Stop Press: how the models worked out in data

There has been surprisingly little mention in the news (ie none that I have seen) of the first major analysis of the world data from the COVID-19 pandemic, published in The Lancet on July 21st, before H.M. Government started to panic over increased positive PCR tests, executed local lockdowns, and threatened national ones if the “R-number” virtual canary begins to look green around the gills.

Perhaps, though, it’s not surprising we have not heard about it from the BBC, given what we have come to expect, since the paper seems to undermine much of what the governments of the world have been doing from the start.

The paper repays careful reading, and isn’t too technical to make sense, so I won’t discuss the details too much. Basically, morbidity and mortality were closely related with the risk factors with which we are all now familiar – age, obesity, rates of various illnesses like cardiovascular disease and diabetes, and so on. Analysis seems to show that, although the overall pattern of the disease looks similar across nations, the complication and death rates match the different incidences of these factors pretty closely. They also map, as one might expect, to the availability of health resources and organisation, and certain other criteria for, in effect, the “good order” of nations.

What they don’t match, with certain provisos, is the degree to which society was shut down by governments. As the abstract says:

Rapid border closures, full lockdowns, and wide-spread testing were not associated with COVID-19 mortality per million people.

Let the significance of that sink in. Lockdowns did not, in the end, save lives, nor did banning international travel, nor did the frantic rush to ramp up testing. On the other hand, there is incontrovertible proof that lockdowns have caused many deaths, and that through their damage to the economy they will cause many more for years to come. Quarantine restrictions have damaged world trade and ruined the travel industry, not to mention their psychological and financial effect on stranded and quarantined travellers. And testing with PCR, as I explained recently, may mainly be serving to prolong an epidemic that ended weeks ago.

The most confusing aspect of the paper, to me, is an apparent contradiction of this conclusion, expressed in the very next sentence of the abstract:

However, full lockdowns (RR=2.47: 95%CI: 1.08 5.64) and reduced country vulnerability to biological threats (i.e. high scores on the global health security scale for risk environment) (RR=1.55; 95%CI: 1.13 2.12) were significantly associated with increased patient recovery rates.

On the face of it, it is hard to see how lockdowns would increase recovery rates if they didn’t improve mortality. Indeed, it is hard to see how lockdowns, which were designed to prevent infection, would help anyone recover. But I think the answer is partly that lockdowns, in some countries, did help health services cope with the initial surge of patients, and so help ensure optimal care. Another possible explanation is that, since most western countries dutifully pledged to follow WHO guidelines, and were for the most part equally panicked by the Ferguson non-peer-reviewed terror models, “full lockdown” correlations might simply be a proxy for ability to organise, as reflected in the phrase “reduced vulnerability to biological threats.”

But the bottom line in the abstract is that all the things we didn’t like about our governments’ responses, that is to say the hubristic attempts to “defeat the virus” rather than cope with it as has always been the case previously, did no good anyway:

Interpretation: In this exploratory analysis, low levels of national preparedness, scale of testing and population characteristics were associated with increased national case load and overall mortality.

That’s not to say we have to be fatalistic about such novel diseases and await our inevitable demise. The actual causes of avoidable death, from obesity to inadequate planning, are things we knew about before. Some of them (like the rates of heart and lung disease and diabetes) appear to be the results of previous government interference in individuals’ lives, if the falsity of the “animal fats bad, carbohydrates good” dietary hypothesis we’ve all been pushing for decades is indeed the case. But it’s not earth-shattering to say that if we sort out chronic health problems, eliminate extreme poverty, organise health services well and do some contingency planning, we’ll cope best when viruses arrive.

Paradoxically, the study seemed to confirm an earlier impression that smoking might be somewhat protective against COVID mortality, so we can blame public health efforts at cutting smoking too – though that would, I think, be rather unfair since smoking causes nearly everything else, including the heart and lung disease.

Joking aside, even good policies can have unforeseen downsides under special circumstances like this. But lockdowns are a different kettle of fish. They were always a vast, unprecedented, uncontrolled and unvalidated experiment on the world’s population. They could be reliably predicted to have severe untoward effects (and were, even in this blog, from the start).

This paper gives strong evidence that that experiment failed in every important way. Given the gung-ho way lockdown and the rest were adopted, every government that enforced them is morally culpable, as are the scientific bodies that advocated them – and, of course, the “world authorities” that lubricated and facilitated the whole business, of which the corrupt UN and WHO are at the forefront.


Others may draw some other useful lessons from this paper. Work of this kind has its limitations, in that looking at such a wide range of parameters was bound to show multiple correlations which may be entirely fortuitous, or at least only proxies for the real causes. There’s always the danger that it’s a case of “chasing p-values.” (Remember William Briggs helpful statistical principle: “Die, p-value, Die Die Die.”)

But non-correlation is pretty significant, and that is what the study shows between the acute policy responses to COVID-19, and the one figure that really matters, the overall mortality.

It’s nice when the data confirms ones initial instincts, even as bodies like BioLogos still appear to be suggesting we ditch them in favour of trusting the models that failed to predict the actual outcomes.

But hey – science is self-correcting, and in the meantime it’s only damaging 7 billion people.

Avatar photo

About Jon Garvey

Training in medicine (which was my career), social psychology and theology. Interests in most things, but especially the science-faith interface. The rest of my time, though, is spent writing, playing and recording music.
This entry was posted in Medicine, Politics and sociology, Science. Bookmark the permalink.

4 Responses to Stop Press: how the models worked out in data

  1. Avatar photo GD says:

    Hi Jon,

    I understand the complexities in formulating models of events such as these, but I am curious to know how you regard second waves and the apparent association with formation of crowds and spreading the virus? Down here (Victoria) we have experienced a dramatic increase in deaths (compared to the initial death count) from the secondary wave, so called.

    • Avatar photo Jon Garvey says:

      Hi GD

      My guess is that much of the pattern suggests a seasonal component – Australia may have begun to cop it, after being largely spared, because it is in the Southern hemisphere and it’s winter. Likewise Brazil. There seem to be two peaks in North America, affecting the North and South respectively – it’s a big country in terms of latitude.

      Looking at today’s stats, you’re still well below the average total deaths (500 in 25m, compared to our 40,000 in 75m), and there’s even a hint in the curve that the deaths may have peaked already. You scarcely had a first wave that I can see, compared to Europe. That probably reflects both geography and demographics. All the negative headlines about Brazil neglect to say that the deaths per million are still lower than the European average.

      That seasonal component (and the pattern of most Coronavirus strains as well as flu) makes it likely that there will be a genuine increase of cases (ie illness!) in the northern hemisphere come winter, and hence an increase in deaths.

      However, I don’t see anything in what I’ve read to suggest it will be on the scale of the first wave, firstly because that “cleaned out” many of the most susceptible people, and secondly because there is normally a degree of herd immunity after the first year.

      The wild card is that if lockdown did achieve anything at all, it will have both hindered herd immunity and made the population less physically and psychologically well, as well as preventing the spread of ordinary Coronaviruses that would give cross-immunity. In that case, we might well have problems.

      But if the paper cited above is accurate, it looks like lockdown has not really achieved anything (you may be aware that in most countries, the peak of new lethal cases was reached about a week before lockdown even started, and had started to decline). In that case, we’re more likely to see little illness and death, but feverish testing producing enough false positives to give the excuse to panic!

      As to spread by crowds, I’m very skeptical. They’re talking about that here, too, saying that the low death rate is because only the young are getting the virus. I don’t buy that – the elderly are not hermetically sealed, and are well-represented in shops, on the beach, and so on. If it were a real “second wave,” at least some vulnerable folks would be succumbing, but the death rate is inexorably downwards.

      Another cause for doubt in the “spread by crowds” is that I live in a holiday region crowded out because of restriction on foreign travel. There is scarcely a mask to be seen on the crowded beaches, pubs and restaurants are full, and my bed and breakfast owning friend says her guests are not bothering with any hand-washing etc. But the official Devon website says that of the handful of “cases” during the summer, not a single one has been found to originate from incoming torists – they’re all sporadically within households.

      • Avatar photo GD says:

        Hi Jon,

        There are many variables on a global scale. Here however, we have a smaller number of cases in hospitals and deaths so it may be easier to see some sort of causality. Regarding winter, the south is relatively cold and wet and the flu should be dominant – yet one state (Vic) has more deaths and infections than the other two states with similar weather. In fact Vic now has more deaths than the rest of the country and the sources appear to have been identified.

        I am convinced that it is in our interest to take precautions. The US has shown how bad things can get, for a myriad of reasons summarized as a crazy political system.

        I wish we had a complete answer but than ….?

        • Avatar photo Jon Garvey says:

          Relatively low-scale outbreaks certainly are useful in gaining knowledge, as our experience in the southwest of England shows. In theory we’ll be up for for the biggest second wave, having been largely spared the first – but again it depends on how much asymptomatic spread has been going on over the summer.

          One of the docs interested in this statistically showed how all the “big” outbreaks were, in fact, separate smaller outbreaks, each with their own patterns, which I guess is the case where you are.

          At the same time, I think it’s likely that the limiting factors are susceptible populations (to harm, that is, rather than to positive tests) and herd immunity. The problem throughout has been the “mission creep” from optimizing the latter to the vain hope (in my view) of eradicating the virus. In the end all those vulnerable will either get it, or be protected by the immunity of the majority. Twas ever thus, and I think twill always be, rather than the pipe-dream of wonder-vaccines.

          I gather (and may have posted before) that the original 1917-18 H1N1 Spanish Flu is still with us – we just became less susceptible to it, as a world population, after the first bad years.

Leave a Reply