Jeremy Samuel Faust is an emergency physician in the Division of Health Policy and Public Health at Brigham and Women’s Hospital in Boston and an instructor at Harvard Medical School. Carlos del Rio is a professor of medicine and global health at Emory University. The views expressed are solely their own.


There’s a metric that can indicate when it is safe to reopen society that does not depend on politics or guesswork. It’s called excess mortality.


Excess mortality is the number of deaths from any cause that both occur in a given time period and surpass the expected number. Deaths in the United States have been carefully counted for more than a century. These “all-cause mortality” numbers are extraordinarily stable. We know to a remarkable extent how many Americans are expected to die every day.


Excess deaths in Massachusetts

Total deaths in the state each year since 2015 compared with deaths in 2020.

Deaths in 2020






Annual deaths from 2015 to 2019

Week 1

Week 15

Week 52

Source: Massachusetts Department of

Public Health data via Jeremy Faust



As The Post reported Saturday, there were an estimated 37,100 excess deaths across the United States in March and the first two weeks of April — nearly 13,500 more than are currently attributed to covid-19. The number of such deaths can be useful in indicating when the coronavirus threat may be less potent.


Remember, excess mortality is a metric that does not depend on the number or percentage of positive SARS-CoV-2 tests. Those statistics depend on policies: How many tests are being done, and on whom? Similarly, the “case fatality rate” of confirmed or suspected cases of covid-19 is subject to forces resulting from testing decisions.


Excess mortality does not depend on counting the number of covid-19 deaths, which ultimately relies on the subjective opinion of physicians and medical examiners proffering their best guesses on death certificates (and whose minds might be understandably steered by the day’s news — “Did this patient with advanced cancer die of the coronavirus, or with the coronavirus?”).


In graphs that track the number of deaths per week, month and year, entire generations blend into the next. For years, nothing much seems to happen. Then there’s an unusual rise in deaths, say, among young men in the 1990s. It tapers off by the end of the century. There’s a sudden spike in New York and New Jersey in September 2001. Mostly, though, the death counts drone on with the march of time, without much deviation and without fail.


These graphs make visible the mundane reality that death is a part of life, quantified. They also show when something unusual is happening. In Massachusetts, for example, the week-to-week data show that this year began like usual. Then, on the week ending March 29, there was a blip: 10 percent more deaths than the usual number. Nothing unprecedented, but it looked like a fluke. Rates returned to normal the next week.

Then history began to unfold in the graphs. The following week, 11 percent more deaths occurred than expected. The next week, 35 percent more. The week after that, there were 73 percent more deaths than normal. Then the number climbed again, to 119 percent more deaths than expected. By April 19, there had been 2,946 more deaths than expected since the beginning of March. By then, the state had reported 1,800 deaths from the coronavirus.

Now, are these undercounted deaths directly caused by covid-19? Did they occur indirectly due to covid-19, because patients who needed medical attention were scared to seek it?

The reliability of excess mortality lies in eschewing this question. Put another way, using excess mortality as a barometer of this pandemic involves being deliberately agnostic to such questions. Excess mortality cares not why anyone died. It simply observes the fact.


This is also why excess mortality presents an unusual opportunity. By closely monitoring excess mortality, which is occurring all over the United States, it is possible to determine when it is safe to reopen the economy and when is too soon.


As long as excess mortality rates are observed, the effect of SARS-CoV-2 remains too substantial to return to normal. Conversely, as excess mortality abates, it’s possible that physicians will continue to observe that some people who die have also tested positive. Even so, if death rates remain at expected levels, the virus is not posing an unusual threat to our normal way of life.


This will be especially potent if new cases spike in several months. Many cases may occur in people who — knowingly or not — already had the virus. If this coronavirus behaves like other respiratory illnesses, second exposures should be milder. If most people who test positive in the coming months have already had the virus, the death rate will be very low — assuming that our immune systems behave as expected. Excess mortality rates would therefore be indispensable in helping health officials to contextualize future spikes in covid-19 cases.



Excess mortality should form the core of evaluation around reopening the economy. Excess mortality can, however, lag behind caseloads because covid-19 deaths start cropping up a couple of weeks after infection, so comprehensive testing is still necessary. Without tracking excess mortality closely, there is a risk that officials might see the results of universal testing and interpret a handful of new cases as bigger threats than they truly are. This could paralyze society for too long.


While excess mortality information in the absence of adequate testing could inform policymakers that another shutdown is necessary, such information might come too late to prevent another major outbreak. Employing these tools together, however, would allow us to determine whether this pandemic has subsided — and likely detect any resurgences.