Dr No’s last post showed there were substantial humps in non-covid–19 mortality in care homes and private homes in the first wave of the pandemic, and that these humps cannot by fully explained by a displacement of what would have been hospital deaths to care and private home deaths. Something else must be at work, which Dr No rather loosely suggested might be ‘neglect, or worse’, in the absence of other obvious explanations. The ‘or worse’ is to cover the ‘midazolam thing’ — the excess use of midazolam last year, and by implication, excessive, and inappropriate, use of end-of-life care to ‘help the old folk on their way’. Dr No used ‘or worse’ because he is not fully persuaded this did (or for that matter did not) happen. There was a rise in aggregate midazolam prescribing — for example, see the striking April 2020 spike here — but there was also a rise in the numbers of deaths, so it becomes a chicken and the egg problem: did more deaths cause more midazolam prescriptions, or did more midazolam prescriptions cause more deaths? We don’t know. Perhaps it was a bit of both.
Not mentioned in the post itself, though inevitably, and relevantly, it came up in the comments, is the extent to which the covid death count is inflated. This matters, because, to the extent that hot stiff bias — an overexcited tendency to put covid on the death certificate — applies, then the blue lines on the post’s charts are higher than they should be, and the red lines, and so humps, are lower than they should be. Bear in mind the ONS counts are ‘deaths involving covid’, that is, deaths where covid is mentioned anywhere on the death certificate, rather than PHE’s comedy stats based on deaths with 28 days of a positive test, which are frankly no better than saying any death within 28 days of having had a whiff of tobacco smoke about them is a smoking death. The problem is ‘deaths involving covid’ also includes deaths where covid was not the underlying cause — it was coincidentally on the certificate — and so, even by ONS definitions, some of these deaths are not true covid deaths. To pick up on other comments to the post, this is more one of Scott’s tangled webs, than the more general contemporary wicked web of covid deception.
This tangled web ONS ‘deaths involving covid’ data — how many are true covid deaths, how many are ‘deaths with’ rather than ‘deaths from’ and how many are just plain wrong, because of hot stiff bias — set Dr No wondering what other biases might be at work. If wielding a spanner is the core task of the traditional mechanic, then spotting bias is the core task of the traditional epidemiologist. Here, it is not computers that do the work, but the little grey cells; computers only become of use once you have spotted the potential bias, because you can’t include an ‘unknown unknown’ in the analysis. And it is of course the case biases can work both ways. Some might increase an effect, such as the number of deaths attributed to covid, but it is just as possible another bias, perhaps operating at a different time, might decrease the number of deaths attributed to covid.
Now, the most striking thing about deaths attributed to covid is just how few there have been in recent weeks. Despite the July wave of test positives rising almost to January levels, the rise in the weekly number of covid deaths has remained remarkably low. Naturally, the authorities attribute this decoupling between test positives and deaths as the vaccines doing their job. But, we have to remember, association is not causation. Post vaccini ergo patiens vivum is pure post hoc ergo propter hoc stuff and nonesense. Furthermore, there is growing evidence that the vaccines are not all they are cracked up to be. A number of recent studies have suggested the vaccines may have little or no effect on transmission of the now dominant Delta scariant. Could there be another bias at work that explains the remarkably low numbers of covid deaths, despite vaccine breakthrough cases and high prevalence levels?
Dr No suggests there might be. He can’t prove, or disprove, anything yet, if ever, so he puts this forward as nothing more than a hypothesis. But it is nonetheless an intriguing one. Perhaps the vaccine doesn’t work only through the immune system, but also by re-programming the expectations of doctors, in particular those completing death certificates. Most doctors are heavily committed to the vaccine narrative, and the reductions in covid deaths they must surely bring about; and then there are all those wonderful Big Parma and PHE studies showing how many lives have been saved by vaccines. In short, in comparison to the pre-vaccine era, when hot stiff bias meant even the slightest hint would ensure covid got on the death certificate, could it now be that a new hot vaccine bias means that, for the doubly vaccinated, covid is far less likely to get on the death certificate, not necessarily because of any immune effect, but simply because the doctor knows the patient has been vaccinated.
Before you dismiss the suggestion as absurd, ask why do we blind observers in clinical trials to whether the patient has had active treatment or placebo? We do it because we know knowledge of treatment given affects assessment of treatment effects. Dr No merely suggests a similar effect for vaccination: if you know your patient has been vaccinated against covid, then you are simply less likely, in uncertain cases, to attribute the death to covid, because of hot vaccine bias; in effect, an expression of a form of cognitive dissonance. In the pre-vaccine era, quite the opposite happened: in uncertain cases, you were more likely, through hot stiff bias, to attribute the death to covid.
The reality, Dr No suspects, is somewhere in the middle. He is not suggesting for a moment that the vaccines have no effect through the expected immune channels. All the evidence points to the fact they do have some effect, though perhaps rather more modest than advocates would have you believe. But he also thinks that both hot stiff bias and now hot vaccine bias have also had their effects. In the pre-vaccine era, hot stiff bias inflated the number of deaths attributed to covid; in the post-vaccine era, knowledge that a patient has been vaccinated prompts a more sober assessment of the cause of death, with hot vaccine bias reducing the chances of the death being attributed to covid. Why, it’s even possible the two in a roundabout way cancel themselves out, and true covid mortality is closer to what we see now, than it ever appeared to be at the height of the first and second waves. Now, there’s a thought.
Footnote: The use of The Cow-Pock by James Gillray (1756–1815) as the image for this post is not meant to endorse general anti-vaccine sentiment; it is merely to remind us, as Gillray seeks to do, that we should always be sceptical of any exaggerated claims.