After climbing the grim upward slope of an epidemic curve that seemed destined to rise forever, we find ourselves somewhat unexpectedly sliding down the other side, with cases in the U.S. declining at such a steep rate it almost seems too good to be true: They’ve gone from a peak of 300,000 new cases per day on Jan. 2 to 62,000 on Tuesday. Just as with each steep rise in cases in the spring and fall, scientists and armchair epidemiologists alike are offering up their favorite explanations (herd immunity! Bad weather! The Democrats switched off the virus right after Inauguration Day!).
The search for understanding makes sense; the loss of a sense of control during the pandemic has been unsettling, and the early battle cry of “flatten the curve” was our way of wresting that control back. Remember those confident predictions back in April that had us at zero cases by July 1? As with the upward trend in the fall (college kids! Seasonality! Pandemic fatigue!), we want answers for the steep downward trend we are seeing not just in the U.S. but in many parts of the world. Surely our mask diligence and social sacrifices are now being rewarded?
Like a wildfire that suddenly changes course or burns out, epidemic curves are something even our best data science can’t fully predict or explain. Epidemiologist Michael Osterholm has spoken of the humility we should have in trying to explain the virus’s every move. The regional ebbs and flows of the virus across the U.S. have often defied obvious explanation. Spikes in the upper Midwest in the fall started dropping dramatically in unison in early November and did not appear to surge around Thanksgiving, when cases in California and the Sunbelt started picking up again. Europe saw eerily similar increases in the fall, despite widely differing severity of previous waves and policy responses. India has seen collapsing cases for unknown reasons.
We do know that an epidemic will grow when the reproductive number (Rt) is above 1, meaning each infection on average is infecting more than one additional person. When it’s below 1, the epidemic shrinks. While one’s ability to infect others is obviously a function of human physical contact, many different factors can interact to push the Rt above or below that critical tipping point.
Let’s review some of the most popular hypotheses for the dramatic decline in cases.
1) Vaccinations. We have been waiting for months for the vaccine cavalry to come over the hill and save us. It’s possible that the U.S. is seeing dividends from the early vaccination of front-line health care workers and those in residential care settings, both settings with high transmission potential. But given the slow (albeit accelerating) vaccine rollout combined with weeks of lag time in building immunity, it’s unlikely that the sudden drop in cases is due largely to vaccinations. Israel has outpaced the rest of the world with more than 45 percent of the population receiving at least one vaccine dose (compared with 12 percent in the U.S.) and has national lockdown restrictions—but even there, cases are only starting to come down. Cases are also falling quickly in areas with low vaccination levels, including Europe and South Africa. This isn’t to say the vaccine effect is zero, but it’s unlikely to be causing a dramatic drop by itself.
2) Less testing. The mobilization of health care resources toward vaccination could have reduced testing and thus measured cases, but this does not seem to be a big contributor. While the number of tests per capita has declined a bit since early January, the share of positive tests has also declined, pointing to a real decline in cases rather than an artifact of less testing. Hospitalizations, which are much less subject to this testing bias, are also falling fast, a strong indication that the decline in infections is real.
3) Behavior and policy change. Given the importance of superspreading for SARS-CoV-2 transmission, with a small percentage of people responsible for a large percentage of onward infections, small changes in behavior that reduce these opportunities could have a big impact. When a fire is raging all around, people tend to voluntarily adjust their behavior to avoid getting burned. Plus, in reaction to the runaway transmission we saw in November through January, many states and local areas stepped up their stay-at-home orders and other restrictions. The patchwork nature of these reactive policies makes it challenging to identify their specific effects, and cases are dropping in almost all states even with a wide range of policies. Google mobility trends show we are still well below pre-pandemic trends for travel related to retail, recreation, transit, and workplaces, but there are no obvious declines in mobility since the beginning of the year, apart from brief weather-related dips in some states. But mobility trends may miss more subtle behavior changes such as fewer meetups with people outside one’s household or increases in mask wearing outside the home, and we have seen some favorable trends for those things, according to an ongoing survey conducted by Carnegie Mellon University. Together, all of this likely did contribute to reductions in the Rt—for one thing, we know these prevention measures have effectively crushed the flu (thought to be less transmissible than SARS-CoV-2) this year in both hemispheres.
4) Seasonality. We are accustomed to seeing a seasonal rise and fall for many respiratory viruses, including influenza. While seasonality is less scientifically understood than one might expect, it is thought to be a combination of social contact patterns (return to school, moving activities indoors) as well as an effect of temperature and humidity on virus transmissibility. While many assumed a seasonal component was at play with the steep rise in COVID-19 cases in the autumn in the U.S. and Europe, the fact that we are still in the dead of winter in many places, coupled with significant increases over the summer in many U.S. regions, makes this far from a slam-dunk case. The calendar-driven social mixing dynamics may be a better fit for the current drop in cases, with busy autumn activities and holidays giving way to a typical January slump fueled by dark, cold weather and fewer social events.
5) Herd immunity. One key variable in epidemic modeling is the number of susceptible individuals remaining in the population—the human kindling that keeps an epidemic fire burning. As more and more people become infected and immune, the virus cannot spread. The best estimates suggest that we are still far from the herd immunity threshold that would keep the epidemic from growing again, generally believed to be about 70–80 percent of the population. There have been almost 28 million confirmed COVID-19 cases in the U.S. The Centers for Disease Control and Prevention estimates that we are picking up 1 in 4.6 actual cases, meaning almost 130 million people may have been infected, or close to 40 percent of the U.S. population. Considering the additional protection of those newly vaccinated, it’s possible that population immunity is helping to slow transmission as SARS-CoV-2 encounters more firewalls than fresh kindling. This dynamic may be aided by the fact that not all people are as likely to be exposed or transmit the virus. Those with more social contacts or jobs that don’t allow social distancing are more likely to have already been infected, leaving the remaining susceptibles harder for the virus to reach. But with 50–60 percent of the population still vulnerable, those embers could easily catch fire again.
6) The known unknowns. As Michael Osterholm emphasizes, despite our best scientific efforts, we must humbly admit that human understanding of SARS-CoV-2 infection dynamics in the real world is limited. The uniformity of the recent drops across U.S. states as well as globally points to something—a rhythm, a natural ebb and flow, a viral boom-and-bust cycle. This “natural” cyclicality is surely a complex interaction of the factors above, each contributing to push the reproductive rate below that critical threshold for which the exponential momentum starts to work in our favor. Things that are cyclical like the economy or epidemics are cyclical because we don’t fully understand how to control them. If we did understand them, we’d always have 7.2 percent GDP growth and low unemployment. Human behavior is beautifully adaptive but with such individual diversity that it is not easily modeled.
While the current decline in cases and hospitalizations is extremely welcome news, the specter of new variants means we might be seeing a pullback of the tide before the next big wave comes crashing down. Is the next wave inevitable? Possibly, but the steep decline in cases in the U.K. and South Africa demonstrates that the new variants themselves are not immune to changes in our behavior.
So, while we should avoid the temptation to cherry-pick our favored explanations for every twist and turn of the epidemic curve, neither should we be fatalistic. The epidemic curve is a dance between virus and hosts, so we always have a say in where we end up on the dance floor.