Breastfeeding Practices in Patients with Heart Disease Stratified by Area Deprivation Index

Abstract

Objective:

We sought to evaluate breastfeeding (BF) practices in patients with maternal cardiac disease (MCD) stratified by area deprivation index (ADI) to identity communities at risk.

Study Design:

Retrospective cohort of patients managed by the University of Alabama at Birmingham (UAB) Cardio-Obstetrics Program. Patients were included if they had ≥1 prenatal visit with the Cardio-Obstetrics team, delivered at UAB, and had a street address on file. The primary outcome was BF rate at hospital discharge. Secondary outcomes included BF intent on admission and BF at the postpartum (PP) visit. ADI reports socioeconomic disadvantage at the census tract level; 1 = least deprived and 100 = most deprived. Baseline characteristics and BF rates were compared by ADI categories: Low (ADI 1–33), medium (ADI 34–66), and high (ADI 67–100).

Results:

One hundred and forty-eight patients were included: 14 (10%) low, 42 (28%) medium, and 92 (62%) high ADI. Patients in the high ADI category were younger relative to those in the medium or low ADI (26 versus 28 versus 32 years; p < 0.01) and less likely to be married or living with a partner (30.4% versus 58.5% versus 71.4%; p < 0.01), There was no difference in BF intent between the lowest, medium, and highest ADI categories (85.7% versus 85.4% versus 81.6%; p = 0.38) or BF rates at hospital discharge (100% versus 92.7% versus 85.6%, p = 0.23). However, there was a significant difference in BF rates at the PP visit (90% versus 63.0% versus 38.6%; p < 0.01) even after controlling for differences in baseline characteristics (odds ratio = 0.11 (95% confidence interval [0.01–0.93]), p = 0.043).

Conclusions:

There was an association between living in a resource-poor community and early cessation of BF in our population of patients with MCD. Community-based interventions targeting mothers with heart disease living in high ADI communities may help these individuals achieve higher BF rates.

Introduction

Breastfeeding (BF) provides not only neonatal benefit, reducing risks of asthma, Type 1 diabetes, ear infections, sudden infant death syndrome, and necrotizing enterocolitis,^1,2 but also positively affects maternal health. It has been shown to decrease risks of ovarian and breast cancer, obesity, Type 2 diabetes, hypertension, and hyperlipidemia in mothers who breastfeed.³ Importantly, the composite maternal cardiovascular benefit of BF has been well described in the literature.⁴ Given that cardiac disease in women is increasing in prevalence and severity in the United States⁵ and has short-term adverse effects on pregnancy outcomes and long-term adverse effects on the mother,⁶ it is important to examine potential barriers to BF in the specific population of women with cardiac disease.

Despite the benefits of BF, according to Centers for Disease Control and Prevention (CDC) census data from 2019 to 2022, only 1 in 4 infants is exclusively breastfed through six months of age in the United States. In Alabama specifically, the rate is even lower, closer to 1 in 5.⁷ When broken down by social and demographic factors, the census data demonstrate decreased BF rates in women of non-Hispanic Black race, lower maternal age, and unmarried status.

Similarly, 2019 National Vital Statistics System birth certificate data, representing a larger sample size, demonstrate racial/ethnic disparities in BF initiation at national and state levels.⁸ In yet another recent study analyzing the relationship between women's status and BF practices, exclusive BF at 6 months was positively associated with employment and earnings, among other factors, and negatively associated with poverty.⁹ Decreased BF rates in women from lower socioeconomic backgrounds, most often non-Hispanic Black women, compound the fact that these women are already subject to worse health outcomes,¹⁰ particularly related to cardiovascular disease.¹¹

While both the risk of maternal cardiac disease (MCD) in pregnancy and the social factors that serve as barriers to BF are separately established, little is known about the intersection between social determinants of health and BF in women with heart disease. At our institution in Alabama, disadvantaged mothers¹² and infants¹³ have higher rates of morbidity and death, and mothers with heart disease and their infants in particular are subject to worse outcomes.¹⁴ However, it is unknown if or how disadvantaged status, as defined by the Area Deprivation Index (ADI), is associated with BF rates. Therefore, our objective is to examine the association between disadvantaged status—defined by the ADI—and BF in women with heart disease, hypothesizing that a higher ADI is associated with lower rates of BF. We aim to identify at-risk communities to inform community-based interventions and resources to improve outcomes for mothers and their infants.

Materials and Methods

We conducted a retrospective cohort study of pregnant patients with cardiac disease from the University of Alabama at Birmingham (UAB). The UAB Institutional Review Board provided approval for the study, with informed consent waived, given the retrospective nature of the study. Patients were selected from a database of those treated by the Cardio-Obstetrics Program at UAB between March 1, 2015, and June 30, 2019.

The Cardio-Obstetrics Program provides prenatal care in a comprehensive, multidisciplinary manner to patients with both congenital and acquired heart disease. Patients in the program are typically seen by a Maternal-Fetal Medicine specialist and a Cardiology attending and all Cardio-Obstetric patients are discussed at monthly Cardio-Obstetric meetings. Inclusion criteria for the study included ≥1 prenatal visit with the Cardio-Obstetrics team, delivery at UAB, and a street address on file to assign a numerical ADI classification. Patients were excluded if they delivered at another institution or if the neonate died before discharge from the delivery-associated hospitalization.

We compared baseline demographics as recorded in the medical record, including maternal age, self-reported race, tobacco use, alcohol use, married or living with partner, employment status, insurance status, body mass index (BMI), chronic hypertension (CHTN), diabetes mellitus (DM), MCD, gestational hypertension (GHTN)/preeclampsia (per American College of Obstetrics and Gynecology diagnostic criteria¹⁵), parity, gestational age (GA) at delivery, mode of delivery, and neonatal intensive care unit admission. Severity of MCD in our cohort was designated using the modified World Health Organization (mWHO) classification and independently confirmed by an adult congenital cardiologist (M.G.C.), as previously reported.¹⁶ The mWHO classification describes the impact of specific cardiac lesions ranging from class I, in which patients have a low risk of maternal morbidity or mortality, to class IV, conditions that pose a high risk of maternal mortality or severe morbidity and in which pregnancy is contraindicated due to threat to the life of the mother.¹⁴

The primary outcome of the study was BF at discharge from the delivery-associated hospitalization. BF was defined as maternal expression of human milk and could be administered to the neonate through direct feeding or pumping. Secondary outcomes included self-reported BF intent at admission for the delivery-associated hospitalization and self-reported BF at the postpartum (PP) visit between 4 and 8 weeks PP. All outcomes were abstracted from the medical record as these are standing fields in the history and physical for the delivery-associated hospitalization, PP hospital rounding notes, as well as PP clinic notes.

Patients were stratified by National ADI, a measure of socioeconomic disadvantage derived from the census tract level with 1 being the least deprived and 100 being the most deprived.¹⁷ Each patient's ADI was determined using their home address, as listed in the medical record. ADI was divided into three evenly distributed categories; the lowest category was designated an ADI score of 1–33, the medium category an ADI score of 34–66, and the highest category an ADI score of 67–100. Baseline characteristics and BF rates were compared by ADI categories.

To compare differences between ADI categories, one-way analysis of variance (ANOVA), chi-square, and Fisher's exact tests were utilized as appropriate. A multivariable logistic regression model was conducted adjusting for differences in significant baseline characteristics. SAS version 9.4 (Cary, NC) was used with a significance level designated at p < 0.05 for all analyses.

Results

One hundred forty-eight patients managed by the Cardio-Obstetrics Program at UAB met eligibility criteria and were included in data collection. The cohort was heavily skewed toward those living in resource-poor communities: 14 (10%) patients were categorized as low ADI, 42 (28%) patients as medium ADI, and 92 (62%) patients as high ADI. Baseline characteristics between groups differed in several notable ways (Table 1). First, patients with the highest or medium categories of ADI were younger relative to the lower category (32 versus 28 versus 26 years from lowest to highest ADI; p < 0.01).

Table 1.

Baseline Maternal Characteristics of Patients Managed by the University of Alabama at Birmingham Cardio-Obstetrics Program

Characteristic	Entire cohort n = 148	Lowest ADI (0–33) n = 14	Medium ADI (ADI 34–66) n = 42	Highest ADI (ADI 67–100) n = 92	p
Maternal age (years)	27 ± 6	32 ± 4	28 ± 7	26 ± 7	<0.01
Parity	1.0 ± 1.3	1.4 ± 2.2	0.9 ± 1.1	1.0 ± 1.2	0.53
BMI	32.2 ± 8.5	29.0 ± 5.6	31.8 ± 7.9	32.8 ± 9.2	0.28
Race					0.11
NHB	58 (39.5)	3 (21.4)	12 (28.6)	43 (47.3)
NHW	81 (55.1)	10 (71.4)	28 (66.7)	43 (47.3)
None of the above	8 (5.4)	1 (7.1)	2 (4.8)	5 (5.5)
Married or living with partner	62 (42.2)	10 (71.4)	24 (58.5)	28 (30.4)	<0.01
Insurance status					0.049
Private	56 (37.8)	10 (71.4)	18 (42.9)	28 (30.4)
Public	85 (57.4)	4 (28.6)	23 (54.8)	58 (63.0)
Self-pay	7 (4.7)	0 (0.0)	1 (2.4)	6 (6.5)
GA at delivery	37.0 ± 3.3	37.8 ± 2.7	37.3 ± 2.6	36.7 ± 3.7	0.44
Mode of delivery					0.35
Vaginal	82 (55.4)	6 (42.9)	21 (50.0)	55 (59.8)
Cesarean	66 (44.6)	8 (57.1)	21 (50.0)	37 (40.2)
NICU admission	33 (22.5)	1 (7.1)	10 (24.4)	22 (23.9)	0.35
GHTN/preeclampsia	37 (25.0)	3 (21.4)	8 (19.1)	26 (28.3)	0.49
CHTN	22 (14.9)	3 (21.4)	3 (7.1)	16 (17.4)	0.23
DM	12 (8.1)	0 (0.0)	2 (4.8)	10 (10.9)	0.32
mWHO class
I	12 (8.1)	1 (7.1)	1 (2.4)	10 (10.9)	0.28
II	36 (24.3)	4 (28.6)	13 (31.0)	19 (20.7)	0.40
II–III	36 (24.3)	3 (21.4)	8 (19.1)	25 (27.2)	0.58
III	37 (25.0)	3 (21.4)	13 (31.0)	21 (22.8)	0.57
IV	27 (18.2)	3 (21.4)	7 (16.7)	17 (18.5)	0.92

Bolded text indicates statistical significance.

ADI, area deprivation index; BMI, body mass index; CHTN, chronic hypertension; DM, diabetes mellitus; GA, gestational age; GHTN, gestational hypertension; mWHO, modified World Health Organization; NHB, non-Hispanic black; NHW, non-Hispanic white; NICU, neonatal intensive care unit.

Second, they were less likely to be married or living with a partner (71.4% versus 58.5% versus 30.4%; p < 0.01) compared to those in the low ADI category. Finally, there were more patients with public insurance in the highest or medium ADI group (28.6% versus 54.8% versus 63.0%; p < 0.05). On initial analysis, self-described race was not different between groups; however, when comparing non-Hispanic white race to non-Hispanic black (NHB) race alone, patients in high and medium categories were more likely to describe as NHB race (21.4% versus 28.6% versus 47.3% from lowest to highest ADI; p < 0.04). Parity, BMI, GA at delivery, medical comorbidities, including GHTN/preeclampsia, CHTN, and DM, as well as MCD severity defined by mWHO classification were not different between groups.

BF rates at discharge from delivery-associated hospitalization were not different between the lowest, medium, and highest ADI categories (100% versus 92.7% versus 85.6%, p = 0.23). There was also no difference in BF intent at admission for delivery-associated hospitalization between groups (85.7% versus 85.4% versus 81.6%, p = 0.38). More specifically, exclusive BF intent was 71.4% versus 68.3% versus 53.5% from lowest to highest ADI categories, respectively, while both formula feeding and BF were 14.3% versus 17.1% versus 28.3% in the respective groups.

However, there was a significant difference in BF rates at the PP visit between 4–8 weeks PP between the lowest, medium, and highest ADI categories (90% versus 63.0% versus 38.6%; p < 0.01). Specifically, BF rates among patients from the high ADI group, representing the greatest socioeconomic disadvantage, dropped 47% from 86% at hospital discharge to 39% at the PP visit (Table 2 and Fig. 1). Attendance at the PP visit in the overall cohort was 64%, which was not different between groups.

Table 2.

Breastfeeding^a Rates in Patients Managed by the University of Alabama at Birmingham Cardio-Obstetrics Program

Outcomes	Entire cohort n = 148	Lowest ADI (0–33) n = 14	Medium ADI (34–66) n = 42	Highest ADI (67–100) n = 92	p
BF intent at admission for delivery					0.36
Formula feeding	14 (9.5)	0 (0.0)	3 (7.1)	11 (12.0)
Any BF	122 (82.4)	12 (85.7)	35 (83.3)	75 (81.5)
Both	35 (23.7)	2 (14.3)	7 (16.7)	26 (28.3)
Breastfeed	87 (58.8)	10 (71.4)	28 (66.7)	49 (53.3)
Unknown/undecided	12 (8.1)	2 (14.3)	4 (9.5)	6 (6.5)
BF rates at discharge from the delivery-associated hospitalizationb	129 (89.0)	14 (100.0)	38 (92.7)	77 (85.6)	0.23
BF rates at the PP visitb	48 (51.1)	9 (90.0)	17 (63.0)	22 (38.6)	<0.01

Bolded text indicates statistical significance.

^a Breastfeeding is defined as maternal expression of human milk administered to the neonate through direct feeding or pumping.

^b Denominator used in rate calculations based on available data; missing data: BF rate at discharge, n = 3; BF rate at PP visit, n = 54.

ADI, area deprivation index; BF, breastfeeding; PP, postpartum.

FIG. 1.

BF rates in patients managed by the UAB Cardio-Obstetrics program based on ADI. BF intention at admission for delivery-associated hospitalization, BF rate at discharge, and BF rates at the PP visit were abstracted from the medical record. Patients were grouped to compare low, medium, and high ADI using one-way ANOVA, chi-square, and Fisher's exact tests (significance level designated as p < 0.05). ADI, area deprivation index; ANOVA, analysis of variance; BF, breastfeeding; PP, postpartum; UAB, University of Alabama at Birmingham.

Results did not differ in the adjusted analysis (Table 3). A multivariable analysis was conducted controlling for age, marital status, and insurance. In the adjusted analysis, women from the highest tertile ADI had lower odds of BF at the PP visit (adjusted odds ratio 0.11 (95% confidence interval 0.01–0.93); p = 0.04).

Table 3.

Unadjusted and Adjusted Odd Ratio and Confidence Intervals for Prevalence of Self-Reported Breastfeeding at Postpartum (4–6 Weeks from Birth) (Breastfeeding at Postpartum—Yes/No)

	Unadjusted OR			Adjusted OR
	Variable	OR	CI	p	OR	CI	p
Area deprivation index
Lowest ADI (Ref)	—	—	—	—	—	—
Medium ADI	0.19	0.02–1.72	0.14	0.25	0.03–2.35	0.22
Highest ADI	0.07	0.01–0.59	0.015	0.11	0.01–0.93	0.043
Maternal age	1.06	0.99–1.13	0.11	1.02	0.95–1.10	0.56
Married/living with partner (Ref)	2.18	0.96–4.97	0.06	1.14	0.43–3.01	0.79
Insurance status
Private (Ref)	—	—	—	—	—	—
Public	0.30	0.13–0.70	0.005	0.44	0.17–1.16	0.10
Self-pay	0.48	0.03–8.30	0.62	0.69	0.03–14.18	0.81

Bolded text indicates statistical significance.

Model adjusted for age, marital status, and pay or source.

CI, confidence interval; OR, odds ratio.

Discussion

In a cohort of patients managed by a single-center comprehensive cardio-obstetric program in the Southeast United States, those residing in areas with higher socioeconomic disadvantage reported significantly lower BF rates at the PP visit, despite equal BF intent and rates of BF during the delivery-associated hospitalization. We previously found that severity of MCD did not affect BF rates in this same population.¹⁸

The CDC list the rate of BF (any BF, not exclusive BF) collected in the 2019 National Immunization Survey as 83.2%.⁷ In our study, intent to breastfeed in some capacity was documented at a rate almost identical to national rates. Initiation of BF by hospital discharge in our cohort was greater than average rates in our state of Alabama (69.6%).⁷ However, disparities were found in BF rate at the PP visit (4–8 weeks) between patients from the highest and lowest ADI categories.

Patients in the lowest ADI category in our cohort reported a BF rate at PP visit (4–8 weeks) of 74.6%, similar to the rates in the 2019 National Immunization Survey of 78.6% at 1 month and 73.6% at 2 months. However, this was not the case for patients in the medium and highest categories. BF rates among patients in our cohort from the highest ADI category (as reported at the 4–8-week PP visit) were almost half of the average national rate. Given there was no difference in mWHO classification between groups, we suspect that socioeconomic disadvantage, represented by ADI, is primarily responsible for difference in BF rate noted.

Clinic- and hospital-based BF interventions have not eliminated disparities in BF outcomes, and our group seeks to identify the highest-risk communities in which to conduct community-based interventions. Therefore, we initially present unadjusted results demonstrating that women in the highest ADI communities have the lowest BF rates. These results do not adjust for key differences in baseline characteristics between ADI groups, specifically, age, marital status, and payor source.

We did not originally adjust the model out of concern that these maternal characteristics are likely included in the causal pathway. It is documented that poorer, unmarried women funded by Medicaid are more likely to live in poor communities, and we wanted to avoid overadjustment.¹⁹ However, we also conducted an adjusted analysis to better illustrate the association with ADI and BF alone. The results remain robust after adjusting for patient-level characteristics. In this cohort, those living in areas with higher socioeconomic disadvantage do report lower rates of BF in the PP period.

Strengths of this study are that it represents a well-characterized cohort of patients and data derived from reliable databases. In addition, it provides insight into an increasingly prevalent subset of cardio-obstetric patients in which BF rates have not been previously characterized. To our knowledge, this is the first study to investigate the relationship between BF and socioeconomic determinants of health in a cardio-obstetrics population. However, we also acknowledge limitations of this study. As a single-center analysis, this study is limited by the sample size of the cardio-obstetric database at UAB.

The retrospective nature of the study may also limit the availability of comprehensive data, as it is possible that BF data were not consistently collected, and as many women did not return for follow-up at 6 weeks as recommended. Furthermore, the use of ADI is imperfect. ADI is a surrogate measure for socioeconomic status and may have contributed to some misclassification. Demographic data on household income, occupation of mothers, and education levels may have provided a more accurate estimate of socioeconomic status at individual levels. Further multicenter prospective observational studies are needed to corroborate the validity of data gathered.

While our study identifies lower BF rates in high-risk communities in the PP period, the primary question is what community-based interventions are available to improve BF rates among the most vulnerable. Racial and ethnic minorities, for example, who often suffer from disproportionate levels of socioeconomic disadvantage due to the egregious legacy of racism, may face unique barriers, citing lack of social, work, and cultural support, in addition to a host of other factors, as reasons for not BF.²⁰ Previous studies have demonstrated that peer counseling-based interventions such as behavioral education may be effective in increasing exclusive BF rates among these minority groups, specifically Black and Latina patients.^20,21

Of particular interest are community-based doula programs, which provide support for women during the perinatal and PP period and may also improve BF outcomes.^22,23 Our group has an ongoing project through the American Heart Association Health Equity Research Network to test whether community health workers can improve outcomes such as BF rates. Given the potential health implications of BF in our studied population of women with cardiac disease, and the noted BF disparity in BF at the PP visit, such interventions are imperative.

In summary, PP mothers with cardiac disease who lived in resource poor communities were less likely to breastfeed their newborns than PP mothers with cardiac disease who live in communities with more resources, as defined by ADI. We look forward to engaging with our community advisory board to identify community-based solutions to improve outcomes for these mothers and their infants.

Authors' Contributions

Conceptualization, I.C.C., A.O., and R.S.; formal analysis, C.T.B. and M.T. Investigation, I.C.C. and A.R. Writing—original draft preparation, I.C.C. and R.S.; writing—editing, I.C.C. and R.S.; administration, I.C.C. Supervision, R.S. All authors have read and agreed to the published version of the article.

Disclaimer

The content is solely the responsibility of the authors and does not necessarily represent the official views of any funding agency.

Disclosure Statement

No competing financial interests exist.

Funding Information

This research was not supported by any specific grant from a funding agency. R.S. is funded by NHLBI K23HL159331, as well as the American Heart Association Health Equity Research Network AHA22HERNPMI985239.