key: cord-0716402-zmpkj264
authors: Alroy, Karen A; Crossa, Aldo; Dominianni, Christine; Sell, Jessica; Bartley, Katherine; Sanderson, Michael; Fernandez, Steven; Levanon Seligson, Amber; Lim, Sungwoo; Wang, Shu Meir; Dumas, Sarah E; Perlman, Sharon E; Konty, Kevin; Olson, Donald R; Gould, L Hannah; Greene, Sharon K
title: Population-Based Estimates of Coronavirus Disease 2019 (COVID-19)–like Illness, COVID-19 Illness, and Rates of Case Ascertainment, Hospitalizations, and Deaths—Noninstitutionalized New York City Residents, March–April 2020
date: 2021-01-18
journal: Clin Infect Dis
DOI: 10.1093/cid/ciab038
sha: 3e885c103fb9864cece1a1dfd324e8cfb67f77a0
doc_id: 716402
cord_uid: zmpkj264

Using a population-based, representative telephone survey, ~930 000 New York City residents had COVID-19 illness beginning 20 March–30 April 2020, a period with limited testing. For every 1000 persons estimated with COVID-19 illness, 141.8 were tested and reported as cases, 36.8 were hospitalized, and 12.8 died, varying by demographic characteristics.

Using a population-based, representative telephone survey, ~930 000 New York City residents had COVID-19 illness beginning 20 March-30 April 2020, a period with limited testing. For every 1000 persons estimated with COVID-19 illness, 141.8 were tested and reported as cases, 36.8 were hospitalized, and 12.8 died, varying by demographic characteristics.

Keywords. COVID-19; burden; population-based survey; New York City; mortality.

New York City (NYC) was an early epicenter of the United States' coronavirus disease 2019 (COVID-19) outbreak. COVID-19 cases (ie, laboratory-confirmed COVID-19 diagnoses reported to the NYC Health Department) do not reflect the total COVID-19 burden of illness [1] . We conducted a populationbased survey [2] to characterize the burden of COVID-19 during the initial phase of the epidemic in NYC, when healthcare and testing capacity were limited and just before the New York State Governor issued an executive order closing all nonessential businesses. We estimated percentage and number of NYC residents with COVID-19-like illness (CLI), and the subset with COVID-19 illness. We used COVID-19 surveillance and Vital Statistics data to calculate rates of COVID-19 case ascertainment, hospitalizations, and deaths per 1000 residents with COVID-19 illness.

We added questions to the NYC Community Health Survey (CHS) [3] . In this population-based, representative, annual, cross-sectional landline and cellular telephone survey, noninstitutionalized NYC adults (≥18 years) were interviewed in English, Spanish, Russian, and Chinese and asked about the adult survey respondent or a randomly selected child in their household. Questions included demographics, new COVID-19 symptoms 30 days or less before interview, and illness onset date. The CHS methodology is described elsewhere [3] ; data were additionally weighted by person-time (Supplementary Methods).

The percentage of NYC residents with CLI was estimated using a NYC Health Department definition, defined as reporting 1 or more of the following new symptoms 30 days or less before interview: cough, subjective or measured fever (≥100.4°F or 38.0°C), shortness of breath/difficulty breathing, sore throat, loss of taste, and loss of smell. For interviews conducted 20 March-29 May 2020, we categorized people as having CLI if symptoms began during 20 March-30 April 2020. To account for people with CLI but not COVID-19, we used emergency department (ED) syndromic surveillance chief complaint and discharge diagnosis data [4] . Persons who sought ED care for CLI (Supplementary Table 1 ) were matched to persons reported via electronic laboratory reporting with a positive severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) polymerase chain reaction (PCR) test. We calculated the percentage of people with CLI who visited NYC EDs, were tested, and had a positive test. This percentage was multiplied by the weighted number of survey respondents with CLI to estimate the citywide number of residents with COVID-19 illness. Uncertainty intervals (UIs) were calculated, accounting for variability when combining this percentage with population estimates using Monte Carlo replicates [5] to simulate this product's distribution (Supplementary Methods).

We estimated case ascertainment, hospitalization, and death rates per 1000 residents with COVID-19 illness using reportable disease and Vital Statistics data. Institutionalized individuals who resided or died in congregate living facilities (eg, nursing homes, shelters, correctional settings) were excluded from all analyses. To align the survey data's illness onset timing with administrative data, we applied time lags between illness onset, date of diagnosis, hospitalization, and death.

We used 3 death definitions: decedents with confirmatory laboratory evidence of COVID-19 ("confirmed deaths"), decedents with "COVID-19" or related terminology on the death certificate for cause of death but without laboratory confirmation within 60 or fewer days of death ("probable deaths"), and the estimated number of deaths beyond seasonally expected levels ("excess deaths") (Supplementary Methods) [6] . Excess deaths represent an upper possible boundary of deaths caused by COVID-19 illness, although this might reflect deaths attributable to reduced care seeking and non-COVID-19 causes. Analyses were conducted using SAS version 9.4.1 (SAS Institute), SUDAAN version 11.0.1 (RTI International), and R version 3.5.2 (R Foundation for Statistical Computing).

Estimates accounted for complex sample design. We conducted 2-sided t tests of survey data. Small sample sizes prohibited age adjustment of stratified data. The CHS annual cooperation and response rates were calculated (Supplementary Methods).

During 20 March-29 May 2020, we interviewed 2757 respondents (Supplementary Figure 1 ). An estimated 13.8% (95% confidence interval [95% CI]: 11.4%, 16.7%) had CLI with onset during 20 March-30 April 2020 (Supplementary Table 2 ). The overall age-adjusted estimate was 13.6% (95% CI: 11.2%, 16.2%). Children aged 17 years or younger (8.3%; 95% CI: 5.1%, 13.2%) and adults aged 65-74 years (5.8%; 95% CI: 2.7%, 12.4%) and 75 years or older (3.5%; 95% CI: 2.0%, 6 

We used population-based survey data to estimate that more than 900 000 NYC residents had COVID-19 illness beginning during 20 March-30 April 2020. Our case ascertainment rate was 14.2%, meaning that for each confirmed COVID-19 case, there were an estimated 7 residents with COVID-19 illness. Asymptomatic SARS-CoV-2 infections represent as many as 40-45% of all infections [7] and would not be detected by this study. Antibody testing captures asymptomatic and symptomatic cases, so it follows that our estimate is lower than that from a serologic study of NYC residents during 23 March-1 April (11.9 infections/case) [8] .

The COVID-19 outbreak has illuminated substantial health inequities [9] . No differences were noted in CLI by race/ethnicity, an unexpected finding considering Black and Latino/a residents are overrepresented among essential workers and multigenerational households [9, 10] . Our point estimates of hospitalization and death rates among Black and Latino/a persons, however, were highest despite wide UIs. Systemic racism and social determinants of health, including access to care, underlying health conditions, and age distribution differences, might contribute to varying disease severity across NYC populations [9] .

COVID-19-like illness definition accuracy could not be assessed; however, we calculated a similar CLI percentage, 13.9% (95% CI: 11.4%, 16.7%), using the Council of State and Territorial Epidemiologists definition (Interim-20-ID-02) with our survey data [11] . For specificity, our CLI definition did not include rare COVID-19 manifestations (eg, neurologic or multisystem inflammatory symptoms). Institutionalized adults, who represent populations with substantial COVID-19 burden, were not included, likely decreasing our estimates [12, 13] .

Potential nonresponse bias could not be assessed. NYC residents who died outside of NYC were not reflected in the Vital Statistics data. We assumed the same percentage of test positivity among people with CLI who received a SARS-CoV-2 test and those who did not visit an ED, likely inflating our estimates, because people with severe illness were more likely to seek emergency care. Small sample sizes in stratified data limited point estimate comparisons and age adjustments. Lastly, these data did not contribute to real-time situational awareness because of the time needed to collect survey data, calculate survey weights, and estimate time lags between illness onset, date of diagnosis, hospitalization, and death.

Burden of disease studies help quantify how many people became sick during an outbreak and the differential burden among populations. The NYC Health Department rapidly added questions to an existing general health surveillance survey to capture populationbased data on symptoms and healthcare seeking during a period of limited testing. When responding to future emerging infectious diseases, adding questions to an existing survey platform can aid in measuring the burden of illness when testing is limited. We recommend asking about illness onset in the prior calendar month instead of the past 30 days to simplify calculating the percentage and number of residents with new symptoms. An ongoing serosurvey linked to this population-based survey will aid in more completely characterizing the direct and indirect effects of the COVID-19 outbreak; findings are forthcoming.

Substantial underestimation of SARS-CoV-2 infection in the United States

Case fatality rates based on population estimates of influenza-like illness due to novel H1N1 influenza

Advancing the use of emergency department syndromic surveillance data

Annual estimates of the burden of seasonal influenza in the United States: a tool for strengthening influenza surveillance and preparedness

Preliminary estimate of excess mortality during the COVID-19 outbreak

Prevalence of asymptomatic SARS-CoV-2 infection: a narrative review

Seroprevalence of antibodies to SARS-CoV-2 in six sites in the United States

The disproportionate impact of COVID-19 on racial and ethnic minorities in the United States

Differential occupational risk for COVID-19 and other infection exposure according to race and ethnicity

Interim COVID-19 position statement

COVID-19 outbreak

Flattening the curve for incarcerated populations-Covid-19 in jails and prisons

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.