Thursday 30 April 2020

We're long past the peak but they don't seem to want us to know that

Readers will know I got sufficiently bemused by how Covid deaths in hospital are shown on the daily briefing to do my own analysis (Spreadsheet Phil, 17 April). At first I couldn't understand why they didn't show a rolling 7 day figure to even out the acknowledged reporting delays at weekends. But I realised that there are much more general delays in reporting. For example, of the 546 deaths in hospitals in England reported on Tuesday 28th April, 93 occurred on the  27th, 213 on the 26th, 79 on the 25th, 50 on the 24th,  23 on the 23rd and 16 on the 22nd. Beyond that there are single figures of deaths on various dates stretching all the way back to 19 March.

Some deaths will be unexplained at the time and only attributed to covid later but, in the absence of being able to ask Chris Whitty, one can only wonder how they can discover that a hospital death from over a month ago should now be attributed to covid-19.  But then, as my own local health board in Wales recently realised it was only reporting covid deaths monthly, producing an enormous blip in the figures, there are obviously lots of things that can and do go wrong. One wonders what the folk in Betsi Cadwaladr health trust, from the CEO downwards, were thinking as they watched the bulletins about the daily figures. Day after day on Wales Today they were saying that the number of deaths in the trust area was not being announced because it was so small it might identify the victims, which is a comment that gets stranger the more you think about it. Quite a few folk in the trust must have been aware that there had been many more deaths in its three major hospitals than were showing in the published figures. One can see why it is a failing body that has been in special measures for 5 years.

Whatever the quality of the data, you can see the raw figures for England each day on the NHS England website* broken down by region and trust. In order to try to understand what is actually happening I've prepared my own chart, showing English hospital deaths by the day they occurred and the day they were reported, the latter being the main COBR graph that we see on TV each day. I stop at 23 April as the deaths by day of occurrence are still totting up significantly for 4 or 5 days. After that you could add sometimes 10 or 20 but it wouldn't affect the overall picture very much.


What's clear is that we passed the peak around April 8-10. Which is actually a little before many  experts were predicting the peak would occur back at the start of the month - they were saying it would be Easter Sunday, April 12th. The real decline won't be quite as steep as my red line above as extra deaths for the days towards the right hand side get announced with the daily figures over the next month, but it will probably only affect the slope a small amount.

I can understand that the government don't want to broadcast the fact that we are already more than  two weeks past the peak. They need to be sure the decline is sustained (it is) but there are also other factors. There is a lag in patients coming through hospital at different rates, for example. They won't want behaviours to slip in anticipation of relaxation of restrictions. And deaths is only one of the five tests: it would be embarrassing if they had to say the tests were all met apart from security about provision of PPE, which most of the cabinet presumably used to think meant Politics, Philosophy and Economics, ironically the subject of Matt Hancock's Oxford degree.

Embarrassing but not actually surprising when I read that the demand for PPE has been 20 times that of a seasonal flu epidemic and China is currently exporting a fifth of its capacity because of its own needs. Hindsight is all very well but I challenge anyone to say that, if asked in advance how much PPE should be stocked for a pandemic, they would have said "oh, at least 20 times that we could need for a major flu outbreak".

One thing puzzling me is why journalists aren't asking whether the peak has been passed more clearly, indeed asserting that we have passed the peak when questioning ministers, especially given that I assume they have access to somewhat more research and analysis capability than I have. Some journalists have certainly realised it: David Smith noted last weekend that "the peak in deaths and new infections seems to have been around April 8th"**. Indeed, Smith goes on to say that the peak came "before the effects of the lockdown and reflected the voluntary social distancing that preceded it".

Wow! Smith says the government's initial response was slow, uneven and lacked leadership. But then, as he also notes, when China locked down Hubei province most folk thought that was something only an authoritarian regime could do. After all Public Health England moved the risk to the British public from "very low" just one notch to "low" on 22 January. Even when Italy moved to lockdown on 9 March there was still an element of "it could never happen here". Smith thinks the government lurched suddenly from complacency to lockdown as no-one would blame it for putting public health first. And, as I've suggested, the response to lockdown and the chancellor's measures has been a bigger shutdown than the government thought would occur, as many businesses that were not required to shut still decided to do so. I would add that the message reinforced every day has been "thanks for what you're doing, keep it up, stay at home" not "stay home but go to work if it's not something we've specifically banned and as long as you can maintain social distancing".

One can read into Smith's words (he doesn't explicitly say it) that he thinks the government panicked and went further than, with hindsight, they should have done. And, because they put the "fear of God into people" the public responded to an even greater degree. The result? The biggest hit to the economy not just since the financial crisis but since the Great Frost of 1709.

Do I agree? I'm not sure yet and I've already criticised many commentators for excessive use of hindsight together with some rewriting of history. The first real social distancing restrictions other than exhortations weren't until schools were closed on 20 March and pubs and restaurants were closed on the 21st, only three days before the first effective day of lockdown, the 24th. So I think it's a bit difficult to assert with confidence that the peak at 8 April would have happened without the lockdown. In due course we may be able to tell more by looking at examples like Sweden, where they haven't gone as far with lockdown and have relied more on social distancing advice, the position we were in from March 20-23.

The new step announcing all deaths in the daily press conference from yesterday will further disguise the fact that we have passed the peak and buy ministers time for the lockdown review next week. It will do that because the "all covid deaths" data must be even more erratic than the surprisingly erratically hospitals only figure. I thought they were right to stick to hospital deaths as they are more readily measured and represent the vast majority of cases. And it's the trend that matters.

But as with testing, where Hancock felt he had to pluck his 100,000 number out of thin air presumably on the basis that it's a big number, I think ministers felt the pressure about deaths in care homes in particular. They know as we do that many of those deaths have been advanced by very short periods but they daren't say so because of the inevitable media storm that would result. So when Jonathan Van-Tam carefully answered a question yesterday by saying that, in time, the excess deaths figure will prove the most reliable and meaningful I'm sure he was factoring in that, looked at with the perspective of time, many of the care home deaths would effectively drop out of the statistics as they would have happened in the same year anyway. So when you compare the total deaths with the 5 year average the excess deaths figure should show all of the impact of the epidemic, both directly from covid and indirectly from delayed cancer treatment and the gamut of other knock-on effects, but care home deaths that have effectively been advanced just a few months wouldn't cause a blip.

Unfortunately, troubling though it is, the care home issue is not the big story or issue here that the journalists have made it. I say that having had elderly three close relatives in die in care homes over the last decade, all having had minimal quality of life in their closing years. (I may in due course write about how I think the elderly really can be shielded, which is relevant to a broader population than care homes).

Of course, passing the peak doesn't mean we're out of the wood. In most previous pandemics the second wave has been more deadly. This is probably why the government is signalling that we'll take baby steps out of lockdown. As it seems we have to wait 6 weeks to know whether we can go further it could be a very slow exit.

I also have to declare that I got something wrong (what do you mean that's normal?). I said only 3 days ago that the current situation was outside any living person's experience. I was forgetting Hong Kong flu, even though I lived through it. What, you don't really remember it either? There were two waves, in 1968-9 and 1969-70 which caused some 80,000 deaths in the UK. Much of the post stopped being delivered and trains and buses ran to heavily reduced timetables because so many drivers were sick, people of working age being the most severely affected. Clinical staff also caught it and, for a while, the NHS was overwhelmed. The second wave was deadlier with the biggest weekly death toll,in the first week of 1970, being over 20,000, bigger than any weekly total so far in the coronavirus epidemic. It's all very being wise after the event and saying that care homes were going to be the problem - lessons learned papers from flu epidemics note that a lesson from 1970 was"not to ignore the likelihood of a seriously depleted workforce to implement policy"***. Of course there are proportionally more old people now, but that epidemic hit working age people harder, others have hot children hardest.

I'd heard of Hong Kong flu but I don't remember any consternation in January 1970. I was probably preoccupied with Everton's imperious procession towards the League title and my impending A-levels. Of course 1970 was very different from now. We didn't live with 24 hour news and the death figures weren't briefed out daily. We had hospitals and you didn't have to pay for health care, which probably still struck our grandparents (if not our parents) as a novelty. If they got overwhelmed because of something unforeseen it wasn't "somebody's fault". There appeared to be no consideration given to special public health measures (presumably the cabinet minutes have long been released) and certainly not to banning sporting fixtures. Phew! Over history it's been Everton, not Liverpool, that has been worst impacted by events (champions in 1939, team broke up; champions in 1985, Heysel stadium disaster precluded them from competing for the European cup won by the mighty Steaua Bucharest and then leading to loss of the manager and key players including Gary Lineker). It hadn't occurred to me that 1970 was a near miss......

Now that would have been a disaster.

https://www.england.nhs.uk/statistics/statistical-work-areas/covid-19-daily-deaths/

** David Smith's Economic Outlook column is in the Sunday Times each week. His column "Lockdowns are difficult, ending it will be the hard part" appeared on 26 April. The stuff on Hong Kong flu also mainly comes from Smith's column, though he got it from a book by Nicholas Timmins: The Five Giants, A Biography of the Welfare State.

*** Journal of Epidemiology and Infection, July 2008.   https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2870877/





No comments:

Post a Comment