Geographic disparities in peripartum cardiomyopathy outcomes

Abstract

BACKGROUND:

Cardiomyopathy causes more than a third of late postpartum pregnancy–related deaths in the United States, and racial disparities in outcomes among pregnant individuals with cardiomyopathy exist. Underlying community factors may contribute to disparities in peripartum cardiomyopathy outcomes.

OBJECTIVE:

This study aimed to identify the geographic distribution of and disparities in peripartum cardiomyopathy outcomes, hypothesizing that patients living in communities with higher social vulnerability may have worse outcomes.

STUDY DESIGN:

This was a retrospective cohort study of patients with peripartum cardiomyopathy per the National Heart, Lung, and Blood Institute definition from January 2000 to November 2017 at a single center, excluding those with a post office box address as a post office box address may not reflect the census tract in which a patient resides. Severe peripartum cardiomyopathy (vs less severe peripartum cardiomyopathy) was defined as ejection fraction <30%, death, intensive care unit admission, left ventricular assist device or implantable cardioverter defibrillator placement, or transplant. The US census tract for the patient’s address was linked to the Centers for Disease Control and Prevention Social Vulnerability Index, a 0 to 1 scale of a community’s vulnerability to external stresses on health, with higher values indicating greater vulnerability. The Social Vulnerability Index includes social factors divided into socioeconomic, household composition, minority status, and housing type and transportation themes. The Social Vulnerability Index and Social Vulnerability Index components were compared among patients by peripartum cardiomyopathy severity.

RESULTS:

Of 95 patients in the original cohort, 5 were excluded because of the use of a post office box address. Of the remaining 90 patients, 56 met severe peripartum cardiomyopathy criteria. At baseline, individuals with and without severe peripartum cardiomyopathy had similar ages, marital status, payor type, tobacco use, gestational age at delivery, and mode of delivery; however, individuals with severe peripartum cardiomyopathy were more likely to be Black (vs White) (59% vs 29%; P<.007) and less likely to recover ejection fraction (EF) to ≥55% by 12 months (36% vs 62%; P=.02) than individuals with less severe peripartum cardiomyopathy. Patients with severe peripartum cardiomyopathy were more likely to live in areas with a higher Social Vulnerability Index (0.51 vs 0.31; P=.002) and with more residents who were unemployed, impoverished, without a high school diploma, in single-parent households, of minority status, without a vehicle, and in institutionalized group quarters than patients with less severe peripartum cardiomyopathy. The median income was lower in communities of individuals with severe peripartum cardiomyopathy than in communities of individuals with less severe peripartum cardiomyopathy.

CONCLUSION:

Patients with severe peripartum cardiomyopathy outcomes were more likely to live in communities with greater social vulnerability than patients with less severe peripartum cardiomyopathy outcomes. To reduce disparities and maternal mortality rates, resources may need to be directed to socially vulnerable communities.

Introduction

Although peripartum cardiomyopathy (PPCM) is rare and affects only 1 of 2230 births in the United States, the mortality estimates for PPCM range from 3% to as high as 40%.¹ PPCM is responsible for more than a third of late postpartum pregnancy–related deaths in the United States.² PPCM is defined by the US National Heart, Lung, and Blood Institute (NHLBI) as idiopathic heart failure in the last month of pregnancy or within 5 months of delivery in patients with no prior history of heart disease.³ Certain populations bear a disproportionate burden of PPCM. Racial disparities in PPCM outcomes are well established, with Black individuals suffering higher rates of morbidity and mortality related to cardiomyopathy.^4–8

It is unknown why some birthing individuals experience worse outcomes than others, even when care is received at a single site. The Centers for Disease Control and Prevention (CDC) recommend that maternal mortality review committees (MMRCs) be established to assess the preventability of maternal deaths and to make recommendations to prevent future pregnancy-related causes of death.⁹ MMRCs have identified that community factors may significantly contribute to maternal mortality, which led us to investigate which community factors may underly disparities in PPCM outcomes using the CDC Social Vulnerability Index (SVI).

The CDC SVI database uses US Census data to assess social vulnerability at the census tract level.¹⁰ SVI is divided into 4 themes: (1) socioeconomic, (2) household composition and disability, (3) minority status and language, and (4) housing type and transportation.¹⁰ Initially, the SVI was created for disaster planning and mitigation. However, recent research has used the CDC SVI to assess the relationship between social vulnerability and various health outcomes, including teen pregnancy rates,¹¹ surgical outcomes,^12–15 and COVID-19 incidence.^16,17

There is a gap in the literature regarding whether geographic—and resulting socioeconomic—disparities in PPCM exist and whether these geographic and community disparities can help us better understand well-documented racial disparities. Therefore, our objective was to identify the geographic distribution of PPCM cases and to identify geographic and community disparities in PPCM outcomes, hypothesizing that living in an area with high social vulnerability, defined by the CDC SVI, is associated with worse PPCM outcomes.

Materials and Methods

We performed a retrospective cohort study of patients with PPCM (per the National Institutes of Health NHLBI definition)³ from January 2000 to November 2017 at a single, tertiary healthcare center. Institutional review board approval was obtained from the University of Alabama at Birmingham (UAB). To identify patients, the electronic medical record (EMR) system was queried for female patients with International Classification of Diseases, Ninth Revision (ICD-9), and International Classification of Diseases, Tenth Revision (ICD-10), codes for PPCM, female patients ≥18 years of age being cared for by the Heart Failure Clinic or the Heart Failure Hospital Service at the UAB Hospital, or an obstetrical patient with an inpatient heart failure consultation. The date range was January 2000 to November 2017. This cohort was previously identified to examine racial disparities in health outcomes among people with PPCM.⁴ Among those identified, patients were included if they self-reported either Black or White race, reported a residential address, were managed by cardiology at the UAB Hospital, and met the NHLBI criteria for PPCM. Patients were excluded if they only listed a post office (P.O.) box address as US census tract data do not apply to P.O. box addresses.

Baseline demographic and delivery data were obtained, and key outcomes of interest were abstracted from the EMR, including left ventricular EF at the time of diagnosis, recovery of EF to ≥55% within a year of delivery, months from diagnosis to recovery, the EF of those patients at 6 to 12 months who had not recovered, and maternal death. Additional outcomes included the need for intensive care unit (ICU) admission, left ventricular assist device (LVAD), implantable cardioverter defibrillator (ICD), or heart transplant. Further details regarding the identification of patients with PPCM and data collection have been previously published.⁴ Severe PPCM (vs less severe PPCM) was defined as EF <30%,⁸ death, or need for ICU admission, LVAD, ICD, or heart transplant.

Social vulnerability and certain community factors for each patient’s home community were identified as follows. Patients’ residential addresses were identified in the EMR system using the address from the patient encounter in the closest time proximity to their first visit for PPCM. The US census tract for each address was identified and linked to the CDC SVI. The SVI quantifies a community’s vulnerability to external stresses on health, and SVI data are publicly available from the CDC.¹⁰ The SVI is reported on a scale of 0 to 1. The higher the SVI, the greater the community’s vulnerability. The SVI corresponds to a percentile ranking. For example, an SVI of 0.9 means that 90% of US census tracts are less vulnerable, and only 10% of US census tracts are more vulnerable. The SVI includes social factors divided into socioeconomic, household composition and disability, minority status and language, and housing type and transportation themes.

Baseline characteristics, SVI, and SVI subcomponents were compared among patients by PPCM severity. Differences between the groups were assessed using chi-square tests of association and Fisher exact tests for categorical data and the Student t test or Wilcoxon rank-sum tests for continuous variables, as appropriate. All analyses were performed using SAS (version 9.4; SAS Institute; Cary, NC) and evaluated at a .05 significance level.

Results

Of note, 95 patients with PPCM met the inclusion criteria in the original cohort.⁴ Of those patients, 90 had physical (non-P.O. box) addresses that could be linked to a US census tract. Of the 90 patients included, 56 met the severe PPCM criteria. At baseline, patients with and without severe PPCM had similar ages, marital status, payor type, tobacco use, gestational age at delivery, and mode of delivery. Patients with severe PPCM were more likely to be Black (vs White) (59% vs 29%; P<.007). Patients in the severe PPCM group were less likely to recover their cardiac EF to ≥55% by 12 months than patients in the less severe PPCM group (36% vs 62%; P=.02) (Table 1). The Figure 1 shows the geospatial distribution of patients (yellow indicates the highest density). The visual overlap apparent in the Figure is a result of using an intentionally large radius for each data point to ensure patient privacy. Most patients in the cohort lived in Alabama; patients with severe PPCM showed a larger geographic distribution, including neighboring states, than patients with less severe PPCM.

TABLE 1.

Demographic, delivery, and cardiomyopathy characteristics of patients with PPCM

Characteristics	More severe PPCM (n=56)	Less severe PPCM (n=34)	P value
Race			.007
Black	33 (59)	10 (29)
White	23 (41)	24 (71)
Age at PPCM diagnosis (y)	28.6±5.9	29.7±7.0	.46
Age≥35 y	12 (21)	8 (24)	.82
Marrieda	27 (49)	22 (65)	.15
Payor source			.25
Public	25 (45)	11 (32)
Private	31 (55)	23 (68)
Smokingb	16 (29)	9 (27)	.90
Delivered at UABc	6 (11)	6 (19)	.34
Gestational age at delivery (wk)d	37.0±3.8	38.3±1.8	.11
Mode of deliverye			.37
Vaginal	14 (32)	13 (42)
Cesarean	30 (68)	18 (58)
EF at diagnosis	21.7±6.8	38.8±8.5	<.0001
EF<35% at diagnosis	54 (96)	11 (32)	<.0001
Clinical outcomes
Recovered EF to ≥55% within 12 mo of diagnosis	20 (36)	21 (62)	.02
If recovered, months from diagnosis to recovery	6 (1–12)	6 (1 –6)	.47
Persistent LV dysfunction defined as:f
EF≤45% at 6–12 mo after delivery	16 (46)	3 (14)	.02
EF≤40% at 6–12 mo after delivery	11 (31)	1 (5)	.02
EF≤35% at 6–12 mo after delivery	10 (29)	1 (5)	.03
EF≤30% at 6–12 mo after delivery	3 (9)	0 (0)	.28
EF≤25% at 6–12 mo after delivery	1 (3)	0 (0)	1.00

Data are presented as number (percentage), mean±standard deviation, or median (interquartile range) unless otherwise indicated.

EF, ejection fraction; LV, left ventricular; PPCM, peripartum cardiomyopathy; UAB, University of Alabama at Birmingham.

^aMissing 1 more severe

^bMissing 1 less severe

^cMissing 1 more severe and 2 less severe

^dMissing 25 more severe and 13 less severe

^eMissing 12 more severe and 3 less severe

^fMissing 21 more severe and 13 less severe.

FIGURE 1. Geographic distribution of patients with PPCM by disease severity.

PPCM, peripartum cardiomyopathy.

Patients with severe PPCM were more likely to live in areas with higher social vulnerability indices (mean±standard deviation) (0.51±0.30 vs 0.31±0.28; P=.002) (Table 2). Patients with severe PPCM were more likely to score higher—and therefore be more vulnerable—than patients with less severe PPCM in the areas of socioeconomics, household composition and disability, and housing type and transportation (Table 2). There were statistically significant differences in social vulnerabilities of patients with severe PPCM, including more residents living in poverty, unemployed, without a high school diploma, in single-parent households, of minority status, without a vehicle, and in institutionalized group quarters (Table 2). The median income was approximately $6500 lower in communities of people with severe PPCM than in those with less severe PPCM ($21,261±$6807 vs $27,719±$10,876; P=.003) (Table 2).

TABLE 2.

SVI and select components in patients with PPCM

Variable	More severe PPCM (n=56)	Less severe PPCM (n=34)	P value
SVI	0.51±0.30	0.31±0.28	.002
Socioeconomic	0.49±0.28	0.32±0.28	.005
Household composition and disability	0.51±0.29	0.35±0.26	.008
Minority status and languagea	0.49±0.29	0.46±0.33	.72
Housing type and transportation	0.54±0.31	0.34±0.28	.003
Proportion of population living in poverty	0.18±0.11	0.11 ±0.09	.003
Proportion unemployed (age ≥16 y)	0.10±0.06	0.07±0.04	.03
Per capita income (US dollars)	21261±6807	27719±10876	.003
Proportion of population with no high school diploma	0.19±0.09	0.13±0.09	.003
Proportion of single-parent households with children under 18 y	0.14±0.05	0.11±0.06	.02
Proportion of population of minority race	0.39±0.27	0.25±0.22	.02
Proportion of persons (age ≥5 y) who speak English “less than well”	0.01±0.01	0.02±0.03	.14
Proportion of housing in structures with ≥10 units	0.05±0.08	0.06±0.09	.73
Proportion of mobile homes	0.12±0.13	0.13±0.14	.91
Proportion of households with more people than rooms	0.02±0.02	0.01±0.02	.19
Proportion of households with no vehicle available	0.08±0.09	0.03±0.03	.0005
Proportion of persons in institutionalized group quarters	0.02±0.04	0.004±0.01	<.0001

PPCM, peripartum cardiomyopathy; SVI, Social Vulnerability Index.

^aIncludes proportion of the population of minority status and proportion of persons ≥5 years who speak English “less than well.”

Discussion

Principal findings

PPCM is the leading cause of late post-partum pregnancy–related death.² We demonstrated that patients with severe PPCM were more likely to reside in areas with high social vulnerability than those with less severe PPCM. Specifically, patients with severe PPCM were more likely to live in areas with more residents in poverty, unemployed, without a high school diploma, in single-parent households, of minority status, without a vehicle, and in institutionalized group quarters than those with less severe PPCM.

Results

The novel use of the CDC SVI as it relates to understanding contributing factors to maternal morbidity and mortality should be highlighted. Few studies have used the CDC SVI to further understand pregnancy complications. Yee et al¹¹ investigated environmental and social influences on teen pregnancy rates and found that socioeconomic status, household composition, and minority and language status were positively correlated (P<.001) with teen births. In particular, an additional 11.5 per 1000 births were seen with each increase in the SVI quartile.¹¹

Several studies have examined the relationship between county-level vulnerability and postoperative outcomes using the CDC SVI. Diaz et al¹⁸ reported worse surgical outcomes after colectomy and coronary artery bypass graft among patients from high SVI counties. Hyer et al¹² found that high SVI was associated with poorer textbook outcomes, such as surgery complications, extended stay, readmission, and mortality. Furthermore, an increased SVI was associated with a greater risk of worse postoperative outcomes in patients with cancer.¹² In addition, high SVI was correlated with a greater likelihood of undergoing emergent as opposed to elective cholecystectomy and colorectal surgery.^13,14 These studies are similar to ours in that the CDC SVI was used to measure the uneven distribution of social vulnerability within the context of a single medical condition.¹⁷

We did not identify any articles investigating the role of SVI and PPCM outcomes, but we acknowledge existing literature highlighting the interplay between social determinants of health and PPCM outcomes. Specifically, Getz et al¹⁹ reported that increased composite neighborhood disadvantage index (NDI) (relative risk [RR], 1.29) and Black race (RR, 1.63) were each separately correlated with worse PPCM outcomes. In particular, of the 6 NDI components, low education defined as adults with less than a high school education had the strongest association with worse PPCM outcomes. Similarly, we found that patients with severe PPCM were more likely to be of Black race (P=.02) and without a high school diploma (P=.003). Overall, the emerging literature suggests that community factors influence disparities in PPCM outcomes.

Clinical implications

Maternal health outcomes are the result of a complex confluence of many clinical and nonclinical factors. Although the medical care patients receive in a clinical setting is of critical importance to outcomes, there are many influences on health outcomes that originate outside of the clinical setting, including individual, social, community, economic, and systemic factors. From a clinical perspective, outcomes may improve if healthcare teams can account for and address variations among patients in social determinants of health. From a policy perspective, investing resources into strengthening low-resource communities may help to reduce morbidity in patients with PPCM.

Research implications

Because this was a retrospective study, we have revealed an association, but our study design did not prove that living in a socially vulnerable community directly causes worse outcomes for people with PPCM. Further studies to assess causality are needed, and additional research is needed examining whether diverting resources to socially vulnerable communities can improve outcomes.

Strengths and limitations

There are limitations to consider when interpreting the results of this study. Because of a change in the medical record system during the period of our study, the first recorded encounter with an address associated was greater than 1 year apart from the date of the diagnosis for 18 patients. Therefore, it is possible that some of these patients lived in a different community at the time of PPCM diagnosis. Although patients with less severe PPCM may receive care locally, patients with severe PPCM are more likely to be referred to a higher level of care; therefore, we acknowledge that our cohort may be influenced by selection bias. However, in contrast, there may be an inherent selection bias against patients without adequate means of transportation. Furthermore, because of the retrospective nature of the study, some data are missing as detailed in the table; longer-term outcomes, such as persistence of left ventricular dysfunction, are not available for all patients. Any changes in the definition or clinical management of PPCM over this 18-year period could not be accounted for in this analysis, and the use of alternate definitions of PPCM may yield different results. Lastly, we acknowledge that geography may serve as a proxy for race or environmental factors that may lead to poorer underlying health in this cohort; however, this was unavoidable given the methodology of this study.

This study has several strengths. This was a well-characterized, previously published cohort⁴ examined through the lens of the SVI, a publicly available, CDC-endorsed tool. Furthermore, elucidating geographic disparities in PPCM outcomes can inform and guide public health interventions to promote health equity and reduce rates of maternal morbidity and mortality. We believe that this study will incite further investigation of risk factors associated with maternal morbidity and mortality in other communities within the United States and help uncover other potential health inequities in this arena.

Conclusions

Maternal mortality review committees have identified that community factors may significantly contribute to maternal mortality.⁹ These data further support that geographic and social factors are associated with worse PPCM outcomes. Causal relationships remain unknown, but it is imperative that as key stakeholders develop policies to reduce maternal morbidity and mortality, investing resources into and strengthening infrastructure in low-resource communities may help to reduce morbidity in patients with PPCM.

AJOG MFM at a Glance.

Why was this study conducted?

There are known racial disparities in perinatal morbidity, including peripartum cardiomyopathy (PPCM) outcomes. To eliminate disparities, we need to better understand what underlies the disparate outcomes so that interventions to eliminate disparities can be appropriately targeted.

Key findings

Patients with severe PPCM were more likely to live in areas with a higher Social Vulnerability Index. The median income was lower in communities of patients with severe PPCM than in communities of patients with less severe PPCM.

What does this add to what is known?

Community factors may be inextricably linked to patient health outcomes. To reduce disparities and maternal mortality, resources may need to be directed to socially vulnerable communities.