When a new virus emerges, there is naturally a reaction to worry in the absence of data. Epidemiologists will then typically use theoretical models to project the impact of the new virus, and a response can be formulated based on those results.

As the new virus spreads, actual data emerge and becomes more credible every day. Over this period, more weight should be given to the actual data when compared to the theoretical projections.

We’ve reached a point with Covid-19 that the actual data starts telling an interesting story. As an actuary, I really like digging into the data, so this thread will unpack it one step at a time, and potentially draw some conclusions.

The United States has done by far the most COVID tests - in particular in New York State (where by all accounts things have been bad). Their data is really great too, so I’m going to focus on New York state for this analysis.

As of 25 April, New York had carried out 777,568 COVID tests, of which 282,143 were positive (36% of those tested were positive). As the quantum of tests increase we can reliably extrapolate these figures into the untested population.


Data source: https://covidtracking.com/data/state/new-york

The population of New York State is 19.45 million people. Extrapolating, 36% of 19.45 million could suggest that as many as 7 million New Yorkers could have contracted COVID. Naturally, in the early days of the virus, those more likely to have the virus will have the test...

..so 7 million is most likely an overstatement. This is why antibody tests are needed to confirm a more accurate estimate of COVID cases. Two days ago, New York State’s early antibody testing predicted 13.9% of the population may have had COVID (about 2.7 million people).

This is the more conservative number, so let’s use this. The 36% positive test rate suggests that the 13.9% total prevalence is realistic. It’s VERY important to use this as the virus exposure, rather than actual total positive tests since the deaths will happen regardless.

Now we can look at the total reported COVID deaths. For New York State this currently sits at 16,599. I believe this number is a conservative measure since a COVID death is counted even when COVID wasn’t the primary cause of death.

Other reports suggest the COVID deaths may be higher ( https://www.nytimes.com/interactive/2020/04/21/world/coronavirus-missing-deaths.html ) but see this thread for my theory.
 https://twitter.com/peter_castleden/status/1254207778895400967?s=20

The most accurate number we have is 16,599 so let’s use that. Dividing this number by the estimated number of positive cases of 2.7 million suggests that the actual mortality rate of COVID is 0.6% or 6 people per thousand.

New York City has some fantastic data here ( https://www1.nyc.gov/site/doh/covid/covid-19-data.page) which breaks the death and cases data down into more granular detail. In the link, they show the death rate per 100,000 by age band.

It is clear how much age plays a role, with 13.59 people per 100,000 between age 18 and 45 dying with COVID. For those over 75, this number is 70 times higher (at 970 deaths per 100,000). The average for the city is 130 deaths per 100,000.

I overlayed this data on top of NYC demographic data and can conclude that for every 100 COVID Deaths, 47% of the deaths would be people over age 75 and 71% of total deaths being people over age 65, and only 28% of the death being with people under age 65...

(despite them making up 86.5% of the total population).


Flu has a widely quoted fatality rate of 0.1% (although if we counted all deaths of people who happened to have flu at the time of death as flu deaths, that number would likely be significantly higher).

Regardless, taking this age data into account the likely COVID mortality rate of those under age 45 is 0.1% (precisely the same as flu, except COVID, kills zero children). For those age over 75, it’s scarier with a mortality rate suggesting 7% (or 70 times more deadly than flu)

Given that it’s clear that this virus is neither as bad a predicted by the theoretical models, and it’s overwhelmingly affecting the elderly, isn’t is time to end global lockdown and focus ALL resources saving the lives of the elderly.