Thursday 12 October 2023

COVID-19: The Post-Mortem (2)

This is a follow-up to my “COVID post-mortem” post of 27th September.

I realized that, having assessed average excess mortality against vaccine take-up and cumulative deaths per case in each country, I could use a similar technique to assess average excess mortality against the severity of different types of lockdown interventions through the course of the epidemic.

The sample of countries was necessarily small, since only 35 out of 50 countries had reported excess mortality figures within the 90 days prior to the last date covered by the data I took.

Stringency

In my “magic spreadsheets,” I kept track of the Blavatnik School of Government “stringency” measure, a percentage which gives a measure of how far a country was locked down overall on a particular day. I also kept detailed track of how far each country was locked down each day, under each of nine headings: schools, workplaces, events, gatherings, public transport, stay at home, travel restrictions, international travel restrictions and face coverings.

Measures of stringency

When I came to do this assessment, I found that I could not use the official stringency measure for comparison, because the detailed data behind it, on which I based my own metrics, has not been reported since late 2022. Fortunately, I had already worked out my own alternative measures, both derived from the stringency.

What I call “average lockdown” was calculated in the same way as the official stringency, but using the detailed data, excluding “public information” status and including face covering restrictions instead.

I had also defined a “harshness” metric, which was intended to give an idea of how subjectively unpleasant a given level of lockdown was. This included only mandatory measures, and gave more weight to intrusive and inflexible measures than the official stringency calculation did.

Both of these averaged the values day-by-day up to the cut-off date, at which I had decided that COVID was no longer a serious threat, so stopped collecting the Blavatnik data. That date was 17th August 2022.

Findings not broken down by lockdown type

What I found was that, for both my metrics, excess mortality since the start of the pandemic tended to increase as the average severity of the interventions (from the start of the epidemic up to the cut-off date) went up. But the slope of the trend line for the harshness metric was steeper than that for the lockdown metric.

Over the full ranges of both metrics, the trend line for the harshness metric ran from about 7% to 11% excess mortality, whereas that for the lockdown metric went only from about 8% to 10.5%. This suggests that the psychological effects of mandatory, harsh lockdowns may have been responsible for a significant proportion of the excess deaths seen.


Findings broken down by lockdown type

Rather than show all the individual graphs, I will give the slopes of the trend lines for the various different types of lockdown interventions, and for the two metrics graphed above, in the form of a table.

Lockdown type

Slope

Schools

+0.2189

Face Coverings

+0.1550

Average Harshness

+0.1130

Gatherings

+0.0848

Workplaces

+0.0830

Events

+0.0827

Average Lockdown

+0.0629

Travel Restrictions

+0.0587

Stay at Home

+0.0427

Public Transport

-0.0411

International Travel Restrictions

-0.2150


Clearly, comparing countries across Europe, by far the most destructive type of lockdown in its effect on excess mortality has been locking down schools. It’s understandable why this would have had adverse psychological effects on both parents and children; and maybe other family members as well. It was also, according to my researches earlier in the epidemic, the least effective of all the types of lockdowns at controlling the spread of the virus. Hardly surprising, since children were not in general at great risk from COVID-19. The second most destructive type of lockdown has been face-mask mandates; an imposition which was at the same time annoying, demeaning and, on this evidence, worse than ineffective at saving lives.

Lockdowns on gatherings, workplaces and events are all very comparable in the size of their effects on excess mortality. All three of these types of lockdowns seriously detract from the quality of life by restricting people’s social contacts, or even taking them away altogether. Workplace closures have also a very negative economic impact, most of all on small businesses.

Travel restrictions and stay at home mandates have had a similar, but smaller, effect to lockdowns on gatherings, workplaces and events. This may be because they were not generally used until less intrusive means had already been tried without success. But the trend still suggests that they have increased excess mortality rather than decreasing it.

The only two of these interventions which appear, on this evidence, actually to have saved any lives since the epidemic began are public transport closures and international travel restrictions. As I worked out some time ago, they also seem to have been the two most effective among the measures tried, in terms of controlling the spread of the virus.

This seems to me to be simple common sense, in both cases. After all, the safest way to get around in an epidemic of an airborne infectious disease is in your own car, without passengers. For those who don’t have their own cars, a taxi is the next best option. Moreover, screening or more stringent measures at borders can protect a country – for a time – from large-scale incursions of the virus.

To give an idea of the magnitudes involved, the graph for school lockdowns is at the head of the article. Across the range of average lockdowns used in Europe, excess mortality roughly doubles, from about 6.5% to about 13.5%, from the lowest to the highest locked down. Very significant! At the other end of the scale, here is the graph for international travel restrictions:

Here, the trend is even more spectacular, falling from 15% to only just over 5% excess mortality from the lightest international lock-down to the harshest.

To sum up

The technique I have used to assess average excess mortalities in Europe against average lockdown levels of different types is, of course, rudimentary. And the sample size is small, with only 35 countries reporting up-to-date excess mortality figures.

Nevertheless, the trends in my scatterplots do suggest that average excess mortality since the beginning of the epidemic may well have been made worse by many types of lockdown interventions.

School lockdowns have been the worst, with excess mortality roughly doubling from one end of the lockdown scale to the other. Face mask mandates are second. Restrictions on gatherings, workplaces and events have had a lesser effect. Travel restrictions and stay-at-home mandates have had less effect still. But all these correlations between higher lockdown levels and higher excess mortalities are still positive. Of all the types of lockdowns, only public transport closures and restrictions at national borders show any evidence of having saved lives since the epidemic began.

Furthermore, when the different degrees and types of lockdowns are weighted in such a way that mandatory measures and subjectively harsher lockdowns have more weight, the positive trend between average excess mortality and average lockdown severity increases. This suggests that the psychological effects of lockdowns may be at least a contributor to the recent surges in excess mortality in many European countries, including the UK.

No comments: