Abstract
The seroprevalence of COVID-19 in the self-reported well US population is currently unknown. In September we tested a convenience sample of 63,106 life insurance applicants for COVID-19 and found a prevalence of 6.6%. This population was specifically selected because they were not being evaluated clinically but solely being tested for mortality risk. Using 2019 US census estimates this equals more than 11.1 million (bootstrap 95% CI: 10.8 – 11.5 million) asymptomatically infected patients, which is double the number of cases reported to CDC as of September 1st.
Introduction
With the imminent introduction of vaccines for COVID-19 it is important to identify patients that have previously been asymptomatically infected; they represent two-thirds of the infected population. They would be expected to have the least benefit from vaccination while also be the group at higher risk of adverse side effects. The prevalence of this group in the initial vaccine trials is not adequate to determine the residual risk of vaccine induced enhanced respiratory disease1. FDA has previously recognized this risk in their Direction to Vaccine Developers2.
In the setting of a pandemic, clinical decisions regarding testing and treatment are based on symptoms and the prevalence of the disease in the population. With SARS-CoV-2 the spectrum of symptom severity is wide. In the most severe cases testing may not be needed to determine the need for hospitalization, while in less severe illness testing becomes more necessary to determine the proper care setting and the need for isolation. A testing strategy is necessarily informed by knowledge of the disease prevalence in the population. In the case of SARS-CoV-2 these estimates have largely been based on the number of laboratory-confirmed cases reported to the CDC as per their mandate3. However, because SARS-CoV-2 is known to produce asymptomatic infection, and not all who are symptomatic obtain testing, these reported cases necessarily underestimate the total burden of infection in the population. Since SARS-CoV-2 antibodies persist for weeks or months after infection, the prevalence of these antibodies in a population sample may produce a better estimate of the overall cumulative prevalence of infection.
Methods
A national convenience sample of 63,106 life insurance applicants was tested for the presence of antibody to SARS-CoV-2. Blood was obtained as part of the usual underwriting requirements for life insurance. This population was selected because they were not being evaluated clinically to diagnosis an illness and were individuals that self-reported they were healthy.
Results
The rate of seroprevalence varied widely by state, was slightly higher for females than males, and was strongly negatively associated with age. To estimate the total burden of SARS-CoV-2 infections in the US the total 2019 estimated census population was multiplied by the US population proportion between the ages of 16 and 80 (75.5%). Then, the state-specific proportion of positive tests was applied from our sample. Confidence limits were estimated by generating 5000 bootstrap samples (with replacement) of our data and recalculating the total number of US cases. All statistical analyses were performed using R (version 3.6.1) and R-studio (version1.2.1335)4.
Characteristics of the study population are displayed in Table 1. The distribution by state is displayed in Table 2. Our seroprevalence-based estimate suggests that approximately twice the number of SARS-CoV-2 infections have occurred than have been reported to the CDC. This suggests a much more widespread pandemic, but with lower case rates of hospitalization, complications and deaths. Weaknesses of the study include an imbalanced representation of the US states, as well as the lack of samples from those under age 16 or over age 80. Even with these stated limitations the study validates the need for population wide surveillance; for the practicing physician the prevalence data provides a guide to relative risk of infection in their minimally or asymptomatic patient population.
Characteristics of study population by SARS-CoV-2 antibody status. Numeric values shown as median [IQR]. 1 p-value by Pearson Chi-square test.
While the debate continues about the degree and durability of protection afforded by prior infection the concern about adverse events may discourage many patients from vaccination. With the limited availability of vaccine it is important to preselect the population at highest risk that is likely to have the greatest benefit. Individuals that have had SARS-CoV-2 or that are antibody positive should defer vaccination until the vaccine is more widely available.
Additional information
Statement of conflicting interest
Neither Robert Stout, PhD nor Steven Rigatti, MD has any competing interest. Both authors have complete access to all the data in the study and take responsibility for the integrity of the data and analysis.
Western IRB reviewed the study under the Common Rule and applicable guidance and determined it is exempt under 45 CFR § 46.104(d)(4) using de-identified study samples for epidemiologic investigation.
This article has been submitted and approved as a preprint on medRxiv5.https://medrxiv.org/cgi/content/short/2020.11.10.20215145v1.
Data Availability
Data will be available to non-profit organizations.
Statement of author contribution
Robert Stout directed the testing of samples, combined the demographic data and then deleted all identifiable personal information. Reviewed all patient data for completeness and prepared the initial prevalence estimates.
Steve Rigatti Analyzed the prevalence for the data set statistically prepared the boot-strap adjusted individual state prevalence. He prepared the estimates for national and state prevalence of asymptomatic COVID-19 infection. Both authors prepared the text
Footnotes
Revised at the request of Editor
