Last week, the New York Times published a story with the headline “Europe’s Deadly Second Wave: How Did It Happen Again?” The essential theme of the article is that European governments relaxed their restrictions too quickly and thus brought this second wave upon themselves. The lead image is the chart below, showing daily deaths peaking in early April, falling to a long trough from late spring through early fall, then rising to an equivalent peak by the end of November.
I’m not interested, for the purposes of this article, in litigating the issue of whether or not Europe opened too soon. There’s a much more interesting and less widely discussed question that is prompted by the European numbers: why is the size of the wave, which is so similar when looking at Europe as a whole, so different when viewed on a country by country basis.
That this question is not only ignored by the NYT article, it is obfuscated, makes it doubly intriguing to me. In describing this second wave, the article claims that “Western European countries such as Italy and Belgium, which were hit hard the first time, are suffering nearly as badly now. Portugal is doing even worse.” These sentences give the impression, especially when combined with the potent visual image of the matching waves, that Western Europe on the whole is having a similar experience now as it had in the spring. As we’ll see, that’s terribly misleading.
The following chart shows, for the 31 European countries tracked in the NYT article, the peak daily deaths in the first wave and in the second wave. The third column is the ratio between the two peaks.
I’ve sorted the chart by the size of the first wave. What is immediately apparent, if you even glance at the third column, is that there’s an inverse relationship between the size of the first peak and the ratio of the two peaks. Those countries who had a small first wave were much more likely to have a second wave bigger than their first.
In fact, of the eight countries with the largest first wave, none of them have had a second wave as large as the first. Only one, Italy, had a second wave even seventy percent as large as the first. The median peak deaths per 100,000 in the top eight countries was 1.44 in the first wave. In the second wave, it’s 0.83. More than forty percent lower.
By contrast, the median peak deaths per 100,000 in the bottom 23 countries, was 0.16 in the first wave and 0.75 in the second wave. The second wave in these countries is about five times larger than the first wave was!
The real story is not that Europe is facing a second wave that is identical to its first. The real story is that many of the countries who were spared by the first wave are getting hammered by this one, but those countries who faced the biggest wave the first time around are doing substantially better now.
Okay, understanding the reality of the situation, we can begin to ask the truly interesting question: why is it that countries who did worse in the spring are doing better in the fall, and vice versa?
To begin to think about that question, it helps to look at the factors that distinguish the countries in the two groups. Here is the list of the eight hardest hit countries: Belgium, Spain, France, Ireland, United Kingdom, Italy, Sweden, Netherlands.
A few observations:
1. They are mostly contiguous. With unrestricted travel within Europe, the virus moved easily across the borders of these countries until restrictions were put in place.
2. They are both populous and densely populated. Greater density makes transmission rates higher.
3. They are among the most economically developed countries in Europe, with tight connections to the global economy. The virus came to these countries first.
These factors easily explain why the first peak was so much higher in these countries than it was in other European countries. The restrictions that brought the first peak down quickly came too late for these countries to avoid substantial COVID-related deaths, but much of the rest of Europe was spared the worst of the pandemic.
When the second wave came this fall, would one logically expect those same factors to matter? I think the answer is yes, though to a lesser extent. The virus had the entire summer to slowly seep into the rest of Europe, so global connectivity mattered less. But population density and travel within Europe still ought to have lead, all things being equal, to substantially larger waves of COVID in the same eight countries that suffered the most in the spring.
Why didn’t they?
Two answers appear plausible: either the countries who were hit hard in the spring have better protected themselves now by more rigorous restrictions and greater compliance with distancing and mask-wearing, and/or they are benefiting from some level of herd immunity.
Before we dive into the question of which explanation appears most important, let’s take a look at the experience in the United States. I repeated the same analysis for the U.S., looking at the relative peak within each state. The chart below is set up in the same manner as the European chart.
It doesn’t take long to see that the pattern in the U.S. is much the same as in Europe. Those states who hit a high peak before this fall have been “rewarded” with a lower peak now. Those states who were spared before this fall are having a relatively harder time now.
In fact, the pattern in the U.S. is even more extreme than in Europe. There were nineteen states who has peak deaths per 100,000 over 0.8 in the fall. (The same cutoff I used for Europe.) The median in those states was 1.40. It is now 0.58. For those nineteen states, the current wave is less than half as large as it was before.
In the states which were not hit as hard before the fall, the median has jumped threefold: from 0.28 to 0.86.
In the U.S., as in Europe, the first wave was worst in its most dense, globally connected, geographically contiguous area: the northeast. The rest of the country was relatively spared as lockdowns clamped down on transmission before the virus became as prevalent in other regions as it was in the large, northeastern cities.
The U.S., unlike Europe, saw a second wave, smaller in size but still significant, during the summer. This wave came mostly in the south, as restrictions were eased and hot weather forced people inside in states like Florida, South Carolina, Texas and Arizona.
As we look at the current wave in the U.S., let’s consider again the question of whether we would, all things being equal, expect those states who had the highest waves prior to the fall to have them again now.
I think the answer to that question is a clear yes. The density and interconnectedness of states like New York, New Jersey, Connecticut, Massachusetts, Washington D.C., Maryland and Pennsylvania make them ideal locations for viral transmission. And yet, the size of the current wave, as measured by deaths per 100,000, is greater in North Dakota, South Dakota, Iowa, Kansas, Nebraska, Wyoming, and New Mexico than it is in any of those states.
As with Europe, it is not clear from this analysis whether restrictions or herd immunity is more responsible for the observed outcomes. I doubt it’s a coincidence that the former list is made up mostly of blue states and the latter mostly of red states. But a suggestion that restrictions (and compliance with them) entirely explains this pattern probably overstates the case.
If herd immunity were not a factor at all, we would expect to see at least a handful of European countries and American states with large first waves and equally large second waves. The mishmash of regulations in Europe and across the U.S. is not consistent enough to produce the very distinct patterns seen here. And yet, of the eight European countries and seventeen American states who had a peak daily death rate of greater than 0.8 per 100,000 prior to this fall, only two, Illinois and Indiana, have had fall peaks as large as their earlier peak.
Back in July, I wrote an article suggesting that there was evidence to suggest that the hardest hit parts of Europe and the U.S. might be close to achieving a level of herd immunity that would mitigate the impact of any future waves. The gist of the argument was that some segment of the population may be naturally not susceptible to SARS-CoV-2, and that when an additional 20–25% of the population develops immunity through exposure to the disease, the sum of those two segments brings the community close enough to the herd immunity threshold to slow the virus’ spread. The data discussed above appears to corroborate that theory.
If it is true that herd immunity is playing a material part in restraining the size of the second wave in some locations, this bodes very well for the future. In Europe, 19 of the 31 countries have now had peaks of 0.8 per 100,000 or greater. In the U.S., 36 out of 50 states have reached that level. On both continents, it is, for the most part, only the least populous places that haven’t reached that level. (In the U.S., California, and in Europe, Germany, stand out as exceptions.)
The vast majority of people, therefore, live in places that have built up some level of herd immunity. As vaccines become available, we will not be starting from scratch. While in a fully susceptible population, it might take vaccinating 60–70% to eradicate the virus, we are likely to begin to see the prevalence of COVID-19 fall long before that many people receive the vaccine.