Is There a “Magic Bullet” for COVID-19?
Thirty-one years ago, I did my first night of call as a new surgery intern in Atlanta. It was intimidating. We were often evaluated in medical school by how well we could identify the one right answer on a multiple-choice test. I remember I would try to have a single solution for whatever the nurse called me for: one anti-hypertensive for a patient with high blood pressure, one sliding scale of insulin for a patient with a high glucose level, and so on.
I’ll never forget the very first call from a nurse – probably 10 minutes into my night. The nurse said she had a post-op patient with constipation and asked if I would I give her an order for a suppository. I froze. (I hadn’t yet learned how valuable nurses are in helping residents on call, especially in their first year.) I said to wait just a minute and I would call back. I went to the senior resident call room where the chief resident on call and another senior resident were talking. I interrupted them to ask if it was OK to order a suppository for a patient with constipation. They looked at each other and one said, “Uh, yeah.” I’m sure they had a good laugh after I left and thought, “That intern won’t last long.”
Having one primary treatment for any medical disease or symptom makes sense on the surface, until you realize that all disease, and everything in our biologic world, is complex. That means there will never be one right treatment or a “magic bullet” for any disease. As much as we would like Reductionist Science to be true (that we can identify a single cause leading to an effect; that things are black and white; that a drug or device is good or bad; that there is one right answer) it’s not how our real, biologic world works.
Over 100 years ago, Paul Ehrlich, a German physician, scientist and 1908 Nobel prize winner, promoted the concept of a “magic bullet” cure for any infectious disease. The drug Ehrlich and his team eventually discovered was called Salvarsan, a treatment for Syphilis. It was tolerated better than the previous mercury-based treatment, but it didn’t work for everyone. It also had some severe side effects, including fever, rash, liver damage, limb loss, and even death. Salvarsan was clearly an improvement in the treatment of Syphilis, but it was no “magic bullet”.
In a recent article published by FDA researchers in the journal Frontiers in Immunology, they presented lessons learned about COVID-19. In their conclusion, they noted, “We are learning day after day, that patients may be affected by SARS-CoV-2 differently and that many factors influence the outcomes of the disease.” This is true for any disease, whether it’s obesity, cancer, addiction or depression; all are complex and are the result of many factors and will impact different subpopulations differently. In the same way, any potential treatment for COVID-19, such as Hydroxychloroquine, will have different outcomes for different subpopulations. There may be benefit for some and it may contribute to unintended harm for others. If there is no harm or benefit, then giving the drug would be wasteful.
The following illustrations are intended to help demonstrate the principles of Systems Science applied to the COVID-19 virus:
And, for one proposed treatment, Hydroxychloroquine:
These figures demonstrate the same principles of Systems Science that Netflix uses to identify subpopulations of viewers – what they call “taste groups”. Using data from viewers, data from each show, and data from each local environment (language and cultural variables), the algorithms can provide show recommendations for each subpopulation to drive customer retention and satisfaction.
The same scientific principles could be applied in healthcare by first defining the whole patient process (COVID-19 infection in this example). Then working with clinical teams in each local environment, patient and treatment data that matter the most could be collected, and outcomes could be measured and collected as well. As more and more data is collected, algorithms could be generated, and their accuracy would improve over time.
I wish I had been able to use Systems Science tools when I was in training. With the appropriate systems and data science infrastructure in place, my response to my first call from a nurse could have been very different. Based on several factors specific to the patient with constipation, an algorithm may have suggested that magnesium citrate by mouth would be more likely to have the best value for this patient rather than a suppository. And instead of asking the chief resident if it was OK to order it, I would ask the nurse and the patient to make sure we all agreed – a data-informed, shared decision process. But without an appropriate data and analytics infrastructure in place in healthcare, we will continue to wait for that “magic bullet”. How long do we have to keep waiting?