Geographic Variation in Trends and Disparities in Heart Failure Mortality in the United States, 1999 to 2017

Abstract

Background

Cardiovascular disease mortality related to heart failure (HF) is rising in the United States. It is unknown whether trends in HF mortality are consistent across geographic areas and are associated with state‐level variation in cardiovascular health (CVH). The goal of the present study was to assess regional and state‐level trends in cardiovascular disease mortality related to HF and their association with variation in state‐level CVH.

Methods and Results

Age‐adjusted mortality rates (AAMR) per 100 000 attributable to HF were ascertained using the Centers for Disease Control and Prevention's Wide‐Ranging Online Data for Epidemiologic Research from 1999 to 2017. CVH at the state‐level was quantified using the Behavioral Risk Factor Surveillance System. Linear regression was used to assess temporal trends in HF AAMR were examined by census region and state and to examine the association between state‐level CVH and HF AAMR. AAMR attributable to HF declined from 1999 to 2011 and increased between 2011 and 2017 across all census regions. Annual increases after 2011 were greatest in the Midwest (β=1.14 [95% CI, 0.75, 1.53]) and South (β=0.96 [0.66, 1.26]). States in the South and Midwest consistently had the highest HF AAMR in all time periods, with Mississippi having the highest AAMR (109.6 [104.5, 114.6] in 2017). Within race‒sex groups, consistent geographic patterns were observed. The variability in HF AAMR was associated with state‐level CVH (P<0.001).

Conclusions

Wide geographic variation exists in HF mortality, with the highest rates and greatest recent increases observed in the South and Midwest. Higher levels of poor CVH in these states suggest the potential for interventions to promote CVH and reduce the burden of HF.

Nonstandard Abbreviations and Acronyms

AAMR

age‐adjusted mortality rate

BRFSS

Behavior Risk Factor Surveillance System

CVH

cardiovascular health

Clinical Perspective

What Is New?

• While increases in age‐adjusted mortality rates for cardiovascular deaths related to heart failure have been observed in all census regions since 2011, increases are greatest in the Midwest and Southern United States.

• Large disparities between US states in cardiovascular health are associated with age‐adjusted mortality rates for cardiovascular deaths related to heart failure.

What Are the Clinical Implications?

• Differences in the burden of heart failure mortality are largely attributable to modifiable risk exposures and emphasize the need and potential for interventions to target cardiovascular health to minimize the burden of heart failure mortality.

Over the past several decades, advances in the management of cardiovascular disease (CVD) have led to substantial declines in CVD mortality in the United States. However, recent data have shown a significant slowing in this trend since 2011. ¹ , ² Among heart disease subtypes, ischemic heart disease mortality has continued to decline, ³ while heart failure (HF) mortality has experienced a significant reversal with increases in mortality related to HF since 2011. ⁴ Some of this increase may be driven by the rapid aging of the US population. ⁵ While HF mortality rates are increasing nationally, there is significant regional variation in HF prevalence, ⁶ HF hospitalization rates, ⁷ , ⁸ and outcomes after HF hospitalization. ⁷ , ⁹ It is therefore essential to understand how the burden of HF mortality is borne at regional and state levels, as well as the underpinnings of any observed variation. Prior studies that have looked at HF mortality rates by state have found that underlying risk factors such as obesity, diabetes mellitus, and hypertension are significantly associated with HF mortality rate. ¹⁰

Cardiovascular health (CVH) incorporates both biological risk factors (total cholesterol, blood pressure, body mass index, and fasting plasma glucose) as well as behavioral risk factors (smoking, physical activity, and diet) into one comprehensive measure of CVH. ¹¹ Prevalence of poor CVH increased nationally from 2003 to 2011, preceding the recent rise in HF mortality. ¹² CVH also varies significantly by state, with higher rates of poor CVH clustered in Southern states. ¹³ , ¹⁴ These factors suggest that geographic variation in the distribution of HF mortality may be attributable to underlying geographic variation in CVH.

The present study seeks to (1) define geographic differences in contemporary trends in cardiovascular mortality related to HF (abbreviated throughout as HF mortality) and (2) examine the relationship between HF mortality and underlying risk factors, as measured by the American Heart Association's CVH score.

Methods

Study Population and Data

We undertook a serial cross‐sectional analysis of data from the 4 US census regions (Northeast, South, Midwest, West) as well as all 50 states and Washington DC using annual data from 1999 to 2017. The states and census regions included in this analysis were the Northeast (CT, MA, ME, NH, NJ, NY, PA, RI, VT), the Midwest (IA, IL, IN, KS, MI, MN, MO, NE, ND, OH, SD, WI), the South (AL, AR, DC, DE, FL, GA, KY, LA, MD, MS, NC, OK, SC, TN, TX, VA, WV), and the West (AZ, CA, CO, ID, MT, NM, NV, OR, UT, WA, WY). Within regions, age‐adjusted mortality rates (AAMRs) were quantified for each race‒sex group. Data were not available to calculate HF mortality among Black men in ID, ME, MT, ND, NE, NH, NM, RI, SD, UT, VT, WV, and WY and among Black women in ID, ME, MT, ND, NE, NH, NM, OR, RI, SD, UT, VT, and WY because of the small Black populations in these states.

Race‐sex specific AAMRs for cardiovascular deaths with any mention of HF were calculated for states and census regions using the Centers for Disease Control and Prevention's Wide‐Ranging Online Data for Epidemiologic Research (CDC WONDER), standardized to the 2000 US population. ¹⁵ We used the US Behavior Risk Factor Surveillance System (BRFSS) to calculate state‐level CVH. ¹⁶ All data used in the study are de‐identified and released publicly by the Centers for Disease Control and Prevention for researchers and therefore this study did not require review by the Institutional Review Board at Northwestern University.

Outcome Ascertainment

HF AAMR were ascertained from 1999 to 2017 among US Black and White adults aged 35 to 84 years using the multiple cause of death files from CDC WONDER, which includes the underlying and contributing cause of death from all death certificates in the United States. ¹⁵ Because HF is considered an intermediate cause of, or mode of, death, the cause of death coding instructions from the International Classification of Disease suggest that other plausible heart conditions should be listed as the underlying cause of death instead of HF. In a study of death certificate data from the ARIC (Atherosclerotic Risk in Communities) Study, HF was >3.3 times likely to be listed as a multiple cause of death than the underlying cause of death. ¹⁷ Thus, measuring HF mortality by including any cardiovascular death in which HF is listed as a contributing cause helps to capture the broad burden of HF‐related death without including non‐CVD deaths that list HF where it less likely to be contributing (eg, neoplasm). Specifically, for the primary analysis, cardiovascular deaths related to HF were identified among those with CVD (I00–I78) listed as underlying cause of death and HF (I50) listed as contributing cause. This includes those who died with an underlying cause of death of coronary heart disease, myocardial infarction, and stroke, among other causes of CVD. We also examined 2 additional definitions whereby HF was listed as the underlying cause of death as well as all deaths with any mention of HF (as underlying or contributing cause) in sensitivity analyses.

Assessment of Cardiovascular Health Exposure

CVH was estimated at the state‐level using data from BRFSS ¹⁶ according to American Heart Association definitions and standards. ¹¹ BRFSS is a telephone‐based self‐reported health surveillance system that collects sociodemographic data and tracks health status and behaviors in the United States. We used questions from the core component of the BRFSS on hypertension, high cholesterol, diabetes mellitus, body mass index, tobacco use, physical activity, consumption of fruits and vegetables, as well as demographic information including age, sex, and race/ethnicity. Data from the core component are available from every state. However, questions for several factors are not asked every year, therefore obtaining complete data to estimate CVH are only available in odd years (eg, 2015, 2017). Participants who reported a history of coronary heart disease, myocardial infarction, or stroke were excluded as tracking CVH at the population‐level is intended for use in a primary prevention sample.

Ideal CVH for each metric was assessed included the following: responses of “no” when asked if a doctor has told a participant that he or she has high blood pressure, high cholesterol, or diabetes mellitus; reporting a body mass index of 18.5 to 25.0 kg/m²; reporting <100 lifetime cigarettes smoked, or 100 lifetime cigarettes smoked but are not currently smoking; reporting ≥150 minutes a week of moderate‐intensity activity, or ≥75 minutes of vigorous‐intensity activity, or an equivalent combination of aerobic physical activity; and ≥5 daily servings of fruits or vegetables (Table 1). Though the American Heart Association's healthy diet score consists of multiple more components than fruits and vegetables (intake of whole grains, sodium, sugar‐sweetened beverages, and fish), fruits and vegetable intake were used as a proxy, as has been done previously. ¹³ , ¹⁸ CVH is considered “ideal” when an individual met criteria for “ideal” for 7 factors, and is considered “poor” for 2 or fewer factors as has been done previously. ¹³

Table 1.

Quantification of State‐Level American Heart Association Definition of Cardiovascular Health Using the Behavioral Risk Factor Surveillance System

Measure	BRFSS Question/Variable	Definition for Ideal Cardiovascular Health
BMI	About how much do you weigh without shoes? About how tall are you without shoes?	BMI (kg/m2)=18.5–24.9
Diabetes mellitus	Have you ever been told by a doctor that you have diabetes?	Answered “no”
Cholesterol	Those who have been cholesterol screened—have you ever been told by a doctor, nurse, or other health professional that your blood cholesterol is high?	Answered “no”
Hypertension	Have you ever been told by a doctor, nurse, or other health professional that you have high blood pressure?	Answered “no”
Dietary Pattern	Not counting juice, how often do you eat fruit? How often do you eat a green leafy or lettuce salad, with or without other vegetables? During the past month, how many times did you eat dark green vegetables? How often do you eat potatoes, not including French fries, fried potatoes, or potato chips? How many times did you eat orange‐colored vegetables such as sweet potatoes, pumpkin, winter squash, or carrots? How many times did you eat other vegetables?	Consumed 5 or more servings of fruits and vegetables per day
Physical Activity	Respondents who reported doing 150+ min (or vigorous equivalent) of physical activity	150+ min (or vigorous equivalent min) per week of physical activity.
Smoking Status	Have you smoked at least 100 cigarettes in your entire life? Do you now smoke cigarettes every day, some days, or not at all?	Had not smoked at least 100 cigarettes in their lifetime; or reported smoking 100 cigarettes in their lifetime, but not currently smoking

BMI indicates body mass index; and BRFSS, Behavioral Risk Factor Surveillance System.

Statistical Analysis

We performed Joinpoint trend analysis to identify inflection points in overall AAMR trends and linear regression to quantify annual rates of change in AAMR. We performed these analyses for the overall population and stratified by region, sex, and race/ethnicity subgroups. Separately for 2011 and 2017, linear regression was used to quantify the relationship between state CVH and HF mortality, with a state's percentage of residents with poor CVH as the independent variable and HF AAMR as the dependent variable. All analyses were performed in SAS version 9.4 (SAS Institute, Cary, NC) and Joinpoint version 4.7.0.0. ¹⁹ , ²⁰

Role of the Funding Source

The funding sponsor did not contribute to design and conduct of the study, collection, management, analysis, or interpretation of the data or preparation, review, or approval of the article. The authors take responsibility for decision to submit the article for publication. Dr. Khan had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Results

Regional Differences in Cardiovascular Mortality Related to HF, 1999 to 2017

The South and Midwest regions had higher HF AAMRs than the Northeast or West across the study period (Table 2). AAMR for HF mortality experienced a significant inflection point in 2011, generally declining before and increasing after 2011 across all 4 regions (Tables 2 and 3, Figure 1). Annual increases in AAMR per 100 000 after 2011 were greatest in the Midwest (β=1.14 [95% CI, 0.75, 1.53]), indicating an increase of 1.14 deaths per 100 000 per year. In the South, annual AAMR increase was 0.96 per 100 000 per year (0.66, 1.26) followed by the West (0.72 [0.05, 1.39]) and Northeast (0.35 [0.03, 0.68]).

Table 2.

Total Number of Cardiovascular Deaths Related to Heart Failure and Heart Failure Age‐Adjusted Mortality Rate by US Census Region From 1999 to 2017 Among Black and White Adults Age 35 to 84 Years

Y	Northeast		Midwest		South		West
	No. of Deaths	AAMR (95% CI)	No. of Deaths	AAMR (95% CI)	No. of Deaths	AAMR (95% CI)	No. of Deaths	AAMR (95% CI)
1999	20 052	71.2 (70.3‒72.2)	26 181	82.3 (81.3‒83.3)	38 955	82.1 (81.3‒82.9)	18 714	75.8 (74.7‒76.9)
2000	20 229	71.4 (70.4‒72.4)	25 727	80.6 (79.6‒81.6)	39 305	82.1 (81.3‒82.9)	18 177	72.7 (71.6‒73.8)
2001	19 589	68.6 (67.7‒69.6)	25 164	78.1 (77.1‒79.1)	39 223	80.4 (79.6‒81.2)	18 350	71.9 (70.9‒73.0)
2002	18 785	65.4 (64.4‒66.3)	24 297	74.8 (73.8‒75.7)	38 397	77.5 (76.7‒78.2)	18 489	71.2 (70.2‒72.3)
2003	18 146	62.9 (62.0‒63.8)	23 939	73.1 (72.1‒74.0)	38 281	75.9 (75.2‒76.7)	18 471	70.1 (69.1‒71.1)
2004	17 621	60.9 (60.0‒61.8)	22 994	69.7 (68.8‒70.6)	37 501	73.3 (72.5‒74.0)	17 693	66.2 (65.3‒67.2)
2005	17 129	59.3 (58.4‒60.2)	22 714	68.5 (67.6‒69.4)	37 956	72.9 (72.2‒73.6)	17 928	66.2 (65.2‒67.2)
2006	16 028	55.7 (54.9‒56.6)	21 464	64.4 (63.5‒65.3)	36 443	68.9 (68.2‒69.6)	17 199	62.8 (61.8‒63.7)
2007	15 104	52.5 (51.7‒53.3)	19 890	59.4 (58.6‒60.3)	35 215	65.6 (64.9‒66.3)	16 345	59.2 (58.2‒60.1)
2008	14 522	50.5 (49.7‒51.3)	19 991	59.3 (58.5‒60.1)	34 074	62.4 (61.7‒63.0)	16 142	57.6 (56.7‒58.5)
2009	13 883	48.5 (47.7‒49.3)	19 231	56.9 (56.1‒57.7)	33 558	60.5 (59.8‒61.1)	15 354	54.0 (53.2‒54.9)
2010	13 633	47.5 (46.7‒48.3)	18 951	55.9 (55.1‒56.7)	33 750	60.0 (59.4‒60.7)	14 918	51.9 (51.1‒52.8)
2011	13 780	48.0 (47.2‒48.8)	19 068	55.7 (54.9‒56.5)	32 953	57.3 (56.6‒57.9)	15 510	52.7 (51.8‒53.5)
2012	13 577	46.8 (46.0‒47.6)	19 202	55.3 (54.5‒56.1)	34 189	57.9 (57.3‒58.5)	15 266	50.3 (49.5‒51.1)
2013	14 023	48.0 (47.2‒48.8)	20 064	56.9 (56.1‒57.7)	36 447	60.1 (59.5‒60.7)	15 810	50.7 (49.9‒51.6)
2014	14 200	48.1 (47.3‒48.9)	20 716	57.8 (57.0‒58.6)	38 174	61.3 (60.6‒61.9)	16 368	50.8 (50.0‒51.6)
2015	14 729	49.3 (48.5‒50.1)	21 790	60.0 (59.2‒60.8)	40 935	63.9 (63.3‒64.5)	18 109	54.7 (53.9‒55.5)
2016	14 851	49.0 (48.2‒49.8)	22 454	60.6 (59.8‒61.4)	41 987	63.6 (63.0‒64.2)	19 190	56.1 (55.3‒56.9)
2017	15 278	48.9 (48.1‒49.7)	23 600	62.0 (61.2‒62.8)	43 793	64.4 (63.8‒65.0)	19 648	55.7 (55.0‒56.5)

AAMR indicates age‐adjusted mortality rate.

Table 3.

Heart Failure Age‐Adjusted Mortality Rate by Region Among Black and White Men and Women Age 35 to 84 Years Between 1999 and 2017

Region	Total Deaths, n	AAMR (95% CI)			Slope β (95% CI)
		1999	2011	2017	1999–2011	2011–2017
Northeast	305 159
Black men		84.2 (77.9‒90.5)	61.4 (56.7‒66.1)	69.4 (64.9‒73.9)	−2.35 (−2.95 to −1.74)	1.29 (0.69‒1.90)
Black women		62.7 (58.5‒66.9)	42.5 (39.3‒45.6)	43.1 (40.3‒46.0)	−2.09 (−2.60 to −1.59)	0.51 (−0.23 to 1.25)
White men		90.0 (88.2‒91.8)	59.9 (58.5‒61.4)	62.2 (60.8‒63.6)	−2.79 (−3.08 to −2.51)	0.51 (0.23 to 0.80)
White women		57.1 (55.9‒58.3)	37.8 (36.8‒38.8)	36.5 (35.5‒37.4)	−1.94 (−2.16 to −1.72)	−0.05 (−0.36 to 0.26)
Midwest	417 437
Black men		108.2 (101.2‒115.2)	91.4 (85.7‒97.2)	106.0 (100.3‒111.7)	−1.79 (−2.36 to −1.22)	2.45 (1.03‒3.87)
Black women		88.0 (82.9‒93.2)	64.0 (60.0‒67.9)	74.0 (70.0‒78.0)	−2.38 (−2.76 to −2.01)	1.76 (0.21‒3.31)
White men		102.0 (100.2‒103.8)	66.2 (64.9‒67.6)	75.1 (73.7‒76.4)	−3.14 (−3.47 to −2.82)	1.58 (1.20‒1.96)
White women		65.5 (64.3‒66.7)	43.6 (42.7‒44.6)	46.4 (45.4‒47.3)	−2.10 (−2.32 to −1.87)	0.59 (0.28‒0.91)
South	711 136
Black men		119.6 (115.2‒124.1)	88.6 (85.4‒91.9)	110.3 (107.2‒113.5)	−2.53 (−3.08 to −1.98)	3.79 (2.61‒4.96)
Black women		89.3 (86.3‒92.3)	62.0 (59.8‒64.2)	72.4 (70.2‒74.6)	−2.62 (−3.01 to −2.22)	1.85 (1.43‒2.27)
White men		97.3 (95.9‒98.8)	67.0 (65.9‒68.1)	75.2 (74.2‒76.3)	−2.73 (−2.96 to −2.51)	1.51 (1.16‒1.85)
White women		64.4 (63.4‒ 65.4)	43.3 (42.5‒44.0)	45.5 (44.8‒46.3)	−1.94 (−2.14 to −1.74)	0.45 (0.06‒0.85)
West	327 681
Black men		120.7 (110.0‒131.4)	83.1 (75.8‒90.3)	106.4 (99.1‒113.7)	−2.93 (−4.02 to −1.84)	3.65 (2.41‒4.88)
Black women		94.4 (86.3‒102.5)	63.8 (58.1‒69.6)	62.5 (57.5‒67.5)	−2.26 (−2.99 to −1.53)	0.48 (−1.26 to 2.21)
White men		92.0 (90.2‒93.9)	64.5 (63.1‒65.9)	69.1 (67.8‒70.4)	−2.61 (−2.88 to −2.33)	1.30 (0.23‒2.38)
White women		60.3 (59.0‒61.6)	40.5 (39.4‒41.5)	41.0 (40.0‒41.9)	−1.74 (−1.93 to −1.55)	0.31 (−0.44 to 1.07)

AAMR indicates age‐adjusted mortality rate; and β, change in deaths per 100 000 per year.

Figure 1. Geographic variation in regional and state‐level age‐adjusted cardiovascular mortality rates related to heart failure in 2011 and 2017.

States are color‐coded according to their age‐adjusted mortality rate (per 100 000). States represented in deeper red have higher age‐adjusted mortality rates. Numbers in the map correspond to census region: 1 (Northeast), 2 (Midwest), 3 (South), 4 (West). AAMR indicates age‐adjusted mortality rate.

Geographic patterns were consistent for each race‒sex group (Figure 2). Specifically, Black men and women had consistently higher AAMRs and steeper increases in AAMR than their White peers across all census regions. White women consistently had the lowest HF AAMRs across regions and White women in the Northeast were the only group to experience a negative rate of change (−0.05 [−0.36, 0.26]) between 2011 and 2017. In sensitivity analyses whereby HF was identified as either the underlying cause or any mention in all causes of death, similar regional patterns and race‒sex differences were observed (Table S1).

Figure 2. Geographic variation in regional and state‐level age‐adjusted cardiovascular mortality related to heart failure by race‐sex group in 2017.

States are color‐coded according to their corresponding race‐sex age‐adjusted mortality rate (age‐adjusted mortality rate, per 100 000). States represented in deeper red have higher age‐adjusted mortality rates. Numbers in the map correspond to census region: 1 (Northeast), 2 (Midwest), 3 (South), 4 (West). In several Mountain West and upper Great Plains states, there was insufficient data to calculate age‐adjusted mortality rates for Black men and women attributable to small Black populations in those states. AAMR indicates age‐adjusted mortality rate.

State‐Level Differences in Cardiovascular Mortality Related to HF, 1999 to 2017

In 1999, 2011, and 2017, the states in the highest quintile of AAMRs came exclusively from the South and Midwest census regions (Table 4). Four states, all from the South region, consistently ranked among the 5 highest AAMRs in 1999, 2011, and 2017: Arkansas (5th, 4th, 3rd), Alabama (4th, 3rd, 4th), Oklahoma (3rd, 2nd, 5th), and Mississippi (1st, 1st, 1st). Only 3 states consistently ranked among the 10 lowest AAMRs during these years: Arizona (48th, 43rd, 44th), Connecticut (46th, 45th, 49th), and Florida (51st, 50th, 48th). A minority of states experienced a decrease in AAMR both between 1999 to 2011 and 2011 to 2017: Alaska, Mississippi, Nebraska, New Jersey, New York, North Dakota, Vermont, and West Virginia. All other states saw decreases between 1999 to 2011 and increases between 2011 to 2017. The ratio of the state with the highest AAMR to the state with lowest AAMR went from 2.5 in 1999 (Mississippi [133.8], Florida [54.6]), to 3.4 in 2011 (Mississippi [112.7], Hawaii [32.8]), to 2.8 in 2017 (Mississippi [109.6], Alaska [38.6]).

Table 4.

Total Number of Cardiovascular Deaths Related to Heart Failure and Heart Failure Age‐Adjusted Mortality Rate by US State in 1999, 2011, and 2017

State	1999		2011		2017
	No. of Deaths	AAMR (95% CI)	No. of Deaths	AAMR (95% CI)	No. of Deaths	AAMR (95% CI)
Alabama	2360	106.6 (102.3‒110.9)	2044	79.5 (76.0‒82.9)	2044	81.9 (78.6‒85.3)
Alaska	72	62.2 (48.1‒79.1)	89	49.9 (39.5‒62.2)	89	38.6 (30.9‒47.6)
Arizona	1466	58.9 (55.8‒61.9)	1394	42.5 (40.2‒44.7)	1394	45.4 (43.3‒47.5)
Arkansas	1457	102.6 (97.3‒107.8)	1219	75.9 (71.6‒80.2)	1219	82.8 (78.5‒87.1)
California	10 379	82.4 (80.8‒84.0)	7782	55.3 (54.0‒56.5)	7782	56.7 (55.6‒57.9)
Colorado	1118	70.1 (66.0‒74.2)	1066	49.3 (46.2‒52.3)	1066	51.7 (48.9‒54.5)
Connecticut	1112	60.5 (56.9‒64.1)	762	41.1 (38.1‒44.0)	762	41.1 (38.2‒43.9)
Delaware	354	91.8 (82.2‒101.4)	210	43.3 (37.4‒49.2)	210	52.5 (46.6‒58.4)
District of Columbia	158	58.8 (49.6‒68.0)	104	40.4 (32.5‒48.2)	104	54.3 (45.8‒62.8)
Florida	5796	54.6 (53.2‒56.0)	4267	34.8 (33.7‒35.8)	4267	41.5 (40.5‒42.6)
Georgia	2917	94.6 (91.2‒98.1)	2734	65.9 (63.4‒68.4)	2734	74.7 (72.2‒77.1)
Hawaii	107	66.3 (53.7‒78.9)	70	32.8 (25.3‒41.7)	70	44.3 (35.8‒52.8)
Idaho	387	68.8 (61.9‒75.6)	420	57.5 (51.9‒63.0)	420	68.9 (63.5‒74.4)
Illinois	4587	79.7 (77.4‒82.0)	3269	54.3 (52.4‒56.2)	3269	61 (59.1‒63.0)
Indiana	2797	95.7 (92.1‒99.2)	2181	67.4 (64.6‒70.3)	2181	72.8 (70.0‒75.6)
Iowa	1079	64.0 (60.2‒67.9)	823	48.0 (44.7‒51.3)	823	60.1 (56.5‒63.7)
Kansas	1034	75.9 (71.2‒80.5)	763	53.4 (49.6‒57.2)	763	58.6 (54.7‒62.4)
Kentucky	1946	99.8 (95.3‒104.2)	1525	67.5 (64.1‒70.9)	1525	74.4 (71.1‒77.8)
Louisiana	1694	85.4 (81.4‒89.5)	1490	67.2 (63.8‒70.7)	1490	88.3 (84.6‒92.1)
Maine	514	73.7 (67.3‒80.1)	375	46.2 (41.5‒50.9)	375	56.1 (51.2‒61.1)
Maryland	1561	67.7 (64.3‒71.0)	978	37.1 (34.7‒39.4)	978	44 (41.6‒46.4)
Massachusetts	2085	62.9 (60.2‒65.6)	1411	42.3 (40.1‒44.6)	1411	48.5 (46.2‒50.8)
Michigan	3881	81.9 (79.4‒84.5)	2902	56.2 (54.2‒58.3)	2902	62.6 (60.5‒64.7)
Minnesota	1469	64.3 (61.0‒67.6)	1037	40.6 (38.1‒43.0)	1037	45.8 (43.3‒48.3)
Mississippi	1749	133.8 (127.5‒140.1)	1690	112.7 (107.2‒118.1)	1690	109.6 (104.5‒114.6)
Missouri	2527	87.7 (84.3‒91.1)	1864	58.2 (55.5‒60.8)	1864	69.2 (66.5‒72.0)
Montana	338	73.5 (65.7‒81.4)	275	51.8 (45.6‒58.0)	275	59.1 (53.0‒65.2)
Nebraska	723	82.1 (76.2‒88.1)	530	56.3 (51.4‒61.1)	530	54.7 (50.1‒59.3)
Nevada	632	78.6 (72.4‒84.8)	442	37.8 (34.2‒41.4)	442	50.3 (46.6‒53.9)
New Hampshire	335	59.0 (52.7‒65.3)	352	51.6 (46.2‒57.1)	352	55.3 (50.0‒60.6)
New Jersey	2943	69.2 (66.7‒71.7)	2008	46.9 (44.9‒49.0)	2008	45.7 (43.8‒47.7)
New Mexico	440	57.7 (52.3‒63.0)	435	43.9 (39.7‒48.0)	435	48.1 (44.0‒52.1)
New York	6543	71.8 (70.1‒73.5)	4450	47.2 (45.8‒48.6)	4450	42.9 (41.6‒44.2)
North Carolina	2831	76.3 (73.5‒79.1)	2584	54.1 (52.0‒56.2)	2584	63.8 (61.7‒65.9)
North Dakota	272	76.4 (67.3‒85.5)	202	55.5 (47.8‒63.2)	202	45.9 (39.0‒52.7)
Ohio	5505	93.8 (91.3‒96.3)	3838	61.7 (59.8‒63.7)	3838	67.2 (65.2‒69.2)
Oklahoma	1764	107.2 (102.2‒112.2)	1484	80.5 (76.4‒84.6)	1484	79.9 (76.0‒83.9)
Oregon	1390	81.9 (77.6‒86.2)	1239	63.8 (60.2‒67.4)	1239	72.2 (68.7‒75.7)
Pennsylvania	5954	79.5 (77.5‒81.6)	3991	53.9 (52.2‒55.5)	3991	59.2 (57.5‒60.9)
Rhode Island	373	62.5 (56.2‒68.9)	259	46.6 (40.9‒52.3)	259	57.5 (51.4‒63.7)
South Carolina	1750	93.1 (88.8‒97.5)	1640	66.5 (63.3‒69.8)	1640	74.5 (71.4‒77.7)
South Dakota	266	66.2 (58.2‒74.2)	192	44.5 (38.2‒50.8)	192	51.1 (44.5‒57.6)
Tennessee	2422	89.4 (85.8‒92.9)	2001	60.2 (57.5‒62.9)	2001	73.2 (70.5‒75.9)
Texas	6538	83.2 (81.2‒85.2)	6153	60.2 (58.7‒61.7)	6153	66.8 (65.3‒68.3)
Utah	524	72.3 (66.1‒78.5)	577	60.6 (55.6‒65.6)	577	62 (57.4‒66.6)
Vermont	193	65.3 (56.1‒74.5)	172	49.7 (42.2‒57.2)	172	40.1 (33.8‒46.5)
Virginia	2461	80.9 (77.7‒84.1)	1975	53.7 (51.3‒56.0)	1975	60.3 (58.0‒62.7)
Washington	1679	67.0 (63.8‒70.2)	1578	53.3 (50.6‒56.0)	1578	56 (53.5‒58.5)
West Virginia	1197	112.2 (105.8‒118.5)	855	74.7 (69.6‒79.7)	855	73.3 (68.4‒78.1)
Wisconsin	2041	75.4 (72.2‒78.7)	1467	49.4 (46.9‒52.0)	1467	56 (53.4‒58.6)
Wyoming	182	83.3 (71.2‒95.4)	143	53.6 (44.7‒62.6)	143	56.3 (47.9‒64.7)

AAMR indicates age‐adjusted mortality rate.

Association of State‐Level Differences in CVH and Cardiovascular Mortality Related to HF

The percentage of individuals meeting criteria for “poor” CVH (2 or fewer ideal factors) for each state in 2011 and 2017 is shown in Table S2. In 2011, the percentage of residents with poor CVH ranged from 8.4 (Colorado) to 22.4 (Mississippi). In 2017, poor CVH ranged from 6.5% (District of Columbia) to 19.7% (Kentucky). In 2011 and 2017, the percentage of state residents with poor CVH was significantly associated with HF mortality (P<0.001) (Figure 3). In 2017, the model β estimate was 3.13, indicating ≈3 additional deaths per 100 000 associated with every 1% higher in the prevalence of poor CVH at the state level.

Figure 3. Correlation of state‐level prevalence of poor cardiovascular health score with cardiovascular mortality related to heart failure in (A) 2011 and (B) 2017.

Poor cardiovascular health was calculated according to American Heart Association criteria with state‐level data from the Behavior Risk Factor Surveillance System. CVH indicates cardiovascular health.

Discussion

Principal Findings

AAMR for HF mortality experienced an inflection point in 2011 nationally with similar trends across all 4 regions: generally declining before and increasing after 2011. Wide geographic variation exists in HF mortality rates. The South and Midwest experienced the highest rates and the largest increases observed since 2011. Black men in each region had the highest HF mortality rates and experienced the greatest increases between 2011 and 2017. Only 8 states saw decreases in their HF mortality rates between 2011 and 2017, while all others saw increases. No state from the West region saw decreases between 2011 and 2017. States from the South and Midwest census regions consistently comprised the 10 highest AAMR. A higher proportion of residents in a state with poor CVH was associated with higher rates of HF mortality in that state.

Current Study in Context

The current study adds to this literature by demonstrating the significant geographic heterogeneity in the burden of HF mortality and highlights opportunities for targeted prevention efforts on a state and local level. HF AAMR in the South and Midwest are higher than other regions. The South has also seen the greatest increases in HF AAMR since 2011. This is consistent with historical work that has demonstrated geographic variation in HF and stroke mortality, with higher rates clustered in Southern states leading to the region being labeled the “stroke belt”. ⁸ , ¹⁰ Others have also demonstrated higher rates of HF‐related morbidity reflected by hospitalization clustered in Southern and Midwest states. ⁷

Our study also confirms significant disparities in HF mortality that are pervasive across regions and states and are consistent with prior data from population‐based cohort studies, including the Multi‐Ethnic Study of Atherosclerosis that demonstrated Black participants had higher rates of developing incident HF (4.6 per 1000 person‐years) compared with Hispanic participants (3.5), White participants (2.4), and Chinese participants (1.0). ⁴ , ²¹ Multiple factors underlie these geographic and demographic trends. For instance, risk factors such as hypertension, ¹⁰ , ²² obesity, and diabetes mellitus ¹⁰ , ²³ have previously been shown to cluster in Southern states, where HF mortality is high. Similarly, we found that state‐level variation in poor CVH is significantly associated with HF mortality, which is consistent with prior publications demonstrating higher rates of poor CVH and CVD mortality in Southern states. ¹⁴ Ample epidemiologic evidence demonstrates that Black men and women have higher rates of poor CVH related to a variety of upstream social determinants of health, which include structural and systemic racism. ¹² , ¹⁴ A separate study of county‐level variation in total CVD mortality showed that demographic factors account for 36% of CVD mortality variation and economic/social conditions accounted for another 32%. ²⁴ Combined, healthcare indicators, healthcare usage, and features of the environment accounted for 6%.

Given the rising rates of HF mortality as well as clear variation in rates across the United States, state‐level policies and programs are needed to address the growing burden of HF. These programs must function on multiple levels. First, programs must target ideal CVH promotion and treat underlying CVD risk factors as they develop. Current estimates indicate that only 1% to 3.3% of the population meets criteria for ideal CVH, ¹³ , ²⁵ and as re‐demonstrated in this study, a high proportion of the population are classified as poor. If we are to stem the growing burden of HF morbidity and mortality, we must address the modifiable risk factors to improve CVH. Early identification and treatment of risk factors should be a priority, as should integrated programs focusing on the management of chronic conditions that lead to HF and other CVD. ²⁶ In addition to focus on individual health behaviors are important, we must also examine regional socioeconomic and political/policy infrastructures that underlie these trends to enact structural and environmental changes. ²⁶ , ²⁷ For example, local policy measures such as taxation of tobacco products or sugary beverages and availability of healthy foods may affect health risk behaviors and ultimately CVH. ²⁸ , ²⁹ , ³⁰ Local infrastructure, such as the structure of state and local boards of health, may influence public health expenditures and indirectly health outcomes. ³¹ However, interventions that focus on proximate causes alone are unlikely to mitigate the increasing Black‐White HF mortality disparities that reflect structural and systemic barriers to access to high quality care.

Crucially, we must consider the role that social determinants of health play in these health disparities as well. In this regard, state‐level policies are vitally important. One illustrative example of this is the different approaches states have taken to Medicaid expansion. When the Affordable Care Act went into effect, it included provisions for the expansion of Medicaid to all adults with a family income <138% of the federal poverty level; however, a Supreme Court ruling in 2012 essentially made the expansion optional to individual states. ³² As of May 2020, 36 states (including the District of Columbia) have implemented the expansion, 1 state (Nebraska) has adopted but not yet implemented, while 14 states have not adopted expansion. ³³ Subsequent research has demonstrated that implementation of the Affordable Care Act not only increased the overall rate of insurance coverage in the United States, but it also reduced race and ethnicity related disparities in health insurance. Coverage gains and disparity improvements were greater in states that implemented Medicaid expansion compared with states that did not expand Medicaid. ³² Unfortunately, ≈46% of Black working‐aged adults live in non‐expansion states and have thus been disproportionately impacted by non‐expansion. ³⁴ The states that have not expanded are clustered predominantly in the South and Midwest where rates of HF mortality are also highest.

Strengths and Limitations

The current nationwide study builds on the literature by highlighting geographic trends specifically of cardiovascular mortality related to HF (abbreviated here as HF mortality) over time in the US population. By measuring HF mortality in this way, we capture all cardiovascular deaths in which HF is listed as a contributing cause. This is significant because HF is more likely to be listed as a contributing cause of death than an underlying cause of death. ¹⁷ While prior studies have published geographic trends in HF mortality previously, ¹⁰ , ³⁵ we were able to more fully capture the burden of HF mortality using this approach.

Our study has several limitations. First, our findings are based on death certificate data. Therefore, there is the possibility for misclassification of deaths because of poorly defined underlying cause of death and/or lack of inclusion of HF as a contributing cause of death. While it is possible that miscoding may affect race–sex groups disparately, this alone does not likely completely explain the disparities observed. ¹⁷ To address the potential role for alternate coding of HF on the findings, we performed sensitivity analyses examining alternate definitions (HF as the underlying cause or HF as any contributing cause to all causes of death) and regardless of which definition used, the race‒sex and geographic patterns described above persisted. Additionally, leveraging national death certificate data provides the most comprehensive evaluation of state and regional burden of HF mortality. Second, limited numbers across states in other key race/ethnic groups (eg, Asian Americans, Hispanic/Latino Americans) and concern for misclassification of race/ethnicity led to our focus on only Black‒White differences. Even so, in several states the number of deaths among Black men and women was so small that AAMRs could not be reliably calculated. This limits our ability to infer about HF mortality rates among Black men and women in these states. Third, data on type, severity, and treatment of HF, such as left ventricular ejection fraction, presence of comorbidities (eg, diabetes mellitus), and guideline‐directed medical therapy use are unavailable in the CDC WONDER data set. Fourth, quantification of CVH using BRFSS may be subject to under‐estimation given reliance on self‐report. However, this likely biased our results towards the null. As CVH is a tool in the primary prevention of CVD, we excluded individuals with a known history of coronary heart disease, myocardial infarction, or stroke from the CVH calculations. BRFSS does not ask about other chronic CVD (such as heart failure, peripheral arterial disease, or history of revascularization), so we are unable to exclude those individuals. Finally, increasing awareness of HF could contribute to increases in reporting of HF as a contributing cause of death, in which case, recent data better reflect true burden of HF mortality in the United States.

Conclusions

In summary, we demonstrate that there is significant geographic variation in HF mortality, which is associated with state‐level CVH. Highest rates of HF mortality and greatest increases occurred in the South and Midwest. Black men are disproportionately affected by HF mortality and are experiencing the most rapidly increasing rates. Interventions at the regional and state level, particularly those equitably targeting CVH and HF prevention, are urgently needed.

Sources of Funding

Khan is funded by American Heart Association #19TPA34890060, KL2TR001424, P30AG059988, and P30DK092939. Research reported in this publication was supported, in part, by the National Institutes of Health's National Center for Advancing Translational Sciences, Grant Number KL2TR001424 (Khan). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Disclosures

None.

Supporting information

Tables S1–S2

(J Am Heart Assoc. 2021;10:e020541. DOI: 10.1161/JAHA.120.020541.)