What Do 'Missing Deaths' Imply About the Impact of COVID-19?

A New York Times analysis has little bearing on the infection fatality rate or the number of life-years typically lost to the disease.


New York Times analysis of mortality data from 11 countries suggests that deaths associated with the COVID-19 epidemic have been undercounted. The analysis, which counts 28,000 "missing deaths" in those countries, includes fatalities caused by other illnesses, on the assumption that hospitals overwhelmed by COVID-19 cases were unable to treat patients who might otherwise have been saved.

While this analysis adds to our understanding of the epidemic's impact, the significance of the undercounting described by the Times should not be exaggerated. It has little bearing on the overall fatality rate among peopled infected by the COVID-19 virus or the number of life-years typically lost to the disease.

Comparing deaths during the last month to deaths during the same period last year, the Times finds a total of 193,000 "excess deaths," of which 165,000 were recorded as COVID-19 fatalities. Based on that comparison, the death toll attributable to the epidemic (even if not to COVID-19 itself) was 17 percent higher than the official numbers indicate.

Since this analysis includes deaths that were not actually caused by COVID-19, it is not directly relevant to calculating the infection fatality rate. Even leaving that point aside, an undercount of this magnitude pales in comparison to the gap between total infections and official tallies of COVID-19 cases, which may be off by a factor on the order of 40 or so (judging from a recent study in Los Angeles County). The error in the denominator, in other words, is apt to be far bigger than the error in the numerator. And while the official tallies no doubt miss some COVID-19 deaths (especially those involving people with other serious medical conditions who die at home), they may also include fatalities caused by other illnesses among patients who tested positive for the virus.

The Times claims "these numbers undermine the notion that many people who have died from the virus may soon have died anyway." That depends on what you mean by "many" and "soon."

British epidemiologist Neil Ferguson, who led the Imperial College team responsible for dire COVID-19 projections that had a powerful impact on policy makers around the world, has estimated that "as much as half to two-thirds" of people who are killed by COVID-19 in the U.K. "would have died anyhow" by the end of the year. The Times analysis, since it is limited to a single month, does not address the accuracy of such predictions.

Even if Ferguson's estimate is off, it is clear that COVID-19 deaths are concentrated among people who are elderly and/or have serious preexisting conditions. That means the impact of the disease, in terms of life-years lost, is bound to be much less dramatic than it would be if COVID-19 were primarily killing otherwise healthy children, teenagers, and younger adults. That observation remains true whether or not people who are especially vulnerable to the disease would have died in the same month or year in the absence of the epidemic.

"This is killing mostly older people," observes Princeton bioethicist Peter Singer. "I think that's really relevant. I think we want to take into account the number of life years lost—not just the number of lives lost. The average age of death from COVID in Italy is 79½. So you do have to ask the question: How many years of life were lost? Especially when you consider that many of the people who have died had underlying medical conditions. The economist Paul Frijters roughly estimates that Italians lost perhaps an average of three years of life. And that's very different from a younger person losing 40 years of life or 60 years of life."