Georgia very controversially lifted their lockdowns on April 24, 3 weeks ago, raising concerns of a second "surge" and with some warning of a return to exponential growth. It does take a few weeks for new infections to emerge and be counted, so it takes time to see the effect of this policy.
Now that 3 weeks have passed, it is worth taking a look at the data for Georgia. Thankfully, this state has more complete data displays than most other states, so there is a lot of good information on their website. Here is the key figure showing incidence over time.
Looking at these data, we see that the incidence rate seems to have decreased since opening up April 25th. But we have to think very carefully about how to interpret these data.
As I mentioned in a previous post, Georgia tries to assess infection date more carefully than other states -- most states count an incident case on the day when the positive SARS-CoV-2 test result is obtained, which is biased late given (1) test results typically take 2-8 days after sample taken, and (2) many people do not even get SARS-CoV-2 test until they've had symptoms for a while. This makes it difficult to assess the effect of interventions and policies, since the positive tests today typically come from infections started ~2 weeks ago.
To their credit, Georgia tries to account for this, in that when they obtain a positive test, they date that case based on (1) when symptoms reportedly started, and if unknown, (2) date sample is obtained and, if unknown, (3) date of positive test. This appropriately shifts cases earlier and closer to the actual infection date, and makes it easier to assess effect of policies.
However, as I commented in my previous post, this practice can result in serious undercount of cases in the past 2 weeks, since the numbers reported for today will not count individuals whose symptoms started today or whose testing sample were taken today but do not have a positive test yet -- these counts will be added to today's counts as these people receive positive test results in the next 2 weeks or so. I thought this was a fairly serious issue, and privately worried if this might be used by Georgian officials to try to inappropriately argue incidence was decreasing more than it was to justify their policy.
As can be seen in the plot above, to their credit Georgia has adapted their display such that the most recent 2 weeks data sort of "greyed out" with a comment that these data may be incomplete. These are the days whose incidence counts are likely to increase in the coming days. Kudos to them for transparency -- rare in today's world -- and especially given the heat they are under based on the bold strategy they have taken.
However, the data from 2 weeks and previous (up to April 30) are pretty mature, and worth looking at. This includes cases from the first week after the lockdown was lifted. We do not see any increase yet, no uptick evident yet, and their 7-day moving average even shows a slight decrease in incidence. The past 14 days show a sharp decline, but as mentioned above these data are incomplete so should not be overinterpreted. I eagerly will watch this as the next 2 weeks will tell us a lot, but so far this is very encouraging.
How can a state that has the most extreme lifting of lockdowns not see an increase in cases? One key factor is that just because the government lifts lockdowns does not mean everyone immediately returns to their pre-pandemic activities and practices. Some yahoos might immediately go to crowded parties and bars without any type of protection, but a vast majority of the people have fear or concern about the virus and are taking personal precautions to avoid exposure and infection. It is likely that relative to pre-pandemic times, people are avoiding large crowds, practicing social distancing in public, avoiding personal contact with others, avoiding touching public surfaces, practicing better hand hygiene -- some are wearing masks (I wish more did) and a certain proportion of the society is so scared of the virus they still won't leave their house even with lockdowns lifted, even to do things with minuscule risks of infection.
This matches what has been reported in Sweden. Government-induced lockdowns were not necessary for individuals to practice caution to reduce exposure and spread of the virus, as many people in Sweden are avoiding crowds, practicing social distancing, some wearing masks, etc. That is why I strongly believe the "Swedish strategy" is broadly misrepresented in the media. It is NOT trying to quickly move to herd immunity, which would suggest people purposefully exposing themselves to the virus to speed the spread -- it is just avoiding overly restrictive government-imposed lockdowns and shuttering of businesses, educating the public, and trusting them to behave reasonably in avoiding the major risks.
This also applies to the current situation in Georgia.
This is a KEY QUESTION for policymakers to think about -- what is the counterfactual to state-induced lockdowns? That is, without the lockdowns, what would the viral spread be like? Those that seem (to me) to be suggesting indefinite lockdowns seem to believe that without the lockdowns, we would see a major surge and possibly even exponential growth throughout society. However, based on the factors mentioned above, I don't think that is accurate. Given what they have learned, and the healthy fear of the virus they have acquired, much of society will practice mitigation strategies even if they are not mandated by the government. This would suggest a strong degree of mitigation will be seen even as lockdowns are lifted. Thus, I think the accurate counterfactual to state-induced lockdowns is more like what we see in Sweden and what we will see in Georgia (and now Texas and other places that are opening up) and should be taken into account.
That is why we need to watch these numbers as they emerge with AN OPEN MIND. We should not dismiss the strategies of Georgia, Texas and others, being so certain of their foolishness and the carnage they will produce, that we don't even look at the data and see what they tell us. We are slow to update our understanding of reality shaped by doomsday scenarios when the truth is not quite so cataclysmic, and some people don't seem interested in even looking at that data..
My strongly believed prediction has been that lifting of lockdowns will produce some natural increase in incidence, but it will not spike up into a new surge, and certainly will not approach anything close to exponential growth because of the abundance of caution practiced by a vast majority of those in society (a small minority of yahoos aside). Thus, it seems possible to lift these restrictions and open more of society and still maintain a reasonable level of viral spread mitigation.
This virus spreads like crazy and is very dangerous, so we need to respect the danger it poses and take steps to mitigate its spread. But I think the lockdowns are overkill and the collateral damage they are inducing, some of which will be permanent/long term, outweigh their benefits relative to what I believe to a more targeted mitigation strategy that can be practiced by limited government guidelines accompanied by an organized, clear education program for how people can mitigate their risk of infection while reentering society.
I believe we need to move to LIFT LOCKDOWNS now, or very soon, and the politicians that do not start doing so will pay an extreme political price and their states will pay a greater economic price.
コメント