It’s been ten days since I last ran the numbers on covid diagnoses and deaths in the US. Has there been an appreciable trend, either up or down, since then?
Here are the official diagnostic totals by 10-day interval:
- June 12 – June 22: 27.1K new test-positives per day
- June 2 – June 12: 22.0K new test-positives per day
- May 23 – June 2: 21.8K new test-positives per day
- May 13 – May 23: 24.2K new test-positives per day
- May 3 – May 13: 24.5K new test-positives per day
So there’s been a 23% bump in test-positives over the last ten days, reversing the modest decline experienced over the prior month. Is this a true increase in test-positive cases, or an artifact of increased testing? Likely it’s a bit of both: per national data, testing rates are up over the past ten days, but so too are test-positive rates.
There’s no intrinsic reason why an increased testing rate should result in a lower test-positive rate. Having had months of experience dealing with this new disease, physicians might be getting better at spotting cases based on clinical presentation even without testing. Also, flu season is over, making differential diagnosis easier.
As I’ve noted frequently, the official diagnostic tallies grossly underestimate the actual infection rate. Consequently I’ve been using a proxy estimate of infection rates based on death rates. Covid-related deaths are probably undercounted too, but not as drastically as are new test-positive diagnoses. Based on well-designed studies, I’ve been estimating a US age-adjusted mortality rate of 0.6% for those who are covid-infected. Here are the 10-day interval calcs:
- June 22 — June 12: 598.5 deaths/.006 = 100K new cases per day
- June 2 – June 12: 852.6 deaths/.006 = 142K new cases per day
- May 23 – June 2: 939.5 deaths/.006 = 157K new cases per day
- May 13 – May 23: 1316.0 deaths/.006 = 219K new cases per day
- May 3 – May 13: 1700.4 deaths/.006 = 283K new cases per day
Using the death proxy estimate, new cases have decreased 30 percent over the past ten days — a trend precisely opposite to that of newly diagnosed test-positive cases. And this downward trend has been persistent, with a decrease of 65% from a month and a half ago.
What might account for the discrepancy in these two methods of estimating infection rate trends? Maybe the covid mortality rate is decreasing. Has treatment improved substantially? While the overuse of ventilators has decreased, and while some treatments show modest improvements in symptoms, there have been no breakthroughs dramatic enough to decrease deaths by two-thirds over the past month and a half.
Maybe the elderly — those most vulnerable to severe illness and death from covid — have so successfully isolated themselves from contagion that the age-adjusted estimated mortality rate needs to be revised downward. That contention isn’t supported by available evidence: the longitudinal data show stable proportions of older people in newly diagnosed cases and in deaths. The 0.6% mortality rate estimate still holds, and so does the estimated reduction in new infections.
Maybe the increased testing is identifying people who are experiencing less severe symptoms and so are less likely to die from the disease. This seems the most likely resolution to the paradox. More widespread testing in clinical practice brings infected people in for treatment and testing sooner. While earlier treatment might improve outcomes appreciably, it’s more likely that the causality is reversed: a reduction in severe cases frees up capacity in doctors’ practices and in hospitals for treating the less severe cases. And let’s not forget the financial incentives: empty schedules and empty beds translate into reduced revenues.
There’s the natural time sequence to be considered: infection, then diagnosis and treatment, then recovery/death. It’s certainly possible that a jump in test-positives today will work its way through to a corresponding spike in deaths a week or two from now. If that logic sounds familiar, here’s what I wrote ten days ago:
Deaths are a lagging indicator of infection: someone who dies today likely became infected two weeks ago. There are reports that hospitalizations are on the rise, portending a new surge in deaths. We’ll have to wait another week or two to see how that shakes out.
Ten days have elapsed since then. Instead of a surge, there’s been a marked decrease in deaths. A couple of state-specific results of interest:
Minnesota. George Floyd was killed by Minneapolis police on May 25. Protests were expected to generate a surge in new cases. Instead:
- June 12-22: 343 new test-positives and 11 deaths per day
- May 15-25: 652 test-positives and 20 deaths per day
Arizona — recognized as the spikiest state. Over the past ten days the rate of test-positives has doubled, while the death rate has remained the same.
Based on the body of evidence, it seems to be the case that, despite the reopening of America and contrary to my expectations, the coronavirus epidemic is waning.
Whether the downward trend in infections and deaths persists or reverses itself remains to be seen. Public health initiatives for quarantining and contact tracing have met with indifference and at times aggressive resistance. Social distancing and mask-wearing seem to be declining. Schools, sharing superspreader potential with nursing homes and religious services and saloons, are likely to open in the fall.
I’ll redo these calcs in another ten days.