The Illusion of Control

Evidence is important but there are limits

Before John Lloyd went to medical school, he studied to be a Christian minister. When he started practice, furosemide (Lasix) had yet to be developed; before the electronic record, he wrote his notes in story-form with big cursive letters, and, how he looks in that picture is how he looked in the middle of a call night.

In one short conversation, eight years ago, the now retired and beloved-by-everyone lung doctor changed my view of medicine—and life.

It had to do with control, or the lack of it.

Dr Lloyd and me

In 2013, my friend Vinay Prasad published a study of medical reversals—which occurs when a proper trial reverses an established practice. My first newsletter described the CAST trial, perhaps the biggest reversal of a generation.

Prasad found shocking results: in their review of trials in the New England Journal of Medicine over a decade, they found that, of 363 trials that tested an accepted medical practice, 146 (40%) reversed the practice. Pause there for a moment. Four in ten accepted medical practices did not survive the test of a proper trial.

I was fired up. At the time, I wrote a blog post calling for change to the culture of American Medicine—by removing hubris. Our local medical society published the essay in their monthly.

One morning in the doctor’s lounge, Dr. Lloyd told me that he liked my essay. But he added that there was little new there. He said doctors have been doing dumb stuff for centuries, citing the blood-letting that killed George Washington.

I paused. Then came the seminal moment.

“John, we don’t control outcomes.”

“Dr. Lloyd, I am a cardiologist; I surely control outcomes.”

Then he told me a story of two patients with sepsis due to bacterial infection. Sepsis is a condition that is treated with strict protocols without much ad-libbing.

Lloyd spoke first of a patient who was terribly sick; the man had awful lab values and required oodles of meds to support the blood pressure. Lloyd expected the patient would die that night. But the next morning the patient was sitting up eating breakfast and looking great.

He then told me of a patient later that same week who also had sepsis. This patient had far better labs and looked a lot less sick. Lloyd used the same protocol, but sadly that patient was dead the next morning.

Dr. Lloyd’s stories highlight two important concepts of using medical evidence.

One is the stochastic nature of medical illness. The other is that trials give us average effects of a treatment on a select group of patients, but how that average effect applies to each patient is unpredictable and variable.

Stochastic is a big-dollar word that means something has a statistical probability or pattern of happening in the future, but that pattern may not be predicted precisely.

An example is that a cardiologist who is exposed to radiation over a career has a higher probability of getting cancer, but you can’t predict which cardiologist will get cancer. Another: a smoker has a higher risk of getting cancer or heart disease, but some smokers live to age 90. IOW— stuff happens.

The second concept that Lloyd taught with his two patients is the idea that average effects help us predict outcomes in 1000s of patients but not so much the one in front of us.

A look at the (labeled) graph of an unquestionably beneficial therapy—carvedilol in heart failure—shows what I mean:

---

You see the problem Dr. Lloyd has created:

First is that we cannot predict the future. Second is that even highly effective therapies have small average effects, which may or may not affect the patient in front of us.

A cynic might take from this a nihilistic approach. I do not.

In future writings, I will explain how these observations have affected my view of helping people who seek my help.

A teaser: think how these lessons might influence the decision to be screened for a low-incidence disease.