It’s always been clear that coronavirus appeared in the United States sooner than it appeared in official testing, and was spreading through communities on both coasts before the first official case of “community spread” disease was confirmed in the last week of February. The real question is how soon that community spread began, and to what extent it had progressed before state governments woke up to take steps against an outbreak that the federal government had been assuring them was “fully contained.” At the point when Donald Trump was saying there were “only 15, soon to be down close to zero,” how many cases of COVID-19 were actually moving across the nation?
A new model from Northeastern University provides an answer that describes a “silent explosion” in a number of American cities during February, with cases coming in from Italy and elsewhere forming the seeds of thousands of cases already spreading person-to-person at a time when the government recognized only 100 confirmed cases. The model is provocative—and it absolutely fits with the soon-to-be tragic situation in New York City, where belated efforts kept showing that the boundaries of the outbreak were outpacing the hamstrung ability to test. It’s an insightful look at what could have really been going on while testing waited on the expertise of Labradoodle Dude. However, the model is not without its issues. In particular, the silent explosion seems a bit too explosive, and much too silent.
On Feb. 18, South Korea had 31 official cases of COVID-19. Four days later, it had 600. That didn’t happen because the disease was doubling every few hours. It happened because South Korea—now regarded as the gold standard in handling the first wave of this pandemic—was also caught unaware by community spread that went undetected by early testing focused on travelers from known hot spots. Italy had a similar explosion of numbers just a day later, as what had become three known cases turned into more than 1,000 before the end of the month.
Considering those numbers, it shouldn’t be surprising that the United States had significant amounts of community spread well in advance of the date when the numbers began to genuinely move forward in the last week of the month. In fact, the announcement on Tuesday that researchers had uncovered three previously undetected deaths due to coronavirus infection, one as early as Feb. 6, shows that COVID-19 was absolutely moving through some areas of the country around the same time that the first official case was logged on Jan. 19. When that first known death happened on Feb. 6, there were only a dozen known cases in the United States, 18 in South Korea, and three in Italy. But all three countries were on their way to a massive surge that clearly showed the disease was anything but locked up “airtight.”
What’s surprising about the Northeastern University projections, as reported in The New York Times, is how large that behind-the-scenes growth may have been. The study, directed by Alessandro Vespignani of the university’s Network Science Institute, suggests that on March 1, as New York City was just becoming aware that there were cases “in the wild” within the city, there were not just a few hundred people infected with COVID-19 circulating the city’s streets, but 10,000. It’s a daunting number, one that makes efforts to isolate early hot spots like New Rochelle seem all the more futile and frustrating.
The Northeastern data is very much aligned on the time frame with other projections of when community spread really began in America. Assuming a handful of cases made their way to the United States by mid-January—well before Trump “closed the border” with China—by the first of February there could have been a population of several hundred cases in the country, divided between the East and West coasts. Over the next weeks, those cases were bolstered by visitors from Europe and elsewhere that weren’t yet on the radar as potential hot spots. Meanwhile, the federal government was testing only a very few who met very specific criteria, and the FDA was studiously refusing to approve any privately produced tests. In essence, the virus was given a full month during which no one was even looking. It would have been hard to do worse had the White House pumped virus into a football stadium and cheered it to grow, grow, grow.
Still, there are also reasons to question the numbers generated by this Northeastern model, in the sense that they seem too large, and too even. It’s not just New York City that Vespignani projects to be virus-ridden by the time the calendar page flips to March. The model also projects 9,300 cases in San Francisco, and over 2,000 each for Chicago, Boston, and Seattle. That seems problematic. The Times report brushes off concerns about the extraordinary size of the numbers with a single sentence saying: “Some scientists cautioned that the new report’s estimates of an enormous, unseen wave of infections could be too high — even though testing surveillance lagged at the time” and a single quote from an epidemiologist saying that the numbers are “on the high side.” The numbers are definitely on the high side.
Even looking at the fastest possible rates of doubling, the numbers that Vespignani’s team are suggesting would have put over 5,000 cases around the United States by the middle of February. By the start of March, those cases should have produced somewhere between 100 and 200 deaths. It’s easy to see how small numbers, like the individual deaths recorded in California on Feb. 6 and Feb. 17, may have failed to raise red flags at a time when Americans, including American doctors, were being reassured that COVID-19 was contained. Triple digits is another thing.
For comparison, Italy had logged 150 deaths on 3,000 confirmed cases by March 5. There’s no doubt Italy also had a large number of unreported cases at that point, but subsequent testing in Italy—which has included 100% testing in at least two towns and nationwide testing at twice the rate of the United States—would certainly mediate against the idea that this number was off by orders of magnitude. If the United States had 28,000 cases of COVID-19 by March 1, there would absolutely have been a large number of deaths to go with that number. It’s difficult to see how 100 or more cases of death through acute respiratory distress could have evaded notice in a nation watching epidemics unfold in several nations. It’s even harder to see how the thousands of patients who would have struggled into emergency rooms before March 1 would have failed to raise a nationwide howl.
Still, there is a lot to respect in the Northeastern model. These numbers don’t require any claims to coronavirus starting months earlier than known, don’t imply that a large percentage of the population is already infected, don’t require transmission rates that are off the charts, and begin with reasonable numbers of cases sourced in the early January time frame. The lack of real-world validation in the form of more recorded deaths and hospitalizations (which may yet come, as more records are reevaluated in coming months) suggests that these numbers are the high end of possible, and the distribution of cases across cities is too “smooth” when it comes to explaining the development of cases after March 1. But they’re certainly in the realm of possible.
What seems more likely is that that all those times reporters (including yours truly) made a pronouncement about how the United States was “just two weeks behind Italy” … we were wrong about those weeks. As cases were increasing in South Korea and Italy, they were also going up in the United States. We were looking over there, but it was increasing right here. Officially, there were three “secondary epicenters” of the COVID-19 pandemic—Italy, Iran, and South Korea—which developed in mid- to late-February. In reality, there was a fourth, the United States. It’s just that no one was testing.
The most important point about Vespignani’s projections isn’t their size, it’s their date. The Northeastern team took their projections of exploding community spread to both federal and state officials in mid-February. At the time, they weren’t modeling what had happened; they were warning against what was ahead.
“They told me, ‘OK, that’s happening on your computer,” said Vespignani, “not in reality,”
It wasn’t reality. Reality comes with pain, terror, and exhaustion. It comes with kids grieving their parents. Or parents grieving their child. What the Northeastern team offered was just numbers. The officials who ignored the warning made it into reality.
Notes
A couple of quick notes.
From the contents of the last few days, it may seem that I get a kick out of poking holes in serious scientific work. Honestly … that’s sometimes true. But I’d like to think I’m more interested in examining these studies and testing them against an image of this crisis, in a way that’s both coherent and cohesive. I’ve never liked the old saw of “extraordinary claims require extraordinary proof,” but studies that generate headlines often do so because they’re going against the grain of existing wisdom, and they certainly demand a second look. In the case of this Northeastern study, the only issue I have is how to make the numbers they suggest mesh with the numbers of cases/deaths logged. It’s really a remarkably interesting result for a number that, after all, was a projection before the fact.
Also, there’s been a general reaction in comments (and my overflowing mailbox) that I’m hostile to any idea that suggests any way in which COVID-19 might be less threatening. And … yes, the gist of many things I’ve written, especially in the last few weeks, has been to insist that COVID-19 is as bad as it looks. The hospitalization rate is awful. The rate of deaths is awful. The prospect of putting it behind us anytime soon is awful. However, it’s not like I get any joy out of this. I’m am very, very prepared to eat crow on any of these topics. So very. I have a plate ready and whole bottle of ketchup. Please give me a trustworthy report showing that 99% of the population already had COVID-19, that the fatality rate is really less than toenail fungus, and that tomorrow we can all go to the park. I want to believe.