The initial COVID-19 outbreak in the US caused disruptions in usual behavioral patterns.1-3 To assess associated changes in external causes of death, we analyzed monthly trends from 2015 to 2020 in deaths resulting from drug overdoses, homicide, unintentional injuries, motor vehicle crashes, and suicide in the first 6 months of the pandemic.
We measured monthly excess mortality (the gap between observed and expected deaths) from 5 external causes using provisional national-level underlying cause death certificate data published by the National Center for Health Statistics (NCHS) through August 2020 (released March 2021). Data from March to August 2020 were aggregated by the NCHS into 5 groups: drug overdose (all intents), assault (homicide), unintentional injuries, motor vehicle crashes, and intentional self-harm (suicide) (see the Supplement for ICD-10 codes).4,5
To forecast all-cause and cause-specific expected monthly deaths from March to August 2020, we used seasonal autoregressive integrated moving average (sARIMA) models developed with cause-specific monthly mortality counts and US population data from January 2015 to February 2020. We plotted observed and expected deaths monthly with 95% CIs estimated from sARIMA models.
We estimated the contribution of individual cause-specific mortality to all-cause non–COVID-19 excess mortality by dividing cause-specific mortality by total non–COVID-19 excess mortality from March to August 2020 (see the Supplement). Confidence intervals for the percent contribution to non–COVID-19 excess mortality were determined by subtracting the observed number of deaths from the upper and lower 95% thresholds for the expected number of deaths. For excess mortality counts, any figure not crossing 0 was considered statistically significant. For observed-to-expected ratios (OERs) of cause-specific mortality, statistical significance was defined as a 95% CI that excluded the null value of 1.00.
Analyses were conducted using R version 4.0.2. This study used publicly available data and was not subject to institutional review approval per HHS regulation 45 CFR 46.101(c).
From March to August 2020, there were 256 635 (95% CI, 161 450-351 823) all-cause excess deaths (1 661 271 observed; 1 404 634 expected) and 174 334 COVID-19 deaths (underlying cause). For the study period, OERs for 3 external causes of death were significantly higher than expected (drug overdoses, homicides, unintentional injuries), 1 unchanged (motor vehicle crashes), and 1 lower (suicides) (Table).
There were 10 443 excess drug overdoses (95% CI, 6115 to 14 771; Figure, A), accounting for 12.7% of non–COVID-19 excess mortality (95% CI, 7.4% to 17.9%); 2014 excess homicide deaths (95% CI, 1086 to 2942) (Figure, B), accounting for 2.4% of non–COVID-19 excess mortality (95% CI, 1.3% to 3.6%); and 7497 excess deaths due to unintentional injuries (95% CI, 694 to 14 300) (Figure, C), accounting for 9.1% of non–COVID-19 excess mortality (95% CI, 0.8% to 17.4%). There was no significant change in motor vehicle crash deaths overall (725; 95% CI, −1090 to 2540) but fewer than expected motor vehicle crash deaths occurred in April (−523; 95% CI, −815 to −231), and significant increases were recorded monthly from June to August (1550; 95% CI, 611 to 2489) (Figure, D). Suicide deaths were statistically significantly lower than projected by 2432 deaths (95% CI, 1071 to 3792 fewer deaths) (Figure, E).
Provisional mortality data showed that deaths from some but not all external causes increased during the pandemic, representing thousands of lives lost and exceeding prepandemic trends.
Explanations for these changes are unknown. Drug overdoses and homicides may have been related to economic stress. Pandemic-associated changes in access to substance use disorder treatments may have exacerbated mortality from overdoses.6 Decreases in motor vehicle crash deaths in April coincided with less traffic, despite increases in drivers testing positive for drugs and alcohol and lower seatbelt use.3 Increases in motor vehicle crash deaths in June to August occurred as traffic increased (though still below 2019 levels), likely reflecting higher-risk behaviors.3 Lower than projected suicide deaths are paradoxical with reported increases in depressive and other mental health symptoms during the pandemic. Additional data are needed to understand the mechanism behind this finding.
This study has limitations, including death certificate accuracy and that 2020 data published by NCHS are considered preliminary. However, substantial changes to March to August 2020 data are unlikely. Also, the true number of non–COVID-19 medical deaths may have been lower than projected during the pandemic period, as evidenced by the observation that in May, the total excess deaths due to drug overdoses, assaults, and unintentional injuries exceeded the apparent number of all non–COVID-19 excess deaths.
Corresponding Author: Jeremy S. Faust, MD, MS, Department of Emergency Medicine, Brigham and Women’s Hospital, 10 Vining St, Boston, MA 02115 (firstname.lastname@example.org).
Accepted for Publication: May 3, 2021.
Published Online: May 21, 2021. doi:10.1001/jama.2021.8012
Author Contributions: Dr Faust had full access to all of the data in the study and takes responsibility for the integrity of the data and accuracy of the data analysis.
Concept and design: Faust, Du, Lin, Krumholz.
Acquisition, analysis, or interpretation of data: Faust, Du, Mayes, Li, Lin, Barnett.
Drafting of the manuscript: Faust.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Faust, Du, Lin.
Administrative, technical, or material support: Faust, Mayes.
Supervision: Li, Lin, Krumholz.
Conflict of Interest Disclosures: Dr Lin reported working under contract from the Centers for Medicare & Medicaid Services (CMS) to develop and maintain measures for hospital performance that are publicly reported. Dr Krumholz reported receiving personal fees from UnitedHealth, IBM Watson Health, Element Science, Aetna, Facebook, Siegfried & Jensen Law Firm, Arnold & Porter Law Firm, Martin/Baughman Law Firm, F-Prime, and National Center for Cardiovascular Diseases, Beijing; being the co-founder of HugoHealth, a personal health information platform, and Refactor Health, an AI-augmented data management company; receiving contracts from CMS through Yale New Haven Hospital to develop and maintain measures of hospital performance; and receiving grants from Medtronic and FDA, Medtronic and Johnson & Johnson, and Shenzhen Center for Health Information. No other disclosures were reported.
et al. Trends in US emergency department visits for mental health, overdose, and violence outcomes before and during the COVID-19 pandemic. JAMA Psychiatry
. 2021;78(4):372-379. PubMedGoogle ScholarCrossref