By Dr. John M. Livingston | Medical Policy Adviser
I was surprised yesterday when Governor Little decided to extend his statewide “stay-at-home” order. However, I was more surprised to read Betsy Russell’s “Eye on Boise” article about various state and local officials — two of whom own their own businesses — who supported the governor’s decision. I will attempt to wrap my mind around the new strategy of continuing the “stay-at-home” order, and for the support given to that policy by people in government. I am sure if Betsy went out and interviewed restaurant owners and small businesses that employ fewer than 50 people, she would find very different answers. And this group of small business owners — along with their families, their employees, and employees’ families — is the very group of people government officials say they represent!
In statistics and logic, there are several types of bias. Among these are selection bias, information bias, and confounding bias. I was first introduced to selection bias in church school in the seventh grade, when we read in the book of Matthew, chapter 7, which reads, “And why worry about a speck in your friend’s eye when you have a log in your own?” I see selection bias, wherein people select examples that lead to the outcome they desire, in the general scientific literature. Within the last year, there have been several articles admonishing clinicians, public health specialists, and bench researchers about the misuse of data, and the failure to offer alternative hypotheses when describing outcomes.
The best example of selection bias extrapolated to the media would be in Betsy Russell’s article mentioned above. She asked questions and got the results she knowingly or unknowingly selected for. I think she sees the log in her own eye — careful selection of those she interviews — and doesn’t care.
Next is information bias. The process of acquiring data can itself produce information bias, and over, time this bias can become a self-fulfilling prophecy. The process of acquiring information needs to be constantly scrutinized, so that preexisting prejudices regarding outcomes aren’t preordained. This idea could also be applied to the press reporting from the same sources over the years, when both the source and the reporter have a preexisting bias.
Confounding bias occurs when two variables acting independently affect an outcome, but their effect is obscured because the relationship of the variables is not understood. In statistics and graphic theory, an almost analogous concept is described by the collide-three independent variable theory, and in chemistry by “collision theory” of molecules. The point is: When outcomes are being evaluated, the way in which the independent events impact each other needs to be taken into account.
An example of confounding factors affecting epidemiological studies would be the impact of body weight and co-morbidities on clinical and statistical outcomes. Is everyone who is given a CPT diagnostic code of COVID-19 dying of the virus or of their other conditions? Are their social factors being taken into account, such as proximity of living conditions, use of public transportation, air travel, and immunologic demographic abnormalities? Are these factors considered in an analysis and the formulation of mitigation strategies? Should the same mitigation strategy be used in Boise, Idaho as in New York City? What about Ada and Blaine
Counties? Should local officials be given more say in how mitigation strategies are applied?
I bet the local mayors in Betsy Russell’s article would provide a different strategy to the coronavirus problem, especially if they routinely shared morning coffee in a local restaurant with their buddies who owned small businesses in their municipalities. Restaurant owners and their customers would advise the local public officials very differently than the local public officials are advising the governor. That’s probably because local business owners don’t see Governor Little during the course of their everyday activities, and they are more likely to try and court his favor than he is to court their vote for office — an issue of agency further discussed below.
I identify three questions that have not been asked by scientists or the media. These unasked questions prove my hypothesis about selection and confounding biases.
First question: What happened to 500,000 doses of hydroxychloroquine delivered to New York? I don’t know, but every health care provider I have talked to — both civilian and military personnel, including registered nurses, physicians, and technicians currently serving in the tristate area — is taking hydroxychloroquine for prophylaxis. I have also been told that, starting three days ago in New York City, every patient diagnosed and admitted to the hospital with COVID-19 has been offered hydroxychloroquine. I bet — knowing the creativity of the drug market in New York City — this drug is being offered on the street. Is it just social distancing that is causing New York’s drop in coronavirus cases, or are there other factors involved? Or both? Is the black market for hydroxychloroquine an example of private enterprise saving the day?
Second question: Why is the disease so different in different locales? Is it just because of confounding factors, or could there be other causes? Three days ago, in the New England Journal of Medicine, the “Icelandic Study” reported Iceland’s experience with COVID-19. Iceland is a country very similar to Idaho regarding demographics and access to medical care.
An answer to my second question was hinted at in the Icelandic article, when they discussed an extension to the PCR testing using haploid pair analysis. Maybe the virus coming into New York from Europe was different than the virus — minus or plus a couple of mutations — that came from the West into California. And, knowing this, should mitigation be different?
My third question: Testing. The Icelandic study focused on random testing of symptomatic cases, and all comers demonstrated similar incidence rates, around 1%. Testing is important in certain situations, like for health care providers and food workers. But the public health strategies based on symptoms would not differ if an antigen or antibody test were positive or negative.
There are clinical and public health reasons for testing. The public health reason is not an emergency. And if hospitals and employers want to test their employees, they can pay for the test. That is what Treasure Valley Hospital is doing locally, and they also test family members. That is what the airlines will be offering passengers.
I know there are people on the governor’s panel who know these technicalities described above. Hopefully, their knowledge is one of the reasons they are advising him. The same is true about the people interviewed by Betsy Russell.
In the end, there is one final reason for their opinions about the stay-at-home order. Adam Smith described this as “the good fellows theory.” Despite what famous politicians have told us, we cannot feel another person’s pain. Even losing a child yourself does not enable you to feel the loss of your best friend’s child the same way. The emotions of empathy and sympathy are qualitative, but putting yourself in another’s shoes — a form of fiduciary duty — is quantitative. Because of this, a relationship is established whereby the despair of an upside failure is never approached by the joy of a downside gain. Advisor agents, either because of money or reputation, will always skew to cautiously mitigating the downside.
This is not real or hopeful empathy. If those same advisors never put on an entrepreneurial hat when assessing risk, or if doctors were to never treat a cancer patient with less than a 50% chance of survival, or if engineers were assessing the risk of a space shot, progress would never happen.
Political advisors and our governor, when mitigating coronavirus risk, are putting their own professional and political reputations ahead of their constituents, who are small businessmen and women, employers and employees. So is Betsy Russell, by the way she only interviews government workers and lobbyists.
In that regard, it is my opinion that President Trump intuitively understands risk in the same terms as Adam Smith. He may not be able to put it into the same words, but he has been living with risk — winning and losing — all his adult life. He is unique amongst the leaders in our country in this regard. I wish our governor understood concepts of risk and opportunity cost the same way.