Characteristics of Persons Who Died with COVID-19 — United States, February 12–May 18, 2020

Case-based surveillance

Demographic and clinical data about COVID-19 cases are reported to CDC from 50 states, the District of Columbia, New York City, and U.S. territories using a standardized case-report form (case-based surveillance) or in aggregate. Data on 52,166 deaths from 47 jurisdictions among persons with laboratory-confirmed COVID-19 were reported individually to CDC via case-based surveillance during February 12–May 18, 2020. Among the 52,166 decedents, 55.4% were male, 79.6% were aged ≥65 years, 13.8% were Hispanic/Latino (Hispanic), 21.0% were black, 40.3% were white, 3.9% were Asian, 0.3% were American Indian/Alaska Native (AI/AN), 0.1% were Native Hawaiian or other Pacific Islander (NHPI), 2.6% were multiracial or other race, and race/ethnicity was unknown for 18.0%. (Table 1). Median decedent age was 78 years (interquartile range (IQR) = 67–87 years). Because information about underlying medical conditions was missing for the majority of these decedents (30,725; 58.9%), data regarding medical conditions were not analyzed further using the case-based surveillance data set. Because most decedents reported to the supplementary data program were also reported to case-based surveillance, no statistical comparisons of the decedent characteristics between the data sets were made.

Supplemental surveillance

To collect more complete data on race/ethnicity, selected underlying medical conditions* by age, and clinical course, CDC solicited supplementary information from medical charts and death certificates of decedents with laboratory-confirmed COVID-19 from state, territorial, and local public health departments. The supplementary data request also sought information on locations of death, which is not collected routinely on the CDC case-report form. Among 56 public health departments contacted by CDC, 16^† provided supple­mentary data on 10,647 COVID-19 deaths that occurred during February 12–April 24, 2020.

Among the 10,647 COVID-19 decedents for whom supple­mentary data were collected, 60.6% were male, 74.8% were aged ≥65 years, 24.4% were Hispanic, 24.9% were black, 35.0% were white, 6.3% were Asian, 0.1% were AI/AN, 0.1% were NHPI, 2.9% were multiracial or other race, and race/ethnicity was unknown for 6.3% (Table 1). Decedent age varied by race and ethnicity; median age was 71 years (IQR = 59–81 years) among Hispanic decedents, 72 years (IQR = 62–81 years) among all nonwhite, non-Hispanic decedents, and 81 years (IQR = 71–88 years) among white decedents. The percentages of Hispanic (34.9%) and nonwhite (29.5%) decedents who were aged <65 years were more than twice those of white decedents (13.2%) (Figure).

At least one underlying medical condition was reported for 8,134 (76.4%) of decedents for whom sup­plementary data were collected, including 83.1% of decedents aged <65 years. Overall, the most common underlying medical conditions were cardiovascular disease (60.9%), diabetes mellitus (39.5%), chronic kidney disease (20.8%), and chronic lung disease (19.2%) (Table 2). Among decedents aged <65 years, 83.1% had one or more underlying medical conditions. Among decedents aged ≥85 years, 69.5% had one or more underlying medical conditions. Diabetes was more common among decedents aged <65 years (49.6%) than among those aged ≥85 years (25.9%).

Among decedents for whom supplementary data were reported, 8,976 (84.3%) were hospitalized. Among 3,021 (28.4%) with dates of illness onset and death reported, the median interval from illness onset to death was 10 days (IQR = 6–15 days); among 7,794 decedents with hospital admission and death dates, the median interval from hospital admission to death was 5 days (IQR = 3–8 days). Among the decedents, 62.0% died in hospitals. By age group, the largest percentage who died in the emergency department (6.8%) or at home (1.0%) was aged <65 years (combined total = 7.8%), and decreased with increasing age group, whereas the percentage who died in long-term care facilities increased with increasing age and was highest among decedents aged ≥85 years (12.6%).

Among the decedents during February 12–April 24, 2020, for whom supplementary information was provided, 9,997 (93.9%) resided in New York City, New Jersey, or the state of Washington, three areas with early widespread circulation of SARS-CoV-2; the median age among decedents in these three jurisdictions was 75 years, (IQR = 64–84 years). The median age among decedents residing in the other 13 jurisdictions was similar (78 years, [IQR = 68–85 years]).

Discussion

Using national case-based surveillance and supplementary data reported from 16 jurisdictions, characteristics of >10,000 decedents with laboratory-confirmed COVID-19 were described. More than one third of Hispanic decedents (34.9%) and nearly one third (29.5%) of nonwhite decedents were aged <65 years, but only 13.2% of white decedents were aged <65 years. Consistent with reports describing the characteristics of deaths in persons with COVID-19 in the United States and China (2–5), approximately three fourths of decedents had one or more underlying medical conditions reported (76.4%) or were aged ≥65 years (74.8%). Among reported underlying medical conditions, cardiovascular disease and diabetes were the most common. Diabetes prevalence among decedents aged <65 years (49.6%) was substantially higher than that reported in an analysis of hospitalized COVID-19 patients aged <65 years (35%) and persons aged <65 years in the general population (<20%) (5–7). Among decedents aged <65 years, 7.8% died in an emergency department or at home; these out-of-hospital deaths might reflect lack of health care access, delays in seeking care, or diagnostic delays. Health communications campaigns could encourage patients, particularly those with underlying medical conditions, to seek medical care earlier in their illnesses. Additionally, health care providers should be encouraged to consider the possibility of severe disease among younger persons who are Hispanic, nonwhite, or have underlying medical conditions. More prompt diagnoses could facilitate earlier implementation of supportive care to minimize morbidity among individuals and earlier isolation of contagious persons to protect communities from SARS-CoV-2 transmission.

The relatively high percentages of Hispanic and nonwhite decedents aged <65 years were notable. The median age of nonwhite persons (31 years) in the United States is lower than that of white persons (44 years); these differences might help explain the higher proportions of Hispanic and nonwhite decedents among those aged <65 years. The median ages among Hispanic and nonwhite decedents (71 and 72 years, respectively) were 9–10 years lower than that of white decedents (81 years). However, the percentage of Hispanic decedents aged <65 years (33.9%) exceeded the percentage of Hispanic persons aged <65 years in the U.S. population (20%); the percentage of nonwhite COVID-19 decedents aged <65 years (40.2%) also exceeded the overall percentage of nonwhite decedents aged <65 years (23%) in the U.S. population (8). Further study is needed to understand the reasons for these differences. It is possible that rates of SARS-CoV-2 transmission are higher among Hispanic and nonwhite persons aged <65 years than among white persons; one potential contributing factor is higher percentages of Hispanic and nonwhite persons engaged in occupations (e.g., service industry) or essential activities that preclude physical distancing (9). It is also possible that the COVID-19 pandemic disproportionately affected communities of younger, nonwhite persons during the study period (10). Although these data did not permit assessment of interactions between race/ethnicity, underlying medical conditions, and nonbiologic factors, further studies to understand and address these racial/ethnic differences are needed to inform targeted efforts to prevent COVID-19 mortality.

The findings in this report are subject to at least five limitations. First, despite >90% completeness for age and race/ethnicity variables in the supplementary data set, the proportion of missing data for some variables, such as underlying medical conditions, clinical course, and race/ethnicity in case-based surveillance, and location of death, was higher than that for other variables; accordingly, the proportions reported for these variables should be considered minimum proportions rather than robust estimates. Second, reporting practices varied by jurisdiction, and several states bundled underlying medical conditions into organ system–specific categories (e.g., hypertension was included as cardiovascular disease) or did not code specifically for a given condition (e.g., immunosuppression was only specifically coded in 10 of the jurisdictions). These differences in reporting structure precluded evaluations of specific conditions other than diabetes using the entire data set. Third, generalizability of the findings from either data set to all deaths among persons with COVID-19, either within the individual jurisdictions or across the United States, is unknown; COVID-19 testing practices for decedents might differ among jurisdictions. Fourth, information from the supplementary data set provides additional insight into decedent demographic and clinical characteristics; however, these data are a convenience sample from 16 public health jurisdictions. Therefore, because the age-race structure of the underlying population is not known, age-standardized mortality rates could not be calculated. Although more than 90% of decedents resided in just three jurisdictions, and most are represented in case-based surveillance, they represent a subset of deaths reported during this period. Therefore, neither calculations of mortality rates nor statistical comparisons between the demographic characteristics of the decedents with available supplementary data and those from case-based surveillance were possible. Finally, these data were collected during a period before dexamethasone was shown to reduce deaths among ventilated patients; implementation of dexamethasone and other therapeutics, as well as shifts in the ages of patients and geographic locations of cases might affect the generalizability of these data to the current period.

Despite these limitations, this report provides more detailed demographic and clinical information on a subset of approximately 10,000 decedents with laboratory-confirmed COVID-19. Most decedents were aged >65 years and had underlying medical conditions. Compared with white decedents, more Hispanic and nonwhite decedents were aged <65 years. Additional studies are needed to elucidate associations between age, race/ethnicity, SARS-CoV-2 infection, disease severity, underlying medical conditions (especially diabetes), socioeconomic status (e.g., poverty and access to health care), behavioral factors (e.g., ability to comply with mitigation recommendations and maintain essential work responsibilities), and out-of-hospital deaths. Regional and state level efforts to examine the roles of these factors in SARS-CoV-2 transmission and COVID-19-associated deaths could lead to targeted, community-level, mortality prevention initiatives. Examples include health communication campaigns targeted towards Hispanics and nonwhite persons aged <65 years. These campaigns could encourage social distancing and the need for wearing cloth face coverings in public settings. In addition, health care providers should be encouraged to consider the possibility of disease progression, particularly in Hispanic and nonwhite persons aged <65 years and persons of any race/ethnicity, regardless of age, with underlying medical conditions, especially diabetes.