Comment: Why the covid-19 death toll of 200,000 may be low

By Faye Flam / Bloomberg Opinion

The official covid-19 death toll in the U.S. is about to pass 200,000. There’s a lot behind that number.

Is 200,000 an undercount, as some contend, or an overcount? What does it say about the ongoing risk to us and our families? Were these preventable? How should we see these in context of the approximately 3 million of those in the U.S. who die every year?

The 200,000 number represents an accumulation of educated guesses and approximations. What it’s trying to represent is the number of people who would still be with us today if not for the pandemic. But people can die in complicated ways and doctors have some leeway in what’s written on a death certificate. Some people might have tested positive for covid-19 but died of other causes, and others died of heart failure or pneumonia caused by an undetected infection.

A different number can help shed light on the question of how many people have truly died from covid-19: the total deaths this year compared to the same period in previous years. A few weeks ago, when the official covid-19 death toll was 172,000, there had been an excess 245,000 deaths, says Yale University epidemiologist Daniel Weinberger. That leaves a large number unexplained. Many are likely to be from the coronavirus, though a few might be the result of people not seeking care for heart attacks or other acute health problems.

Weinberger and his colleagues gained a little more insight by looking at when and where those unaccounted deaths happened. Many occurred during the summer in states that saw big increases in virus cases, while states with few cases over the summer, such as Massachusetts, showed a typical number of deaths. That’s lead him to the conclusion that covid-19 deaths are somewhat undercounted and the U.S. actually reached 200,000 mark a few weeks ago. He and colleagues published their interpretation of the excess deaths in a July in the Journal of the American Medical Association, and he’s recently updated the analysis.

Looking into the numbers can also help identify who is vulnerable, says Cambridge University statistician and risk expert David Spiegelhalter. The risk of dying once you get infected goes up about 13 percent with each year of age, he says, which means the risk rises fast, like compound interest. Compared to a 20-year-old, a 40-year-old has 10x the risk of dying once infected, a 60-year-old 100x the risk and an 80-year-old 1,000x the risk.

In a paper published earlier this month in the British Medical Journal, Spiegelhalter compared this with the way the normal risk of dying goes up with age. That risk goes up the same way but not as steeply, with a 10 percent increase in risk for each year. “Covid punishes age even more.”

Most people vastly overestimate the risk of dying once infected, guessing it’s around 20 percent, which is only correct if you’re in your 90s. Another way to think about it, he says, is that getting the virus doubles your normal risk of dying in a given year. If your risk of dying from other causes is low, then doubling that is still not very much. For most of us, it’s a small fraction of a percent. But for older people, it’s substantial.

Digging into the numbers can also shed light on how many years of life the virus is taking from people. Were victims already at death’s door? This question, he says, can be approached by looking at the death rate after the big pandemic peaks hit. If there’s a big dip in the death rate, that suggests that people who died during the pandemic would have died in subsequent weeks. In that case, pandemic might have simply pushed all the deaths that would have occurred during a six-month period into a smaller three-month span.

He’s looking at UK data, where he says there is a small depression in deaths over the weeks after the spring peak, but less than he’d expected, leading him to conclude that even those in their 80s probably had a few years left. “Many of these people were frail and elderly but that doesn’t mean they were about to die.”

The mortality numbers can also reveal something about the way the health care system was disrupted. In the UK, the number of non-covid deaths in hospitals went way down during the spring peak, which corresponded with people being cleared out of those hospitals. During the same period, many more people died at home from non-covid causes. Would these people have lived longer had they been able to stay in the hospital? If so, they are indirect casualties of the pandemic policies. He summarized some of his statistical analysis in a recent issue of New Scientist Magazine.

University of California Berkeley demographer Ronald Lee posed broader questions about the way the pandemic would affect the U.S. and world populations. “I wanted some sense of what it would mean: Is life expectancy going to drop 20 years and is the population going to shrink?” he explains.

His calculations, made with colleague Joshua Goldstein, looked at the effects on life expectancy if the pandemic eventually kills a million people in the U.S. That looks pessimistic now, though it wasn’t last March, he says.

That would take about three years off the average life expectancy — down from 78.9 to about 75.9, or to put it another way, he says, it effectively gives people the life expectancy of someone three years older — or one year older if the death toll is a more optimistic 250,000.

In terms of years of life lost, covid-19 has stolen much less than the 1918 flu or HIV, which tended to kill much younger people. That’s little comfort, though, when so many people have died. “200,000 is still a horrifying number,” he says.

While a number can seem bloodless, numbers are necessary to understand what’s happening in this pandemic. Understanding these numbers is important for shaping policy goals, like keeping the health care system from becoming overwhelmed. They should also make it easier to forecast the trajectory of the pandemic going forward, and do what needs to be done to save lives.

Faye Flam is a Bloomberg Opinion columnist.