What the COVID-19 death toll tells us about the next 2 weeks
TL;DR: The confirmed case number is not helpful in understanding today’s COVID-19 pandemic. The death toll is a more informative measure, but still lags the actual infection number by 3 weeks. Given that mitigation efforts in the U.S. started only in the last 1–2 weeks, we will continue to see an exponential growth of the death toll in the next couple of weeks. The number of American deaths will likely exceed 20,000 by Easter Sunday. However, this doesn’t mean social distancing and shelter-in-place have not been effective; it simply means that their impacts take time (~3 weeks) before being reflected in the death number. We should continue to prepare for the tsunami of COVID-19 patients in the coming weeks, and persist with social distancing and shelter-in-place. Otherwise hundreds of thousands more Americans will die in the coming months.
The confirmed COVID-19 case number is not a useful measure in the U.S.
On Thursday, March 26th, the U.S. overtook China and Italy as the country with the highest number of COVID-19 cases, at more than 85,000 cases. Today, there are more than 100,000 confirmed cases in the U.S.
While these are significant milestones, the number of confirmed cases in the U.S. tells us very little about the current state of the COVID-19 pandemic in this country. I discussed in my previous article that, because of (1) the lag between infection and reporting and (2) a high prevalence of untested and unreported cases, the confirmed case number drastically underestimates the extent of the pandemic in the U.S. today. The number of active infections in the U.S. is estimated to be ~100 times or more of what is confirmed today, likely in millions to tens of millions.
You may be wondering, “if the confirmed case number is not a helpful measure of the COVID-19 pandemic in the U.S., what else should we pay attention to?”
The number of COVID-19 deaths remains the most reliable indicator of the pandemic.
The death toll number, unlike the confirmed case number, does not depend on the availability of testing, therefore is a more reliable signal for the severity of the pandemic (we will assume here that a negligible number of COVID-19 deaths goes unreported).
Yet, the death toll number still suffers from the same issue of delay as the confirmed case number. Given that COVID-19 victims die on average 18 days after they experience their first symptom and a typical incubation period for COVID-19 is 5 days, the death figure lags the number of infections by 23 days. In other words, the cumulative death toll represents where we were in the spread of this pandemic approximately 3 weeks ago (*I will use 3 weeks for both the infection-to-death lag and the intervention-to-impact-on-death lag going forward).
So, what do the current and historical COVID-19 death numbers tell us about where we are in this pandemic and where we are headed in the near future?
The cumulative death toll has been increasing by 31% each day, or doubling every 2.6 days
If we assume that patients die on average 3 weeks after their infection, then we can conclude that the 108 people who have died by March 17th were infected on or before February 25th. In other words, the cumulative death toll on March 17th reflects the extent of our COVID-10 pandemic, 3 weeks earlier, on February 25th.
There is a significant delay between any social interventions and their impact on the death toll.
It is unlikely that we will be able to develop, produce, and distribute an effective treatment for COVID-19 over the next 2–3 weeks. Without new COVID-19 treatments, the only way to slow the exponential increase in the death toll is to reduce the transmission of COVID-19.
But those interventions to reduce transmission — social distancing, sheltering-in-place — affect the rate of disease spread today. Because of the delay between infection and death, we will not see the effects of these social interventions on the death rates for 3 weeks.
Social interventions did not start until March 10th, therefore, we will continue to see the cumulative death number double at the current rate until March 31st.
The U.S. did not take COVID-19 seriously until mid-March. On March 10th, “social distancing” and “flatten the curve” started trending on social media. On March 13th, Trump declared a national emergency. California was the first state to announce the statewide shelter-in-place on March 19th. Other states followed with their shelter-in-place orders between March 21st and 25th.
Based on the events above, I use the following parameters for my projection:
- Death numbers continue to rise at the doubling time of 2.6 days (daily increase rate of 31%) as it has been.
- On March 31st — 3 weeks after the March 10th social media campaign — we begin to see the infection transmission slow down to a 5.2 day doubling time (daily increase rate of 14%).
Cumulative death numbers are projected to exceed 4,500 by the end of March, and 20,000 by Easter Sunday, April 12th.
Compare these death numbers to the current (March 27, 5PM EDT) death toll of 1,581 in the U.S., 3296 in China, and 9,134 in Italy .
This exponential growth in deaths does NOT mean social distancing and shelter-in-place are not working.
It simply means that it is too early for the impact of these social interventions to be reflected in the death toll number due to the ~3 week lag.
So what now?
We must all prepare for the incoming tsunami of COVID-19 patients.
My alma mater Yale University has converted its gym into a hospital floor to care for COVID-19 patients. On March 26th, Governor Cuomo of New York approved a single ventilator to be shared by two patients.
Obviously the mismatch between the demand and the supply of hospital resources including hospital beds, ICU beds, and ventilators, as well as the estimated times of peak demand will vary by state as seen below.
We should continue to be resourceful, flexible, and decisive as we create more capacity to care for exponentially growing COVID-19 patients. If we do not have adequate resources to care for these patients, the death rate will be much higher.
Given this delay between intervention and their impact on the death toll, we must act now to prevent additional hundred of thousands of Americans from dying
The Trump administration is discussing “opening up” the country by Easter. This New York Times article suggests that this is premature and will lead to 720,000 deaths in the U.S. by the end of October.
COVID-19 social interventions are not like driving, where the car responds immediately to your turning of the steering wheel. These social interventions are more akin to gardening; you plant your seed, and have to wait a few weeks before you can see it germinate.
Despite the exponential increase in deaths we are about to see, we should not relent from our commitment to social distancing and shelter-in-place. We will see the fruits our efforts, just not for another couple of weeks. To prevent additional hundreds of thousands of our loved ones from dying in the coming months, we must continue to stay home.