Could over 20% of individuals living in New York City really already have been exposed to COVID-19?

Biology/Immunity Infection and Spread

A: Possibly, but there are a few things to consider about the sample of individuals included in the study (and all sero-surveys assessing past exposure to the infection via detection of antibodies targeted against SARS-CoV-2).

In this study, ~3000 individuals (~1300 in NYC specifically) were recruited to be tested for presence of antibodies targeted against COVID-19 (i.e., an indicator of past exposure) among those standing in line at selected grocery and big box stores.

Some have argued that the sample may over represent those that are out and about in public (i.e., those doing the grocery shopping for their family may have a greater likelihood of past exposure), which might exclude older home-bound individuals and/or those who are more strictly sheltering in place who have less opportunity for past exposure, meaning the 21.2% prevalence in NYC found by this study could be an over-estimation. On the other hand, health care workers who may have higher rates of exposure but may not be visiting the grocery store due to working, may be under-represented (i.e., this could lead to an under-estimate of the true prevalence). Children and teenagers are also likely to be under-represented among those sampled. In addition, the specific store selected for recruitment of individuals may only represent those living in the geographic area nearby the store (i.e., those within walking distance). The population living in proximity to one grocery store may not be representative of a given borough as a whole in terms of a range of factors that could influence exposure to COVID-19 (i.e., ability to work from home vs. outside the home, degree of overcrowding, and/or extent of intergenerational households).

This estimate of the prevalence of COVID-19 in NYC is notably higher than that which was observed in a study in Santa Clara County (~2.5-4%) and Los Angeles County (~2.8-5.6%). Sampling strategies were notably different in these two studies, compared to NYC. In Santa Clara County, individuals were recruited primarily from Facebook, with some attempt to recruit individuals living in both wealthy and poorer zip codes. In Los Angeles County, researchers attempted to obtain a more population-representative sample by contacting individuals from a random sample of email addresses and telephone numbers of residents.

As studies attempting to determine the underlying prevalence of COVID-19 in various geographic areas continue to emerge, the extent to which the study population is or is not representative of the overall population will be important to consider in the interpretation of results. As will differences in the sensitivity and specificity of tests being used across various studies.

While higher prevalence of COVID-19 in any given population than previously estimated is good news for mortality rates (i.e., more people may have had the disease and not have died than is currently estimated by the fatality rate among known cases), it is also still important to remember that the vast majority of local populations (as well as global) are still susceptible to this infection.

Last, there is still much to learn about what the presence of antibodies targeted against COVID-19 means as an indicator of immunity and the extent to which long-term immunity occurs after infection (particularly mild versus severe infections).

Stay tuned as more studies are carried out and we learn more about what is likely the true prevalence in the population as well as relevance of presence of antibodies for immunity over time! In the mean time, here is a good primer for thinking critically about such studies by STAT News.

Experts are raising concerns about the validity of some of the studies, cautioning policymakers not to put too much weight on any one finding.