A new preprint study estimates that COVID-19 deaths in the United States cut lives short by a total of 2.5 million years as of early October. The author, Harvard Medical School geneticist Stephen Elledge, says he did the analysis to correct "the false impression that the impact on society of these deaths is minimal" because they are concentrated among the elderly.
According to the most recent data from the Centers for Disease Control and Prevention (CDC), people 65 or older, who represent 17 percent of the U.S. population, account for nearly 80 percent of COVID-19 deaths. But as Elledge emphasizes, people in that age group are not necessarily on the verge of death. The average life expectancy at 65 in the United States is about 83 for men and 86 for women.
"Because the great majority of COVID-19 deaths occur among the elderly," Elledge writes, "the false impression that the impact on society from these deaths is minimal may be conveyed since these individuals were closer to a natural death. Aside from any troubling ethical implications associated with rationalization of COVID-19 mortality along these lines, such a conclusion is unwarranted for at least two reasons. First, as individuals age, their life expectancies increase too, well beyond the life expectancy at birth, which is the value most familiar to the general public. Second, a significant number of relatively young individuals have also died from COVID-19 and had decades of remaining life expectancy."
Based on actuarial data on life expectancy and the age distribution of COVID-19 fatalities, Elledge roughly calculated that the 194,087 deaths reported by CDC as of October 3 amounted to 2,572,102 years of potential life lost (YPLL). The average loss based on that calculation was about 13 years and three months. Because the CDC breaks COVID-19 deaths into 10-year age ranges and the risk of dying from COVID-19 rises with age, Elledge performed an adjustment that reduced the YPLL number by 3.5 percent, to 2,486,160. That implies an average loss of nearly 12 years and 10 months.
Elledge notes that his analysis did not adequately account for "the effect of comorbidities on life expectancy." Since people who are less healthy to begin with are more likely to die from COVID-19, a calculation based on average life expectancies by age group is apt to exaggerate the years of potential life lost due to the disease. Elledge takes a stab at adjusting for comorbidities by including a calculation that reduces COVID's YPLL toll by 15 percent, from about 2.5 million to about 2 million. But this is really just a guess.
Other metrics of death-related costs, such as disability-adjusted life years (DALY) and quality-adjusted life years (QALY), try to take into account how healthy people would have been during their remaining years. That consideration is obviously relevant when comparing deaths at a young age to deaths at an advanced age, even leaving aside the difference in years of life expectancy.
"We did not undertake those analyses but note that there is a growing awareness of lasting effects on those infected with SARS-CoV-2 that lead to serious medical consequences," Elledge says. The implication is that a QALY or DALY approach would cut both ways: discounting years lost by people who were already in poor health while taking into account the lasting effects of nonfatal infections.
Leaving those issues aside and taking Elledge's numbers at face value, how do they compare to the losses associated with other causes of death?
In 2018, according to CDC data, unintentional injuries among Americans 85 or younger, including traffic accidents, were responsible for about 5.3 million years of potential life lost. Although those injuries caused fewer deaths (about 167,000) than COVID-19 has, the average loss was much bigger: more than 31 years, compared to about 13 years for COVID-19 per Elledge.
The average YPLL is only slightly higher for cancer than Elledge's estimate for COVID-19: 14 vs. 13 years. But because cancer causes more deaths (nearly 600,000 in 2018, per the CDC), the total loss is much bigger: about 8.5 million years. And while heart disease, according to the CDC's numbers, claimed an average of 10 years in 2018—less than the figure Elledge calculated for COVID-19—the total loss was still much higher: 6.6 million years of potential life.
In 2008, the CDC calculated that smoking was responsible for about 443,000 deaths per year (many of those from cancer and heart disease), amounting to 5.1 million YPLL, or more than 11 years on average. This year the CDC estimated that "excessive alcohol use" causes about 93,000 deaths a year (including about 7,000 motor vehicle deaths), resulting in a YPLL toll of 2.7 million annually, or an average of 29 years. By contrast, the 2.5 million YPLL toll calculated by Elledge for COVID-19 is higher than the annual loss that the CDC attributes to suicide (1.8 million), homicide (944,000), chronic lower respiratory disease (1.5 million), and diabetes (1.2 million)—all of which are considered serious problems worthy of national attention.
Contrary to Elledge's implication, however, people who emphasize the age distribution of COVID-19 deaths are not saying those deaths are "minimal" or don't matter. They are saying that years of potential life lost, and perhaps also the quality of those years, should be considered when weighing the costs and benefits of policies aimed at curtailing the epidemic. Elledge seems to agree.