Furthermore, smoking cessation during pregnancy shows initial promise to increase breastfeeding initiation. Future research is needed to better understand the role of smoking, relapse, and cessation interventions on breastfeeding rates among low-income, predominantly minority populations.

Introduction

Breastfeeding is recommended by the World Health Organization (WHO) and the American Academy of Pediatrics (AAP) because of the many infant and maternal health benefits conferred.^1,2 Specifically these recommendations include exclusive breastfeeding for the first 6 months of life and then breastfeeding in addition to complementary foods for an additional 1 to 2 years or until weaning is desired.^1,2 For infant health, breastfeeding is associated with reductions in a variety of conditions such as respiratory tract infections, otitis media, gastrointestinal infections, overweight and obesity, diabetes, and Sudden Infant Death Syndrome.^1–5 Furthermore, benefits to the mother associated with breastfeeding include reduced postpartum weight,⁶ lower risk of diabetes,⁷ cardiovascular disease, hypertension,^8,9 and breast and ovarian cancer.^3,10

In the United States, breastfeeding rates for low-income women receiving benefits from the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) demonstrate current and historically lower breastfeeding rates compared to nonparticipant counterparts.^11,12 Breastfeeding rates also differ by race/ethnicity with African American women consistently demonstrating lower breastfeeding rates compared to other racial/ethnic groups.^13,14 According to the Association of State and Territorial Health Officials (ASTHO) and the District of Columbia (DC) Department of Health, breastfeeding initiation among low-income African Americans in DC was only 57.1%, and the sustained breastfeeding rate until 6 months of age was about 26%.¹⁵

Other known factors associated with lower breastfeeding rates for mothers include less education, younger age, unmarried status, lower income, and return to work.^14,16–18 Studies examining “Baby-Friendly” hospitals or birth facilities (BFH), which emphasize policies to support breastfeeding, suggest favorable outcomes for breastfeeding initiation, duration, and exclusivity.^19,20

Smoking during pregnancy and the postpartum period is another factor known to decrease breastfeeding initiation and duration in addition to its known negative impacts on health.^21–24 The latest AAP guidance indicates that smoking is not a contraindication to breastfeeding and that women who smoke can still breastfeed (while simultaneously working to reduce or quit cigarettes).¹ As smoking represents one of few modifiable behavioral risk factors to affect breastfeeding, understanding its influence on breastfeeding rates is critical. Furthermore, promoting smoking cessation during pregnancy and relapse prevention postpregnancy mark vital interventions to improve women's health during a critical time period for both mother and infant health. The Healthy People 2030 goal of 95.7% of smoking abstinence during pregnancy underscores the need for such efforts.²⁵

According to the National Center for Health Statistics, 7.2% of women with live births reported smoking during their pregnancy in 2016.²⁶ Smoking during pregnancy was higher among women aged 20–24 (10.7%) and those with a high school education (12.2%).²⁶ Higher smoking rates during pregnancy are also found among low-income and unmarried women.²⁷ Among WIC recipients nationally in 2014, 12.6% reported smoking during pregnancy, more than double the rate in the non-WIC population (5.2%).²⁷

This observational study sought to examine the associations between smoking during pregnancy and breastfeeding initiation, and postpartum smoking and duration, among a population of low-income, predominantly minority DC WIC recipients controlling for sociodemographic factors, low birth weight (LBW), and delivery in a BFH. We hypothesized that women who smoked in this sample would have lower initiation of breastfeeding and shorter breastfeeding duration than women who did not smoke, holding all other sociodemographic factors and BFH constant. Second, we hypothesized that delivering in a BFH would be positively associated with breastfeeding initiation but not with duration holding both smoking status and all sociodemographic factors constant. Third, we hypothesized that recipients self-identifying as Black or African American would have lower breastfeeding initiation and shorter duration compared to those who are not Black, holding smoking status, BFH delivery, and all other sociodemographic factors constant. In addition, we explored whether the number of cigarettes smoked per day (CPD) influenced either breastfeeding initiation and/or duration. Finally, we addressed the question of whether smoking cessation versus smoking continuation during pregnancy was associated with breastfeeding initiation in a subsample of women.

Materials and Methods

Data source

This study is a secondary analysis of deidentified data collected as part of routine WIC program activities in Washington, DC and, thus, not considered Human Subjects Research by the DC DOH's IRB. The District of Columbia's Department of Health's WIC Program collects demographic and health indicators from each recipient over multiple prenatal and postpartum visits along with breastfeeding initiation and duration data.

Sample

The data were extracted for all mothers who had given birth to a singleton infant between January 1, 2014 and June 30, 2016. There were 10,173 women in the dataset after excluding duplicate records, women who had multiples/twins, were pregnant only during the reporting period, had missing race data, or were less than 18 years of age (Table 1). The final models include all women that had breastfeeding outcome data specified and no missing responses for independent variables included in the model.

Table 1.

Sociodemographic, Breastfeeding, and Substance Use Characteristics of Infant WIC Recipients Enrolled in Washington, DC Compared to Overall National WIC Recipients, 2014

Sociodemographic characteristics	DC WIC participants who smoked during pregnancy n = 893	DC WIC participants who smoked postpartum n = 851	DC WIC Sample N (%) ^a Total N = 10,173	National WIC Sample % Total^b
Age
18–34	377 (80.6)	357 (83.8)	7,931 (78.0)	85.9
Average, range (in years)	28.9, 18.0–44.6	27.9, 16.9–44.6	27.3, 13.9–50.5	N/A
Median education level, range (in years)	11.5, 5–16	11.4, 5–16	11.7, 1–18	12.0
Race/ethnicity
Non-Hispanic Black	452 (99.1)	417 (99.8)	7,245 (77.2)	24.1
Hispanic	2 (0.2)	0 (0.0)	2,135 (21.0)	37.9
Non-Hispanic White	8 (0.9)	7 (0.8)	184 (1.8)	31.9
Married	19 (2.1)	18 (2.1)	1,245 (12.2)	32.0^c
Low-birth weight baby	71 (8.0)	65 (7.7)	791 (7.8)	4.4^d
Initiated Breastfeeding	145 (31.0)	82 (19.3)	5,531 (54.4)	69.8^e
Gave birth at a BFH	217 (29.1)	191 (26.7)	5,177 (63.7)	18.3^f
Smoking during pregnancy	—	—	893 (8.8)	12.6^g

Percentages calculated exclude missing data, thus denominator for each variable may be lower than total.

Based on Infant WIC recipients between January 1, 2014 and June 30, 2016.

Based on Infant WIC recipients only (N = 205,684), Source: Thorn et al. (2015).³³

Source: Soneji et al. (2019).³⁴

Based on Child WIC recipients only (N = 4,961,804), Source: Thorn et al. (2015).³³

Based on breastfeeding initiation rates reported for infants aged 6–13 months in 2014, (N = 1,369,284), Source: Thorn et al. (2015).³³

Percentage of births in the United States (WIC and non-WIC populations) taking place in a Baby-Friendly hospital.¹²

Source: Curtin and Matthews, 2016.²⁷

BFH, Baby-Friendly hospital; WIC, Women, Infants, and Children.

Study variables

Main outcome

The outcome variables of interest were: (1) breastfeeding initiation (yes/no) and (2) breastfeeding duration, which was analyzed as a continuous variable (days).

Breastfeeding initiation

WIC staff asked recipients if they had ever breastfed, and this was recorded in the recipient's record. The outcome was determined using WIC food package issuance data, which is a valid indicator for breastfeeding behavior.²⁸

Breastfeeding duration

At each visit to the WIC program, recipients were asked to report if they were currently breastfeeding or to provide the date when they had stopped breastfeeding. If an end date was recorded, the number of days that a recipient breastfed (i.e., duration) was calculated.

Main exposure

Smoking status

Three variables were used to examine smoking behavior.

Smoking during the pregnancy

This was a dichotomous variable reported at each visit to the WIC clinic (yes/no).

Smoking during the postpartum period

This was a dichotomous variable (yes/no) reported at each visit to the WIC site.

CPD during the pregnancy and CPD during the postpartum period were both recorded. CPD was a continuous variable reported by mothers at each visit to the WIC site. The last reported value for each of these two variables was made available to the research team and used in this analysis.

Smoking cessation during pregnancy

A subsample of women reported changes in their smoking status during pregnancy. This was coded as a dichotomous variable (quit/continued smoking during pregnancy).

Confounders

Infant health

LBW status was a dichotomous variable (yes/no).

Breastfeeding environmental support

At the time of data collection, a District-wide initiative to support the transformation of all birthing facilities and hospitals in DC into BFH was happening. Whether delivery occurred at a BFH was included as a dichotomous variable. WIC sites varied in their breastfeeding support programs. The WIC site in which the mother/infant sought care was included as a categorical variable.

Demographics

Sociodemographic and potential confounders included in the analysis were based on a literature review.^14,16–20 Race/ethnicity was dichotomous, non-Hispanic Black versus other races the majority of whom were White and Hispanic. Maternal age and education, both in years, were continuous variables. Maternal marital status was included as a dichotomous variable (married or not married).

Statistical analysis

First, bivariate analyses were conducted between (1) the main outcomes of interest and each of the sociodemographic variables and (2) between each of the sociodemographic variables, to assess their relationships with each other and multicollinearity. A Pearson correlation was conducted between breastfeeding duration and CPD to determine whether they were significantly correlated. Three multivariate logistic regression models assessed the relationship between breastfeeding initiation (yes/no) and (1) smoking during pregnancy (yes/no), (2) CPD during pregnancy, and (3) smoking cessation during pregnancy (yes/no). All models adjusted for the other independent variables. Two multivariate linear regression models assessed the relationship between breastfeeding duration (a continuous outcome, days) and (1) smoking during the postpartum period (yes/no) and (2) CPD during the postpartum period. Both models adjusted for the independent variables (ordinary least squares regression or OLS).

Finally, WIC site was handled with a nested hierarchical approach rather than being controlled for in the regressions above, since each WIC site can differ in the types of breastfeeding support it provides recipients. For example, some WIC sites have greater availability of breastfeeding education programs and/or number of peer counselors than others. To confirm the most appropriate approach, first the Hausman test to help meet asymptotic assumptions was run to formally test both the random effects and fixed effects models^29,30 looking at variance within and between groups. Once all assumptions were met, a random effects model was selected, and generalized latent linear models were used to run subsequent analyses and to strengthen normality and standard errors. In addition, null multilevel models (MLMs) were run first using the maximum likelihood approach to ensure that sufficient evidence for grouping by WIC site was appropriate. Furthermore, based on the variance in breastfeeding behavior associated with WIC site (suggested to be around an intraclass correlation coefficient of 0.1), it was clear that “WIC site” was explaining a sufficient amount of variance to proceed with the full MLM.³¹ Women without a breastfeeding initiation status were removed from the initiation related analyses, and likewise, those without a breastfeeding end date specified were removed from duration related analyses. Women with data missing on the key variables included in the regression analyses were also removed. All analyses were conducted with STATA version 16 (College Station, TX).³²

Results

Study sample and characteristics

The total sample included 10,173 DC WIC recipients with the number included for each analysis dependent on smoking status (Table 1). The sample included women who completed a median of 11.7 years of school, were mostly unmarried (87.8%), identified as Non-Hispanic Black (77.2%), and were on average 27.3 years old. Almost two-thirds (63.7%) of the sample delivered in a Baby Friendly birthing facility, and 7.8% gave birth to a LBW baby (Table 1). The DC WIC population is different from the national WIC population,^12,33,34 with a much larger proportion of Non-Hispanic Black recipients, a greater prevalence of LBW outcomes, and a greater number of births occurring in BFH (Table 1).^12,33,34

Smoking during pregnancy and breastfeeding initiation

Over one-half of the DC WIC population (54.4%) initiated breastfeeding, lower than the proportion reported in the national WIC population (69.8%) (Table 1). Approximately 8.8% of the DC WIC population smoked during their pregnancy, and this is lower than the national proportion of 12.6% (Table 1). Women who smoked during pregnancy had significantly lower odds for initiating breastfeeding than nonsmokers (unadjusted odds ratio [uOR] = 0.32, 95% confidence interval [CI]: 0.27, 0.40) (Table 2). After adjusting for demographic variables (education, age, and race), giving birth at a BFH and LBW, smoking during pregnancy remained significant with lower odds of breastfeeding initiation (adjusted odds ratio [aOR] = 0.47, 95% CI: 0.36, 0.61) (Table 3). In the adjusted model, maternal age, race, marital status, giving birth in a BFH, and LBW were also significantly associated with breastfeeding initiation (Table 3), with Black women at lower odds compared to Hispanics and those babies born with LBW at lower odds compared to non-LBW babies. In contrast older women, married women, and those who delivered in a BFH had higher odds of initiating breastfeeding (Table 3).

Table 2.

Smoking Characteristics of WIC Recipients Enrolled in Washington, DC and Unadjusted Association with Breastfeeding Behavior, 2014–2016

	n (%)	Breastfeeding initiation, n (%)	uOR (95% CI)
Did not smoke during pregnancy	9,280 (91.2)	5,386 (58.0)	Ref.
Smoked during pregnancy	893 (8.8)	145 (31.0)	0.32 (0.27, 0.40)^***
		Average breastfeeding duration in days, SD	Unadjusted coefficient 95% CI
Did not smoke during postpartum period	9,245 (91.6)	127.8, 118.7	Ref.
Smoked during postpartum period	851 (8.4)	79.7, 75.0	−48.1 (−69.5, −26.7)^***

Percentages calculated exclude missing data.

***

Indicates p-value is significant, <0.001.

95% CI, 95% confidence interval; SD, standard deviation; uOR, unadjusted odds ratio.

Table 3.

Multilevel Model of Smoking and Sociodemographic Variables on Breastfeeding Initiation Among WIC Recipients Enrolled in Washington DC (n = 6,876)

Independent variables	uOR	95% CI	aOR	95% CI
Smoking during pregnancy (versus not)	0.32	0.27, 0.40^***	0.47	0.36, 0.61^***
Education (years)	1.00	0.98, 1.02	—	—
Maternal age	1.05	1.04, 1.06^***	1.02	1.01, 1.03^***
Race (non-Hispanic Black versus Hispanic)	0.21	0.18, 0.23 ^***	0.46	0.40, 0.54^***
Married (versus not married)	2.97	2.58, 3.42^***	1.87	1.56, 2.24^***
Giving birth in a BFH (versus not)	1.56	1.41, 1.71^***	1.15	1.04, 1.29^**
LBW	0.72	0.62, 0.83^***	0.82	0.68, 0.98^*

This sample size includes all women who reported they had breastfed for one or more days and had a breastfeeding end date specified and no missing responses for the independent variables included in the model.

Indicates p-value is significant <0.05; ^**indicates p-value is significant <0.01; ^***indicates p-value is significant, <0.001.

aOR, adjusted odds ratio; LBW, low birth weight.

Change in smoking status during pregnancy and breastfeeding initiation

We analyzed a subset of data from women who answered the question about changes to smoking status during pregnancy (n = 1,231). Of these, 371 (30.1%) reported quitting smoking during pregnancy. Smoking cessation during pregnancy was associated with significantly higher odds of breastfeeding initiation (uOR = 1.49, 95% CI: 1.14, 1.99) and remained the only significant variable associated with the outcome, after adjusting for demographic variables (aOR = 1.57, 95% CI: 1.07, 2.29), adjusting for mother's age, race, marital status, giving birth at a BFH, and LBW (data not shown).

Smoking postpartum and breastfeeding duration

About 8.4% of the DC WIC sample reported smoking during the postpartum period. Women who smoked during the postpartum period had a lower duration of breastfeeding; on average, 48.1 days less than women who did not smoke (p < 0.001) (Table 2). Variables that were significantly associated with breastfeeding duration in the unadjusted analysis were included in the multivariate model, namely, smoking during the postpartum period, education, maternal age, race, marital status, and giving birth in a BFH (Table 4). In the multivariate model, postpartum smoking remained significantly associated with lower breastfeeding duration, although the difference decreased to 37 days compared to nonsmokers. In this same model, married women and older maternal age were significant variables for longer breastfeeding duration. In contrast, non-Hispanic Black women had significantly shorter breastfeeding duration than Hispanic women (Table 4).

Table 4.

Multilevel Model of Smoking and Sociodemographic Variables on Breastfeeding Duration Among WIC Recipients Enrolled in Washington DC (n = 2,545)

Independent variables	Unadjusted coefficient	Standard error	95% CI	Adjusted coefficient	Standard error	95% CI
Postpartum smoking (versus not)	−48.06	10.91	−69.46, −26.67^***	−37.96	12.74	−62.92, −13.00^***
Education (in years)	−3.99	0.87	−5.70, −2.28^***	−1.47	1.06	−3.55, 0.62
Maternal age (years)	4.45	0.32	3.82, 5.08^***	3.38	0.40	2.60, 4.16^***
Race (non-Hispanic Black versus Hispanic)	−60.18	4.66	−69.32, −51.04^***	−35.64	7.00	−49.36, −21.94^***
Married (versus not married)	43.87	6.31	31.50, 56.23^***	18.45	7.70	3.38, 33.52^*
Giving birth in a BFH (versus not)	21.49	5.00	11.74, 31.25^***	9.32	5.08	−0.64, 19.29
LBW	−7.74	7.12	−21.68, 6.19	—	—	—

Indicates p-value is significant <0.05; ^**indicates p-value is significant <0.01; ^***indicates p-value is significant, <0.001.

CPD and breastfeeding behaviors

Among smokers who reported the number of CPD during pregnancy (n = 311/893) and postpartum period (n = 416/851) (data not shown), women smoked between 1 to 30 CPD, with an average of 3.97 CPD.

Significantly lower odds of breastfeeding initiation were found with increasing numbers of CPD smoked during pregnancy (aOR = 0.79, 95% CI: 0.72, 0.87) after adjusting for potential confounders (Table 5). Maternal age, race of mother, marital status, LBW status, and delivering in a BFH were also significantly associated with the outcome. Similarly, the number of CPD in the postpartum period was associated with a significant decrease in breastfeeding duration (beta = 0.79, 95% CI: 0.72, 0.87) in the multivariate model after adjusting for potential confounders (Table 5).

Table 5.

Multilevel Model of CPD During Pregnancy and Breastfeeding Initiation and CPD Postpartum and Breastfeeding Duration Among WIC Recipients Enrolled in Washington DC

	Initiation (n = 7,331)		Duration (n = 2,551)
Independent variables	aOR	95% CI	Adjusted coefficient	Standard error	95% CI
CPD	0.79	0.72, 0.87^***	−8.98	3.35	−15.55, −2.41^**
Education (in years)	—	—	−1.50	1.06	−3.59, 0.58
Maternal age (years)	1.02	1.01, 1.03^***	3.41	0.40	2.63, 4.19^***
Race (non-Hispanic Black versus Hispanic)	0.46	0.39, 0.54^***	−36.08	7.00	−49.79, −22.37^***
Married (versus not married)	1.90	1.58, 2.27^***	18.00	7.67	2.96, 33.04^*
Giving birth in a BFH (versus not)	1.15	1.03, 1.28^**	9.17	5.07	−0.76, 19.10
LBW	0.82	0.68, 0.98^*	—	—	—

Indicates p-value is significant <0.05; ^**indicates p-value is significant <0.01; ^***indicates p-value is significant, <0.001.

CPD, cigarettes smoked per day.

Discussion

This research is the first to our knowledge to measure the direct impact of smoking and related sociodemographic factors on breastfeeding initiation and duration in the population of DC WIC mothers and confirms the previous findings of the negative association between smoking and breastfeeding.^12,18,21 Consistent with our first a priori hypothesis, women in our predominantly minority WIC sample who smoked had significantly lower odds of breastfeeding initiation and duration than women who did not smoke, holding BFH and all other sociodemographic variables constant. Our results are consistent with the literature on smoking and breastfeeding, providing additional evidence that even in a population with many known risk factors for lower breastfeeding rates, smoking still matters.^21–24 Other studies in both international²⁴ and U.S. contexts²³ found a strong negative relationship between maternal smoking and breastfeeding initiation. Wallweiner et al.²⁴ sampled a longitudinal cohort of 330 German women during pregnancy and postpartum and found that 4 months after birth, maternal smoking was among the strongest predictors of breastfeeding cessation, consistent with our findings. In the United States, a study of 2,323 women in South Appalachia found that maternal smoking was the strongest predictor of not initiating breastfeeding, and this finding correlated with the number of cigarettes smoked daily.²³ Similar to our study, the authors also found education and marital status (as well as insurance and parity, not measured in our study) as significant predictors of breastfeeding initiation. Moreover, a population-based cohort using the Pregnancy Risk Monitoring System (PRAMS) found a correlation between smoking and early weaning.²¹

Delivering in a BFH significantly improved breastfeeding initiation and duration in our sample, also consistent with our second hypothesis. These findings agree with a review of BFH studies in the United States demonstrating associations with increased initiation, exclusivity, and possibly duration.²⁰ The proportion of Baby Friendly certified hospitals in DC ranks significantly higher than the national average. In 2009, DC DOH and other stakeholders invested heavily in the goal of turning every DC birthing hospital/facility into a BFH, which explains why a significantly larger percentage of our sample delivered in a BFH compared to the national WIC sample.³⁵

Our third hypothesis of Black race being negatively associated with breastfeeding initiation and duration was supported and is consistent with previous studies.^13,14 This highlights the need for interventions to address racial disparities in breastfeeding in DC and elsewhere in the country. The percentage of women reporting tobacco use was lower in our sample compared to the national WIC sample. This aligns with data from a large nationally representative sample from the “Current Population Survey Tobacco Use” supplement which found that Black and Hispanic women smoke at lower rates than White women, with some regional variations.³⁶ Since our sample consisted of largely non-Hispanic Black women, future research looking at larger subgroups of Hispanic and White women and the influence of smoking on breastfeeding is warranted.

In the exploratory analyses, for each unit increase in the number of CPD during pregnancy, a statistically significant 21% decrease in the likelihood of breastfeeding initiation was found, controlling for sociodemographic variables and BFH status. A statistically significant inverse association was also found for the number of CPD postpartum and breastfeeding duration, indicating a 9-day reduction in breastfeeding for each unit increase in number of CPD, controlling for sociodemographic variables and BFH status. The study of U.S. women in South Appalachia described above similarly found lower breastfeeding initiation among women who smoked more cigarettes per day and for longer periods of time,²³ aligning with our findings.

Furthermore, among women who smoked, we found that those who quit during pregnancy showed higher odds for breastfeeding initiation than those who continued to smoke, reinforcing the need for efforts to address smoking during pregnancy, a time period when women may be particularly open to smoking cessation interventions or interventions that improve health. Joseph et al.²² sampled 300 women who quit smoking prenatally with follow-up about breastfeeding at 12-weeks postpartum. Although only one-third of women stated that smoking influenced their breastfeeding decision, of these, 83% reported continued cessation or decreased smoking frequency to breastfeed while the remainder forewent breastfeeding to smoke. A number of smoking cessation programs within WIC have demonstrated success with recipients providing evidence of their importance for this population in the prenatal and/or postnatal periods.^37–39 Applying these or similar cessation programs to the DC WIC context to improve breastfeeding outcomes is supported by these findings.

Together, these results on smoking, sociodemographic variables, and BFH status and their relation to breastfeeding in DC WIC recipients bring important and needed evidence to inform the design of interventions to address this context and population specifically. Given the strong relationship demonstrated between maternal smoking and breastfeeding rates, and the evidence that smoking cessation during pregnancy positively impacts breastfeeding,²² continued attention to programs that support pregnant women to quit smoking on local and national levels and those specifically serving WIC recipients^27,37,39 is warranted. The strong impact of BFH on breastfeeding, policies, and hospital environments that support breastfeeding merits continued expansion to other hospitals/facilities to help mitigate risk factors for low breastfeeding rates. In addition, given the limited information about smoking within the BFH programming/policies, further work to expand on this complementary evidence-based risk within the program is needed. Our results add evidence to WIC programs nationally by studying variables of interest across programs and their influence on breastfeeding rates to better understand programs, policies, and supports most relevant to recipients served by WIC and related breastfeeding programs.

Limitations

This dataset using longitudinal programmatic data provided an opportunity to assess the impact of smoking behavior over time on breastfeeding behavior in a real-world setting. WIC supports interventions to promote breastfeeding nationally, and implementation often depends on the individual WIC site, with some sites, including more individual, peer, or lactation counseling services than others.⁴⁰ Thus, we accounted for WIC site with multilevel modeling.

As with most data coming from large programs, inherent limitations exist. In the case of WIC, data are collected to certify WIC recipients, provide appropriate WIC benefits, and evaluate the impact of WIC on health and nutrition outcomes (i.e., not for research purposes). While these data are specific to a DC sample of largely non-Hispanic Black WIC recipients, understanding the relationship of smoking, breastfeeding, and race/ethnicity in the larger national context is critical to reduce racial disparities in critical health outcomes.

Recipients reported their behaviors to WIC staff members at each visit. Self-report bias may include the possibility of underreporting breastfeeding behavior to obtain formula vouchers or overreporting breastfeeding behavior to receive additional maternal food vouchers (DC WIC uses a system of vouchers to provide benefits to recipients based on reported breastfeeding). Social desirability bias is also possible and includes overreporting of breastfeeding behaviors to gain approval from WIC staff members or underreporting of socially undesirable behaviors like smoking. Information bias could occur from potential variation in how data were collected over time by different WIC staff and may also lead to differences in missing data.

For recipients who did not report a breastfeeding end date or were currently breastfeeding, breastfeeding duration could not be calculated, and these women were excluded from the duration analysis. Other potentially influential variables such as employment status of mothers and potential theoretical influences (i.e., provider influence or social norms) may exist but were not measured. Data on peer counseling received by individual WIC recipients were collected separately and not available to be linked with the deidentified data included. Further data on insurance information were also not collected, but most of the women eligible for WIC were also eligible for Medicaid services.

Conclusion

Consistent with the literature, we found a statistically significant negative association between smoking during pregnancy and postpartum and breastfeeding initiation and duration in WIC recipients, respectively, after adjusting for sociodemographic factors and delivery in a BFH, which was further exacerbated by increased CPD. However, additional analysis remains to better tease out the complex relationship between smoking and breastfeeding, as well as the directional and biological effects of nicotine on breastfeeding behaviors, in a longitudinal cohort. Promisingly, women who quit smoking during pregnancy had a 50% greater likelihood of initiating breastfeeding compared to those who continued smoking. Together these results underscore the importance of smoking cessation programs to improve breastfeeding rates among WIC recipients.

Footnotes

Authors' Contributions

Concept and design: J.S.-R., A.R., R.C.R. Acquisition, analysis, or interpretation of data: J.S.-R., A.R., R.R., D.K., E.W., S.V., P.T. Drafting of the article: J.S.-R., A.R., R.R. Critical revision of the article for important intellectual content: J.S.-R., A.R., R.R., D.K., E.W., S.V., P.T. Statistical analysis: J.S.-R., A.R. Obtained funding: J.S.-R., A.R.

Disclosure Statement

No competing financial interests exist.

Funding Information

Clara Schiffer Fellowship, Sumner M. Redstone Global Center for Prevention and Wellness Pilot Research Grant, George Washington University.

References

Eidelman

, Schanler

, Johnston

, et al. Breastfeeding and the use of human milk. Pediatrics, 2012; 129:e827–e836.

World Health Organization [WHO] (2016). Breastfeeding. Available at www.who.int/topics/breastfeeding/en/ (accessed April 1, 2021).

, Chung

, Raman

, et al. Breastfeeding and maternal and infant health outcomes in developed countries. Evid Rep Technol Assess (Full Rep), 2007; 153:1–186.

Patro-Golab

, Zalewski

, Kolodziej

, et al. Nutritional interventions or exposures in infants and children aged up to three years and their effects on subsequent risk of overweight, obesity, and body fat: A systematic review of systematic reviews. Obes Rev, 2016; 17:1245–1257.

Horta

, Loret de Mola

, Victora

. Long-term consequences of breastfeeding on cholesterol, obesity, blood pressure and type 2 diabetes: A systematic review and meta-analysis. Acta Paediatr, 2015; 104:30–37.

Tahir

, Haapala

, Foster

, et al. Association of full breastfeeding duration with postpartum weight retention in a cohort of predominantly breastfeeding women. Nutrients, 2019; 11:938.

Schwarz

, Brown

, Creasman

, et al. Lactation and maternal risk of type 2 diabetes: A population-based study. Am J Med 2010;123:863.e1–863.e6.

Schwarz

, Ray

, Stuebe

, et al. Duration of lactation and risk factors for maternal cardiovascular disease. Obstet Gynecol, 2009; 113:974–982.

Rameez

, Sadana

, Kaur

, et al. Association of maternal lactation with diabetes and hypertension. JAMA Netw Open, 2019; 2:e1913401.

10.

Collaborative Group on Hormonal Factors in Breast Cancer. Breast cancer and breastfeeding: Collaborative reanalysis of individual data from 47 epidemiological studies in 30 countries, including 50302 women with breast cancer and 96973 women without the disease. Lancet, 2002; 360:187–195.

11.

Jensen

. Participation in the supplemental nutrition program for women, infants and children (WIC) and breastfeeding: National, regional, and state level analyses. Matern Child Health J, 2012; 16:624–631.

12.

Centers for Disease Control and Prevention [CDC]. Breastfeeding Report Card: Progressing Toward National Breastfeeding Goals. Available at https://www.cdc.gov/breastfeeding/pdf/2016breastfeedingreportcard.pdf (accessed April 5, 2021).

13.

Ruowei

, Perrine

, Anstey

, et al. Breastfeeding trends by race/ethnicity among US children born from 2009 to 2015. JAMA Pediatr, 2019; 173:e193319.

14.

Stough

, Khalsa

, Nabors

, et al. Predictors of exclusive breastfeeding for 6 months in a national sample of US children. Am J Health Promot, 2019; 33:48–56.

15.

Association of State and Territorial Health Officials [ASTHO]. (2015). State Story: District of Columbia Department of Health Supports Peer Counseling to Improve Breastfeeding Rates. Available at:2015. Available at www.astho.org/MCH/BF/Peer-and-Professional-Support/DC/HYS/ (accessed April 1, 2021).

16.

Xiang

, Chow

, Mora-Marquez

, et al. Breastfeeding persistence at 6 months: Trends and disparities from 2008 to 2015. J Pediatr, 2019; 208:169–175.

17.

Jones

, Kogan

, Singh

, et al. Factors associated with exclusive breastfeeding in the United States. Pediatrics, 2011; 128:1117–1125.

18.

Thulier

, Mercer

. Variables associated with breastfeeding duration. J Obstet Gynecol Neonatal Nurs, 2009; 38:259–268.

19.

Hawkins

, Stern

, Baum

, et al. Evaluating the impact of the Baby-Friendly Hospital Initiative on breast-feeding rates: A multi-state analysis. Public Health Nutr, 2015; 18:189–197.

20.

Munn

, Newman

, Mueller

, et al. The impact in the United States of the Baby-Friendly Hospital Initiative on early infant health and breastfeeding outcomes. Breastfeed Med, 2016; 11:222–230.

21.

Liu

, Rosenberg

, Sandoval

. Breastfeeding duration and perinatal cigarette smoking in a population-based cohort. Am J Public Health, 2006; 96:309–314.

22.

Joseph

, Emery

, Bogen

, et al. The influence of smoking on breast feeding among women who quit smoking during pregnancy. Nicotine Tob Res, 2017; 19:652–655.

23.

Bailey

, Wright

. Breastfeeding initiation in rural sample: Predictive factors and the role of smoking. J Hum Lact, 2011; 27:33–40.

24.

Wallweiner

, Muller

, Doster

, et al. Predictors of impaired breastfeeding initiation and maintenance in a diverse sample: What is important?. Arch Gynecol Obstet, 2016; 294:455–466.

25.

Office of Disease Prevention and Health Promotion. Healthy People 2030: Pregnancy and Childbirth. Available at https://health.gov/healthypeople/objectives-and-data/browse-objectives/pregnancy-and-childbirth/increase-abstinence-cigarette-smoking-among-pregnant-women-mich-10 (accessed April 1, 2021).

26.

Drake

, Driscoll

, Matthews

. Cigarette smoking during pregnancy: United States, 2016. NCHS Data Brief, 2018; 305:1–7.

27.

Curtin

, Matthews

. Smoking prevalence and cessation before and during pregnancy: Data from the birth certificate. Natl Vital Stat Rep, 2016; 65:1–13.

28.

Whaley

, Koleilat

, Jiang

. WIC infant food package issuance data are a valid measure of infant feeding practices. J Hum Lact, 2012; 28:134–138.

29.

Arellano

. On the testing of correlated effects with panel data. Journal of Econom, 1993; 59:87–97.

30.

Wooldridge

JM.

Econometric Analysis of Cross Section and Panel Data. Cambridge, MA: MIT Press, 2002.

31.

Lee

. Using hierarchical linear modeling to study social contexts: The case of school effects. Educ Psychol, 2000; 35:125–141.

32.

StataCorp. Stata Statistical Software: Release 13. College Station, TX: StataCorp LP, 2013.

33.

Thorn

, Tadler

, Huret

, et al. WIC Participant and Program Characteristics 2014. Available at www.fns.usda.gov/sites/default/files/ops/WICPC2014.pdf (accessed April 1, 2021).

34.

Soneji

, Beltran-Sanchez

. Association of Special Supplemental Nutrition Program for Women, Infants, and Children with preterm birth and infant mortality. JAMA Netw Open, 2019; 2:e1916722.

35.

Long

, Young

, Tender

JAF

, et al. The DC breastfeeding-friendly hospital initiative: An evaluation of hospitals' support for breastfeeding in the capital of the United States. J Hum Lact, 2013; 29:465–468.

36.

Osypuk

, Kawachi

, Subramanian

, et al. Are state patterns of smoking different for different racial/ethnic groups? An application of multilevel analysis. Public Health Rep, 2006; 121:563–577.

37.

Gadomski

, Adams

, Tallman

, et al. Effectiveness of a combined prenatal and postpartum smoking cessation program. Matern Child Health J, 2011; 15:188–197.

38.

Olaiya

, Sharma

, Tong

, et al. Impact of the 5As brief counseling on smoking cessation among pregnant clients of Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) clinics in Ohio. Prev Med, 2015; 81:438–443.

39.

Kurti

, Tang

, Bolivar

, et al. Smartphone-based financial incentives to promotion smoking cessation during pregnancy: A pilot study. Prev Med, 2020; 140:106201.

40.

National WIC Association. Breastfeeding Peer Counselors: A Successful Program that Should be Expanded. Available at https://thewichub.org/breastfeeding-peer-counselors-a-successful-wic-program-that-should-be-expanded/ (accessed April 1, 2021).