Impact of Neighborhood Factors on Heart Failure Outcomes

Abstract

Background

Few studies have examined the impact of neighborhood-level factors on outcomes for patients with heart failure with reduced ejection fraction (HFrEF).

Objectives

The purpose of this study was to understand the impact of neighborhood factors on readmission and mortality risk hospitalized patients with HFrEF.

Methods

We analyzed data from the Hopeful Heart Trial that evaluated the impact of blended collaborative care for treating HFrEF and depression among patients discharged from 8 Pittsburgh-area hospitals from March 2014 to October 2017. Using patients' home address at discharge to determine neighborhood Walk Score (WS; 0-100 scale) and Area Deprivation Index (ADI; 0-100), we examined the incidence of 12-month all-cause and cardiovascular-related hospital readmissions and vital status up to 5 years postdischarge through June 2022 using multivariable-adjusted Cox proportional hazards models.

Results

Hopeful Heart enrolled 756 people with HFrEF (baseline mean age 64.0, 44% female, 73% White race, 28% ± 9.1% mean left ventricular ejection fraction, mean 9-Item Patient Health Questionnaire score 12 ± 5.7, median WS 69 (IQR: 49-88), and median ADI 12 (IQR: 10-15) and followed them for a median of 57.7 months (IQR: 25.0-68.4). Individuals from the least walkable neighborhoods experienced greater 12-month all-cause mortality (HR: 1.70 [95% CI: 1.11-2.61]; P = 0.016), while those from the most deprived neighborhoods had higher 12-month cardiovascular-related hospital readmission (HR: 1.39 [95% CI: 1.09-1.78]; P = 0.008). Neither WS nor ADI predicted 12-month all-cause readmission and cardiovascular-related mortality or 5-year all-cause mortality.

Conclusions

Among recently hospitalized HFrEF patients, neighborhood factors affect 12-month rehospitalization and mortality risk but not 5-year mortality. (Blended Collaborative Care for Heart Failure and Co-Morbid Depression; NCT02044211)

Central Illustration

Social determinants of health (SDOH)—the environmental and personal factors that impact health¹—influence risk among people living with heart failure (HF) and are implicated in the association between depression and poorer clinical outcomes. Indeed, depression is associated with an increased risk for hospitalization,² readmission,³ and all-cause mortality among people with HF independent of the severity of the underlying cardiovascular disorder.4, 5, 6 Poverty and living in an under-resourced neighborhood have been associated with an increased risk of depression among HF patients;⁷ yet the impact of the lived environment on patients with HF and comorbid depression remains unclear.

We conducted a secondary analysis of data from the National Institutes of Health-funded Hopeful Heart Trial to understand the relationship of neighborhood deprivation and walkability on hospital readmission rates and mortality among HF patients with comorbid depression. Hopeful Heart studied the impact of a nurse-delivered collaborative care strategy for treating depression among patients with heart failure with reduced ejection fraction (HFrEF).⁸ We earlier reported that our intervention improved mental health-related quality of life and mood symptoms among those receiving the collaborative care strategy compared to those randomized to their doctors' usual care for HFrEF and depression, although participants across both arms had similar rates of 12-month mortality and readmission after adjustment for baseline treatment assignment and depression status.⁸

Hypothesis and purpose

We now hypothesize that neighborhood deprivation and walkability is predictive of later risk of hospital readmission and death independent of randomized treatment assignment and baseline depression status. This study will help to examine the impact of neighborhood factors on outcomes for patients with HFrEF with and without depression.

Methods

Patient cohort and study protocol

We published the Hopeful Heart Trial's methods⁹ and primary outcomes⁸ previously and included our Institutional Review Board-approved protocol in Supplemental Appendix 1. Briefly, between March 2014 and October 2017, Hopeful Heart enrolled 756 HF patients with left ventricular ejection fraction (LVEF) of ≤45% from 8 Pittsburgh-area hospitals who were screened for depression with the 2-item Patient Health Questionnaire (PHQ)¹⁰ at discharge and with the 9-item PHQ¹¹ 2 weeks after discharge. The trial included 629 depressed HFrEF patients who screened PHQ-2 positive at discharge and scored ≥10 on the PHQ-9 later who were then randomized to either: 1) their doctors' “usual care” for HFrEF and depression; 2) “collaborative care” for HFrEF alone; or 3) “blended care” for both conditions. The trial also included 127 nondepressed control patients (PHQ-2 negative and 2-week PHQ-9 score ≤5).

Baseline measures

Prior to hospital discharge, we collected participants' sociodemographic characteristics including age, self-reported race and gender, marital status, education level, home address, and subjective HF symptoms on the NYHA HF class scale¹² and abstracted information on comorbid hypertension, diabetes, tobacco use, systolic and diastolic blood pressure prior to hospital discharge, and LVEF from a detailed chart review.

Following confirmation of protocol eligibility at 2 weeks with the PHQ-9,¹¹ study assessors then administered the Kansas City Cardiomyopathy Questionnaire (KCCQ-12) assessing disease-specific health-related quality of life¹³ and the 12-item Short Form Health Survey (MCS-12 and PCS-12) assessing generic mental and physical health-related quality of life.¹⁴

Area Deprivation Index

Area Deprivation Index (ADI) is a validated measure of neighborhood disadvantage that integrates 17 weighted indicators of poverty, education, housing, and employment at the Census tract level.^15,16 To determine the ADI score for each patient, we entered their home address at the time of trial enrollment into the Neighborhood Atlas mapping function (https://www.neighborhoodatlas.medicine.wisc.edu/).^15,16 We then grouped study participants into ADI quartiles: very advantaged (national ADI percentile 1st-25th), advantaged (26th-50th), disadvantaged (51st-75th), and very disadvantaged (76th-100th).

Walk Score

The Walk Score¹⁷ incorporates validated^18,19 and publicly available data to assign a score of residence-specific neighborhood walkability based on distance to grocery stores, parks, and public transportation; pedestrian friendliness; and other characteristics.²⁰ The score is normalized on a 0 to 100 scale where 0 reflects the least walkable and most car-dependent location and 100 the most walkable and least car-dependent location. To obtain each patient's Walk Score, we sent their deidentified home address at the time of study entry to Walk Score Data Services (https://www.walkscore.com/). Afterward, we grouped study participants into Walk Score quartiles: least walkable or car-dependent (national Walk Score percentile 1-25), car-dependent (26th-50th), somewhat-to-very walkable (51st-75th), and most walkable (76th-100th).

Outcome Measures

Research assessors blinded to participants' randomization assignments and baseline depression status telephoned participants at 3, 6, and 12 months following randomization and inquired about any hospitalizations since the previous assessment. We adjudicated the cause of readmission to 12 months and all-cause mortality to 5 years or June 30, 2022 (whichever was sooner) through electronic health record review. A flow chart of vital status determination and checklist are included in Supplemental Appendix 2 and 3.

Hospitalization and death records were then reviewed independently by 2 physicians and classified as a cardiovascular or noncardiovascular cause. If reviewers disagreed as to the cause of 12-month hospitalization or death, a third physician reviewed the record to adjudicate the cause of the event.

Statistical analysis

We tested for statistically significant differences in baseline measures using Pearson's chi-square test and for correlation between ADI and Walk Score using the Spearman's rank correlation coefficient. PHQ-9, MCS-12, PCS-12, and KCCQ were tested for skewness and kurtosis tests for normality (Supplemental Appendix 4). Variables used in modeling included depression status (depressed or nondepressed control), randomization arm (usual care, enhanced usual care, or blended care), and hospital type (university, community, community underserved). We then created Cox proportional hazards regression with main effects for ADI, Walk Score, depression status, sex, randomized treatment assignment, hospital type, and their interaction to obtain HRs and P values for 12-month all-cause and cardiovascular-related hospital readmission and mortality and 5-year all-cause mortality. We used Schoenfeld residuals to evaluate the proportional hazards assumption after fitting a Cox proportional hazards model, with no evidence of violation observed in any of the models. No interaction terms (Walk Score, ADI, sex, randomized treatment assignment or depression status, and hospital type) were significant (Supplemental Appendix 5 and 6) and thus were not included in the model. We compared Q4 vs Q1-Q3 for ADI and Q1 vs Q2-Q4 for Walk Score by randomization arm (eg, blended care, enhanced usual care, and usual care) and depression status (depressed vs nondepressed controls). The level of significance was set at an alpha value of 0.05. Statistical analysis was performed using StataNow/SE: Release 18.5. (StataCorp LLC).

Results

Patient cohort

Of the 756 Hopeful Heart study patients, 83% (629/756) screened positive for depression at baseline and 2 weeks following hospitalization; 44% (331/756) were female; 25% identified as Black (189/756); and 73% identified as White (551/756) (Table 1). The mean age of the overall population was 64.0 ± 13.0 years, while the median follow-up time was 52.7 months (IQR: 25.0-68.4). Neither rates of comorbid conditions, NYHA functional class, nor mean LVEF differed by Walk Score or ADI quartile (Table 1). As reported in our primary outcome paper,⁸ there was no difference in goal-directed HF pharmacotherapy (heart failure specific beta-blockers; angiotensin converting enzyme inhibitors (ACE-I)/ angiotensin II receptor blockers (ARB); mineralocorticoid receptor antagonists) at randomization or 12-month follow-up, while participants in the blended care and enhanced usual care arms had similar rates of HF pharmacotherapy use and dose adjustment at 12 months. Regarding depression treatments, the blended care and enhanced care intervention arms in the primary study did not show significant differences in antidepressant (selective serotonin reuptake inhibitor or serotonin-norepinephrine reuptake inhibitor) use or dose adjustment compared with usual care at the time of randomization or 12-month follow-up.

Table 1.

Baseline Characteristics by Walk Score and Area Deprivation Index for All Participants

	Walk Score: All Participants (N = 756)	Least Walkable Areas (Q1) (n = 173)	More Walkable Areas (Q2-Q4) (n = 583)	P Valuea	ADI: All Participants (N = 750)	Most Deprived Areas (Q4) (n = 191)	Less Deprived Areas (Q1-Q3) (n = 559)	P Valuea
Sex
Male	425 (56.2%)	109 (63.0%)	316 (54.2%)	0.04b	422 (56.3%)	95 (49.7%)	327 (58.5%)	0.04b
Female	331 (43.8%)	64 (37.0%)	267 (45.8%)		328 (43.7%)	96 (50.3%)	232 (41.5%)
Race
White	551 (72.9%)	163 (94.2%)	388 (66.6%)	<0.001b	546 (72.8%)	69 (36.1%)	477 (85.3%)	<0.001b
Black	189 (25.0%)	8 (4.6%)	181 (31.0%)		188 (25.1%)	116 (60.7%)	72 (12.9%)
Other/refused	16 (2.1%)	2 (1.2%)	14 (2.4%)		16 (2.1%)	6 (3.1%)	10 (1.8%)
Age
≤65 y	411 (54.4%)	84 (48.6%)	327 (56.1%)	0.08	409 (54.5%)	130 (68.1%)	279 (49.9%)	<0.001b
>65 y	345 (45.6%)	89 (51.4%)	256 (43.9%)		341 (45.5%)	61 (31.9%)	280 (50.1%)
NYHA functional class
II	286 (37.8%)	61 (35.3%)	225 (38.6%)	0.29	286 (38.1%)	72 (37.7%)	214 (38.3%)	0.43
III	392 (51.9%)	98 (56.6%)	294 (50.4%)		388 (51.7%)	95 (49.7%)	293 (52.4%)
IV	78 (10.3%)	14 (8.1%)	64 (11.0%)		76 (10.1%)	24 (12.6%)	52 (9.3%)
Education
≤High school	355 (47.0%)	79 (45.7%)	276 (47.3%)	0.70	351 (46.8%)	92 (48.2%)	259 (46.3%)	0.66
>High school	401 (53.0%)	94 (54.3%)	307 (52.7%)		399 (53.2%)	99 (51.8%)	300 (53.7%)
Married
Single	172 (22.8%)	28 (16.2%)	144 (24.7%)	<0.001b	171 (22.8%)	76 (39.8%)	95 (17.0%)	<0.001b
Married	318 (42.1%)	103 (59.5%)	215 (36.9%)		314 (41.9%)	38 (19.9%)	276 (49.4%)
Other	266 (35.2%)	42 (24.3%)	224 (38.4%)		265 (35.3%)	77 (40.3%)	188 (33.6%)
PHQ-9	12.0 ± 5.7	11.6 ± 5.8	12.1 ± 5.6	0.33	12.0 ± 5.7	12.0 ± 5.9	12.0 ± 5.6	0.95
SF-12 PCS	30.9 ± 10.5	30.6 ± 10.3	30.9 ± 10.6	0.72	30.9 ± 10.5	30.9 ± 10.6	30.8 ± 10.5	0.93
SF-12 MCS	43.5 ± 12.7	44.3 ± 13.3	43.3 ± 12.6	0.34	43.5 ± 12.8	43.2 ± 13.2	43.6 ± 12.7	0.69
KCCQ-12	46.6 ± 23.8	48.3 ± 24.1	46.1 ± 23.8	0.30	46.7 ± 23.9	45.2 ± 24.2	47.2 ± 23.8	0.32
Hypertension	648 (85.8%)	147 (85.0%)	501 (86.1%)	0.71	643 (85.8%)	169 (88.5%)	474 (84.9%)	0.23
SBP	133.9 ± 88.5	134.9 ± 110.9	133.6 ± 81.7	0.88	134.0 ± 88.7	128.5 ± 20.6	136.0 ± 103.0	0.36
DBP	81.5 ± 92.8	84.4 ± 116.9	80.8 ± 85.4	0.70	81.6 ± 93.0	74.1 ± 17.2	84.3 ± 108.2	0.23
Diabetes	388 (51.4%)	81 (46.8%)	307 (52.7%)	0.17	385 (51.4%)	95 (49.7%)	290 (52.0%)	0.60
Ejection fraction	28.3 ± 9.1	28.7 ± 8.7	28.1 ± 9.3	0.51	28.3 ± 9.1	27.9 ± 9.9	28.4 ± 8.8	0.51
Tobacco use
No	252 (33.4%)	59 (34.1%)	193 (33.2%)	0.81	249 (33.2%)	64 (33.5%)	185 (33.2%)	0.003b
Quit >1 y	294 (38.9%)	71 (41.0%)	223 (38.3%)		292 (39.0%)	62 (32.5%)	230 (41.2%)
Quit <1 y	115 (15.2%)	23 (13.3%)	92 (15.8%)		114 (15.2%)	27 (14.1%)	87 (15.6%)
Current	94 (12.5%)	20 (11.6%)	74 (12.7%)		94 (12.6%)	38 (19.9%)	56 (10.0%)

Values are n (%) or mean ± SD.

ADI = Area Deprivation Index; KCCQ = Kansas City Cardiomyopathy Questionnaire; MCS-12 = Mental Component of 12-Item Short Form Survey; PHQ = Patient Health Questionnaire; SBP = systolic blood pressure; SF-12 = Physical Component of 12-item Short Form Survey.

^aPearson's chi-square test or t-test.

^bSignificant result with P value < 0.05.

Neighborhood measures

We obtained Walk Scores for all 756 subjects. The median score was 569 (IQR: 49-88), although the distribution of the scores was highly positively skewed (Figure 1). Quartile 1 for Walk Score (least walkable) included 173 individuals with an average Walk Score of 1.9 ± 2.3, while the aggregate of Quartiles 2 to 4 (more walkable) included 583 people whose average Walk Score was 42.8 ± 22.1. We also obtained ADI scores on 750 of 756 study patients (99%), which had a skewed distribution (Figure 1). The median ADI was 12 (IQR: 10-15). ADI Quartile 4 (most deprived) included 191 individuals with an average ADI of 94.1 ± 3.5, while the aggregate of Quartiles 1 to 3 (less deprived) included 559 individuals whose average ADI was 57.4 ± 19.4. The Spearman correlation coefficient of 0.417 (95% CI: 0.358-0.476) reflected a moderate correlation between Walk Score and ADI.²¹

Figure 1.

Walk Score and Area Deprivation Index Distribution for Study Cohort

The distribution of Walk Scores among participants in the Hopeful Heart trial was positively skewed, while the distribution of Area Deprivation Index scores was negatively skewed. ADI = Area Deprivation Index.

Baseline differences by ADI and Walk Score quartiles

People living in the least walkable areas were more likely to be male, White, and married compared with those from more walkable areas (Table 1). Participants from the most deprived areas were more likely to be female, Black, not married, and <65 years of age compared to those from less deprived areas (Table 1). There were no significant differences in PHQ-9 scores, SF-12 PCS and MCS scores, KCCQ-12 scores, comorbid hypertension or diabetes, LVEF, or tobacco use based on area deprivation or walkability.

Twelve-month hospital readmission

At 12-month follow-up, 433 participants (57%) were readmitted (including 302 for cardiovascular-related issues) among the 756 patients with Walk Scores and 429 (57%) participants were readmitted (including 299 for cardiovascular-related reasons) among the 750 with ADI scores (Table 1). As seen in Table 2, rates of all-cause readmission and cardiovascular-related readmission at 12 months differed significantly among those living in the most deprived areas compared to those living in less deprived locations (Central Illustration). However, living in the most deprived and least walkable areas was not associated with an altered risk of all-cause readmission after adjustment for sex, hospital type, and study arm or depression status (Table 3), whereas living in the most deprived areas, compared to living in less deprived areas, was associated with a higher risk of 12-month cardiovascular-related readmission even after adjustment for sex, hospital type, and study arm (HR: 1.39; 95% CI: 1.09-1.78), as seen in Figure 2 and Table 3. Still, 12-month all-cause and cardiovascular-related readmission rate were similar by ADI and Walk Score quartile.

Table 2.

Descriptive Statistics for Mortality and Readmission Outcomes by Walk Score and Area Deprivation Index

		Walk Score				ADI
		All Participants With Walk Score (N = 756)	Least Walkable (n = 173)	More Walkable (n = 583)	P Value	All Participants With ADI (N = 750)	Less Deprived (n = 559)	Most Deprived (n = 191)	P Value
12-mo all-cause death	Died	98 (13%) [11%-16%]	32 (19%) [13%-25%]	66 (11%) [9%-14%]	0.014a	95 (13%) [10%-15%]	75 (13%) [11%-17%]	20 (10%) [7%-16%]	0.29
12-mo CV-related death	Died	79 (10%) [8%-13%]	24 (14%) [9%-20%]	55 (9%) [7%-12%]	0.09	77 (10%) [8%-13%]	61 (11%) [8%-14%]	16 (8%) [5%-13%]	0.32
12-mo all-cause readmission	Readmitted	433 (57%) [54%-61%]	92 (53%) [45%-61%]	341 (58%) [54%-63%]	0.21	429 (57%) [54%-61%]	307 (55%) [51%-59%]	122 (64%) [57%-71%]	0.031a
12-mo CV-related readmission	Readmitted	302 (40%) [36%-44%]	65 (38%) [30%-45%]	237 (41%) [37%-45%]	0.47	299 (40%) [36%-43%]	203 (36%) [32%-40%]	96 (50%) [43%-58%]	0.001a
60-mo all-cause death	Died	312 (41%) [38%-45%]	76 (44%) [36%-52%]	236 (40%) [36%-45%]	0.42	308 (41%) [38%-45%]	234 (42%) [38%-46%]	74 (39%) [32%-46%]	0.45

Values are n (%) [95% CI].

CV = cardiovascular; other abbreviation as in Table 1.

^aSignificant result with P value < 0.05.

Central Illustration.

Impact of Neighborhood Factors on Heart Failure Outcomes

CV = cardiovascular; HFrEF = heart failure with reduced ejection fraction.

Table 3.

Cox Proportional Hazard Models for ADI and Walk Score for 12-Month All-Cause and Cardiovascular-Related Readmission, 12-Month Mortality, and 5-Year All-Cause Mortality

	Walk Score Adjusted for Sex, Hospital Type, and Randomized Treatment Assignment	Walk Score Adjusted for Sex, Hospital Type, and Depression Status	ADI Adjusted for Sex, Hospital Type, and Randomized Treatment Assignment	ADI Adjusted for Sex, Hospital Type, and Depression Status
12-mo all-cause readmission
Least walkable (ref. = more walkable)	0.95 [0.75-1.20] (0.65)	0.94 [0.75-1.19] (0.62)
Most deprived (ref. = less deprived)			1.17 [0.94-1.45] (0.15)	1.18 [0.95-1.46] (0.13)
12-mo CV-related readmission
Least walkable (ref. = more walkable)	0.98 [0.74-1.29] (0.87)	0.97 [0.74-1.29] (0.85)
Most deprived (ref. = less deprived)			1.39 [1.09-1.78] (0.008)	1.40 [1.09-1.78] (0.008)a
12-mo all-cause mortality
Least walkable (ref. = more walkable)	1.70 [1.11-2.61] (0.016)	1.70 [1.11-2.61] (0.015)a
Most deprived (ref. = less deprived)			0.78 [0.47-1.28] (0.32)	0.78 [0.47-1.28] (0.32)
12-mo CV-related mortality
Least walkable (ref. = more walkable)	1.53 [0.94-2.49] (0.09)	1.54 [0.94-2.50] (0.08)
Most deprived (ref. = less deprived)			0.75 [0.43-1.31] (0.31)	0.75 [0.43-1.31] (0.31)
5-y all-cause mortality
Least walkable (ref. = more walkable)	1.11 [0.85-1.44] (0.43)	1.12 [0.86-1.46] (0.39)
Most deprived (ref. = less deprived)			0.90 [0.69-1.17] (0.43)	0.89 [0.68-1.16] (0.39)

Values are HR [95% CI] (P value).

CV = cardiovascular; other abbreviation as in Table 1.

^aSignificant result with P < 0.05.

Figure 2.

12-Month All-Cause and Cardiovascular-Related Readmission and 5-Year All-Cause Mortality

At 12 months following hospitalization with HFrEF, patients who lived in more deprived areas had a significantly higher risk of cardiovascular related-readmission compared with those who lived in less deprived areas. CV = cardiovascular.

12-month mortality

At 12-month follow-up, we identified 98 deaths (13% all-cause mortality including 79 deaths from cardiovascular causes) among our 756 HFrEF patients including 95 deaths (13% all-cause mortality including 77 deaths from cardiovascular causes) among the 750 that we obtained ADI scores from (as seen in Table 2). Twelve-month all-cause mortality differed significantly among those from the least walkable areas compared with those from more walkable areas.

Living in the least walkable areas was associated with a significantly higher risk of 12-month all-cause mortality when adjusted for sex, hospital type, and depression status (HR: 1.70; 95% CI: 1.11-2.61) (Table 3). Twelve-month all-cause mortality risk (including cardiovascular-related) was similar among participants from neighborhoods with greater and lesser area deprivation (Table 3). Sex, hospital setting, and randomization status were not associated with a significantly higher risk of 12-month mortality.

5-year all-cause mortality

At 5-year follow-up, we confirmed the vital status of 721 of 756 study participants (95% of the sample) and identified 312 deaths (41%: 312/721 [95% CI: 38%-45%]). Of those, we identified 308 deaths among the 750 (41%) (95% CI: 38%-45%) HFrEF patients who had ADI scores. Following adjustment for sex, hospital type, and depression status and including the 35 subjects for whom we were unable to determine vital status as “alive” at 5-year follow-up, we found no significantly higher risk of 5-year all-cause mortality among participants from the most deprived or least walkable areas (Table 3, Figure 2).

Discussion

In this secondary analysis of people with HFrEF and depression who enrolled in a trial to treat depression following hospital discharge, those who lived in the most deprived areas experienced a higher risk of cardiovascular-related hospital readmissions and those who lived in the least walkable areas experienced a higher risk of all-cause mortality at 12-month follow-up. There was no difference in 5-year all-cause mortality by ADI or Walk Score quartile. Our results concur with contemporary evidence relating higher ADI with readmission but not mortality.²²

The built environment of neighborhoods is an important SDOH related to cardiovascular disease risk and outcomes. Greater area deprivation has been associated with increased risk of hospitalization,²³ coronary artery disease,²⁴ HF,²⁵ and overall mortality.²⁶ People with HF living in more deprived areas have been shown to have a higher risk of HF readmission, all-cause mortality, and symptom burden.^27,28 While our analyses extend this evidence, it is important to acknowledge that depression itself has been associated with area deprivation.²⁹ For example, decreased walkability has been associated with increased cardiovascular risk³⁰ and poorer control of cardiometabolic conditions,³¹ while greater walkability is thought to be protective against depression.³² ADI and Walk Score have a moderate correlation in our study cohort, which bolsters the connection between area deprivation, walkability, socioeconomic status, and clinical outcomes.²¹ Deprivation has been associated with lower walkability³³ and being less likely to have a nearby supermarket³⁴ or exercise facility³⁵—thereby impacting how people can act on guideline-based recommendations to exercise, eat healthier foods, or participate in cardiac rehab.^27,36

Our study has important implications for redesigning care for people with HFrEF and depression, since physical activity is an evidence-based treatment for both conditions.^37,38 Physical activity can reduce depression, lower HF symptoms, and improve quality of life for people with HF and depression; therefore, treatment plans should consider neighborhood factors that impact physical activity.^39,40 While we could not find studies directly comparing ADI with Walk Score in people with HF and depression, our results mirror recent retrospective cohort analyses showing that limited vehicle access, increased disability, and less access to healthier food were strongly associated with hospital readmission among HF patients.^41,42 ADI and Walk Score have been tied to other clinical outcomes⁴³ and could be considered in how we support people with HFrEF and depression to engage in physical activity and other evidence-based practices.

Neighborhood deprivation and walkability are proxies for societal forces such as institutionalized racism.⁴⁴ Indeed, study patients from the most deprived areas were more likely to identify as Black, which may reflect the impact of historical redlining and racism.⁴⁵ Black HF patients experience significant disparities in outcomes and care including higher short-term readmission and mortality^46,47and lower likelihood of admission to a cardiology specialty service⁴⁸ or hospital compared with their White counterparts.⁴⁹ We should aim to find better ways to partner with minoritized communities who experience a disproportionate impact of neighborhood factors like area deprivation to cocreate interventions that address unmet needs and reduce disparities.50, 51, 52

Our findings have implications for the ways that providers, health care systems, and policymakers could consider neighborhood factors when caring for people with HF and depression. Interdisciplinary teams could use remote monitoring technology, telehealth, community-based interventions, and informatics tools⁵³ to identify and aid people from more deprived and less walkable neighborhoods. Policymakers and insurers should evaluate the methods that they use to identify underserved areas and be certain that these methods are not excluding communities in need. Policymakers should also invest strategically in infrastructure like supermarkets, community gardens, parks, and exercise facilities in the least walkable and most deprived areas to empower people to live their healthiest lives.

Strengths and limitations

Our study has several strengths including a large cohort with documented HFrEF; near 100% vital status determination; use of objective, validated, and widely used ADI and Walk Scores;15, 16, 17, 18, 19, 20 detailed chart review to assign 12-month causes of readmission and death; and adjustment for baseline depression status and randomization assignment.

Still, our study has important limitations. First, as an observational cohort, causality cannot be determined. Although we reported no differences in the baseline incidence of diabetes and hypertension in Table 1, we did not control for individual SDOH or specific comorbidities common to people with HFrEF⁵⁴ in our modeling. Given the low event rates for some outcomes, we were limited in our ability to adjust for additional confounding factors without risking overfitting of the multivariate models. Second, participants' ADI and Walk Score, which were assigned based on the participant's home address at the time of trial enrollment, did not change even if they later moved. Third, restricting to hospitals in the Pittsburgh region may also have impacted the generalizability of the results to other parts of the country. Fourth, the 64-year mean age of our study cohort is lower than the 72-year mean age for patients admitted to the hospital for HF in 2017,⁵⁵ which may have contributed to the final interpretation of the impact of Walk Score and ADI on participant outcomes. Finally, with patient enrollment occurring from 2014 to 2017, these analyses do not account for more recent goal-directed medical therapy regimens shown to improve hospitalization and mortality among people with HFrEF.⁵⁶ We also did not report guideline-directed procedural interventions such as defibrillator implantation or prior coronary artery bypass graft surgery as these were previously reported in Table 1 of our primary results paper.⁸

ADI does not assess for underlying determinants including community support, structural racism, crime, and violence⁵⁷ and Walk Score does not consider the size of destinations, visit frequency, or such aspects of walkability as recreational walking.^18,58

Conclusions

Our analyses suggest that neighborhood factors may influence clinical outcomes for people living with HF and other complex chronic medical conditions. They have important implications for how health care providers, health systems, and policymakers might aim to improve such neighborhood factors as walkability and area deprivation to address barriers to healthier living for all people.

Perspectives.

COMPETENCY IN SYSTEMS-BASED PRACTICE: Among HFrEF patients, those from more deprived areas had a higher risk of cardiovascular-related hospital readmissions at 12 months following hospital discharge, while those from less walkable areas had a higher risk of all-cause mortality at 12 months follow-up but not 5 years after hospitalization. Providers should consider neighborhood factors when crafting treatment plans and recommendations.

TRANSLATIONAL OUTLOOK: Health systems, insurers, and policymakers could also incorporate neighborhood factors into their population health and risk stratification strategies.

Funding support and author disclosures

Dr Bober is supported through the National Institutes of Health-National Center for Advancing Translational Sciences CTSI 5TL1TR001858-09 training grant and has presented this work at the AHA EPI|Lifestyle Scientific Sessions in Boston, MA, in March 2023 as well as the Society of General Internal Medicine Conference in Aurora, CO, in May 2023. Dr Johnson has received research support from the National Heart, Lung, and Blood Institute (K23HL165110) and has received honoraria from Sanofi and Edwards Lifesciences. Dr Magnani has received National Health, Lung, and Blood Institute grant funding K24HL160527 as a source of funding for the project. Dr Rollman has received National Heart, Lung, and Blood Institute grant funding RO1Hl114016 as a source of funding for the project. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.

Supplementary data

For an expanded Methods section, please see the online version of this paper.