October 27th, 2021Erika Tyagi, Neal Marquez, and Joshua Manson

A crisis of undertesting: how inadequate COVID-19 detection skews the data and costs lives

Earlier this month, our team co-authored an article in the Journal of the American Medical Association on our findings that, during the first year of the pandemic, the COVID-19 infection rate for people incarcerated in state and federal prisons was 3.3 times higher than the rate for the U.S. population as a whole, and the COVID-19 death rate was 2.5 times higher.

These disparities are stark but not surprising — in an earlier study, we found that, in the first months of the pandemic, incarcerated people faced even more disproportionate infection and death rates.

There is reason to believe, however, that actual outcomes have been far worse than these data reveal. That is because calculating infection rates that reflect the true prevalence of COVID-19 requires adequate testing. If tests are not widely administered in prisons and jails, and, by many accounts, they have not been, then infections will go undetected. As a result, infection and death rates will appear lower than they actually are.

Public health experts have noted that frequent and widespread testing is critical to containing outbreaks — especially in congregate facilities where just one case can quickly spread throughout a facility. In these settings, quickly identifying and isolating positive cases is necessary to prevent outbreaks from erupting. This requires testing individuals who are first entering facilities (upon intake or transfer) and individuals who are presenting symptoms, but it also requires regular surveillance testing of asymptomatic individuals and those who may have been in contact with someone who tested positive.

In this report, we break down three important public health metrics — testing rates, test positivity rates, and case fatality rates — that provide critical context to officially reported infection and death data and reveal just how unreliable reported infection and death data may be. These three metrics suggest that, in many places, true infection and death numbers may be much higher than those officially reported.

Click here to read the full report.