Drowning in Coronavirus Research, Part 68

By August 22, 2020 Commentary

Okay, for all you mask proponents, here is a study that may actually provide some support for masking.  (SSRN Paper)   The research comes from Thailand and involves contact tracing to see who got infected and whether or not they wore a mask or engaged in other behavioral modifications that might limit transmission.  So you had a potential confounding problem right from the start in terms of understanding the potential contribution of different interventions.  Over 1000 asymptomatic contacts of active cases in the month of March were interviewed by phone about their use of protective measures.  The outcome was whether they became diagnosed with CV by April 21.  That lag time is long enough to allow for infection after the contact period other than from the index cases, as the authors note.  In addition, you have to look it the appendix to see the definition of a case and asymptomatic and other key terms, and when you do, you see that it is very possible these contacts could have been infected by a person other than the supposed index case.  Finally, these index cases all came from certain clusters of infection, and the contacts were also at the precipitating events in most cases.  211 or 20% of the contacts were diagnosed with CV in the study time frame.  Comparing behaviors across those contacts who were and weren’t infected, the authors concluded that wearing masks all the time when in the presence of the infected person was associated with a lower risk of infection, as was keeping a greater distance from the infected person, limited time in the presence of the contact and frequent hand washing.  Wearing a mask some of the time was not.  People tended to engage in multiple of the protective behaviors, which heightens the confounding issue.  Keeping further away from the infected person and limiting time of contact were as or more protective as wearing a mask all the time.  I would classify this as very weak evidence.  It is self-reported data.  You are asking people to recall behavior in some cases from weeks before and estimate things like how long they were in an infected person’s presence and how close they were to them, in addition to mask wearing behavior.  It isn’t really possible to isolate the effect of a single behavior in this kind of study, despite statistical gymnastics.  And if the contacts were at the same event as the index cases, it is highly likely they were infected at that event, not later.

Okay, for all us proponents of the Swedish approach to the epidemic, here is a column summarizing why.  (ZH Column)   Being careful but rational from the start has left Sweden in an enviable position.  The epidemic appears to be over there.  Perhaps, if meteorological variables are a significant factor there will be a revitalization of spread in the fall or winter, but given that this is already happening in some other European and Asian countries, mostly among the younger population, it seems unlikely that there would later be a significant surge in Sweden.  Would you rather be New Zealand where the people cower in fear and get re-locked down every time a new case appears?

This is actually a pretty important paper explaining from a statistical perspective why research supposedly finding that lockdowns did limit cases and deaths is wrong.  (Medrxiv Paper)   Written by a Swedish statistician, he finds that a study purporting to show that lockdowns worked manipulated the analysis to avoid the fact that Sweden had no lockdown but had a transmissibility rate and other epidemic outcomes similar to those of other countries.  It is sad when science is clearly being misused to justify governmental actions and political positions.

And here is another odd study trying to show that Britain’s lockdown made more sense than Sweden’s avoidance of that tactic.  (Medrxiv Paper)   The first sentence of the abstract gives the odd way this study was set up.  It says that there is a lack of empirical evidence that lockdowns, like England’s, decreased cases or deaths more than herd immunity, like Sweden’s.  That is a weird way to set up a comparison.  Among other things, Sweden did not intentionally set out to create population immunity, it just didn’t engage in an extreme lockdown, because it didn’t think the seriousness of the epidemic warranted that.  If you want to see the comparative result of those strategies, look at relative epidemic outcomes to date, and relative economic and social impacts.  I think these researchers avoided doing that because England’s strategy looks far worse on all counts.  They instead used, or rather misused, what is referred to as time-series analysis to try to show that somehow Britain avoided more cases and deaths than Sweden did.  Again, why do this instead of just looking at the actual outcomes?  The way they set the analysis up and the assumptions that they made ensured that they made it appear that England actually prevented some deaths compared to Sweden.  The only problem with that is that in reality England has a higher per capita death rate than Sweden.  Probably because of the lockdown, England also had a sharper epidemic peak than did Sweden.

Another study from the UK finds that there was a huge drop in use of mental health services during the epidemic, notwithstanding the ability to do virtual visits and research showing that anxiety and other mental health issues rose during this period.  The authors note that this probably means an exacerbation of current problems and a deferral of needed care.  (Medrxiv Paper)

Finally, I don’t know what to make of this study.  A data analyst has done a paper using hospitalization rates and infection test positivity to estimate actual infection rates.  (Medrxiv Paper)   According to this approach, we have had over 150 million cases in the US and some states have very high prevalence rates.  I understand what the author was suggesting, but I am not sure I agree with the method of estimating cases from hospitalization rates.

Leave a Reply