Is it misleading to cite expanded COVID-19 testing as one possible explanation for the post-lockdown increases in confirmed cases that some states have seen? Or is it misleading to cite post-lockdown increases in confirmed cases without taking expanded testing into account?
I am inclined to think the latter is more true than the former. But maybe you should not trust me, since I have long been skeptical of sweeping restrictions on movement and economic activity as a response to the COVID-19 epidemic.
By the same token, however, you should not trust The New York Times, which has long supported such policies in its news coverage as well as its opinion pages. "Pence Misleadingly Blames Coronavirus Spikes on Rise in Testing," says the headline over a recent Times story. But judging from the evidence presented in the story, that is not what Vice President Mike Pence actually did.
"I would just encourage you all, as we talk about these things, to make sure and continue to explain to your citizens the magnitude of increase in testing," Pence said during a call with governors, according to a recording obtained by the Times. "And that in most of the cases where we are seeing some marginal rise in number, that's more a result of the extraordinary work you're doing."
In other words, Pence thinks we should take expanded testing into account when we look at recent COVID-19 trends. He did not say that was the sole explanation for what is happening—just that it is an important factor to consider, especially when an increase in cases is "marginal," which is true.
"Seven-day [case] averages in several states with coronavirus outbreaks have increased since May 31," the Times says, "and in at least 14 states, positive cases have outstripped the average number of tests that have been administered." Since every state that imposed a lockdown has at this point lifted or loosened it, those numbers hardly seem like conclusive evidence in favor of those policies, or against reopening the economy. Nationwide, new cases have been declining since the end of March, and that downward trend continued as lockdowns were lifted.
By contrast, in Texas, where I live, the daily number of new confirmed cases has been rising since late May, from 1,083 on May 26 to 4,246 yesterday—a fourfold increase. The seven-day rolling average of new cases has more than doubled, from 928 on May 26 to 2,386 on June 16. During the same period, daily COVID-19 hospitalizations rose from 1,534 to 2,518—a 64 percent increase. By comparison, the number of viral tests performed each day had risen by just 29 percent as of June 15.
The increase in testing during this period clearly cannot account for the increases in cases and hospitalizations. Furthermore, the percentage of tests that were positive, based on a seven-day average, rose from 4.3 percent on May 26 to 6.7 percent on June 15, which reinforces the point that Texas is seeing a real and substantial increase in infections. So yes, The New York Times is right to suggest that emphasizing increased testing can be misleading, although the paper is wrong to dismiss the issue out of hand.
Unless you think lockdowns were completely ineffectual, it makes sense that COVID-19 infections would rise as restrictions on movement and economic activity were loosened. Yet Texas started doing that on April 30, and it did not see notable upward trends in cases and hospitalizations until this month, weeks after you would expect to see some impact from lifting the lockdown.
One popular explanation of the timing is that the new infections we are seeing now can be traced to gatherings on Memorial Day weekend, when people who may not have seen each other for months got together, often in close quarters without regard to social distancing guidelines. Around the same time, Texas also saw mass protests against police brutality in reaction to the May 25 death of George Floyd in Minneapolis.
Those factors are plausible explanations in light of what we know about the outsized role of "superspreading" events in the epidemic. In Hong Kong, epidemiologists Dillon Adam and Benjamin Cowling found, "just 20 percent of cases, all of them involving social gatherings, accounted for an astonishing 80 percent of transmissions." Another 10 percent of carriers "accounted for the remaining 20 percent of transmissions," meaning that 70 percent of people infected by the virus did not pass it on to anyone. Those findings, Adam and Cowling note, are consistent with the results of other studies.
The implication is that targeted measures aimed at curtailing superspreading—such as masks, social distancing rules, and restrictions on large, crowded gatherings, especially indoors—are likely to be more cost-effective than broad business closures and stay-at-home orders. The latter approach not only imposes enormous economic costs but tests the patience of people whose cooperation is required to make control measures work. If Texas never had a general lockdown, Texans might have been less inclined to flout social distancing rules on Memorial Day or more willing to wear masks in public—precautions that are far less onerous than the mass quarantines nearly all states imposed.
Those orders, especially insofar as they drew distinctions between permissible and prohibited activities that made little or no sense, left people exhausted by arbitrary dictates. In this context, it is hardly surprising that some Texans rebel even at relatively modest precautions when they are recommended by the same public officials who deprived them of their livelihoods and confined them to their homes. That reaction, while understandable, poses a real problem as we settle in for the long haul and wait for a vaccine to save us. Precautions that might have been sustainable without lockdowns may prove less tolerable after that bitter experience.
There is some (relatively) goods news in the data from Texas. The seven-day average of daily COVID-19 deaths has actually fallen since the lockdown was lifted, although it is higher now than it was in late May. That positive trend may reflect improvements in treatment and/or changes in the preexisting health status of newly infected people.
Youyang Gu's epidemiological model, which has an impressive track record of projecting COVID-19 fatalities, shows daily deaths in Texas gradually declining through September. The model's estimate of the reproductive number for Texas—the average number of people a carrier infects—has fluctuated only slightly since April 30, sometimes dipping below one and sometimes rising a bit above that threshold, which indicates a continuing epidemic. The model's projections suggest that the reproductive number in Texas will remain slightly below one during the next few months, indicating an epidemic on the wane.
Whether that actually happens depends on how Texans behave now that they are free to move about and return to work. The lockdowns, whatever their marginal contribution to slowing the spread of the coronavirus (still a matter of much dispute), may simply have delayed COVID-19 deaths rather than preventing them. The eventual death toll will depend on people's willingness to follow sensible guidelines, and it seems that many of them are in no mood to do that.