Why Covid-19 Deaths in New York Were Inevitable
Hint: the reason they were inevitable is the same reason they can’t happen again
No one takes a pandemic seriously until someone dies.
Why should they? We’ve been warned for years about swine flu, bird flu, ebola, murder hornets— it might as well be frogs and locusts. We have enough disasters that are actually happening right now that we can’t afford to spend time worrying about disasters that might or might not happen at some distant point in the future.
This attitude is sensible, and it led us to temper our concerns about Covid-19 — until people started dying. But by then it was too late.
In this article I’m going to explain how we know that, by the time the first few people started dying from Covid-19, more than one million people were already infected. No matter how quickly the states or Federal government acted, it was already too late. The lid of Pandora’s box was open, and there was no way to close it before thousands more died.
But there is hope: the reason that the initial wave of deaths was inevitable is the same reason that it can’t happen again.
Covid-19 seeded hundreds of thousands of infections in New York and more than a million infections nationwide before it caused its first deaths. To say we were caught sleeping would be a tremendous understatement. As of March 15, fewer than 3,000 people had tested positive for Covid-19 nationwide. Fewer than 200,000 had tested positive worldwide.
New York has been the epicenter of the pandemic in the US, and we know some important facts now that we didn’t know on March 15, when the first people died in New York from Covid-19.
The first key fact is that, statistically, it takes about 18 days between the time a person is infected and that person dies (Figure 1). Some individuals die sooner, and some die later, but at a statistical level, for large groups of people, we can count on an average of about 18 days.
The second key fact is that the overall infection fatality rate (IFR) from Covid-19 is approximately 0.5%. This varies greatly among different groups (young people vs. old people, healthy people vs. people with pre-existing conditions, and so on). But, at a statistical level, for large groups of people, we can assume that about 0.5% of the people who contract Covid-19 will die.
These two facts combine more powerfully than you might imagine. They allow us to reconstruct an accurate picture of the status of the disease when the first people started dying in New York and in other states through the country, which in turn provides insight into why the disease has been so deadly.
The first step is to see how many people died 18 days after the first death in a state. In New York, 18 days after its first deaths, that was about 2,400 people. The cornerstone question then becomes, “How many people had to be infected on the day of the first death for 2,400 people to be dead 18 days later?”
We can answer that question using the second fact, that the IFR is about 0.5%. If 2,400 people died on April 2, that means that about 480,000 people (2,400 ÷ 0.5%) had to be infected on March 15 to produce that number of deaths.
That’s right. About half a million people were already infected in New York by the time the first people died.
And because half a million were already infected on the day of the first deaths in New York, 2,400 more people were guaranteed to die, and we’d barely begun fighting the virus.
As a whole, the US totaled about 1.5 million Covid-19 infections as of the day of first death in each state.
This analysis can be extended to estimate the number of infections over the course of the pandemic. Figure 2 shows the history of infections in New York. The dashed line shows the estimates created using the method described in this article. The dotted line shows estimates produced with another method.
Because of the level of infections on the day of New York’s first death (which I’ll call the “initial infection rate”), the death toll in New York became inevitable. As Figure 3 shows, the pandemic crested with New York hitting its maximum contagiousness only two weeks after its first death; after that, contagiousness declined quickly.
Like everyone else I’m deeply saddened by the tens of thousands who died in New York, and I have to ask myself, “What could New York possibly have done to contain the pandemic more quickly than the two weeks it took to reverse the trend?” Because the outbreak started so furtively, I can’t see how it was possible to control the massive outbreak any faster.
Why Didn’t Other States Experience the Problems New York Did?
Other states did experience the same problems as New York, just on a smaller scale. Most states had thousands or tens of thousands of infections before their first deaths. As a whole, the US totaled about 1.5 million Covid-19 infections as of the day of first death in each state. Figure 4 shows the estimated initial infection rates in each state.
Comparing the differences in initial infection rates explains why the virus was so much harder to control in some locations than others.
California has twice the population of New York, but it had less than 10% of the initial infections New York had. Texas and Florida are each somewhat more populous than New York, but those states had much lower initial infection rates.
Twenty four states have larger populations than Louisiana, but in April Louisiana nonetheless had one of the most significant problems in the country because its initial infection rate was the third highest in the country.
Washington state recorded the first Covid-19 death in the US on February 22, three weeks earlier than New York. Why didn’t Washington suffer a similar level of fatalities, or at least similar relative to its population? Washington state had a far lower initial infection rate.
Eighteen days after Washington’s first deaths, its total death count was 57 deaths, compared to New York’s 2,400. That implies that Washington’s initial infection rate was about 12,000 vs. New York’s 480,000.
Figure 5 shows the history of infections in Washington state.
As Figure 6 shows, Washington state took longer to reach peak contagiousness than New York did — 4 weeks rather than 2 — but Washington’s initial infection rate was nowhere near the magnitude of New York’s, and Washington’s outbreak never reached the severity of New York’s.
Why This Can’t Happen Again
Covid-19 seeded hundreds of thousands of infections in New York and more than a million infections nationwide before it caused its first deaths.
To say we were caught sleeping would be a tremendous understatement. As of March 15, fewer than 3,000 people had tested positive for Covid-19 nationwide. Fewer than 200,000 had tested positive worldwide. By the time we realized Covid-19 was a serious threat, it was too late. Its initial 1.5 million stealthy infections in the US made its first 7,500 deaths inevitable. It launched a massive sneak attack, and we were playing catch-up from the beginning.
Today, Covid-19 can never again infect millions of people undetected. Compare the knowledge of the level of infections in mid-March to our knowledge today. In mid-March we had no idea that millions were infected, but the most recent week’s rise in infections has been actively tracked, analyzed, and quantified. We know it’s here, and we know it’s widespread. That fact alone dramatically changes our ability to respond to it.
Covid-19 remains a fearsome enemy, but its days of sneak attacks are over.
I have been focused for 20 years on understanding the data analytics of software development, including quality, productivity, and estimation. The techniques I’ve learned from working with noisy data, bad data, uncertainty, and forecasting all apply to Covid-19.
More Details on the Covid-19 Information Website
For more US and state-level data, check out my Covid-19 Information website.
This article presented a simplified version of the calculation for estimating infections from deaths. The more-involved calculation accounts for irregular frequency in New York’s data reporting, which produces a death total of 2,534, and it accounts for the specific demographics of New York state, which yields an IFR of 0.479%. Those numbers produce an initial infection estimate of 530,000.
For more data state by state as well as US data, see my website.