Confirmed cases of Covid-19 and hospitalizations from the disease have been plummeting for weeks in the US and elsewhere in the Northern Hemisphere. Deaths also are in decline. So it seems the long-dreaded fall-winter wave of the pandemic, which turned out to be just about as terrible as feared (with the exception that it wasn’t accompanied by much of any seasonal influenza), has finally crested.
In more good news, some quite effective Covid vaccines are now available, and after a fitful start, the US effort to inject them into people’s arms is steadily gaining speed. Four times as many people are now receiving first vaccine doses each day as are being infected with the disease, according to the infections estimates of both data scientist Youyang Gu’s covid19-projections.com and the covidestim.org model assembled by researchers at the Harvard and Yale schools of public health.
So is that it? Is the worst of this pandemic behind us, or are there more big waves to come?
By this point the most sensible answer to such questions about Covid-19 is probably. But a few key variables seem most likely to determine what happens next, and running through them may shed at least a little light on the path ahead.
The B.1.1.7 variant that has become predominant in the UK and is expected to take over soon in the US. has been estimated to be 30% to 70% more transmissible than the version of the SARS-CoV-2 virus we have come to know and heartily dislike. Because of the magic of exponential growth, this can translate into a greater than 30%-to-70% increase in the number of cases in just a few weeks.
What seems like only a moderately higher growth rate can bring many, many more cases in short order. In the real world, as in the susceptible-exposed-infectious-recovered (SEIR) models that epidemiologists use, the growth rate does at least slow over time as people become immune and change their behavior to avoid the disease. In Denmark, which is doing the world’s best job of testing for B.1.1.7 and other mutations — and where schools, restaurants, bars and non-essential shops have been closed since Christmas — B.1.1.7 cases are growing modestly amid a collapse in other Covid-19 cases.
Of the 63.4% of Danish Covid-19 cases from the first week in February for which the Statens Serum Institut has been able to sequence the full genome, 30.9% were of the B.1.1.7 lineage. That’s up from 3.8% just four weeks earlier, so B.1.1.7 is rapidly headed toward becoming the dominant variant in the country. The number of B.1.1.7 cases is still pretty small, though, and at the current growth rate it will be months before the variant is as prevalent in Denmark as the old-school version of the virus was in December. But the current growth is occurring amid a strict lockdown that has been extended through the end of this month, and that can’t be maintained indefinitely. “If we look at our models,” Statens Serum Institut epidemiologist Camilla Holten Moller told Science magazine a couple of weeks ago, “this is the calm before the storm.”
Then there are the mutations encountered in Brazil and South Africa that appear able, at least partially, to evade the immunity provided by previous infections and vaccines. Because B.1.1.7 is so contagious, it seems likely to shove most other variants aside. There have been eight cases of the South African B.1.351 variant identified in Denmark, for example, but all were connected to overseas travel, and it doesn’t seem to have spread. However, Danish researchers have identified 72 B.1.1.7 cases in which the virus has a spike-protein mutation similar to the one on the South African variant. It’s too early to determine what impact this will have, but it doesn’t seem good.
In the US, which sequences a far smaller percentage of Covid-19 test samples than Denmark and is thus flying relatively blind on B.1.1.7 and other new variants, state and local authorities have been relaxing Covid restrictions in the face of falling overall case numbers. The effect of these relaxations has probably been limited so far, given how terrible the weather has been across much of the country. According the Dallas Fed’s Mobility and Engagement Index, based on mobile-device location data, Americans have been out and about less (relative to a year earlier) over the past month and a half than at any time since late last spring.
The weather is eventually going to improve. Until now, though, Americans seem to have been slower about jumping back into action when Covid-19 restrictions are lifted than people in some European countries. This is partly just because there’s been such a wide range of Covid policies across different states and cities. But OpenTable Inc.’s city-by-city restaurant data indicate that there may be more to it than that.
Germany was spectacularly successful at warding off the first wave of Covid-19 last spring. After that, though, life in the country went back to something approaching normal, setting the stage for a very tough fall and winter. In the UK, the sudden resurgence of Covid cases in December surely owed a lot to the rise of the more-contagious B.1.1.7 virus variant, but the alacrity with which people there embraced the easing of lockdown restrictions in November must have contributed, too.
In the US, a mix of distrust of official pronouncements and the ease with which many people have been able to switch to an all-remote, all-delivery lifestyle seems to have resulted in a different pattern in which life in much of the country never came close to returning to normal and probably won’t for a while yet. This may help keep things from getting out of hand this spring.
Seasonality
In temperate climates, colds and influenza tend to wax in the fall and winter and wane in the spring and summer. Why exactly this is remains a topic of much research and debate, but it stands to reason that Covid-19 — caused by a coronavirus, as are many colds — would settle into a similar pattern. To some extent it already has: this fall and winter have been the worst of the pandemic in North America and Europe. But as of now, there isn’t enough evidence to make confident pronouncements about what the coming of spring will bring.
Immunity
In the US, close to 30% of the population has probably already been infected with the SARS-CoV-2 virus. Another 12% (there must be some overlap with the first group, but probably not too much yet) has already received at least a first dose of vaccine. About 40% of the population thus has at least some protection. At the current pace of vaccination, this share is increasing by another percentage point every three days, and with the weekly US vaccine supply about to rise from 11 million doses to 13.5 million, it will be increasing faster than that soon.
The threat of new variants, combined with the inevitable imperfections in the immunity provided by both previous infections and vaccines, mean that getting to the full herd immunity that keeps Covid-19 from spreading even if we’re all hanging out together maskless indoors is a daunting and perhaps impossible task. But 40% of the population possessing some immunity should still slow the spread of the disease, and getting to 50% and then 60% should slow it even more.
There’s also the matter of who has been vaccinated. Residents of nursing homes and other long-term care facilities have accounted for an estimated 36% of US Covid-19 deaths. As of mid-January, almost 80% had received at least a first vaccine dose. Americans 65 and older have accounted for 81% of deaths, and they’re the main focus of the current inoculation effort. The vaccines approved for use in the US, as well as a couple of others likely to be approved soon, appear to be 100% effective in preventing deaths from Covid-19. So even if they can’t stop a resurgence of Covid this spring, they may be able to stop it from being nearly as deadly as past waves.
Putting It Together
At the moment one can find quite a wide range of views among experts about what is likely to happen next. Michael Osterholm of the University of Minnesota’s Center for Infectious Disease Research and Policy sees B.1.1.7 causing a surge in the US akin to a “category 5 or higher” hurricane. Michael Mina of Harvard’s T.H. Chan School of Public Health thinks that “the next few months could start to offer a reprieve. Ideally that will last through the summer until we get into next fall, when we’ll probably have another wave of it.” In Germany, Christian Drosten of the Charité medical university’s Institute of Virology contemplates a scenario in which, once the elderly have been vaccinated,
there will be immense economic, social, political and perhaps also legal pressure to end the corona measures. And then, huge numbers of people will become infected within just a short amount of time, more than we can even imagine at the moment. ... It will, of course, be primarily younger people who are less likely than older people to have severe symptoms, but when a huge number of younger people get infected, then the intensive-care units will fill up anyway and a lot of people will die.
Israel, which will be ending its latest lockdown this weekend, looks to be the first test of Drosten’s theory. Thanks to the world’s most successful vaccination campaign, virtually everyone older than 60 has already received two shots. But Israel still has many more Covid cases on a per-capita basis than the US, and there are now more people younger than 60 hospitalized with Covid there than over 60. The next few weeks could be really interesting.
In the US, the “ensemble” of the various forecasts collected for the Centers for Disease Control and Prevention by researchers in biostatistician Nicholas Reich’s laboratory at the University of Massachusetts at Amherst foresees continued declines in cases and in deaths. The ensemble forecast only goes out four weeks, though, and the Reich lab’s handy visualization tool makes clear that it hasn’t been great at calling turning points in the past.
The University of Washington’s Institute for Health Metrics and Evaluation has a forecasting model that looks four months into the future, which got it a lot of attention early in the pandemic but also a lot of criticism because it 1) was repeatedly wrong and 2) eschewed epidemiological modeling for a simple exercise in curve-fitting. Since then, IHME has incorporated an SEIR model and seemingly become much better at forecasting, with its dire September warning of 410,000 total Covid-19 deaths in the US by Jan. 1 not far off from the eventual CDC count of 382,051. At the moment, IHME is projecting a sharp increase in infections in March and April — albeit not quite back to late-December/early-January levels even in the worst-case scenario — and a much-less-severe resurgence in deaths.
IHME’s base-case forecast and the rapid-variant-spread alternative are barely distinguishable at the moment because the main forecast already factors in rapid spread of B.1.1.7, with the alternative adding in the South African B.1.351 variant. The increased-mobility forecast assumes all of that plus mobility moving “toward pre-Covid-19 levels” among those who have already been vaccinated, although it also probably works as a proxy for more widespread, if less sharp, behavioral change.
All in all, the IHME forecast seems like a pretty good starting point for thinking about the next few months. There’s good reason to hope that the worst is over. There’s also good reason to be extremely careful until vaccines are more widely available.
I have actually found myself becoming more cautious lately because I’d feel like such an idiot if I caught Covid mere weeks before becoming eligible to be vaccinated (I’m 57, so I’m thinking it won’t be long). A group of Danish political scientists and computer scientists want to encourage such behavior on a mass scale, arguing in a not-yet-peer-reviewed paper that public health messaging should now focus on “buying time with hope.” That is, the pandemic will be over soon in any case thanks to vaccines, but careful behavior can sharply reduce the intervening toll. The main tool of such messaging, they propose, should be charts similar to the one above. The pink line seems a lot more attractive than the gray one, right?
Bloomberg
