Thread by @christopherruhm

1/12 I’m guessing many of you, like me, are obsessing over the statistics on the COVID-19 (looking at the maps & charts many times per day). Here are some thoughts on how to interpret them.

#econtwitter, #epitwitter, #coronavirus

2/ I’m not an epidemiologist but have considerable experience conducting statistical analyses, including of mortality rates

We currently have two main statistics available to us: 1) # of known cases (24,378 in the US as I write this) & 2) # of deaths (340 in the US currently).

3/ What influences each of these? Confirmed cases will be high due to some combination of: 1) high prevalence of infections and 2) high rates of testing. Mortality will be high due a combination of: 1) high prevalence of infections and 2) the adequacy of medical care provided.

4/ Actually, deaths are also influenced by other factors such as population demographics (the frequency of infections among smokers, the elderly etc.) and special circumstances such as undetected cluster outbreaks like those that initially occurred in Wuhan, China or in the

5/ long-term care facility in Kirkland Washington. & the rise in deaths corresponding to increased infections occurs with a lag.

Subject these caveats, we may gain useful information by looking at the deaths as a percentage of the number of confirmed cases (hereafter PERCENT)

6/ At this writing, some examples are: US: 1.4%; China: 4.0%; Italy: 9.0%; Spain: 5.3%; Iran: 7.5%; Germany: 0.3%; South Korea: 1.1%; Singapore (0.5%).

A likely conclusion is that countries with the high PERCENT’s (e.g. Italy, Spain, Iran) probably actually have many more cases

7/ than have been confirmed & medical systems that are failing to cope well with the crisis. Some other countries, like Germany & Singapore are probably testing extensively & providing excellent medical care. We should learn what they are doing!

8/ Where does the US stand? The 1.4% mortality/confirmed case ratio is fairly good. However, there are two important caveats: 1) the explosion of COVID-19 probably reached the US later than many other countries & so deaths may rise rapidly;

9/

& 2) the health care system, in many place, appears to be at the point of overload (i.e. shortages of masks, hospital gowns, ventilators), again possibly presaging a rapid increase in the coming weeks.

It is also instructive to look at the PERCENTS across states.

10/ Here are some examples for large states (as of this writing): NY: 0.8%; CA: 1.8%; FL: 1.4%; TX: 1.1%; PA: 0.5%; IL: 0.8% OH: 0.4%. All these PERCENTS actually look pretty good. Hopefully, they do not just reflect delayed onset of the disease. We’ll see.

11/ Also, & NY’s high # of cases is at least partially due to more extensive testing than in other places.

More alarming, some of the highest PERCENTS are in places getting less attention. Here are some examples. LA: 2.6%; GA: 3.6%; OR: 2.9%; IN: 3.2%; MO: 3.3% ; KY: 3.6%.

12/ Some of these #’s are small, so the error bounds are large. Still, good reason to be concerned.