Letters to the Governor and Our Modelers

By May 6, 2020 Commentary

As I encourage you all to do, I regularly write to the Governor, who claims he takes all input into account.  I don’t know about others experience but I never even get an acknowledgment of receipt.  And I have written to the people doing the modeling in Minnesota with suggestions and questions.  Initially I got responses, but lately nothing.  So here is the text of my latest email to the Governor and to the modeler, without the paragraph of introduction.

To the Governor:

I am sure you get all kinds of advice, but one danger, of course, is always talking to the same people, who will undoubtedly be most interested in responding to your own biases.  I would strongly encourage you to seek out some other, differing perspectives.  You say you are driven by the data, but you appear to be ignoring the data, very clearly found at the CDC website and in other research, that the virus poses no risk to the vast, vast majority of the population.  It has become increasingly clear, based on numerous antibody studies and intensive infection testing research, that there are far more cases than reported by positive infection testing alone and that all those previously undetected cases were asymptomatic or mild.  For 99% plus of the population, there is essentially no risk.   When I last looked at the CDC statistics, for the entire population of the country age 24 and under, 104 million people, less than 60 deaths had been reported.  There was no reason to close schools, particularly because the “data” increasing suggests that children even after exposure have low rates of infection, so are not significant sources of transmission.
So not only is the comprehensive stay-at-home and business shutdown order unnecessary, it is unwise and will cost more lives in the long run.  You are condemning us to unending stay-at-home and shutdown orders, that will continue the massive economic damage, while limiting the spread of the virus even among no-risk populations.  The failure to allow the development of population immunity, which is the surest way to stop widespread transmission, means the virus persists for longer and is a threat to the vulnerable populations.  The best way to protect those populations is for the rest of us to be infected and develop immunity.
I am 70 years old, but not concerned about coronavirus.  I am very concerned about what we are doing to our children and grandchildren, depriving them of jobs, an education, a normal social life, for absolutely no good reason.
Please don’t be proud or stubborn, reconsider the wisdom of your approach.
And here is what I sent to Ms. Enns, the lead modeler, and to Michael Osterholm, who is advising the Governor on his responses to the epidemic, again without the introductory paragraph, and the last paragraph went only to Mr. Osterholm:
One of my concerns has been the potential front-loading of the epidemic with the most susceptible individuals, which would negate an assumption that early experience with infection, hospitalization and death rates were representative of what would be seen in the entire population.  A new paper here:  https://www.medrxiv.org/content/10.1101/2020.04.27.20081893v1.full.pdf   would appear to validate this concern.  While Medrxiv papers can be drive-by type analyses, this work is done by a stellar international research team.  I would note also the recent publication in Science of a study which also finds very wide variation in the susceptibility of children versus the elderly.  https://science.sciencemag.org/content/early/2020/04/28/science.abb8001/tab-pdf
Models which fail to fully capture this variation in susceptibility or infectability will obviously overstate the impact of the epidemic and will assume it will follow a typical trajectory, when it may have a very different shape.  I know the Minnesota modeling work has attempted to address this issue by the presence of one comorbidity marker.  Since the details of the contact model were not released, I cannot assess exactly how this functioned.  It would appear, however, that it is critical to address this issue in more depth.  The nursing home population is obviously one which has such a high level of susceptibility that it probably should not be included in the general population model but run separately.
It appears that the actual history of the epidemic in the United States and Minnesota is that the virus arrived here much earlier than originally believed.  Given extensive travel contacts with China, the virus likely was infecting people as early as December, and certainly by early January.  This is supported by analyses of data from the influenza-like illness surveillance network.  It began working its way through the population, with the vast majority of cases being asymptomatic or mild, so there was no awareness that the disease was spreading.  Given the limited contact pattern for long-term care facility residents and the community dwelling elderly, it took longer to reach that population, but when it did, it was disproportionately affecting very susceptible individuals and case, hospitalization and death rates appeared to soar. It was not until this time that people began building models and the assumption was that the epidemic had only begun recently.   All along some ongoing level of asymptomatic and mild cases was occurring.  And then there was a spike in serious cases when the most susceptible population was reached.  This creates quite a different shape to the course of the epidemic.
And I am sure you can see that if this is the natural history of the epidemic, what might appear to be the effect of mitigation of spread measures would in fact only be an artifact of what was going to occur in any event, once the highly susceptible portion of the population had been exposed.
I also am puzzled by the Minnesota model’s approach to “detection rate”.  For initializing the model it was assumed only 1% of infections were picked up, but then that rate was elevated to 75%.  This seems completely at odds with all the evidence which suggests that a very large fraction, at least 90%, of infections are not being picked up by the reported positive test results.  This would seem to be particularly true in Minnesota, which has done fairly limited testing to date.  These infections are obviously all asymptomatic or mild.  So I assume the next version of the model will reflect a more accurate likely estimate of the true number of infections and a corresponding reduction in estimated hospitalization rates.
Finally, I am a little surprised that you are not suggesting that allowing some greater controlled spread in the population would not be the wiser course, given the limited sub-populations that are actually at any serious risk.  The fastest way to control any epidemic is to reach population immunity.  Measures which unduly hinder spread among low-risk populations make it take longer to reach that state, which actually creates more danger to the at-risk population for a longer time.  The Swedish approach would seem to validate this and that country will likely be out of danger in the next few weeks, while the rest of the world flounders with indefinite economy-killing severe suppression tactics.  And if the papers cited above are correct in their assessment of the variabilty of susceptibility, we will reach population immunity at much lower levels of infection.
Given all the health and other harms which ensue from severe mitigation measures, I am sure you would consider that an approach of working faster to population immunity might actually be a wiser strategy.  Happy to discuss at your convenience.

Leave a Reply