The Corner

Science & Tech

The Problem with New York City’s COVID-19 Death-Rate Estimates

A medical worker makes a phone call at Maimonides Medical Center during the outbreak of the coronavirus in Brooklyn, N.Y., April 14, 2020. (Caitlin Ochs/Reuters)

More on the continuing saga (see here, here, here and here) of the COVID-19 mortality rate — specifically, on why it is so hard to get accurate statistics, notwithstanding that these statistics are essential to decisions about reopening the economy.

Those of us who have been watching the daily numbers closely could not help but notice the dark cloud that drifted Tuesday over what was otherwise cautiously optimistic news. The number of coronavirus cases seems to be dropping, but deaths are suddenly spiking. Why?

It would overstate the matter to say that the tally of new cases is “plummeting,” but the drop has been noticeable: from a level of over 30,000 new cases per day from April 6 through April 11 (and, on three of those days, over 33,000 new cases), we’ve been down to about 27,000 on each of the three days since Sunday.

Initially, deaths also seemed to be dropping markedly: from around 2,000 per day from April 7 through April 11, down to about 1,500 on Sunday and Monday. Then, suddenly, deaths shot up on Tuesday, to 2,407, by far the highest one-day total yet (surpassing the previous high of 2,035 recorded on April 10).

What gives? Well, the main problem right now is New York. As governor Andrew Cuomo noted yesterday, although daily deaths seemed to be edging downward, below 700 on Monday for the first time in a week, they spiked up over 800 again on Tuesday. (Gov. Cuomo is nevertheless heartened by a decrease in hospitalizations, which will hopefully lead to a trend of declining cases and fatalities.) There were also marked daily death toll increases in New Jersey, Louisiana and Michigan, and less pronounced but noticeable increases in Massachusetts, Illinois, Florida, Pennsylvania, and Texas.

Obviously, the increasing (but still insufficient) availability of testing is affecting the tallies. This should give us some comfort: The reported number of new cases seems to be coming down as testing becomes more widespread. Cause for hope that we are, in fact, turning the corner.

Still, quantifying fatalities and the mortality rate remains elusive. Case in point: New York City. As the New York Times reported yesterday, Gotham’s Health Department abruptly added 3,700 victims to the COVID-19 death toll even though these decedents were not tested.

Despite the lack of coronavirus diagnoses in these cases, the inference that it was a factor in death (or, as the city insists, the proximate cause) is not irrational. The Health Department says that 3,000 more people died in the last month than would ordinarily have been expected in the City this time of year. The City has been vexed by the sparse availability of testing. By counting only people who had tested positive, it was surely undercounting COVID-19 deaths to some degree.

But to what degree? We really don’t know. In truth, we will never know beyond educated supposition.

City health officials deduce that some of the spike in “excess deaths” is only indirectly attributable to the coronavirus. On this theory, COVID-19 infections so overwhelmed the health-care system that some non-infected people are assumed to have died of conditions that would otherwise have been treatable.

Meantime, health officials have been tracking deaths they’ve hypothesized could have been related to the virus, based on symptoms and medical history. But what does that mean? Was the coronavirus present in the decedents (unconfirmable because they weren’t tested)? Are health officials saying COVID-19 was actually the proximate cause of death? That it may have exacerbated underlying health problems? That such comorbidities would not have killed the decedents but for the (unconfirmed) presence of COVID-19?

Who knows? The fact is, they are just making estimates. But, as the Times computes it, this estimate has suddenly increased the overall U.S. death count from COVID-19 by a whopping 17 percent. And if the Big Apple is going to cook the books this way, what is to stop Newark, New Orleans, Philadelphia, Boston, Chicago, Detroit, Los Angeles, and the rest?

New York City is dysfunctional, but this is not a New York issue. The guesstimating is being done at the express invitation of the federal Centers for Disease Control.

The CDC instructs officials to report deaths as COVID-19 deaths whenever the patient has either tested positive or, despite the absence of a test, presents circumstances from which presence of the infection can be inferred “within a reasonable degree of certainty” — such that its contribution to death is “probable” or may be “presumed.” This is drawn from CDC guidance, which directs that COVID-19 be specified in death certificates whenever “COVID-19 played a role in the death.”

It is not my purpose here to intimate that the CDC and other federal health officials, embarrassed by how badly inflated their models of anticipated COVID-19 deaths have proven to be, are scheming to overcount. Undoubtedly, some people will politicize the numbers to serve their own agendas, but I’m confident that health officials would like to see COVID-19 deaths reach zero, not zoom into the stratosphere.

My point remains the policy determinations that must be made. To repeat my refrain, critical decisions are being made based on projections and models cobbled together by federal, state, municipal, academic, and private experts in public health. The more these estimates are scrutinized, the less reliable they appear to be. Yet decisions about whether to reopen the economy, when to do so, and under what restrictions hinge on them.

That is a big problem.


The Latest