The Effect of Opening is Already Showing Up in the Numbers

If you know where to look.

Amid widespread speculation that relaxing stay-at-home orders would reverse the progress made in stemming the transmission of COVID-19, an analysis of recent trends shows that those fears are justified.

Although many states have just begun to re-open their economies and decrease restrictions on businesses and social gatherings, the effect of those changes is already quite clear. For the first time since I began tracking the number cases and deaths in the 84 counties originally hit hardest by the virus, the height and duration of the curve has begun to grow instead of shrink.

This change isn’t readily apparent if you look at the country’s overall numbers: on the surface, things appear to be improving: both deaths and new cases are declining.

In fact, the trends in new cases and deaths are quite alarming. In many counties across the U.S., transmission of the virus is beginning to spin out of control.

To see what’s happening, it will be helpful to rewind to the scene in late March, when it was clear that the virus was spreading throughout the country, but no one yet knew how rapidly or broadly it was spreading. At that time, it appeared that cases would peak at a level of five times the amount at that time, and the peak would occur in late April.

Image for post
Forecast growth of new cases per day in the 85 counties that first reached 150 confirmed cases, based on New York Times database through 3/27/2020. The methodology used for generating this forecast was different that most epidemiological projections, in that it was based entirely on an extrapolation of the growth curves seen across the United States. I will explain the pros and cons of this technique in a different post.

At that time, there was some evidence that social distancing was having an impact, in that, rather than following an exponential rise that would have quickly led to half or more of the country becoming infected, the rate of growth was already beginning to slow. Still, the trends were quite frightening, and implied an outcome far worse than we’ve seen to date.

When I revisited the analysis a week later, the evidence around social distancing’s impact was much stronger:

  • 87% of the counties I’m tracking had improved their trajectory
  • The median county was growing 30% slower than the prior week
  • The total expected number of cases was down by about 1/3

Two weeks after that, by April 16, the improvements were substantially greater, as is clear from the graph below.

Image for post
Forecast growth of new cases per day in the 85 counties that first reached 150 confirmed cases, based on New York Times database through 4/16/2020.

Two weeks after that, on May 1, the improvement continued, though only slightly. The median county had a total case count 8% lower than I’d forecast in mid-April.

By this time it was clear that most of these counties had passed their peaks, even though the number of overall daily cases in the U.S. was still at a plateau. Because these counties were hit hard first, they were further along their growth curve than the U.S. as a whole.

It was clear, therefore, that the daily number of new cases in the U.S. would fall during the month of May, as the decline accelerated in these counties and the rest of the U.S. began to decline as well.

Over the last four weeks, that is exactly what’s happened. And that decline has given our leaders the comfort required to begin loosening restrictions.

But here’s the problem: May and June were on track to be the months of rapid deceleration of case counts and deaths. By the end of June, had the trends of one month ago continued, just about all of these counties would have nearly eradicated the virus. Although the numbers did decline in May, they didn’t decline as quickly as they should have.

First, let’s look at the death data:

  • Forecast Deaths (May 2–28) 20,140
  • Actual Deaths (May 2–28) 21,756
  • Variance: 11%

That may not seem too bad, but the numbers are much worse when you look at cases. This is important because the trends show up first in the case counts.

  • Forecast New Cases (May 2–28) 214,286
  • Actual New Cases (May 2–28) 296,481
  • Variance 38%

And here’s where the numbers get very scary: between May 2 and May 15, the surplus number of new cases was only fourteen percent. The second half of May, the picture changes dramatically.

  • Forecast New Cases (May 16–28) 62,083
  • Actual New Cases (May 16–28) 123,450
  • Variance 99%

The number of new cases doubled the forecast in the last two weeks!

Why isn’t there a widespread uproar over this?

Because the number of new cases still declined from the prior two weeks. It just didn’t decline nearly as rapidly as it should have.

It’s important not to be fooled by the fact that the overall new case counts are decreasing. The sharply slowing rate of decline is the canary in the coal mine. There is a significant danger now that the progress we made over the last two months will be halted and potentially even reversed.

At this point, it is impossible to forecast the path the infection and death rates will take. Up until now there were consistent and predictable trends. Today many parts of the country are seeing those trends reversed. There are, however, still a number of areas where the decline continues to accelerate.

The following chart shows the tremendous differences divergence in the paths taken across the U.S.

Image for post
Variance of actual new cases vs. projected new cases by county for the period from 5/2/20 to 5/28/20

It appears quite likely that, from here on, we’ll see a growing degree of geographic divergence.

Some areas, such as New York and the surrounding areas, as well as New Orleans and much of Michigan do appear to be close to putting the problem behind them. They will reach the point within the next few weeks where infection rates are low enough that testing and contact tracing could keep the virus from ever reappearing in a significant way.

On the other hand, southern California, especially Los Angeles, appears on the verge of losing control. With the relaxation in restrictions on gatherings and the opening of the economy, they could be a month or so away from experiencing the level of devastation witnessed in New York a month ago. Other areas of concern include Milwaukee and Dallas. It’s not too late for them to reverse these concerning trends. But in a few weeks, it might be.

Author, novelist, investor.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store