The last blog I wrote was titled “Did lockdowns work?”. That was more of an impressionist picture based on patterns in Indian data. But over the last few weeks, we analyzed global data for evidence of the effects of preventive restrictions (PRs) of all kinds imposed in all countries. Surprisingly, or not so surprisingly, we found that most of the restrictions worldwide failed in arresting transmission of the virus. The most ambitious objective of lockdowns is to “break the cycle”. A true break in the cycle is expected to arrest the spread completely if the lockdown works for a period slightly more than one infection cycle. This goal was rarely, if ever, achieved. A less ambitious but useful goal would be to reduce the rate of spread of the infection. Testing this is rather tricky because the rate can change spontaneously even if no preventive measure is applied. Therefore it is necessary to separate spontaneous change in slope from the PR induced change in slope.
This problem is like the son-daughter problem. Whether the sex ratio at birth differed from 1:1 cannot be inferred from one or a few families. A given couple can have three consecutive daughters by chance alone. We need a large population to reach a conclusion. Similarly in this analysis which country was successful in reducing the rate by imposing a lockdown cannot be ascertained, because it could be mere chance as well. But the overall success rate can be estimated with confidence. Using this approach, we estimated the success rate of PRs in reducing the transmission of the virus. It turned out that only 4.5 % of the total PRs were successful in reducing the transmission significantly. In a large number of cases the transmission actually increased by imposing a restriction. Quite a number of times the transmission decreased after lifting or relaxing a restriction. This means factors other than restrictions were stronger than the effects of the restrictions. The imposed restrictions could explain only 6.1 % of the total ups and downs of the epidemic curves.
This paper is now available as preprint. https://www.preprints.org/manuscript/202104.0286/v1
Anyone interested in technical details of the analysis can refer to that. Since our inferences are most likely to be viewed as politically incorrect, I don’t know how and how long the peer review will go. But the data are in public domain and the analysis is transparent. So anyone can make an opinion. Just showing two poor correlations here. One is between the stringency of restriction and the expected change in transmission rate, which is not significant despite very high sample size. The other is between change in stringency (i.e. either imposing or relaxing restrictions) and change in transmission rate. This is statistically significant but the strength of the relationship is very poor.
The inefficiency of lockdowns is not surprising. Epidemic is a complex system and simple measures may fail to work. What is surprising is the fact that people are made to believe that this is what is going to work. If the infection spreads, it must be because you were irresponsible, you did not take care. If in some area, the cases went down, we say it was well managed. Then what about countries where there was excellent control in one phase and an uncontrolled surge in another? Saying that people followed a restraint in one phase and did not follow it in another is a circular statement, unless there is an independent and well quantified measure of the restraint.
To me what is more important is the psychology behind this. Medicine, public health as well as political administration does not like to say that we can’t do anything. There is a need to pretend that we are doing the right thing and we are doing our best. Lockdown is an ideal demonstration that we did something. Whether it was effective or not, is immaterial.
There are examples in medicine other than lockdown ddemonstrating the “must do something” phenomenon. Remdesivir, hydroxychloroquine and convalescent plasma completely failed in clinical trials. The trials were conducted by reputed organizations and published in flagship journals. But in spite of completely failing in clinical trials, remdesivir is sold as the leading drug everywhere. In my own city, the stocks are nearly exhausted, people are mad after getting it and it is being sold at a high price. This is because, as long as there is no effective antiviral drug, we need to pretend that there is something that works and need to show we are doing our best. Control of blood sugar to arrest diabetic complications (in the case of type 2), blood pressure control to avoid stroke, cholesterol lowering to avoid heart attack have all performed poorly in randomized clinical trials, but these are the most widely sold drugs.
The “must do something” phenomenon is not restricted to medicine. We see it in so many examples including state administration, crime control, business crisis, parental behavior, child behavior and so on. It must be giving a social advantage to the individuals or agencies in control. This advantage predominates and overcomes actual concern. The concern and the criteria of success itself are then shifted. Rather than avoiding complications or deaths, reducing sugar or cholesterol itself becomes a measure of success. This is an interesting phenomenon and I am sure people doing this do it honestly and often with good intensions. They don’t want to be aware that what they do does not work in reality, because working in reality is no more the concern. I did something is the feeling giving satisfaction. This is understandable as a social phenomenon, but my worry is that all this is being sold under the name of science. People are never made aware that the “something” did not work, or worked very poorly, hasn’t worked so far and if it works in your case it might be nothing else but chance. I wish that at least people of science should be aware of this and avoid the trap.