As the number of reported US coronavirus cases climbs past 600,000 and reported deaths past 26,000 as of April 15, there is a little bit of good news in the data: Things may be getting somewhat better, especially for areas that were hot spots, like New York and Washington state. For the past week, new case numbers have been growing steadily, rather than exponentially. And daily new deaths are lower this week than last week.
New York and Washington state, which saw cases and deaths skyrocket in the final week of March, have seen their numbers get a little more under control in the first week of April. “We are flattening the curve,” said New York Gov. Andrew Cuomo, pointing out that new hospitalizations are at “the lowest number we’ve had since this nightmare started.” Feared hot spots like Florida aren’t yet overwhelmed, either (though in all of these states, spikes in all-cause mortality suggest that the situation is worse than the numbers suggest).
A body of preliminary research suggests that social distancing and lockdowns have worked to flatten the curve and, at a minimum, greatly slowed down the spread of the disease. “We’re seeing less cases,” professor of health metrics sciences Ali Mokdad at the University of Washington told me. “Social distancing is working.”
But while that’s good news, it’s limited good news. Case numbers aren’t growing as fast or as exponentially as they were in March, but it is not apparent yet that they’re shrinking, either. Some models project a nationwide peak soon (or argue the US has already reached it) and should then see a decline in the number of cases, but these models are highly uncertain, and it remains a possibility that case numbers will continue to grow, even if they’re growing more slowly than previously.
Since it takes several weeks for new case numbers to reflect changes in policy, experts have been waiting for the last few weeks to learn how stay-at-home orders have worked. Now the data’s in. It suggests that social distancing (not just the lockdowns, but all steps that people took to avoid other people) is working — but perhaps not necessarily working well enough to drive new cases down to zero.
That puts the country in a challenging position going forward, even if it should rightly take some pride in having successfully tempered the exponential spread of the virus.
Social distancing works
First, the good news: Social distancing absolutely does work to slow the spread of the coronavirus. In the states that started social distancing the earliest, including Seattle and San Francisco, the growth of new cases has slowed noticeably. In places that implemented them a little later, like New York, the growth of new cases began to slow later.
Testing is still limited and the criteria for testing varies in different places, making direct comparisons difficult, but there are enough cities seeing improvement that a real trend can be observed.
Modeling done by the Center for Mathematical Modeling of Infectious Diseases (CMMID) at the London School of Hygiene & Tropical Medicine attempts to use confirmed case numbers to estimate the “effective reproductive number” or R0 — the number of people the average person infects. If that number is less than 1, the disease will die out.
Early studies found that the R0 for the Covid-19 coronavirus was in the 2-2.5 range (by comparison, the seasonal flu’s R0 is 1.3). In the US, the CMMID model found that the number was between 2 and 3 when the disease first took off, but it has been steadily declining in the weeks since. It is now around 1, meaning the average person is infecting about one additional person. That decline started before stay-at-home orders went into effect in most states and reflects the impact of those orders and the voluntary social distancing efforts before them.
Other recent studies point toward a similar conclusion, though there’s still a lot of uncertainty. An April 6 article in The BMJ (The British Medical Journal) argued that Seattle’s success at controlling the virus is a cause for hope:
Two reports released on 30 March by the Seattle based Institute for Disease Modelling suggested that the early implementation of successively stronger measures, starting with a 4 March recommendation that workers be allowed to work from home, effectively reduced the reproductive value of the coronavirus from 2.7 in late February to 1.4 by 18 March.
Then came the lockdown, after which, the article says, “modellers suggested the region was ‘on the cusp’ of pushing the viral reproductive value to 1, below which the virus wouldn’t be able to sustain its spread over time.”
Other states now appear to be on the same trajectory. New reported confirmed cases in the US hovered for a week between 30,000 and 34,000, and have at last fallen under a reported 30,000 in the past few days (it’s not just limited testing; the percentage of tests that are positive is falling as well). This is a big change from previous weeks, where by Friday there’d be twice as many new cases as there were on Monday.
New York, the hardest-hit state in the country, is seeing that trend. “The evidence is very strong that the curve is leveling off in New York,” Jeffrey Harris, a medical doctor and MIT economist, told me. His new working paper, “The Coronavirus Epidemic Curve Is Already Flattening in New York City,” explores what that means. Through March 20, the paper argues, New York’s confirmed case numbers were doubling every 1.3 days. That’s when they started leveling off — still growing, but with a much slower doubling time.
Some of this is attributable to reduced testing — the number of tests being conducted has fallen slightly over the last few weeks. But it cannot all be attributed to that. The percentage of tests that return a positive result has also been falling. The number of new hospitalizations has been growing more slowly.
The most remarkable thing is that much of these gains showed up in case statistics too early to be just the result of lockdowns and stay-at-home orders. Harris, for example, finds that New York’s slowdown in new case numbers started on March 20 — before the state even put such an order into place.
“In many places, many people started staying at home before the orders came out,” Mokdad told me. “Many companies moved to working from home,” he said, including Seattle giant Microsoft, which switched to work from home several weeks before formal stay-at-home orders were put in place in the city.
A few weeks after those companies first started ordering employees to work from home, the virus was no longer spreading exponentially throughout the Seattle population. The improvement began well before measures started in cities like Los Angeles, Detroit, and New Orleans, suggesting — though all this data is still preliminary — that it’s early social distancing that made the difference.
Key things we still don’t know yet
But while the slowing spread of the virus is good news, a lot depends on exactly how much the spread has slowed. And that’s where there’s still a lot of uncertainty — which makes things difficult for epidemiologists, modelers, and policy planners who want to start planning the next phase of the attack on the virus.
Here’s the dilemma: Ideally, case numbers would now start falling. If the effective reproductive number (the R0) of the virus has been pushed below 1 with the shelter-in-place orders around the country, then every day should see fewer infections than the day before, and case numbers should fall, hopefully as quickly as they rose.
Most schemes to reopen the country rely on this: They require that case numbers fall for 14 days before the US starts loosening restrictions. The idea is that two weeks of falling cases is enough that it can’t just be a coincidence, and enough to lower the overall case count so regions can trace contacts and use more intensive monitoring approaches in a targeted way.
But even in the parts of the country that have now been living under extensive restrictions for several weeks, case numbers aren’t falling across the board — though in some areas (most crucially New York) they do seem to be. More reliable measures like hospitalizations and deaths aren’t falling, either. That’s why the CMMID estimates the RO in the US at about 1 — each sick person is infecting about one more person.
But that estimate has a lot of uncertainty, too, and that uncertainty is significant. If the number is a little smaller than 1, new case numbers will decline. If it’s a little larger than 1, new case numbers will keep increasing. And there are some signs that it might, indeed, be larger than 1, considering that case numbers aren’t consistently decreasing anywhere. “We are still moving in an uphill direction,” Becca Bartles, Providence St. Joseph Health’s executive director of infection prevention, told The BMJ.
“It is almost completely flat in the borough of Manhattan,” Harris told me of the New York numbers. “The doubling time in the other boroughs is much slower but is still not leveling off.” It has been more than three weeks since New York’s stay-at-home order, so these slowdowns in the rate of infections likely reflect entirely new infections that occurred under the order.
In other words, social distancing is definitely working, but the question of whether it is working well enough remains to be seen, and the fact that numbers are plateauing rather than falling isn’t a great sign.
“It seems that the press has been eager to push the narrative of ‘we are near the peak!’ and ‘the end is in sight,’ but given the strong uncertainty about the future and lack of clear consensus among modelers, I think these messages are premature,” UMass infectious disease researcher Nicholas Reich argued.
The more accurate message: There are some promising signs, but nothing definitive, and the range of possible futures is very wide.
The long road ahead
Even if case numbers are falling, that just brings us face to face with another uncertainty: What next?
“We cannot lift all the restrictions at once,” Mokdad told me. He favors a cautious, gradual approach that lets us keep the R0 below 1 (assuming that it is below 1 right now): “Relax a little, see that the virus is not circulating, relax a little more.”
But the current uncertainty is fatal to such an approach. If we’re going to calibrate our restrictions around the R0 of the virus, we need really good and timely data on exactly what that number is. “All of the conditions to go back to opening business depend on being able to do testing,” Mokdad says. And US testing, which grew rapidly for most of March, has been plateauing lately.
It’s a vexing situation. The US has succeeded at its first task of slowing the wildfire spread of the virus through our communities, but it may have achieved only a standstill, not a victory. And as pressure grows to open more businesses and get the country back to work, there is still limited (and not encouraging) data about what will be needed to keep the virus from roaring back. Even worse, if the data suggests that the country actually needs more stringent restrictions than the current ones, it’s hard to imagine who’d muster the political will to implement them.
That bad news must be grappled with alongside the good if the country is to take sensible next steps.