Abstract
This cohort study evaluates recent reversals in declines in cardiovascular mortality and whether they vary across sociodemographic categories.
Recent data suggest plateaus or reversals after long-standing declines in cardiovascular mortality, particularly for heart failure (HF)–related mortality.1,2,3 There is little data on the association of the COVID-19 pandemic with these trends and the net change in contemporary HF mortality rates and disparities. Accordingly, we investigated the current status of HF mortality rates in relation to historical declines and whether they vary across sociodemographic subgroups.
Methods
This cohort study used data from the Centers for Disease Control and Prevention Wide-Ranging Online Data for Epidemiological Research. Because data were anonymized and publicly available, approval by an institutional review board and informed consent were not required in accordance with 45 CFR §46. Deaths were classified as HF-related if HF was listed in the death certificate (using the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision code I50) as the underlying or contributing cause of death. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines.
Age-adjusted mortality rates (AAMR) per 100 000 were extracted for the overall population and stratified by age, sex, race and ethnicity (as reported on death certificates and assessed because of known racial and ethnic disparities in socioeconomic determinants of health), urbanization status (data only available until 2020), and census region. We used the Joinpoint Regression Program, version 5.0.1 (National Cancer Institute), to calculate the annual percentage change (APC) for AAMR via the nih.joinpoint package on R version 4.2.0 (R Project for Statistical Computing).4,5,6 This software identifies time points at which a trend changes (inflection points) and produces different linear trends for segments before and after inflection.
Reversals were defined as the ratio between mortality increase (if any) after the inflection point and the decrease preceding the inflection point. To account for uncertainty in mortality estimates and construct 95% CIs around reversal estimates, we used Monte-Carlo simulations with 1 million iterations to sample parameter estimates from a multivariate t-distribution. The 2.5th and 97.5th percentiles were used as the lower and upper bounds of the 95% CI. Results were significant if the 95% CI excluded the null. Data were analyzed from June to September 2023.
Results
HF-related mortality declined with APCs of −1.62% (95% CI, −2.15 to −1.09%) from 1999 to 2005 and −3.29% (95% CI, −4.84 to −1.71%) from 2005 to 2009. From 2009 to 2012, declines nearly plateaued with an APC of −0.95% (95% CI, −4.11 to 2.32%). From 2012 to 2019, mortality rates increased (APC, 1.82%; 95% CI, 1.30 to 2.34%). During 2020 and 2021, increases in mortality rates accelerated (APC, 7.06%; 95% CI, 4.24 to 9.96%) (Figure, A and Table).
Figure. Temporal Trends in Heart Failure-Related Mortality in the US, 1999 to 2021.
Data by urbanization status was only available until 2020.
Table. Trends in the Rates of Age-Adjusted Heart Failure-Related Mortality Rates (per 100 000) for the Overall US Population and Across Different Sociodemographic Groups, 1999 to 2021.
| Stratification and trend years | Annual percentage change for time segment (95% CI)a | Mortality rate in 1999 (95% CI) | Mortality rate at inflection point (95% CI)b | Mortality rate in 2021 (95% CI) | Decrease in mortality from 1999 until inflection (95% CI) | Increase in mortality after inflection (95% CI) | Reversal percentage (95% CI)c |
|---|---|---|---|---|---|---|---|
| Overall | |||||||
| 1999-2005 | −1.62 (−2.15 to −1.09) | 105.49 (105.11 to 105.88) | 81.43 (81.13 to 81.73) | 106.25 (105.93 to 106.58) | 24.06 (23.57 to 24.55) | 24.82 (24.38 to 25.26) | 103.2 (101.1 to 105.3) |
| 2005-2009 | −3.29 (−4.84 to −1.71) | ||||||
| 2009-2012 | −0.95 (−4.11 to 2.32) | ||||||
| 2012-2019 | 1.82 (1.30 to 2.34) | ||||||
| 2019-2021 | 7.06 (4.24 to 9.96) | ||||||
| Age | |||||||
| <45 y | |||||||
| 1999-2012 | −0.88 (−1.25 to −0.51) | 1.76 (1.7 to 1.82) | 1.6 (1.54 to 1.66) | 3.05 (2.97 to 3.13) | 0.16 (0.08 to 0.24) | 1.45 (1.35 to 1.55) | 906.3 (605.6 to 1873.7) |
| 2012-2019 | 4.82 (3.62 to 6.03) | ||||||
| 2019-2021 | 17.40 (11.17 to 23.97) | ||||||
| 45 to 64 y | |||||||
| 1999-2011 | −1.86 (−2.05 to −1.67) | 31.98 (31.53 to 32.43) | 26.41 (26.07 to 26.75) | 47.84 (47.39 to 48.29) | 5.57 (5.01 to 6.13) | 21.43 (20.86 to 22.00) | 384.7 (351.5 to 425.3) |
| 2011-2019 | 4.68 (4.28 to 5.08) | ||||||
| 2019-2021 | 14.22 (11.57 to 16.93) | ||||||
| ≥65 y | |||||||
| 1999-2005 | −1.66 (−2.20 to −1.10) | 769.43 (766.49 to 772.37) | 588.86 (586.58 to 591.15) | 740.91 (738.51 to 743.31) | 180.57 (176.85 to 184.29) | 152.05 (148.74 to 155.37) | 84.2 (82.3 to 86.1) |
| 2005-2009 | −3.36 (−4.97 to −1.73) | ||||||
| 2009-2012 | −1.03 (−4.31 to 2.36) | ||||||
| 2012-2019 | 1.49 (0.96 to 2.03) | ||||||
| 2019-2021 | 6.15 (3.22 to 9.17) | ||||||
| Sex | |||||||
| Female | |||||||
| 1999-2005 | −1.59 (−2.21 to −0.96) | 94.48 (94.02 to 94.94) | 71.22 (70.86 to 71.58) | 90.3 (89.91 to 90.69) | 23.26 (22.68 to 23.84) | 19.08 (18.55 to 19.61) | 82.0 (79.7 to 84.4) |
| 2005-2009 | −3.74 (−5.57 to −1.86) | ||||||
| 2009-2012 | −1.30 (−5.10 to 2.66) | ||||||
| 2012-2019 | 1.45 (0.82 to 2.09) | ||||||
| 2019-2021 | 6.72 (3.19 to 10.38) | ||||||
| Male | |||||||
| 1999-2005 | −1.65 (−2.03 to −1.26) | 121.99 (121.29 to 122.69) | 95.79 (95.27 to 96.31) | 126.98 (126.43 to 127.53) | 26.2 (25.33 to 27.07) | 31.19 (30.43 to 31.95) | 119.0 (115.3 to 123.0) |
| 2005-2008 | −3.25 (−5.44 to −1.01) | ||||||
| 2008-2012 | −1.02 (−2.12 to 0.10) | ||||||
| 2012-2019 | 2.08 (1.73 to 2.42) | ||||||
| 2019-2021 | 7.19 (5.33 to 9.09) | ||||||
| Race | |||||||
| Hispanic | |||||||
| 1999-2012 | −2.62 (−2.99 to −2.26) | 72.51 (70.91 to 74.11) | 54.17 (53.22 to 55.12) | 69.85 (68.98 to 70.71) | 18.34 (16.48 to 20.20) | 15.68 (14.39 to 16.97) | 85.5 (77.2 to 95.0) |
| 2012-2019 | 1.74 (0.78 to 2.71) | ||||||
| 2019-2021 | 8.67 (3.88 to 13.69) | ||||||
| Non-Hispanic Black | |||||||
| 1999-2012 | −1.92 (−2.21 to −1.62) | 108.78 (107.39 to 110.17) | 87.93 (86.85 to 89.01) | 122.64 (121.51 to 123.78) | 20.85 (19.09 to 22.61) | 34.71 (33.14 to 36.28) | 166.5 (154.5 to 180.3) |
| 2012-2019 | 2.64 (1.77 to 3.52) | ||||||
| 2019-2021 | 9.42 (4.93 to 14.11) | ||||||
| Non-Hispanic Other | |||||||
| 1999-2014 | −2.08 (−2.52 to −1.63) | 59.61 (57.62 to 61.6) | 42.09 (41.08 to 43.1) | 52.55 (51.59 to 53.51) | 17.52 (15.29 to 19.75) | 10.46 (9.07 to 11.85) | 59.7 (51.2 to 69.5) |
| 2014-2021 | 2.96 (1.92 to 4.00) | ||||||
| Non-Hispanic White | |||||||
| 1999-2005 | −1.57 (−2.09 to −1.04) | 107.23 (106.8 to 107.65) | 84.04 (83.69 to 84.38) | 111.98 (111.6 to 112.37) | 23.19 (22.65 to 23.73) | 27.94 (27.43 to 28.45) | 120.5 (117.8 to 123.3) |
| 2005-2009 | −3.15 (−4.71 to −1.57) | ||||||
| 2009-2012 | −0.67 (−3.86 to 2.63) | ||||||
| 2012-2019 | 1.98 (1.45 to 2.50) | ||||||
| 2019-2021 | 7.18 (4.25 to 10.20) | ||||||
| Urbanizationd | |||||||
| Rural | |||||||
| 1999-2012 | −1.94 (−2.23 to −1.64) | 127.29 (126.32 to 128.25) | 101.69 (100.89 to 102.5) | 125.31 (124.45 to 126.16) | 25.6 (24.34 to 26.86) | 23.62 (22.45 to 24.79) | 92.3 (87.6 to 97.2) |
| 2012-2020 | 2.22 (1.62 to 2.83) | ||||||
| Urban | |||||||
| 1999-2012 | −2.21 (−2.51 to −1.92) | 100.32 (99.9 to 100.74) | 77.17 (76.85 to 77.49) | 95.14 (94.81 to 95.46) | 23.15 (22.62 to 23.68) | 17.97 (17.51 to 18.43) | 77.6 (75.6 to 79.7) |
| 2012-2020 | 2.38 (1.80 to 2.96) | ||||||
| Census region | |||||||
| Midwest | |||||||
| 1999-2011 | −2.36 (−2.64 to −2.07) | 115.05 (114.23 to 115.86) | 90.64 (89.97 to 91.31) | 118.78 (118.04 to 119.52) | 24.41 (23.35 to 25.47) | 28.14 (27.14 to 29.14) | 115.3 (110.5 to 120.4) |
| 2011-2019 | 2.08 (1.43 to 2.73) | ||||||
| 2019-2021 | 7.64 (3.21 to 12.25) | ||||||
| Northeast | |||||||
| 1999-2011 | −2.48 (−2.81 to −2.15) | 97.61 (96.81 to 98.41) | 75.8 (75.15 to 76.46) | 87.7 (87.03 to 88.37) | 21.81 (20.78 to 22.84) | 11.9 (10.96 to 12.84) | 54.6 (50.7 to 58.5) |
| 2011-2021 | 1.60 (1.17 to 2.04) | ||||||
| South | |||||||
| 1999-2003 | −1.00 (−2.28 to 0.29) | 106.64 (105.98 to 107.3) | 83.28 (82.77 to 83.8) | 110.9 (110.36 to 111.44) | 23.36 (22.52 to 24.20) | 27.62 (26.87 to 28.37) | 118.2 (114.3 to 122.4) |
| 2003-2011 | −2.75 (−3.29 to −2.22) | ||||||
| 2011-2019 | 1.75 (1.24 to 2.27) | ||||||
| 2019-2021 | 7.93 (4.45 to 11.52) | ||||||
| West | |||||||
| 1999-2005 | −1.32 (−2.15 to −0.49) | 99.75 (98.9 to 100.61) | 77.07 (76.44 to 77.71) | 101.35 (100.68 to 102.02) | 22.68 (21.62 to 23.74) | 24.28 (23.36 to 25.20) | 107.1 (102.2 to 112.2) |
| 2005-2012 | −2.56 (−3.36 to −1.76) | ||||||
| 2012-2019 | 1.80 (1.03 to 2.57) | ||||||
| 2019-2021 | 7.83 (3.67 to 12.15) |
An annual percentage change (APC) is shown for each of the time periods identified by the Joinpoint software as having different trends. Thus, each group has 2 or more APC estimates corresponding to different time segments.
The inflection point was the year at which mortality rates reached their nadir (and after which they started increasing) as identified by the Joinpoint software.
This represents the percentage ratio between the initial decline (1999 until the inflection point) and the subsequent increase (from inflection onwards).
Age-adjusted mortality data stratified by urbanization status were only available until 2020.
The increases from 2012 to 2021 were larger than the initial declines, such that the reversal in mortality was 103.2% (95% CI, 101.1 to 105.3%) (Table). Reversals were observed across all subgroups but varied in magnitude (Figure, B-F and Table). The greatest reversals were seen in those aged younger than 45 (906.3%) and 45 to 64 years (384.7%), males (119.0%), Non-Hispanic Black individuals (166.5%), rural areas (92.3%), and the South (118.2%) and Midwestern US (115.3%) (Table). Results were similar when the analysis was restricted to HF as the underlying cause of death.
Discussion
Our analysis shows that declines in HF-related mortality from 1999 to 2012 have been entirely undone by reversals from 2012 to 2021, meaning that contemporary HF mortality rates are higher than in 1999. The origins of these reversals preceded the COVID-19 pandemic, although the larger increases in 2020 to 2021 indicate that the pandemic may have accelerated them due to limitations to health care access and possible cardiac involvement. An important limitation of this analysis is the reliance on death certificate data, which may misattribute some deaths, particularly in cases where symptoms of HF cannot be easily differentiated from those of other causes (eg, COVID-19) and the inability to disaggregate race and ethnicity in a more granular manner. Additionally, changes in diagnostic methods, coding patterns, and longer survival for patients with conditions that predispose to heart failure (eg, ischemic heart disease) may lead to an increase in the prevalence of HF, thereby increasing the observed HF mortality rate.
Data Sharing Statement
References
- 1.Shah NS, Molsberry R, Rana JS, et al. Heterogeneous trends in burden of heart disease mortality by subtypes in the United States, 1999-2018: observational analysis of vital statistics. BMJ. 2020;370:m2688. doi: 10.1136/bmj.m2688 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Jain V, Minhas AMK, Morris AA, et al. Demographic and regional trends of heart failure-related mortality in young adults in the US, 1999-2019. JAMA Cardiol. 2022;7(9):900-904. doi: 10.1001/jamacardio.2022.2213 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Sidney S, Quesenberry CP Jr, Jaffe MG, Sorel M, Go AS, Rana JS. Heterogeneity in national U.S. mortality trends within heart disease subgroups, 2000-2015. BMC Cardiovasc Disord. 2017;17(1):192. doi: 10.1186/s12872-017-0630-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Joinpoint Regression Program , Version 5.0.1. Statistical Methodology and Applications Branch, Surveillance Research Program, National Cancer Institute. 2023. https://surveillance.cancer.gov/joinpoint/
- 5.R Core Team . R: A language and environment for statistical computing. 2022. https://www.R-project.org/.
- 6.Chaltiel D. nih.joinpoint: R interface for NIH’s Joinpoint regression software. R package version 0.1.0.9001. 2023. https://github.com/DanChaltiel/nih.joinpoint/
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Sharing Statement