Researcher questions studies that suggest Covid-19 fatality rate is lower than expected

By George Q. Daley

The covid-19 pandemic is sowing death, suffering and economic devastation across the world. Containing its deadly march will require answers to many questions, most of which we are only beginning to chip away at. Among the most critical: How many people have been infected, and how many are going to die? Does the virus kill in 1 in 20 or 1 in a 1,000?

Two recent studies from California, using antibody tests designed to look for immune markers of previous infections, seem to suggest that the virus is much less deadly than many previously thought. But beware of these findings: They have not been vetted and should be recognized as such.

The two studies, conducted in Santa Clara and Los Angeles counties, provide a cautionary tale. The results suggest the number of people unknowingly infected may be 50- to 85-fold greater than previously suspected. If true, this rate of infection would imply a significantly lower mortality rate — around 0.12 to 0.2 percent, similar to the flu. That would be great news, but until proved otherwise, these findings should be treated as preliminary at best.

Here’s why: Neither of the two studies has been peer-reviewed. One of the papers appears on a pre-print server; the other one, no longer available online, was summarized in a news release.

The author suggests taking the results of unreviewed Covid-19 studies with a grain of salt.

Unfortunately, such unverified findings have been widely covered in the news and welcomed by those seeking to hasten the relaxation of current social distancing measures in their rush to reopen the economy. This could erode public confidence in current protective measures.

Given the disastrous impact of reigniting a lethal epidemic that could result in millions of deaths, preliminary findings must be handled with great care — no matter how alluring their implications. We call on policymakers and the public to take a deep breath and apply a healthy dose of skepticism to such studies.

We have several concerns about their methodologies. First, both studies used a commercial diagnostic test known to yield false positive results (detecting the presence of antibodies to the virus where there were none). The researchers, relying on a small set of control tests mostly performed by the tests’ manufacturer, claim a false positive rate of 0.5 percent, but independent analyses of the same test have yielded much higher false positive rates that could render their conclusions invalid.

Second, in one of the studies, researchers used Facebook to recruit volunteers. This raises concerns of self-selection bias, as these individuals made the effort to leave their homes and drive to a designated location for a blood test and thus might be motivated to seek testing for various reasons, including recent symptoms or engaging in high-risk activities that boost their risk for infection. For this reason alone, one might expect a misleading higher rate of positive tests.

Third, researchers used different types of blood to establish the accuracy of the diagnostic test than used for the volunteers. They compared frozen pre-pandemic processed blood samples with the fresh blood of volunteers obtained via finger pricks. But not all blood is the same; the biochemical differences between frozen and fresh blood could interfere with the accuracy of the reading.

These findings also obscure an indisputable fact: Hospital intensive care units in hot-spot cities have been overrun with seriously ill covid-19 patients. We do not see this even in bad flu seasons. That means relaxation of social distancing before more of the population is immune measures could end up overwhelming our hospitals.

Consider New York City, where health officials say up to 21 percent of residents may have been infected. While high, this number is far above the false positive rate and is plausible because the city has been an epicenter of the pandemic, with as many as 17,000 deaths. Using expert estimates that 40 to 70 percent of Americans will be infected with the new coronavirus, a death rate of 1 percent would translate into 1.4 million to 2.3 million people succumbing to the disease — a number closely aligned with early predictions of the likely death toll in the absence of mitigation measures such as a vaccine or social distancing. And that does not account for survivors left with serious heart, lung, kidney and neurological damage.

The calculus involved in designing policies that balance lives and livelihoods is astoundingly complex. One cannot be done at the expense of the other. These choices require hard evidence tempered with humility. Scientific evidence may not be the only factor in this equation, but it’s the most critical one.

Making decisions that involve human lives should be based on science, verified and vetted. In our understandable desire to return to normalcy today, we can ill afford to sacrifice our health and well-being tomorrow.
____________________

George Q. Daley is the dean of Harvard Medical School. Stephen Elledge is a professor and researcher at Brigham and Women’s Hospital. Galit Alter is a professor and researcher at the Ragon Institute and Massachusetts General Hospital. Michael Springer is an associate professor at Harvard Medical School.

Credit: Washington Post