Richard North, 02/05/2020  

Given all the variables and uncertainties, not too much store should be set by the ONS interactive map which allows users to identify the number of deaths in different postcode areas.

However, it is a small comfort that there have been no deaths from Covid-19 in my postcode area, up to 17 April. It is of some interest also, given the reported differences in incidence between ethnic types, that in one of the adjoining areas, populated largely by Kashmiris, eight deaths have been recorded.

Those numbers may very well change, as one might expect. One of my immediate neighbours was recently admitted to hospital with severe kidney and liver problems. Admitted to the general ward, she has since been transferred to the Covid ward, after testing positive for SARS-Cov-2.

She had been in self-isolation at home for nearly nine weeks, and no one in the street has reported illness. The inference, therefore, is that this is another case of hospital acquired infection.

Should she die – which, sadly, is expected imminently – will her death be recorded by the postcode of her last known residence, or will it be allocated to the hospital, with its locality the other side of town?

Looking then at my own local area for explanations as to why we should have been so fortunate (in an area of about 50,000 people), one thing that comes to mind is the very limited casual social interchange that we experience.

Working at home, with sometimes limited opportunities for exercise, it had been my habit to take a long, circuitous walk round the urban village, to the park and back – about three miles in length. Unless I misjudged the timing, and I passed the local school at throwing out time, most often I would not see a single person, except for a very small number in the park.

For sure, the closure of pubs, betting shops, hairdressers and the proliferation of nail bars, must have reduced the number of social interactions, but the most obvious effect of the lockdown has been the reduction of vehicle traffic. In terms of pedestrian movements, the difference must have been slight, as there was so little traffic anyway.

It strikes me though that if we are seeing substantial differences between adjoining post code areas, where local knowledge might offer multiple reasons for those differences, how hard it must be to juxtapose national data and draw useful inferences from them.

To that effect, I warm to the piece by David Spiegelhalter, an academic statistician, who cautions against making comparisons. In pointing to the difficulties in measuring the death rates from Covid-19, he points to confounding factors such as the direct and indirect health effects of the epidemic, taking into account reduced road accidents, the benefits of reduced pollution, the effects of recession and so on.

Many studies will try to disentangle all these but, he says, his cold, statistical approach is to wait until the end of the year, and the years after that, when we can count the excess deaths. Until then, he concludes, the "grim contest" of comparing the death rates between different countries "won't produce any league tables we can rely on".

When it then comes to assessing the relative effects of different lockdown strategies (or the lack of them), much the same can be said. And that applies to the multitude of studies seeking to argue that there is little correlation between the severity of "lockdown" and excess deaths in different countries.

Of that ilk is this study, where the author suggests that "no lives were saved" by diverse lockdown strategies, in comparison with pre-lockdown, less restrictive, social distancing policies. Comparison of the epidemic's evolution between the fully locked down countries and neighbouring countries applying social distancing measures only, they say, "confirms the absence of any effects of home containment".

I must say that I am getting a little tired of this sort of claim, and in this case more so as the author refers to the "recent COVID-19 outbreak in Europe", as if this was one event. Yet, as even the slightest familiarity with the situation will affirm, there are many outbreaks. If we take the whole of Europe, we could be talking about several thousands.

In more normal time, for instance, if we had an episode of infectious disease in a care home, with multiple cases, for the purposes of control, that would be treated as an outbreak in its own right. I would go so far as to say, for the purpose of control, such infectious disease episodes in any of these institutions need to be treated as separate outbreaks.

When it comes to hospitals, in terms of dealing with and preventing nosocomial infection, it is common to treat individual wards as epidemiological units, to the extent that one talks about "ward outbreaks". The confines of a district general hospital can support an epidemic in its own right – sadly, not an infrequent occurrence.

Therefore, when it comes to looking at the situation in different countries (and especially in the early stages of a pandemic), it has to be said that they are not valid epidemiological units. Within each country, there are massive variations in case and death rates (as measured by administrative areas).

Here, Sweden is the poster child but, even then the reviews are mixed. Within that country, between the county with the lowest case rate per head of population (Jonkoping County) and the one with the highest (Stockholm County - pictured), there is nearly a 60 percent difference.

The point, of course, is that no country is homogeneous – not even tiny Luxembourg. And one familiar with Malta, with a population (2019) of 493,559 will appreciate that the crowded tenements of Valletta present a wholly different epidemiological scenario to the less-densely populated north-western littoral regions. Incidentally, the population size is very similar to that of the Bradford district.

Making country comparisons on this basis must be considered fraught. And not only does the distribution of sickness and death within each country vary. Policy in some countries is imposed on a national basis, in others regionally and locally. In yet others, there is a composite. Enforcement even within common policy areas is patchy and far from uniform.

Given also massive variations in the way national statistics are collected - and variations within countries with different experience of the epidemics - it is difficult to make comparisons even if all other things are equal - which they are not, as David Spiegelhalter points out.

The only way you are going to get valid comparisons is by assessing the relative effects in specific, defined outbreaks. Those outbreaks are almost certainly going to be sub-national and, as we see with care homes and hospitals, one might be looking at individual buildings – especially if they are isolated from the community in which they are situated.

Then, before comparing different outbreaks, all the parameters have to be standardised (such as case definition, criteria for recording deaths, etc), and all the other policy and sociological influences must be disentangled before it will be possible to be able to detect the effects of the one policy variable - i.e., the rigour (or absence) of the lockdown.

Until we have properly designed studies of this nature, it is far too early to comment, and studies which seek to draw conclusions from different policy regimes are, for the moment, completely without merit.

This notwithstanding, as I pointed out yesterday – and need to emphasise once again – the lockdown is not a control measure. Then I made reference to Dunkirk. As we were running away from the battle, buying time to regroup and re-equip, this could not win the war. It was a (not so) managed retreat. So is the lockdown strategy, and the battle against this virus will be won only by shoe-leather epidemiology or vaccination, or a combination of both.

But the other point we have to live with is that epidemics are times of great uncertainty. The data are not neatly packaged, refined and accurate, policy directions lack clarity and there is no easy way to determine whether any particular actions are having desired effects.

To that extent, an epidemic is as much a political event as a medical emergency – if not more so. That also needs to be emphasised. Thus, actions are taken as much for political effect as they are for strictly scientific reasons – to achieve my so-called "purple banana" effect.

In this context, the lockdown is very much the purple banana. It is government being seen to do something, and the very fact of the "doing" is what matters. Given that this and previous governments had so comprehensively screwed up our response to this pandemic, there was very little else it could do.

Any evaluation of the lockdown policy must be evaluated in that light.

comments powered by Disqus

Log in

Sign THA

The Many, Not the Few