Another day of mixed news.  Let’s get the bad news out of the way first.  Ohio set another daily record for new lab confirmed cases reported with 619, up 8 from the day before.  When you look at the cases by date of onset though (the important trend), the uptick is clearly detectable, but not as sharp.  When looking at the county level numbers, this continues to be mostly driven by the outbreaks at the Marion Correctional Institution (152 new cases yesterday) and the Pickaway Correctional Institution (57 new cases yesterday).  Franklin County also had 108 cases reported which shows up clearly when we look at date of onset as they their record prior to this week was 47 and they have 86 and 71 cases this week.  Not sure what’s driving that one specifically.  The uptick has definitely affected the graph of date of onset.  Our two-week trend is now positive.

The question is should we be concerned about the increase?  I think the answer is probably not.  We expect that the virus spread will have occasional localized outbreaks and the circumstances of those outbreaks matter.  I’d be far more concerned about an outbreak traced to a grocery store vs an outbreak in a mostly self-contained environment like a prison.  As long as the general trend outside of these outbreaks remains down (and it currently is), then we’re on the right track in the state.

Now for the good news.  Yesterday, a paper was submitted for publication based on a study performed by Stanford University.  The study tested a number of people in Santa Clara, CA, for antibodies that indicated they had previously been infected by COVID-19.  Based on their data, they found that the number of cases was potentially 50-85 times more than the number of confirmed cases.  At first glance, this sounds terrible, but if true, it’s great news.  Here’s why – we have strong evidence to believe that a significant number of people have been infected by COVID-19, but have mild or zero symptoms, but it’s really difficult to get an accurate count of these people.  We don’t have the testing resources right now and even if we did, they don’t have symptoms that would drive them to get tested.

If we assume that this virus spreads quickly (high confidence this is true), then the true threat of the virus is based on how many people are susceptible to it and the death and hospitalization rate of those infected.  Right now, we are generally assuming that nearly everyone is susceptible because this is a novel virus (not similar to anything that has circulated previous).  If this has been circulating longer than we think, we are starting to build some level of immunity in the population – this is important.  But even more important is the affect this has on death rates.  The death rate is calculated based on number of deaths divided by number of cases.  Don’t get too hung up on how we attribute deaths to COVID-19 – in the USA, I have found zero evidence we are doing things like counting auto accident deaths as COVID-19 as they test positive.  Regardless, in Ohio, our current naive death rate (knowing that we aren’t testing everyone who is infected) is 418 / 9107 = 4.6%.  Let’s say that we had 50 times more infected people that we have currently counted.  The bottom part of the fraction would increase to 455,350.  Our new death rate would then be 418 / 455,350 = 0.09%.  The difference between those numbers is the former is worth making massive efforts to contain the virus and the latter means go back to work tomorrow because the yearly flu is much worse.


If I could make the word above red flashing lights with sirens I would do it.  There are a LOT of caveats around this study that mean we should avoid getting too excited even as we acknowledge the good news.

  1. Santa Clara is not Ohio.  Santa Clara and California is likely further into this outbreak than we are given their much higher interaction with China.
  2. There are differences in testing availability in California vs. Ohio.  The number of confirmed cases that we would use as our starting point is very different so the math could work out quite differently than my example above.
  3. The testing kits being used for this are new.  If you’re not familiar with lab testing in general, we have to deal with the concept of false negative (really was infected, but test says no) and false positive (really was not infected, but test says yes).  The false negative and false positive rates for this test are not very well known because of the newness of the test.  It adds much uncertainty to the conclusion even as the authors attempt to adjust for that.
  4. The relatively small percentages of confirmed cases and the results of this test means that the false negative and false positive rates are nearly as big as the results themselves.  Again, the authors attempt to adjust for that, but we have to acknowledge the uncertainty in the results.

The bottom line is that we need to see more of these types of studies.  I learned yesterday that Ohio State has started a similar field exercise where they will sample people across our state and continue to retest them over time.  I hope they find that the virus is much more prevalent than we thought, but either way, the data that comes out of that effort will be incredibly helpful in charting our path forward.