Temporal Trends in Substance Use and Cardiovascular Disease–Related Mortality in the United States

Deaths occurring within the United States related to CVD and SU were extracted from the CDC WONDER database.¹⁰ The Multiple Cause‐of‐Death Public Use record death certificates were used to analyze deaths in which SU and CVD were both mentioned as contributing or underlying causes of death on nationwide death certificates. This database has previously been used in several other studies to determine CVD mortality trends.^{11, 12} Patients with SU were identified with International Classification of Diseases, Tenth Revision, Clinical Modification (ICD‐10‐CM) codes provided in supplementary files, and patients with CVD were identified with ICD‐10‐CM codes I00‐I99 (diseases of the circulatory system) in patients ≥25 years of age as previously described.^{12, 13, 14} Smoking/tobacco use was not included in primary analyses as a form of SU. If a patient had multiple SU listed on the death certificate, they would be counted once for SU‐related death. However, in a subgroup by drug category, they would be accounted for all drug category separately listed on the death certificate. ICD codes of intentional substance overdoses were included in our analysis, whereas accidental or assault by substance were not included (Table S1). Institutional Review Board approval was not sought because the CDC WONDER database contains publicly available, anonymized data. All data and materials have been made publicly available and can be accessed at https://wonder.cdc.gov/. Informed consent was not required due to the nature of the publicly available deidentified data.

SU+CVD‐related deaths, population size, and location of death (including medical facilities [outpatient, emergency room, inpatient, death on arrival or status unknown], home, hospice, and nursing home/long‐term care facility) were extracted. Demographics (sex, race and ethnicity, and age) and regional information (urban–rural and state) were extracted from 1999 to 2019. Races and ethnicities were defined as non‐Hispanic (NH) White, NH Black, Hispanic, Native American or Alaska Native, and NH Asian or Pacific Islander patients. These racial and ethnic categories have previously been used within analyses from the CDC WONDER database and rely on reported data on death certificates.¹⁰ Age groups were defined as 25–39, 40–54, 55–69, 70–84, and 85+ years of age. The CDC WONDER database provides age‐adjusted mortality rates in preset 10‐year intervals for adults. Therefore, we were unable to include patients 18 to 24 years of age within the analyses. For urban–rural classifications, the 2013 National Center for Health Statistics Urban–Rural Classification Scheme was used to divide the counties into metropolitan (large central metropolitan, large fringe metropolitan, medium metropolitan, and small metropolitan) and nonmetropolitan (micropolitan and noncore) categories.¹⁵ Regions were classified into Northeast, Midwest, South, and West according to the Census Bureau definitions.

Crude and age‐adjusted mortality rates (AAMRs) per 100 000 population were determined. Crude mortality rates were determined by dividing the number of SU+CVD‐related deaths by the corresponding US population of that year. As previously described, AAMRs were calculated by standardizing the SU+CVD‐related deaths to the year 2000 US population.¹⁶ The Joinpoint Regression Program (Joinpoint V 4.9.0.0, National Cancer Institute) was used to determine trends in AAMR using annual percent change (APC).¹⁷ This method identifies significant changes in AAMR over time by fitting a series of joined straight lines on a log scale where temporal variation occurred. For data containing 17 to 21 time points, the guidelines recommend that the analysis identifies a maximum of 3 inflection points across the study period.¹⁸ Therefore, the Joinpoint regression statistical software was set to determine a maximum of 3 Joinpoints (21 years of analysis were included) where significant temporal variation existed in the trend. APCs with 95% CI for the AAMR were calculated at the identified line segments linking Joinpoints using the Monte Carlo permutation test. The weighted average of the APCs was calculated and reported as average APCs (AAPCs) and corresponding 95% CIs as a summary of the reported mortality trend for the entire study period. APCs were considered increasing or decreasing if the slope describing the change in mortality did not have CIs that crossed zero. Statistical significance was inferred based on nonoverlapping CIs.

Our analysis of SU+CVD trends in the United States between 1999 and 2019 demonstrates several important findings. Despite a reduction in overall CVD mortality during this period, SU+CVD‐related mortality increased at an average of 4% per year. Men, American Indian or Alaska Native individuals, the 55‐ to 69‐year age group, and nonmetropolitan areas demonstrated the highest AAMRs in their demographic subgroups. In 2019, alcohol, opioids, stimulants (methamphetamines), and cocaine were the most common SU contributors to SU+CVD mortality. The temporal increase in SU+CVD during the study period was pronounced in women, American Indian or Alaska Native and NH White individuals, younger individuals, nonmetropolitan areas, and users of stimulants. The increase in SU+CVD‐related AAMR has particularly accelerated since 2012. These results have important public health implications by identifying high‐risk groups that may benefit from targeted efforts to reduce substance‐associated CVD mortality.

Substances evaluated in our analysis, including alcohol, cannabis, cocaine, opioids, and stimulants, have multiple cardiovascular effects and are associated with the development of cardiomyopathy, arrhythmias, microvascular disease, and coronary artery disease, particularly in the case of cocaine and stimulants.^{2, 3, 4, 5, 6, 19, 20, 21} SU may also be associated with the progression of CVD due to reduced health literacy and adherence to medical care.²² Among the substances evaluated in our study, alcohol was the most common to be associated with SU+CVD‐related death, more than 4 times opioids, the second‐highest substance. While cannabis had the lowest SU+CVD‐related absolute AAMR throughout analysis, it demonstrated a large increase in AAPC over the study period. This could be due to a nationwide increase in use in the setting of permissive laws, especially in the younger age group and White individuals,^{7, 23, 24, 25} or due to increased potency of tetrahydrocannabinol in cannabis products in recent years.²⁶ On the other hand, low absolute mortality rates of cannabis+CVD suggest that targeting other SU may yield more substantial improvement in CVD mortality. Large percent increases in SU+CVD were also seen among stimulant use, particularly between 2010 and 2019, where stimulants were the fastest‐growing category of SU+CVD. Stimulants, particularly methamphetamines, are associated with significant cardiotoxicity and have become an epidemic across certain geographic regions in the United States.^{14, 27} These findings further reinforce the growing role of stimulants as a contributor to CVD mortality.

Our results indicate that men have higher SU+CVD‐related AAMR, likely related to a higher SU rate among men and generally lower mortality among women from CVD conditions.²⁸ Both men and women experienced increases in SU+CVD‐related AAMRs throughout the study periods. AAPC was pronounced in women at almost 5% a year, concordant with data from the National Vital Statistics System indicating a notable increase in SU usage and overdose rate especially among middle‐aged women.²⁹ Vulnerabilities leading to SU have been shown to vary between men and women, with greater psychosocial burdens, higher rates of comorbid mood disorders, and higher barriers to SU treatment in women, all potentially contributing to a rise in SU+CVD.³⁰ Therefore, additional efforts to address the growing use of SU geared towards unique sex‐based barriers may be of value to improve SU+CVD mortality trends.

We also highlight substantial racial disparities in the rates of SU+CVD, which expands on the prior literature in this important area. We found that American Indian or Alaska Native individuals had the highest absolute AAMRs and had a pronounced increase in SU+CVD‐related AAMR during the study period, with the AAMR almost double the second‐highest group (NH Black individuals). There may be several reasons for racial differences in SU+CVD. Prior studies have noted that White and American Indian individuals were more likely to use alcohol and develop alcohol use disorder. In contrast, Black individuals were more likely to use illicit drugs (including cocaine or stimulants).³¹ Differences in baseline use may be further exacerbated by disparities in SU treatment, where White individuals were more likely to undergo substance abuse treatment and with higher treatment success rate.^{32, 33} Additionally, racial and ethnic minority groups have higher SU‐related death despite similar or lower SU rates,^{31, 33} which may highlight barriers in health care and other socioeconomic inequalities. Racial disparities may also coincide with economic disparities based on social factors as contributors to both SU and SU+CVD, although prior studies have suggested heterogeneity in risk of SU disorder between individuals from different socioeconomic levels.^{34, 35, 36} Our results suggest that targeting efforts to identify and address causes in racial differences in SU+CVD mortality may be an important component of addressing known racial disparities in overall CVD outcomes.³⁷

Our results also demonstrate that SU+CVD AAMR is more prominent in rural areas than urban areas, with trends particularly diverging over the last decade. Differences in SU+CVD may be one of the explanations for previous studies that have highlighted higher CVD mortality in rural compared with urban regions.³⁸ Rural differences in SU have been attributed to socioeconomic vulnerabilities and access to health delivery systems in general³⁹ and SU treatment in particular,⁴⁰ and will require further evaluation.

Lastly, when examined by age groups, we found that although younger adults had lower SU+CVD AAMR than older individuals, those aged 25 to 39 years had pronounced AAPC increase during the study period. These results are concordant with the data demonstrating a surge in SU use in young adults living in the United States.²⁵ Increases in SU+CVD in younger populations are particularly alarming because younger patients have otherwise low rates of traditional CVD mortality, and targeting SU use in younger adults should remain an essential target of public health intervention.

This study has several limitations. The accuracy of the CDC WONDER data set depends on use of ICD codes, which may have resulted in misclassification of causes of death. Although we included ICD codes of intentional substance overdose, we did not include ICD codes of accidental or assault by substance. Overdose deaths are not specifically addressed in our analysis because the focus was primarily on SU+CVD‐associated mortality, although differentiating overdose from primarily CVD causes may be difficult. Individual drug categories for certain substances such as methamphetamines are not well characterized by ICD‐10 codes. Still, important categories such as ICD 10 codes stimulants have been associated with good accuracy for classifying methamphetamine use.⁴¹ Compared with other studies using ICD codes to identify SU mortality, we have used a broader definition with more inclusive ICD codes to capture more patients with a history of SU,⁴² although we could not independently verify the accuracy of these codes for SU. Using ICD codes, we were not able to specifically identify or differentiate users of alternative form of the substances (for instance, patients who used cannabis extracts). Smoking as a form of SU contributor to mortality is not evaluated in this analysis because the focus was primarily on other SU, and multivariable analyses adjusted for contributing smoking could not be performed in this data set. However, smoking was listed as a contributing source in ≈20% of individuals with CVD+SU mortality and may therefore mediate some of the SU‐related mortality in this analysis. Patients with concurrent smoking+SU will require further evaluation in future studies. Additionally, our results suggest that patients with SU should be screened and offered management for concurrent high‐risk CVD comorbidities including smoking. Reasons for heterogeneity in outcome differences among States were not examined, though they may be related to the higher prevalence of SU because AAMRs are a result of both prevalence and case fatality. Lastly, additional data on variables that may be important contributors to SU and CVD mortality, such as socioeconomic status, income level, mental health, and other comorbid conditions, were not able to be evaluated in our study.

Our analysis demonstrates a prominent increase in SU+CVD‐related mortality in the United States between 1999 and 2019, with the most pronounced increases among women, American Indian or Alaska individuals, younger individuals, those living in nonmetropolitan areas, and users of cannabis and psychostimulants. When looking at the most recent data in year 2019, AAMR was higher among men, American Indian or Alaska Native individuals, those 55 to 69 years old, nonmetropolitan areas, and users of alcohol, opioids, stimulants (methamphetamines), and cocaine. Additional public health efforts targeting SU may be valuable to reduce CVD mortality.

None.

Dr Greene has received research support from the Duke University Department of Medicine Chair's Research Award, American Heart Association (#929502), Amgen, AstraZeneca, Bristol Myers Squibb, Cytokinetics, Merck, Novartis, Pfizer, and Sanofi; has served on advisory boards for Amgen, AstraZeneca, Boehringer Ingelheim/Lilly, Bristol Myers Squibb, Cytokinetics, Roche Diagnostics, Sanofi, and scPharmaceuticals; serves as a consultant for Amgen, Bayer, Bristol Myers Squibb, Boehringer Ingelheim/Lilly, Corteria Pharmaceuticals, CSL Vifor, Lexicon, Merck, PharmaIN, Roche Diagnostics, Sanofi, Tricog Health, and Urovant Pharmaceuticals; and has received speaker fees from Boehringer Ingelheim, Cytokinetics, and Roche Diagnostics. The remaining authors have no disclosures to report.