Quick search in Citations
*Required field
Quick Search in Journals
Quick Search in Journals
AnywhereCitationPNAS NexusProceedings of the National Academy of Sciences
Advanced Search
SUGGESTED SEARCHES:
Covid-19 Artificial Intelligence Mpox Climate Change Gun Violence
Research Article
Political Sciences
Open access

Excess natural-cause mortality in US counties and its association with reported COVID-19 deaths

Eugenio Paglino https://orcid.org/0000-0001-8733-6460, Dielle J. Lundberg, Elizabeth Wrigley-Field https://orcid.org/0000-0003-3489-4279, +8 , Zhenwei Zhou https://orcid.org/0000-0002-9424-7826, Joe A. Wasserman https://orcid.org/0000-0002-9705-1853, Rafeya Raquib, Yea-Hung Chen https://orcid.org/0000-0002-7949-0692, Katherine Hempstead, Samuel H. Preston https://orcid.org/0000-0001-7038-9689, Irma T. Elo https://orcid.org/0000-0003-1858-8239, M. Maria Glymour, and Andrew C. Stokes https://orcid.org/0000-0002-8502-3636 [email protected]Authors Info & Affiliations
Edited by Kenneth Wachter, University of California, Berkeley, CA; received August 11, 2023; accepted December 6, 2023
February 1, 2024
121 (6) e2313661121
https://doi.org/10.1073/pnas.2313661121
Track CitationsAdd to Reading List
PDF/EPUB

Significance

Official COVID-19 mortality statistics have not fully captured deaths attributable to SARS-CoV-2 infection in the United States. While some excess deaths were likely related to pandemic health care interruptions and socioeconomic disruptions, temporal correlations between reported COVID-19 deaths and excess deaths reported to non-COVID-19 natural causes suggest that many of those excess deaths were unrecognized COVID-19 deaths. Efforts to target resources during public health emergencies should consider geographic variation in the quality of mortality surveillance data. Incomplete or delayed cause-of-death reporting may obscure impacts in some areas, leading to ineffective and inequitable responses and evaluations of the pandemic’s effects. To address this, future pandemic preparedness and response efforts should include activities to strengthen the death investigation system.

Abstract

In the United States, estimates of excess deaths attributable to the COVID-19 pandemic have consistently surpassed reported COVID-19 death counts. Excess deaths reported to non-COVID-19 natural causes may represent unrecognized COVID-19 deaths, deaths caused by pandemic health care interruptions, and/or deaths from the pandemic’s socioeconomic impacts. The geographic and temporal distribution of these deaths may help to evaluate which explanation is most plausible. We developed a Bayesian hierarchical model to produce monthly estimates of excess natural-cause mortality for US counties over the first 30 mo of the pandemic. From March 2020 through August 2022, 1,194,610 excess natural-cause deaths occurred nationally [90% PI (Posterior Interval): 1,046,000 to 1,340,204]. A total of 162,886 of these excess natural-cause deaths (90% PI: 14,276 to 308,480) were not reported to COVID-19. Overall, 15.8 excess deaths were reported to non-COVID-19 natural causes for every 100 reported COVID-19 deaths. This number was greater in nonmetropolitan counties (36.0 deaths), the West (Rocky Mountain states: 31.6 deaths; Pacific states: 25.5 deaths), and the South (East South Central states: 26.0 deaths; South Atlantic states: 25.0 deaths; West South Central states: 24.2 deaths). In contrast, reported COVID-19 death counts surpassed estimates of excess natural-cause deaths in metropolitan counties in the New England and Middle Atlantic states. Increases in reported COVID-19 deaths correlated temporally with increases in excess deaths reported to non-COVID-19 natural causes in the same and/or prior month. This suggests that many excess deaths reported to non-COVID-19 natural causes during the first 30 mo of the pandemic in the United States were unrecognized COVID-19 deaths.
Excess mortality calculations have been widely used to assess the mortality impact of the COVID-19 pandemic (1). Excess mortality refers to the difference between the observed number of deaths during a given period and the number of deaths that would be expected based on earlier mortality trends (2). More than 1.1 million excess all-cause deaths occurred in the United States during the first 24 mo of the pandemic, with approximately 635,000 in the first 12 mo and 544,000 in the next 12 mo (3). Most prior studies have found that the number of excess deaths in the United States has exceeded the number of reported COVID-19 deaths during the pandemic (4–8).
The term “excess deaths reported to non-COVID-19 causes” (or alternatively “excess deaths excluding reported COVID-19 deaths”) refers to the number of deaths during the pandemic above (or below) the number expected based on pre-pandemic trends but which were not attributed to COVID-19 on death certificates (7, 9, 10). COVID-19 is typically reported as an underlying cause of death but can be listed elsewhere on the death certificate in some cases. We define “reported COVID-19 deaths” as death certificates with any mention of COVID-19.
There are several possible reasons why some excess deaths during the pandemic were not reported as COVID-19 deaths. Some deaths attributable to SARS-CoV-2 infection may have gone unrecognized by death investigators as COVID-19 deaths due to limited COVID-19 testing, atypical presentation of symptoms, comorbidities, limited resources for death investigation in out-of-hospital settings, and stigma or political beliefs about COVID-19 (11–15). Deaths could also be indirectly related to the pandemic as a result of health care delays and interruptions (16, 17) and/or social and economic impacts of the pandemic such as housing instability, employment loss, food insecurity, social isolation, and increases in poisonings, suicide, homicide, and accidents (18–23).
Most prior studies of excess mortality during the pandemic have examined deaths from all causes (3–6, 8), but estimates of excess deaths from natural causes could be useful to study to what extent reported COVID-19 death counts captured deaths attributable to SARS-CoV-2 infection. Natural causes include diseases and chronic conditions, whereas external causes consist of intentional and unintentional injuries (24). SARS-CoV-2 infection is unlikely to lead to excess deaths from external causes in the short term (25, 26). Therefore, differences between excess natural-cause deaths and reported COVID-19 deaths may represent a more plausible estimate of unrecognized COVID-19 deaths than comparisons that include external causes.
A growing body of research has investigated the spatial and temporal patterning of mortality attributable to COVID-19 across the United States (27–30). A prior study by our team produced monthly estimates of all-cause excess mortality for US counties (3). This study examined spatial and temporal variation in excess mortality, leaving important questions unanswered about excess deaths reported to non-COVID-19 natural causes. Specifically, it is unknown how many excess deaths have occurred during the pandemic that were reported to non-COVID-19 natural causes, which counties and regions had more of these deaths in relative and absolute terms, how these deaths progressed temporally in different areas, and whether temporal patterns provide any evidence as to whether these deaths were unrecognized COVID-19 deaths.
In the present study, we compare monthly estimates of excess natural-cause deaths generated by applying a Bayesian hierarchical model to monthly data on reported COVID-19 deaths for 3,127 counties over the first 30 mo of the pandemic from March 2020 to August 2022. We then explore spatial and temporal patterning of excess natural-cause mortality compared to reported COVID-19 mortality in Census divisions, metropolitan–nonmetropolitan categories, and their combinations.

Results

Over 30 mo from March 2020 to August 2022, 1,194,610 excess natural-cause deaths occurred in the United States (90% Posterior Interval (PI): 1,046,000 to 1,340,204). A total of 1,031,724 (86.4%) were reported to COVID-19, and 162,886 (13.6%) (90% PI: 14,276 [1.2%] to 308,480 [25.8%]) were reported to non-COVID-19 natural causes. Of the reported COVID-19 deaths, 909,380 (88.1%) listed COVID-19 as the underlying cause of death (the disease or injury that initiated the chain of events leading to death), and 122,344 (11.9%) included COVID-19 elsewhere on the death certificate.
The Table 1 presents estimates of excess natural-cause deaths, reported COVID-19 deaths, and excess deaths reported to non-COVID-19 natural causes by Census division, metropolitan–nonmetropolitan category, and their combinations. Excess mortality is also shown in relative terms. Relative excess mortality is the ratio of excess natural-cause deaths to expected natural-cause deaths and can be interpreted as the percentage of observed deaths above (or below) the number expected based on pre-pandemic trends. Relative excess natural-cause mortality excluding reported COVID-19 deaths compares excess natural-cause deaths minus reported COVID-19 deaths to expected natural-cause deaths based on pre-pandemic trends. A positive percentage indicates that even after excluding reported COVID-19 deaths, observed natural-cause deaths exceeded the number expected. Nationally, relative excess natural-cause mortality was 18.2% (90% PI: 15.5 to 20.8%), and relative excess natural-cause mortality excluding reported COVID-19 deaths was 2.5% (90% PI: 0.2 to 4.8%). SI Appendix, Tables S1 and S2 present excess mortality estimates for states and selected counties. Dataset S1 visualizes relative excess mortality estimates for all 3,127 counties via an interactive RShiny App. In this Results section, all estimates refer exclusively to natural-cause deaths even when not directly stated.
Table 1.
Comparison of excess natural-cause deaths, reported COVID-19 deaths, and excess natural-cause deaths excluding reported COVID-19 deaths (also referred to as excess deaths reported to non-COVID-19 natural causes) in Census divisions and metropolitan–nonmetropolitan categories from March 2020 through August 2022
 COVID-19 deathsRatio of excess deaths to COVID-19 deaths Absolute excess mortality total Excluding COVID-19 Relative excess mortality total Excluding COVID-19 
  MedianPI (90%)MedianPI (90%)MedianPI (90%)MedianPI (90%)MedianPI (90%)
East North Central          
Large Metro77,9571.109(0.954 to 1.259)86,441(74,372 to 98,116)8,484(−3,584 to 20,159)0.170(0.143 to 0.197)0.017(−0.007 to 0.041)
Medium or Small Metro43,5821.095(0.936 to 1.244)47,702(40,803 to 54,227)4,120(−2,778 to 10,645)0.161(0.135 to 0.188)0.014(−0.009 to 0.037)
Non-Metro33,2121.163(1.001 to 1.313)38,634(33,236 to 43,617)5,422(24 to 10,405)0.168(0.141 to 0.194)0.024(0.000 to 0.046)
Total154,7511.116(0.963 to 1.263)172,714(149,063 to 195,387)17,963(−5,687 to 40,635)0.167(0.141 to 0.193)0.017(−0.005 to 0.040)
East South Central
Large Metro19,7561.136(0.983 to 1.275)22,448(19,414 to 25,180)2,692(−341 to 5,424)0.180(0.152 to 0.207)0.022(−0.003 to 0.045)
Medium or Small Metro27,9461.232(1.086 to 1.374)34,426(30,360 to 38,393)6,480(2,414 to 10,447)0.196(0.169 to 0.224)0.037(0.013 to 0.061)
Non-Metro29,2491.370(1.232 to 1.509)40,070(36,047 to 44,124)10,820(6,798 to 14,875)0.226(0.198 to 0.254)0.061(0.037 to 0.086)
Total76,9511.260(1.116 to 1.398)96,942(85,856 to 107,568)19,991(8,905 to 30,617)0.203(0.176 to 0.231)0.042(0.018 to 0.066)
Middle Atlantic
Large Metro114,2390.889(0.773 to 1.005)101,560(88,288 to 114,791)−12,680(−25,950 to 552)0.175(0.149 to 0.203)−0.022(−0.044 to 0.001)
Medium or Small Metro30,0540.896(0.739 to 1.046)26,920(22,217 to 31,445)−3,134(−7,836 to 1,391)0.133(0.108 to 0.159)−0.016(−0.038 to 0.007)
Non-Metro11,1151.152(0.992 to 1.316)12,810(11,029 to 14,631)1,695(−85 to 3,516)0.171(0.144 to 0.201)0.023(−0.001 to 0.048)
Total155,4080.909(0.780 to 1.035)141,297(121,185 to 160,867)−14,111(−34,222 to 5,459)0.165(0.138 to 0.192)−0.016(−0.039 to 0.007)
Mountain
Large Metro34,1361.307(1.179 to 1.436)44,620(40,247 to 49,010)10,484(6,111 to 14,874)0.237(0.209 to 0.266)0.056(0.032 to 0.081)
Medium or Small Metro27,0741.277(1.134 to 1.420)34,586(30,703 to 38,438)7,512(3,629 to 11,364)0.208(0.180 to 0.237)0.045(0.021 to 0.070)
Non-Metro14,0851.411(1.271 to 1.546)19,867(17,895 to 21,777)5,782(3,810 to 7,692)0.238(0.210 to 0.267)0.069(0.045 to 0.094)
Total75,2951.316(1.181 to 1.449)99,107(88,940 to 109,075)23,812(13,645 to 33,780)0.226(0.198 to 0.255)0.054(0.030 to 0.079)
New England
Large Metro21,8670.705(0.542 to 0.859)15,422(11,860 to 18,784)−6,445(−10,006 to −3,082)0.103(0.077 to 0.128)−0.043(−0.065 to −0.021)
Medium or Small Metro14,8000.770(0.583 to 0.944)11,396(8,623 to 13,969)−3,404(−6,176 to −831)0.098(0.072 to 0.123)−0.029(−0.052 to −0.007)
Non-Metro3,3521.217(0.908 to 1.550)4,080(3,045 to 5,194)728(−307 to 1,842)0.092(0.067 to 0.120)0.016(−0.007 to 0.042)
Total40,0190.771(0.591 to 0.944)30,858(23,645 to 37,787)−9,162(−16,373 to −2,231)0.099(0.074 to 0.124)−0.029(−0.051 to −0.007)
Pacific
Large Metro84,8291.249(1.098 to 1.393)105,972(93,150 to 118,194)21,143(8,321 - 33,365)0.190(0.163 to 0.217)0.038(0.015 to 0.061)
Medium or Small Metro33,2811.232(1.060 to 1.407)40,988(35,290 to 46,814)7,707(2,009 - 13,533)0.162(0.136 to 0.189)0.030(0.008 to 0.055)
Non-Metro5,8971.472(1.199 to 1.755)8,682(7,070 to 10,349)2,784(1,173 - 4,452)0.128(0.102 to 0.156)0.041(0.017 to 0.067)
Total124,0071.255(1.091 to 1.411)155,572(135,310 to 175,032)31,565(11,303 - 51,025)0.177(0.151 to 0.204)0.036(0.013 to 0.059)
South Atlantic
Large Metro97,3161.167(1.015 to 1.315)113,538(98,793 to 127,935)16,222(1,477 to 30,619)0.176(0.150 to 0.203)0.025(0.002 to 0.048)
Medium or Small Metro77,4221.221(1.068 to 1.381)94,562(82,663 to 106,899)17,140(5,241 to 29,477)0.177(0.151 to 0.205)0.032(0.010 to 0.056)
Non-Metro30,0131.593(1.434 to 1.747)47,818(43,029 to 52,447)17,806(13,016 to 22,434)0.234(0.206 to 0.263)0.087(0.062 to 0.113)
Total204,7511.250(1.097 to 1.401)255,998(224,672 to 286,765)51,248(19,921 to 82,014)0.185(0.159 to 0.212)0.037(0.014 to 0.061)
West North Central
Large Metro20,6471.073(0.912 to 1.230)22,154(18,837 to 25,398)1,507(−1,809 to 4,751)0.155(0.129 to 0.182)0.011(−0.012 to 0.034)
Medium or Small Metro21,1671.017(0.856 to 1.164)21,536(18,124 to 24,642)368(−3,042 to 3,475)0.152(0.125 to 0.178)0.003(−0.021 to 0.025)
Non-Metro19,3121.320(1.116 to 1.523)25,495(21,547 to 29,413)6,183(2,235 to 10,101)0.151(0.125 to 0.178)0.037(0.013 to 0.061)
Total61,1261.133(0.959 to 1.293)69,240(58,643 to 79,056)8,114(−2,482 to 17,930)0.153(0.126 to 0.178)0.018(−0.005 to 0.040)
West South Central
Large Metro61,1261.146(1.020 to 1.271)70,042(62,345 to 77,720)8,916(1,219 to 16,594)0.209(0.182 to 0.238)0.027(0.004 to 0.051)
Medium or Small Metro49,8951.262(1.150 to 1.367)62,987(57,395 to 68,188)13,092(7,500 to 18,293)0.260(0.231 to 0.287)0.054(0.030 to 0.077)
Non-Metro28,3951.419(1.281 to 1.550)40,292(36,361 to 44,007)11,897(7,966 to 15,612)0.238(0.210 to 0.266)0.070(0.046 to 0.094)
Total139,4161.242(1.123 to 1.361)173,206(156,620 to 189,721)33,790(17,204 to 50,305)0.232(0.205 to 0.260)0.045(0.023 to 0.069)
United States
Large Metro531,8731.095(0.956 to 1.231)582,386(508,207 to 654,749)50,512(−23,665 to 122,876)0.180(0.154 to 0.207)0.016(−0.007 to 0.039)
Medium or Small Metro325,2211.154(1.005 to 1.296)375,292(326,835 to 421,492)50,072(1,614 to 96,271)0.176(0.150 to 0.203)0.024(0.001 to 0.046)
Non-Metro174,6301.360(1.203 to 1.512)237,549(210,079 to 263,974)62,919(35,449 to 89,344)0.195(0.168 to 0.221)0.052(0.028 to 0.075)
Total1,031,7241.158(1.014 to 1.299)1,194,610(1,046,000 to 1,340,204)162,886(14,276 to 308,480)0.182(0.155 to 0.208)0.025(0.002 to 0.048)
The estimates of excess natural-cause deaths are also presented in terms of relative excess mortality, which is the ratio of excess natural-cause deaths compared to expected natural-cause deaths (e.g. where a ratio of 0.182 indicates that 18.2% more natural-cause deaths occurred than would have been expected based on pre-pandemic trends). Relative excess natural-cause mortality excluding reported COVID-19 deaths is the ratio of excess natural-cause deaths minus reported COVID-19 deaths compared to expected natural-cause deaths (e.g. where a ratio of 0.025 indicates that 2.5% more deaths reported to non-COVID-19 natural causes occurred than would have been expected based on pre-pandemic trends). Since medians are not additive, the sum of individual point estimates in this table may not exactly match totals.

Geographic Variation in Excess Mortality Reported to Non-COVID-19 Natural Causes.

The Fig. 1 shows ratios of excess deaths to reported COVID-19 deaths in each county and the probability that a county had more excess deaths than reported COVID-19 deaths. There was substantial within- and across-state heterogeneity in these estimates. Overall, there were 15.8 excess deaths reported to non-COVID-19 causes (90% PI: 1.4 to 29.9) for every 100 reported COVID-19 deaths. Regionally, this number was highest in the West (Mountain division: 31.6 deaths [90% PI: 18.1 to 44.9] and Pacific division: 25.5 deaths [90% PI: 9.1 to 41.1]), and the South (East South Central division: 26.0 deaths [90% PI: 11.6 to 39.8], South Atlantic division: 25.0 deaths [90% PI: 9.7 to 40.1], and West South Central division: 24.2 deaths [90% PI: 12.3 to 36.1]).
Fig. 1.
Ratios of excess natural-cause deaths to reported COVID-19 deaths across US counties from March 2020 through August 2022. In panel A, darker counties represent counties with higher ratios of excess natural-cause mortality to reported COVID-19 mortality. In panel B, darker counties represent counties where there is greater certainty that excess natural-cause deaths exceeded reported COVID-19 deaths during the period. Suppressed values reflect counties in which the cumulative number of reported COVID-19 deaths was less than 10 deaths.
The Fig. 2 depicts the relationship between relative excess mortality and relative reported COVID-19 mortality by county in Census divisions, with points above the diagonal axis indicating counties where excess deaths exceeded reported COVID-19 deaths. While there was heterogeneity within divisions, there were more excess deaths than reported COVID-19 deaths in all of them except the New England and Middle Atlantic divisions. In the New England division, there were 9,162 more reported COVID-19 deaths (90% PI: 2,231 to 16,373) than excess deaths, and in the Middle Atlantic division, there were 14,111 more reported COVID-19 deaths (90% PI: −5,459 to 34,222) than excess deaths.
Fig. 2.
Comparison of reported COVID-19 deaths and excess natural-cause deaths across US counties from March 2020 through August 2022. Each point represents a county and reflects its relative reported COVID-19 mortality (horizontal axis) and relative excess natural-cause mortality (vertical axis). Relative excess natural-cause mortality is calculated as the ratio of excess natural-cause deaths to expected natural-cause deaths based on pre-pandemic trends. Relative reported COVID-19 mortality compares reported COVID-19 deaths to expected natural-cause deaths based on pre-pandemic trends. Points above the 45° lines represent counties with more excess natural-cause deaths than reported COVID-19 deaths. Points below the line represent the opposite. The size of each point corresponds to the county’s population. Counties with fewer than 10,000 residents were excluded. The color of each point corresponds to the metropolitan–nonmetropolitan categories: large metropolitan counties (yellow), medium or small metropolitan counties (pink), and nonmetropolitan counties (purple).
Excess deaths exceeded reported COVID-19 deaths by different amounts across metropolitan–nonmetropolitan categories. For every 100 reported COVID-19 deaths, there were 36.0 excess deaths (90% PI: 20.3 to 51.2) reported to non-COVID-19 causes in nonmetropolitan counties. This estimate was 15.4 deaths (90% PI: 0.5 to 29.6) in medium and small metropolitan counties and 9.5 deaths (90% PI: -4.4 to 23.1) in large metropolitan counties. In absolute terms, most excess deaths reported to non-COVID-19 causes occurred outside of large metropolitan areas—either in nonmetropolitan counties (62,919 deaths [90% PI: 35,449 to 89,344]) or medium and small metropolitan counties (50,072 deaths [90% PI: 1,614 to 96,271]).
Stratifying each Census division by the metropolitan–nonmetropolitan categories, we found that excess deaths exceeded reported COVID-19 deaths in all but 4 of the 27 combinations. These were the large metropolitan counties and medium and small metropolitan counties in the New England and Middle Atlantic divisions. Relative excess mortality excluding reported COVID-19 deaths was higher in nonmetropolitan areas compared to the other metropolitan–nonmetropolitan categories in all divisions.

Temporal Variation in Excess Mortality Reported to Non-COVID-19 Natural Causes.

Gaps between excess deaths and reported COVID-19 deaths emerged early in the pandemic and persisted throughout the 30 mo of the study. SI Appendix, Table S3 provides estimates of the ratio of excess deaths to reported COVID-19 deaths by pandemic year. For every 100 reported COVID-19 deaths, there were 15.2 excess deaths reported to non-COVID-19 causes (90% PI: 3.8 to 26.5) in the first year, 15.5 deaths (90% PI: 0.1 to 31.1) in the second year, and 23.5 (90% PI: −44.3 to 85.5) in the final 6 mo. In absolute terms, an estimated 80,136 excess deaths were reported to non-COVID-19 causes (90% PI: 20,034 to 139,676) in the first year, 67,932 deaths (90% CI: 294 to 136,807) in the second year, and 15,305 deaths (90% CI: −28,894 to 55,760) in the final 6 mo. These individual estimates are medians and thus are not exactly additive to the total estimate for the 30 mo period.
The SI Appendix, Figs. S1–S3 visualize the monthly evolution of the ratio for cumulative excess deaths to cumulative reported COVID-19 in regions, divisions, and states, respectively. In the early months of the pandemic, excess deaths substantially exceeded reported COVID-19 deaths. The gap between excess deaths and reported COVID-19 deaths then narrowed partially, as exhibited by declining ratios. However, by the summer of 2021, the ratios once again increased in many areas.

Temporal Correlations between Reported COVID-19 Deaths and Excess Deaths Reported to Non-COVID-19 Natural Causes.

The Fig. 3 depicts monthly variation in excess deaths, reported COVID-19 deaths, and excess deaths reported to non-COVID-19 causes across combinations of Census divisions and metropolitan–nonmetropolitan categories. Excess deaths reported to non-COVID-19 causes exhibited similar temporal patterns to reported COVID-19 deaths, peaking during periods of high reported COVID-19 mortality and declining during the troughs between peaks. The Fig. 4 presents a statistical analysis of temporal correlations between the monthly time series for reported COVID-19 deaths and the monthly time series for excess deaths reported to non-COVID-19 causes.
Fig. 3.
Monthly variation in excess natural-cause deaths, reported COVID-19 deaths, and excess natural-cause deaths excluding reported COVID-19 deaths across Census divisions and metropolitan–nonmetropolitan categories from March 2020 through August 2022. The figure shows relative excess natural-cause mortality (i.e., the ratio of excess natural-cause deaths compared to expected natural-cause deaths based on pre-pandemic trends) using solid brown lines, relative reported COVID-19 mortality (i.e., the ratio of reported COVID-19 deaths compared to expected natural-cause deaths based on pre-pandemic trends) as dashed blue lines, and relative excess natural-cause deaths excluding reported COVID-19 deaths (i.e., the ratio of excess deaths reported to non-COVID-19 natural causes compared to expected natural-cause deaths) as dotted gray lines.
Fig. 4.
Cross-correlation plots examining temporal correlations between the monthly time series for reported COVID-19 deaths and the monthly time series for excess deaths reported to non-COVID-19 natural causes in Census divisions and metropolitan–nonmetropolitan categories from March 2020 through August 2022. Increases in reported COVID-19 deaths may correlate with increases (positive correlation coefficient) or decreases (negative correlation coefficient) in excess deaths reported to non-COVID-19 natural causes. The x axis indicates the time lag in months. Increases in reported COVID-19 deaths may correlate with changes in excess deaths reported to non-COVID-19 natural causes in the same month (lag 0), in future months (lag greater than 0), and/or in previous months (lag less than 0). Correlation coefficients that are outside the gray shaded area indicate significance at alpha = 0.01 and are colored red to assist with interpretation.
Across 7 of the 9 Census divisions and all metropolitan–nonmetropolitan categories, we observed significant positive correlation at lag 0 and/or lag −1. The areas where a positive correlation was observed at lag 0 included: large metropolitan areas and the East North Central, Middle Atlantic, Mountain, New England, and South Atlantic divisions. This indicates that in each of these areas, increases in reported COVID-19 deaths correlated with increases in excess deaths reported to non-COVID-19 causes during the same month. The areas where a positive correlation was observed at lag −1 included medium or small metropolitan areas and nonmetropolitan areas and the Mountain, Pacific, South Atlantic, and West North Central divisions. This indicates that in each of these areas, increases in reported COVID-19 deaths correlated with increases in excess deaths reported to non-COVID-19 causes in the month before the increases in reported COVID-19 deaths.
The second pattern that we observed was a significant negative correlation at lag 1 and/or 2 in 6 of the 9 Census divisions and all metropolitan–nonmetropolitan categories. The areas where a negative correlation was observed at lag 1 included large metropolitan areas and the New England, Pacific, and South Atlantic divisions. This indicates that in each of these areas, increases in reported COVID-19 deaths correlated with decreases in excess deaths reported to non-COVID-19 causes during the next month. The areas where a negative correlation was observed at lag 2 included: medium and small metropolitan and nonmetropolitan areas and the East South Central, Mountain, Pacific, South Atlantic, and West North Central divisions. This indicates that in each of these areas, increases in reported COVID-19 deaths correlated with decreases in excess deaths reported to non-COVID-19 causes that occurred 2 mo after increases in reported COVID-19 deaths. SI Appendix, Table S4 presents estimates of the magnitudes of each of the significant temporal correlations observed in the figure (i.e., the change in excess deaths reported to non-COVID-19 causes in a specific month associated with 100 additional reported COVID-19 deaths).

Discussion

In the present study, we estimated that approximately 1.2 million excess natural-cause deaths occurred in US counties during the first 30 mo of the pandemic. Nearly 163,000 of these excess natural-cause deaths were not reported to COVID-19. The relative gap between excess natural-cause mortality and reported COVID-19 mortality was largest in nonmetropolitan counties, the West, and the South. Contrary to prior literature which indicated that these gaps were mostly limited to the early months of the pandemic (31), we found nearly as many excess deaths reported to non-COVID-19 natural causes in the pandemic’s second year as the first year.
Excess natural-cause deaths could reflect unrecognized COVID-19 deaths, deaths related to health care delays or interruptions, and/or deaths caused by social and economic consequences of the pandemic. Emerging literature indicates that a large share of excess deaths reported to non-COVID-19 causes in the United States during the pandemic were unrecognized COVID-19 deaths (7, 9, 10). One study found that 84% of excess all-cause mortality during the pandemic could be attributed to SARS-CoV-2 infection (7), and another study identified significant temporal concordance between peaks in out-of-hospital reported COVID-19 deaths and excess deaths reported to non-COVID-19 natural causes in California (13).
In the present study, we examined temporal correlations between reported COVID-19 deaths and excess deaths reported to non-COVID-19 natural causes. In nearly all Census divisions and metropolitan–nonmetropolitan categories, we found that increases in reported COVID-19 deaths correlated with increases in excess deaths reported to non-COVID-19 natural causes in the same or prior month (positive correlation at a lag of 0 and/or −1 mo). In many Census Divisions and metropolitan–nonmetropolitan, we also observed that increases in reported COVID-19 deaths correlated with decreases in excess deaths reported to non-COVID-19 natural causes in the subsequent one or 2 mo (negative correlation at a lag of 1 and/or 2).
The temporal correlations we observed suggest that many excess deaths reported to non-COVID-19 natural causes during the first 30 mo of the pandemic in the United States were unrecognized COVID-19 deaths. Community-level awareness of COVID-19 mortality risk changed markedly with local peaks in reported COVID-19 deaths in ways that affected testing and surveillance, despite high awareness of the pandemic overall (32–34). If many of the excess deaths reported to non-COVID-19 natural causes were unrecognized COVID-19 deaths, we would expect to observe the following: a) a contemporaneous temporal correlation between these deaths and reported COVID-19 deaths, b) increases in these deaths prior to peaks in reported COVID-19 deaths while community-level awareness, testing, and surveillance was lower, and c) decreases in these deaths following increases in reported COVID-19 deaths as awareness, testing, and surveillance grew, leading to fewer unrecognized COVID-19 cases and greater suspicion among death investigators about COVID-19 as a potential cause of death. This awareness hypothesis aligns closely with the temporal correlations we observed.
Our study also suggests that health care delays and interruptions were not the primary cause of excess deaths reported to non-COVID-19 natural causes. Since hospitals often reached capacity during peaks in reported COVID-19 deaths, health care delays and interruptions were substantial after peaks (16, 17, 35). Thus, under this hypothesis, we would expect to observe increases in excess deaths reported to non-COVID-19 natural causes in the months after increases in reported COVID-19 deaths. We did not observe this pattern in any Census division or metropolitan–nonmetropolitan category. Despite this finding, health care delays and interruptions may have still contributed to other health effects during the pandemic, including to COVID-19 outcomes, and may have had differential impacts on specific subpopulations of patients (36).
Similar reasoning suggests a limited contribution of socioeconomic disturbances to excess deaths reported to non-COVID-19 natural causes among persons with no recent SARS-CoV-2 infection, as such deaths would not typically rise and fall with reported COVID-19 deaths. Some have also argued that stay-at-home restrictions and physical distancing efforts were primarily responsible for non-COVID-19 excess deaths. Our study did not examine deaths from external causes. However, for natural causes, the large number of non-COVID-19 excess deaths that we observed, especially the approximately 113,000 of these deaths that occurred outside of large metropolitan areas in counties where restrictions were typically more limited or had less impact, is inconsistent with this hypothesis (37–39).
Regions with large discrepancies between estimates of excess natural-cause deaths and reported COVID-19 deaths (e.g., nonmetropolitan areas, the West, and the South) may have experienced more unrecognized COVID-19 deaths due to more limited COVID-19 testing, a greater share of deaths outside of hospitals, and/or a greater reliance on elected death investigators (13, 40–42). Unlike medical examiners who have more extensive training in forensic pathology (43), elected death investigators such as coroners, sheriff-coroners, and justices of the peace often have limited and inconsistent training. Low funding for death investigation offices may have also contributed to inadequate resources for postmortem COVID-19 testing (11, 44). It is also possible that stigma and political beliefs about COVID-19 from death investigators or family members of the deceased influenced cause of death assignment in some settings (12). Prior research has shown that coroners disproportionately underreport opioid-related overdoses and that many deaths in police custody go unrecognized (45–48). Both of these findings highlight the potential for bias in death investigation practices to contribute to underreporting of stigmatized and/or politicized causes of death.
In the New England and Mid-Atlantic states, metropolitan areas experienced a unique pattern of reporting more COVID-19 deaths than excess natural-cause deaths. Several explanations may exist for this pattern. First, other causes of death (e.g., influenza) may have declined in these areas because of physical distancing, masking, and/or economic privilege in some counties that allowed many residents to work from home and avoid infection (49). If these causes of death declined, reported COVID-19 mortality would exceed excess natural-cause mortality. Second, such a finding could be partially explained by mortality selection (i.e., premature mortality from COVID-19 for some individuals who would have otherwise died several months later of another natural cause) (25). Last, some death investigators may have listed COVID-19 on death certificates when a person had COVID-19 but did not die from COVID-19 due to differences in state protocols. For example, COVID-19 death counts in Massachusetts included deaths within 60 d of a diagnosis until March 2022, which differed from the 30 d guidelines in other states (50). If overreporting of COVID-19 deaths did occur, however, it appears to have been infrequent and geographically limited.
The county-level estimates of excess deaths reported to non-COVID-19 natural causes generated in this study for each of the first 30 mo of the pandemic provide valuable insights about irregularities in the death investigation system. Such findings may help to a) inform public health policies and responses that rely on counts of reported COVID-19 deaths or require COVID-19 to be listed on death certificates for eligibility such as the FEMA funeral assistance program (51, 52), b) identify local death investigation offices that performed poorly during the pandemic and would benefit from additional infrastructure, training, and other resources, and c) provide evidence to support larger death investigation system reform to strengthen mortality surveillance in the context of future pandemic preparedness efforts (11, 44).
Future research should work to differentiate unrecognized COVID-19 deaths from other excess natural-cause deaths using hospitalization and other localized data. Studies should also explore potential differences in unrecognized COVID-19 deaths by individual factors such as age, gender, racialized identity, ethnicity, and socioeconomic status and by health system and death investigation system characteristics (6, 40). Studying these differences is critical for identifying populations with underestimated burdens of COVID-19 mortality (20, 30, 53). Moreover, evaluations of mortality attributable to the pandemic should not only consider reported COVID-19 deaths but also unrecognized COVID-19 deaths, other excess deaths from natural causes, and excess deaths from external causes. Consideration of each category of deaths is necessary to gain a complete understanding of the mortality effects of the pandemic.

Limitations.

Our study had several limitations. First, we could not directly distinguish between excess natural-cause deaths that were unrecognized COVID-19 deaths and those that were related to health care interruptions and/or social and economic consequences of the pandemic. Instead, we analyzed temporal correlations between categories of deaths to assess various hypotheses. Second, reported COVID-19 and natural-cause death counts were final in 2020 and 2021 but provisional in 2022. Mortality data in the United States are typically 75% complete after 8 wk, and most of the delayed deaths fall into the external causes category (54). Our data were accessed more than 12 mo after the final death in our study, suggesting they were nearly complete. Third, due to the study’s geographic granularity, our temporal analyses were conducted using monthly mortality data, which offer less power than weekly or daily death counts. Some temporal patterns such as lags of less than 1 mo were not detectable at this level.

Conclusion.

In this study, we used death certificate data to compare excess natural-cause deaths to reported COVID-19 deaths in counties across the United States. We found that many excess deaths were reported to non-COVID-19 natural causes and that this occurred not only in the early pandemic but throughout its first 30 mo. Our results suggest that many of these deaths were unrecognized COVID-19 deaths. Efforts to target resources to the communities most impacted by the pandemic and to prepare for future emergencies should consider geographic variation in the quality of mortality data and the potential for delayed or incomplete data to contribute to ineffective and inequitable responses.

Materials and Methods

Study Design.

We compared estimates of excess natural-cause deaths and reported COVID-19 deaths in counties each month from March 2020 through August 2022 through 5 steps:
(1)
We estimated the number of expected natural-cause deaths for each county–month if the pandemic had not occurred, based on pre-pandemic trends from January 2015 to December 2019.
(2)
We estimated the number of excess natural-cause deaths in each county–month by calculating the difference between observed natural-cause mortality and expected natural-cause mortality.
(3)
We estimated the number of excess deaths reported to non-COVID-19 natural causes (also referred to as excess natural-cause deaths excluding COVID-19 deaths) by calculating the difference between excess natural-cause deaths and reported COVID-19 deaths in each county–month.
(4)
We converted estimates of excess natural-cause mortality to relative excess natural-cause mortality to facilitate comparison across county–months with different populations and mortality burdens. Relative excess natural-cause mortality was calculated as the ratio of excess natural-cause deaths to expected natural-cause deaths and reflects the percentage of natural-cause deaths that occurred in a county–month above (or below) that which would have been expected based on pre-pandemic trends.
(5)
We also generated a relative excess mortality measure for relative excess mortality reported to non-COVID-19 natural causes (also referred to as relative excess natural-cause mortality excluding COVID-19 deaths). This was calculated as the ratio of excess natural-cause deaths minus reported COVID-19 deaths to expected natural-cause deaths and reflects the percentage of natural-cause deaths that occurred in a county–month after excluding reported COVID-19 deaths above (or below) that which would have been expected based on pre-pandemic trends.

Data.

We extracted monthly death counts at the county level from the CDC WONDER online tool. See SI Appendix, Supplementary Text for further details. We extracted natural-cause and reported COVID-19 death counts from the Multiple Cause of Death database using the provisional counts for January through August 2022 and the final counts for 2015 to 2021. We identified reported COVID-19 deaths using the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) code U07.1. Our definition of reported COVID-19 deaths included all death certificates in which there was any mention of COVID-19 (55).
To transform the number of deaths into rates, we used publicly available yearly county-level population estimates from the Census Bureau for 2010 to 2020 (56) and 2021 (57). To obtain monthly population counts and correct for the change of the Census base between the 2015 to 2019 and the 2020 to 2021 estimates, we smoothed the population estimates from July 2015 to July 2021 with a linear spline in time with 4 degrees of freedom before interpolating and extrapolating monthly population estimates through August 2022. We grouped counties into three metropolitan–nonmetropolitan categories (large metropolitan, medium or small metropolitan, and nonmetropolitan) based on the 2013 NCHS Rural-Urban Classification Scheme for Counties (58). We also grouped counties into 9 Census divisions. SI Appendix, Supplementary Text provides further details about these geographic classifications.

Statistical Methods.

We used a spatial Bayesian hierarchical model to predict the number of natural-cause deaths expected in each county–month based on pre-pandemic trends. Our methods replicate those in a previous study estimating excess all-cause mortality and are described there in more detail (3). The major adaptation we made to apply this model to natural-cause mortality was to use a more constraining prior on the ρ parameter of the AR1 process used to model the effect of time around a linear trend. A full description of the model is available in SI Appendix, Supplementary Text.
To assess temporal correlation between the monthly time series of reported COVID-19 deaths and excess natural-cause deaths excluding reported COVID-19 deaths, we fit ARIMA models to both series until reduction to white noise was achieved and then computed the cross-correlation function between the residuals of the two ARIMA models (59). We repeated this process separately for each pair of series by Census division and metropolitan–nonmetropolitan categories. Finally, we fit linear regression models with the number of excess deaths reported to non-COVID-19 natural causes by month as the dependent variable and the number of reported COVID-19 deaths as the explanatory variable including only the significant lags as identified by examining the cross-correlation function.
This study used deidentified publicly available data and was exempted from review by the Boston University Medical Center Institutional Review Board. The programming code was developed using R, version 4.1.0 (R Project for Statistical Computing), and Python, version 3.7.13 (Python Software Foundation), and can be accessed through Dataset S2. A coauthor (J.A.W.) conducted a full review of the project code written by the first author (E.P.) to audit its readability, reliability, and reproducibility.

Data, Materials, and Software Availability

Data used in the study are publicly available from the US Centers for Disease Control and Prevention and US Census Bureau (60, 61). A permanent repository with project replication code, further details about the data sources used, and estimates generated through the study is available at the following link: https://osf.io/sqdbk/ (62). Relative excess mortality estimates for all 3,127 counties are also visualized via an interactive RShiny App in Dataset S1.

Acknowledgments

The content is solely the responsibility of the authors and does not necessarily represent the official views of the study sponsors. We gratefully acknowledge receiving financial support in the form of grants 77521 from the Robert Wood Johnson Foundation to Dr. A.C.S.; R01-AG060115 to Dr. I.T.E., R01-AG060115-04S1 to Drs. I.T.E. and A.C.S., and K00-AG068431 to Dr. Y.-H.C. from the National Institute on Aging; P2C-HD041023 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development to Dr. E.W.-F.; P-6007864-2022 from the W.K. Kellogg Foundation to Dr. A.C.S.; CCF-2200052 from the National Science Foundation to Dr. A.C.S.; and T32HS013853 from the Agency for Healthcare Research and Quality to Ms. D.J.L. E.P. gratefully acknowledges the resources provided by the International Max Planck Research School for Population, Health and Data Science (IMPRS-PHDS). Finally, we wish to thank Paul Shaman from the University of Pennsylvania for his valuable feedback on the temporal correlation analysis.

Author contributions

E.P. and A.C.S. designed research; E.P., D.J.L., E.W.-F., Z.Z., J.A.W., R.R., Y.-H.C., K.H., S.H.P., I.T.E., M.M.G., and A.C.S. performed research; E.P., Z.Z., J.A.W., and R.R. analyzed data; and E.P., D.J.L., and A.C.S. wrote the paper with feedback from all authors.

Competing interests

The authors declare no competing interest.

Supporting Information

Appendix 01 (PDF)

References

1
H. Wang et al., Estimating excess mortality due to the COVID-19 pandemic: A systematic analysis of COVID-19-related mortality, 2020–21. Lancet 399, 1513–1536 (2022).
Crossref
PubMed
Google Scholar
2
T. Beaney et al., Excess mortality: The gold standard in measuring the impact of COVID-19 worldwide? J. R. Soc. Med. 113, 329–334 (2020).
Crossref
PubMed
Google Scholar
3
E. Paglino et al., Monthly excess mortality across counties in the United States during the COVID-19 pandemic, March 2020 to February 2022. Sci Adv 9, eadf9742 (2023).
Crossref
PubMed
Google Scholar
4
D. M. Weinberger et al., Estimation of excess deaths associated with the COVID-19 pandemic in the United States, March to May 2020. JAMA Intern. Med. 180, 1336–1344 (2020).
Crossref
PubMed
Google Scholar
5
C. A. Ackley et al., County-level estimates of excess mortality associated with COVID-19 in the United States. SSM Popul. Health 17, 101021 (2022).
Crossref
PubMed
Google Scholar
6
A. C. Stokes et al., COVID-19 and excess mortality in the United States: A county-level analysis. PLoS Med. 18, e1003571 (2021).
Crossref
PubMed
Google Scholar
7
W.-E. Lee et al., Direct and indirect mortality impacts of the COVID-19 pandemic in the United States, March 1, 2020 to January 1, 2022. Elife 12, e77562 (2023).
Crossref
PubMed
Google Scholar
8
S. H. Woolf, D. A. Chapman, R. T. Sabo, E. B. Zimmerman, Excess Deaths From COVID-19 and Other Causes in the US, March 1, 2020, to January 2, 2021. JAMA 325, 1786–1789 (2021).
Crossref
PubMed
Google Scholar
9
Y. Zhang, H. H. Chang, A. D. Iuliano, C. Reed, Application of Bayesian spatial-temporal models for estimating unrecognized COVID-19 deaths in the United States. Spat Stat. 50, 100584 (2022).
Crossref
PubMed
Google Scholar
10
A. D. Iuliano et al., Estimating under-recognized COVID-19 deaths, United States, March 2020-may 2021 using an excess mortality modelling approach. Lancet Reg. Health Am. 1, 100019 (2021).
PubMed
Google Scholar
11
A. C. Stokes, D. J. Lundberg, J. Bor, K. Bibbins-Domingo, Excess deaths during the COVID-19 pandemic: Implications for US death investigation systems. Am. J. Publ. Health 111, S53–S54 (2021).
Crossref
PubMed
Google Scholar
12
D. Bergin, R. Keller, Death certificates may obscure true toll on nation. USA Today, 27 December 2021, A1.
Google Scholar
13
Y.-H. Chen et al., Excess natural-cause deaths in California by cause and setting: March 2020 through February 2021. PNAS Nexus 1, pgac079 (2022).
Crossref
PubMed
Google Scholar
14
K. Bhaskaran et al., Factors associated with deaths due to COVID-19 versus other causes: Population-based cohort analysis of UK primary care data and linked national death registrations within the OpenSAFELY platform. Lancet Reg. Health Eur. 6, 100109 (2021).
Crossref
PubMed
Google Scholar
15
A. Abobaker, A. A. Raba, A. Alzwi, Extrapulmonary and atypical clinical presentations of COVID-19. J. Med. Virol. 92, 2458–2464 (2020).
Crossref
PubMed
Google Scholar
16
A. B. Friedman et al., Delayed emergencies: The composition and magnitude of non-respiratory emergency department visits during the COVID-19 pandemic. J. Am. Coll. Emerg. Phys. Open 2, e12349 (2021).
PubMed
Google Scholar
17
K. P. Hartnett et al., Impact of the COVID-19 pandemic on emergency department visits—United States, January 1, 2019-May 30, 2020. MMWR Morb. Mortal. Wkly. Rep. 69, 699–704 (2020).
Crossref
PubMed
Google Scholar
18
D. A. Glei, A. Stokes, M. Weinstein, Changes in mental health, pain, and drug misuse since the mid-1990s: Is there a link? Soc. Sci. Med. 246, 112789 (2020).
Crossref
PubMed
Google Scholar
19
J. S. Faust et al., Mortality from drug overdoses, homicides, unintentional injuries, motor vehicle crashes, and suicides during the pandemic, March-August 2020. JAMA 326, 84–86 (2021).
Crossref
PubMed
Google Scholar
20
L. E. Egede, R. J. Walker, Structural racism, social risk factors, and Covid-19—A dangerous convergence for Black Americans. N. Engl. J. Med. 383, e77 (2020).
Crossref
PubMed
Google Scholar
21
E. C. Matthay, K. A. Duchowny, A. R. Riley, S. Galea, Projected all-cause deaths attributable to COVID-19-related unemployment in the United States. Am. J. Publ. Health 111, 696–699 (2021).
Crossref
PubMed
Google Scholar
22
R. Chen et al., Racial and ethnic disparities in estimated excess mortality from external causes in the US, March to December 2020. JAMA Intern. Med. 182, 776–778 (2022).
Crossref
PubMed
Google Scholar
23
M. V. Kiang, R. A. Irizarry, C. O. Buckee, S. Balsari, Every body counts: Measuring mortality from the COVID-19 pandemic. Ann. Intern. Med. 173, 1004–1007 (2020).
Crossref
PubMed
Google Scholar
24
National Center for Health Statistics, “A reference guide for completing the death certificate for drug toxicity deaths” (Tech. Rep. 2, National Vital Statistics System, Hyattsville, MD, 2019).
Google Scholar
25
A. N. Luck, A. C. Stokes, K. Hempstead, E. Paglino, S. H. Preston, Associations between mortality from COVID-19 and other causes: A state-level analysis. PLoS One 18, e0281683 (2023).
Crossref
PubMed
Google Scholar
26
E. W. Lundstrom, C. P. Groth, J. E. Harrison, B. Hendricks, G. S. Smith, Excess US firearm mortality during the COVID-19 pandemic stratified by intent and urbanization. JAMA Netw. Open 6, e2323392 (2023).
Crossref
PubMed
Google Scholar
27
J. T. Chen, N. Krieger, Revealing the unequal burden of COVID-19 by income, race/ethnicity, and household crowding: US county versus zip code analyses. J. Publ. Health Manag. Pract. 27, S43–S56 (2021).
Crossref
PubMed
Google Scholar
28
E. Wrigley-Field, S. Garcia, J. P. Leider, D. Van Riper, COVID-19 mortality at the neighborhood level: Racial and ethnic inequalities deepened in Minnesota in 2020. Health Aff. 40, 1644–1653 (2021).
Crossref
PubMed
Google Scholar
29
J. M. McLaughlin et al., County-level predictors of Coronavirus Disease 2019 (COVID-19) cases and deaths in the United States: What happened, and where do we go from here? Clin. Infect. Dis. 73, e1814–e1821 (2021).
Crossref
PubMed
Google Scholar
30
D. J. Lundberg et al., COVID-19 mortality by race and ethnicity in US metropolitan and nonmetropolitan areas, March 2020 to February 2022. JAMA Netw. Open 6, e2311098 (2023).
Crossref
PubMed
Google Scholar
31
S. M. Moghadas, A. P. Galvani, The unrecognized death toll of COVID-19 in the United States. Lancet Reg. Health Am. 1, 100033 (2021).
PubMed
Google Scholar
32
B. Rader et al., Use of at-home COVID-19 Tests—United States, August 23, 2021–March 12, 2022. MMWR Morb. Mortal. Wkly. Rep. 71, 489–494 (2022).
Crossref
PubMed
Google Scholar
33
S. S. Coughlin, A. Yiǧiter, H. Xu, A. E. Berman, J. Chen, Early detection of change patterns in COVID-19 incidence and the implementation of public health policies: A multi-national study. Publ. Health Pract. (Oxf) 2, 100064 (2021).
Crossref
PubMed
Google Scholar
34
T. M. Holden et al., Geographic and demographic heterogeneity of SARS-CoV-2 diagnostic testing in Illinois, USA, March to December 2020. BMC Publ. Health 21, 1105 (2021).
Crossref
PubMed
Google Scholar
35
B. Boserup, M. McKenney, A. Elkbuli, The impact of the COVID-19 pandemic on emergency department visits and patient safety in the United States. Am. J. Emerg. Med. 38, 1732–1736 (2020).
Crossref
PubMed
Google Scholar
36
A. H. Gertz, C. C. Pollack, M. D. Schultheiss, J. S. Brownstein, Delayed medical care and underlying health in the United States during the COVID-19 pandemic: A cross-sectional study. Prev. Med. Rep. 28, 101882 (2022).
Crossref
PubMed
Google Scholar
37
T. Banerjee, A. Nayak, H. Zhao, A county-level study of the effects of state-mandated COVID-19 lockdowns on urban and rural restaurant visits using consumers’ cell phone geo-location data. J. Publ. Health 31, 249–258 (2023).
Crossref
Google Scholar
38
M. H. Haischer et al., Who is wearing a mask? Gender-, age-, and location-related differences during the COVID-19 pandemic. PLoS One 15, e0240785 (2020).
Crossref
PubMed
Google Scholar
39
D. F. Cuadros, A. J. Branscum, Z. Mukandavire, F. D. Miller, N. MacKinnon, Dynamics of the COVID-19 epidemic in urban and rural areas in the United States. Ann. Epidemiol. 59, 16–20 (2021).
Crossref
PubMed
Google Scholar
40
A. C. Stokes et al., Association of health care factors with excess deaths not assigned to COVID-19 in the US. JAMA Netw. Open 4, e2125287 (2021).
Crossref
PubMed
Google Scholar
41
N. E. Antonio-Villa et al., Socio-demographic inequalities and excess non-COVID-19 mortality during the COVID-19 pandemic: A data-driven analysis of 1 069 174 death certificates in Mexico. Int. J. Epidemiol. 13, 1711–1721 (2022), https://doi.org/10.1093/ije/dyac184.
Google Scholar
42
National Research Council, Strengthening Forensic Science in the United States: A Path Forward (The National Academies Press, 2009), https://doi.org/10.17226/12589.
Google Scholar
43
R. Hanzlick, D. Combs, Medical examiner and coroner systems: History and trends. JAMA 279, 870–874 (1998).
Crossref
PubMed
Google Scholar
44
Institute of Medicine, Medicolegal Death Investigation System: Workshop Summary (The National Academies Press, 2003).
Google Scholar
45
J. M. Feldman, S. Gruskin, B. A. Coull, N. Krieger, Quantifying underreporting of law-enforcement-related deaths in United States vital statistics and news-media-based data sources: A capture-recapture analysis. PLoS Med. 14, e1002399 (2017).
Crossref
PubMed
Google Scholar
46
J. M. Feldman, M. T. Bassett, Monitoring deaths in police custody: Public health can and must do better. Am. J. Publ. Health 111, S69–S72 (2021).
Crossref
PubMed
Google Scholar
47
M.-A. Neuilly, Sources of bias in death determination: A research note articulating the need to include systemic sources of biases along with cognitive ones as impacting mortality data. J. Forensic. Sci. 67, 2032–2039 (2022).
Crossref
PubMed
Google Scholar
48
GBD 2019 Police Violence US Subnational Collaborators, Fatal police violence by race and state in the USA, 1980-2019: A network meta-regression. Lancet 398, 1239–1255 (2021).
Crossref
PubMed
Google Scholar
49
M. G. Baker, Nonrelocatable occupations at increased risk during pandemics: United States, 2018. Am. J. Publ. Health 110, 1126–1132 (2020).
Crossref
PubMed
Google Scholar
50
Massachusetts Department of Public Health, “Department of Public Health updates COVID-19 death definition”, Mass.gov (2022). https://archives.lib.state.ma.us/server/api/core/bitstreams/e364265f-06f9-4634-a1df-410e1531f776/content. Accessed 15 June 2023.
Google Scholar
51
FEMA, “FEMA Tops $2 Billion of COVID-19 funeral assistance, announces new campaign to increase program awareness” (Tech. Rep. HQ-22-034, U.S. Department of Homeland Security, 2022).
Google Scholar
52
E. M. Webster, “FEMA funeral assistance for COVID-19” (Tech. Rep. IN11582, Congressional Research Service, 2023).
Google Scholar
53
G. A. Millett et al., Assessing differential impacts of COVID-19 on black communities. Ann. Epidemiol. 47, 37–44 (2020).
Crossref
PubMed
Google Scholar
54
M. R. Spencer, F. Ahmad, “Timeliness of death certificate data for mortality surveillance and provisional estimates” (Tech. Rep. 1, National Center for Health Statistics, CDC, Washington DC., 2016), p. 8, (National Vital Statistics System: Vital Statistics Rapid Release).
Google Scholar
55
National Center for Health Statistics, Excess deaths associated with COVID-19: Provisional death counts for Coronavirus Disease (COVID-19). CDC Web Archive. https://www.cdc.gov/nchs/nvss/vsrr/covid19/excess_deaths.htm. Accessed 13 October 2023.
Google Scholar
56
U.S. Census Bureau, Datasets: 2010–2020: Counties: Totals (U.S. Census Bureau). U.S. Census. https://www2.census.gov/programs-surveys/popest/datasets/2010-2020/counties/totals/. Accessed 13 October 2023.
Google Scholar
57
United States Census Bureau, Datasets: 2020–2021: Counties: Totals. U. S. Census. https://www.census2.gov/programs-surveys/popest/datasets/2020-2021/counties/totals/. Accessed 13 October 2023.
Google Scholar
58
National Center for Health Statistics, “NCHS urban-rural classification scheme for counties” (2017) (8 August 2023).
Google Scholar
59
P. J. Brockwell, R. A. Davis, Time Series: Theory and Methods (Springer Series in Statistics, Springer, ed. 2, 1987).
Crossref
Google Scholar
60
National Center for Health Statistics (NCHS), Multiple Cause of Death 1999–2020. CDC WONDER Online Database. http://wonder.cdc.gov/mcd-icd10-expanded.html. Accessed 13 October 2023.
Google Scholar
61
National Center for Health Statistics (NCHS), Multiple Cause of Death 2018 - present. CDC WONDER Online Database. https://wonder.cdc.gov/mcd-icd10-provisional.html. Accessed 13 October 2023.
Google Scholar
62
E. Paglino, A. C. Stokes, Replication Kit for “Excess natural-cause mortality in US counties and its association with reported COVID-19 deaths”. OSF. https://osf.io/sqdbk/. Deposited 31 December 2023.
Google Scholar

Information & Authors

Information

Published in

The cover image for PNAS Vol.121; No.6
Proceedings of the National Academy of Sciences
Vol. 121 | No. 6
February 6, 2024
PubMed: 38300867

Classifications

  1. Social Sciences
  2. Political Sciences

Copyright

Copyright © 2024 the Author(s). Published by PNAS. This open access article is distributed under Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND).

Data, Materials, and Software Availability

Data used in the study are publicly available from the US Centers for Disease Control and Prevention and US Census Bureau (60, 61). A permanent repository with project replication code, further details about the data sources used, and estimates generated through the study is available at the following link: https://osf.io/sqdbk/ (62). Relative excess mortality estimates for all 3,127 counties are also visualized via an interactive RShiny App in Dataset S1.

Submission history

Received: August 11, 2023
Accepted: December 6, 2023
Published online: February 1, 2024
Published in issue: February 6, 2024

Keywords

  1. excess mortality
  2. COVID-19
  3. geographic inequities
  4. rural health
  5. death investigation system

Acknowledgments

The content is solely the responsibility of the authors and does not necessarily represent the official views of the study sponsors. We gratefully acknowledge receiving financial support in the form of grants 77521 from the Robert Wood Johnson Foundation to Dr. A.C.S.; R01-AG060115 to Dr. I.T.E., R01-AG060115-04S1 to Drs. I.T.E. and A.C.S., and K00-AG068431 to Dr. Y.-H.C. from the National Institute on Aging; P2C-HD041023 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development to Dr. E.W.-F.; P-6007864-2022 from the W.K. Kellogg Foundation to Dr. A.C.S.; CCF-2200052 from the National Science Foundation to Dr. A.C.S.; and T32HS013853 from the Agency for Healthcare Research and Quality to Ms. D.J.L. E.P. gratefully acknowledges the resources provided by the International Max Planck Research School for Population, Health and Data Science (IMPRS-PHDS). Finally, we wish to thank Paul Shaman from the University of Pennsylvania for his valuable feedback on the temporal correlation analysis.
Author contributions
E.P. and A.C.S. designed research; E.P., D.J.L., E.W.-F., Z.Z., J.A.W., R.R., Y.-H.C., K.H., S.H.P., I.T.E., M.M.G., and A.C.S. performed research; E.P., Z.Z., J.A.W., and R.R. analyzed data; and E.P., D.J.L., and A.C.S. wrote the paper with feedback from all authors.
Competing interests
The authors declare no competing interest.

Notes

The estimates of excess natural-cause deaths are also presented in terms of relative excess mortality, which is the ratio of excess natural-cause deaths compared to expected natural-cause deaths (e.g. where a ratio of 0.182 indicates that 18.2% more natural-cause deaths occurred than would have been expected based on pre-pandemic trends). Relative excess natural-cause mortality excluding reported COVID-19 deaths is the ratio of excess natural-cause deaths minus reported COVID-19 deaths compared to expected natural-cause deaths (e.g. where a ratio of 0.025 indicates that 2.5% more deaths reported to non-COVID-19 natural causes occurred than would have been expected based on pre-pandemic trends). Since medians are not additive, the sum of individual point estimates in this table may not exactly match totals.
This article is a PNAS Direct Submission.

Authors

Affiliations

Eugenio Paglino https://orcid.org/0000-0001-8733-6460
Department of Sociology and Population Studies Center, University of Pennsylvania, Philadelphia, PA 19104
View all articles by this author
Dielle J. Lundberg
Department of Global Health, Boston University School of Public Health, Boston, MA 02118
Department of Health Systems and Population Health, University of Washington School of Public Health, Seattle, WA 98195
View all articles by this author
Elizabeth Wrigley-Field https://orcid.org/0000-0003-3489-4279
Department of Sociology and Minnesota Population Center, University of Minnesota, Minneapolis, MN 55455
View all articles by this author
Zhenwei Zhou https://orcid.org/0000-0002-9424-7826
Department of Biostatistics, Boston University School of Public Health, Boston, MA 02118
View all articles by this author
Joe A. Wasserman https://orcid.org/0000-0002-9705-1853
Research Triangle Institute, Research Triangle Park, NC 27709
View all articles by this author
Rafeya Raquib
Department of Global Health, Boston University School of Public Health, Boston, MA 02118
View all articles by this author
Yea-Hung Chen https://orcid.org/0000-0002-7949-0692
Department of Epidemiology and Biostatistics, University of California, San Francisco, CA 94158
View all articles by this author
Katherine Hempstead
Robert Wood Johnson Foundation, Princeton, NJ 08540
View all articles by this author
Samuel H. Preston https://orcid.org/0000-0001-7038-9689
Department of Sociology and Population Studies Center, University of Pennsylvania, Philadelphia, PA 19104
View all articles by this author
Irma T. Elo https://orcid.org/0000-0003-1858-8239
Department of Sociology and Population Studies Center, University of Pennsylvania, Philadelphia, PA 19104
View all articles by this author
M. Maria Glymour
Department of Epidemiology, Boston University School of Public Health, Boston, MA 02118
View all articles by this author
Andrew C. Stokes1 https://orcid.org/0000-0002-8502-3636 [email protected]
Department of Global Health, Boston University School of Public Health, Boston, MA 02118
View all articles by this author

Notes

1
To whom correspondence may be addressed. Email: [email protected].

Metrics & Citations

Metrics

Note: The article usage is presented with a three- to four-day delay and will update daily once available. Due to ths delay, usage data will not appear immediately following publication. Citation information is sourced from Crossref Cited-by service.


Altmetrics




Citations

Cite this article

  • E. Paglino,
  • D.J. Lundberg,
  • E. Wrigley-Field,
  • Z. Zhou,
  • J.A. Wasserman,
  • R. Raquib,
  • Y. Chen,
  • K. Hempstead,
  • S.H. Preston,
  • I.T. Elo,
  • M.M. Glymour,
  • & A.C. Stokes,
Excess natural-cause mortality in US counties and its association with reported COVID-19 deaths, Proc. Natl. Acad. Sci. U.S.A. 121 (6) e2313661121, https://doi.org/10.1073/pnas.2313661121 (2024).

Copied!

Copying failed.

Export the article citation data by selecting a format from the list below and clicking Export.

Cited by

    View Options

    View options

    PDF format

    Download this article as a PDF file

    DOWNLOAD PDF

    Login options

    Check if you have access through your login credentials or your institution to get full access on this article.

    Personal login Institutional Login

    Recommend to a librarian

    Recommend PNAS to a Librarian
    Save for later Item saved, go to cart

    Purchase options

    Purchase this article to access the full text.

    Single Article Purchase

    Excess natural-cause mortality in US counties and its association with reported COVID-19 deaths
    Proceedings of the National Academy of Sciences
    • Vol. 121
    • No. 6

    Restore content access

    Restore content access for purchases made as a guest

    Figures

    Tables

    Media

    Share

    Share

    Share article link

    Share on social media

    Further reading in this issue

    Trending