Death toll predictions are a mug’s game

John Ware asks if our trajectory is going to be so different to the rest of Europe’s

Artillery Row

Predictions of the UK’s Coronavirus death toll have veered giddily from one extreme to another.

At the weekend the BBC reported “deaths are growing faster, doubling every two or three days.”

And on Monday Professor Alan McNally director of Birmingham University’s Institute of Microbiology and Infections was quoted as saying: “fatalities will continue to double every three days for another two weeks at least, given the lag time between infection and death.”

Prediction models are only as good as their assumptions

That would mean a doubling of the increase in deaths every three days until April 12, or thereabouts. It’s true that yesterday UK hospital deaths doubled in just 24 hours from 180 to 381 – the highest daily jump yet. But is each day’s increase in deaths really set to double every three days for another two weeks? If so, the cumulative total of deaths by 12 April – Easter Sunday – would rise from yesterday’s total of 1789 (hospital deaths) to over 25,000.

Professor McNally made his prediction after criticism of a fellow scientist for underestimating the UK death toll.

Tom Pike, from Imperial College had anticipated only 5700 UK Coronavirus deaths on the assumption that the UK would broadly follow the trajectory in Wuhan, China.

Prediction models are only as good as their assumptions and when on Saturday the death toll reached Pike’s prediction that the UK would max out at around 260 deaths in a day, he acknowledged: “We don’t know where that uptick is going to go, or if it will keep going in the same direction.”

As a former BBC colleague says, death toll predictions have become a mug’s game.

Nonetheless, I’m going to have a go – not by injecting assumptions – but by comparing the UK with the hard numbers of Europe’s worst affected countries.

I do so with this caveat: that deaths between different countries may not be directly comparable without knowing if the death certificates count Covid as the principle cause of death or include deaths where Covid is a contributing cause; that the UK death toll to date of 1789 counts only those who died in hospital and not at home or in residential homes. Still, the hospital toll is the best measure of how the NHS is coping.

The Government says it expects deaths to peak the week beginning 13 April and that seems reasonable because only then will the full effect of the current restrictions have bitten, the infection base having (hopefully) been decimated by three weeks of social distancing.

If, as Prof McNally says, the increase in “fatalities will continue to double every three days” until Easter, the cumulative total of 25,000 by then would by no means be the end of it.

Thousands more deaths would follow while the numbers flatline for several days, (as the peak appears to be doing in Italy and Spain) before becoming established in the descent during which yet more thousands would be dying although at a slowing rate. Whilst in fairness to Professor McNally he has not put a figure on the final toll of his prediction, it is not difficult to imagine that its logic implies a total of some 40,000 deaths.

The 25,000 deaths ascending to the UK peak, I should emphasise, assumes a uniform rate of daily increase in each three day period — which there certainly would not be. Yesterday’s jump from an increase of 180 to 381 shows how the numbers are liable to bounce around. Nonetheless, were there to be doubling “every three days for another two weeks at least” the final period of doubling would end with several thousand deaths a day by Easter, perhaps as many as 5000 deaths before the growth curve flattens.

There is already anecdotal evidence of the ambulance service struggling to cope so if daily deaths did reach anything like that at the peak, not only would the NHS and ambulance service be in meltdown; people would literally be dying in the cars of their relatives trying to get them to hospital that anyway wouldn’t have a spare ventilator, let alone an oxygen mask.

Mercifully, judging by the Italian, Spanish and French trajectories to date that following their respective lockdowns, such apocalyptic scenes seem pretty doubtful.

Italy has been locked down since 9 March when deaths were 463. True, in its first week the daily toll did almost double over two consecutive three day spells, but it has not doubled in any of the five subsequent three day periods. Italy (so far) peaked on 27 March at 919, since when deaths have declined by an average of 96 over the last four days (889, to 756, to 812, to 837). So far so flatlining.

A similar picture of a flattening growth curve is emerging in Spain: deaths doubling twice in the days immediately after lockdown on 17 March, but not doubling now for over a week.

France has seen more doubling, probably because they locked down earlier in their cycle than Italy and the growth curve is always sharper at the start; Likewise UK deaths have been doubling every three days since lockdown on 23 March. Was it really necessary to leave lockdown that late? That will be just one of many key questions at the inevitable inquiry into the government’s woeful lack of preparedness for the crisis.

However, we are respectively still some 700 and 550 deaths behind both Italy and Spain at the same 8 day point after lockdown, so we might hope to begin our levelling off within the next seven days at around or even less than the 800 plus daily toll at which both Italy and Spain now appear to be peaking.

Certainly we are unlikely to see anything like the precipitously large daily rises peaking at several thousand deaths implied (though not stated) in Professor McNally’s prediction.

I am not a statistician or epidemiologist or any kind of expert in these matters. I simple pose the obvious question: why would the UK trajectory be so markedly different from Italy, Spain or France?

It’s hard to see that it would be.  Yes, it’s early days, and as NHS England’s national medical director Dr Stephen Powis stressed yesterday: “We are not out of the woods. We are very much in the woods – and we must not take our foot off the pedal.”

Nevertheless, if we do peak at, or even less than the Italian and Spanish peaks, the total UK death toll in both the ascent up to 12 April and its descent, might well settle at around 20,000 or less – around half of Professor McNally’s implied prediction.

On 17 March the government’s Chief Scientific Adviser Sir Patrick Vallance expressed his hope that UK deaths would be 20,000 or less.

From the hard numbers now tragically falling daily, we can nonetheless begin to glimpse that aspiration becoming a reality, even though deaths on this scale would be “horrible” and “enormous” and the pressure on our valiant doctors and nurses without precedent.

Enjoying The Critic online? It's even better in print

Try five issues of Britain’s newest magazine for £10

Critic magazine cover