Covid-19: Data, data, data (as well as test, test, test)

Let me try to make a single point very briefly (for once).

If you’re trying to manage an epidemic, an exercise which has become almost the sole concern of governments worldwide, what you need is data.

In the UK right now, all the public really have visibility of is of (1) confirmed cases and (2) deaths.

Now, confirmed cases is not very useful because the UK is only able at present to test serious (hospitalised) cases, royalty and politicans. The number of confirmed cases is a huge underestimate of UK infection rates.

The number of deaths is probably reasonably accurate, but there are some problems:
– There is debate about the cause of death in some cases.
– Some commentators suggest that evidence of Covid-19 infection does not prove that the coronavirus was the cause of death.
– There may be at least a delay before the inclusion of some cases that didn’t make it to hospital.
– There may be cases of deaths at home, where the cause is not immediately determined, in part due to the shortage of tests for Covid-19 (I believe there were cases of this happening in Wuhan).
– But most of all, for the purposes of determining the effectiveness of the lockdown, there will be a long time-lag (around 3 weeks) before deaths start to come down.

What we really need is reliable leading data. Here are some metrics that come to mind:
(3) Hospital admissions with (a) confirmed and (b) suspected Covid-19 (though “suspected” will hopefully become unnecessary once more tests come onstream). These should start to come down 10-14 days after an effective lockdown.
(4) The number of hospital beds occupied. Although I was grateful for the data (since I like to be informed), I was shocked to hear in yesterday’s Downing Street press conference (with the last few uninfected members of the response coordinating team!) that 6,200 hospital beds are already occupied by Covid-19 patients with 33,000 available (though how many of the field hospital beds at Excel and elsewhere that includes I’m not sure). With the infection rate doubling every 3-4 days and hospitalisation typically taking place maybe 10-14 days after infection, and the country only having been in lockdown for a few days, you can do the math yourself to judge whether that’s likely to be enough (bear in mind patients are likely to be hospitalised for, say, a week on average – that’s an educated guess).
(5) Number of patients undergoing ventilation. As I understand it, this is a step before true intensive care, but I suspect only a lucky(?) few are going to get the works, e.g. ECMO.

To track the UK epidemic, we’re obviously going to need regional (especially city-region, since that’s where the virus spreads fastest) data, in part because there may be differences across the country in the way the lockdown is implemented.

But I think it would be useful to also break it down by suspected route of infection, specifically in the household or outside.

The point is – and this was well made by my sister, credit where credit is due – infections will continue to occur in the household after lockdown, as those infected before confinement become contagious. Since it will be extremely difficult to avoid household transmission of Covid-19, overall infection rates could even continue to go up after lockdown.

But to determine how effective the lockdown has been we may want to strip out household-acquired infections from the data.

There’s an analogy here with the transmission during the early phase of the epidemic. I heard on Friday 6th March that there were 30 cases of local transmission in the UK. That’s when I knew we really had a problem.

Presumably the NHS and Public Health England (and the equivalents for the other UK nations) have much of this data. I’d like to see more of it put into the public domain for two reasons:
(1) So that the rest of us can draw our own conclusions. We’re all in this together and there should be public debate of the course to take.
(2) That may be the best way to make sure the data reaches the people who need to see it. This is one reason I don’t believe the Chinese data about their epidemic was manipulated, at least not by their central government.

It’s a central tenet of any kind of process management that you can only control what you can measure. That principle must be applied to the Covid-19 epidemic.