My 'Medical Conservative' Talk in San Diego

Some background on how I have adopted a skeptical but not cynical approach to medicine.

This post remains free to all readers. But thanks to all that have chosen to support my work. I am surprised and grateful. JMM

---

This week I traveled to San Diego to give cardiology rounds at the UCSD.

Most ‘cardiology rounds’ come from famous professors who speak about a specific area in cardiology. Mine was different. I was not teaching any one thing.

Instead, I tried to teach an approach to the practice of medicine.

The title of my talk was a Medically Conservative Approach to Cardiology Practice. It’s not a political thing, but the theme has parallels to the famous quote from William Buckley, Jr

A conservative is someone who stands athwart history, yelling Stop, at a time when no one is inclined to do so, or to have much patience with those who so urge it.

It took me more than a decade to reach the ‘yelling, Stop’ approach to medical practice.

Earlier in my career, there wasn’t time to think about evidence. These were the glory days of electrophysiology. Pioneers had just invented catheter ablation of atrial fibrillation and pacing for heart failure. We spent our energy learning these novel approaches. It was exhilarating.

Then three things happened. Innovation plateaued, procedures became (mostly) rote, and I lived long enough to see some serious flubs. I started to look at history and evidence.

Learning how to look at evidence took time. But once you start looking, you see. And then you can’t unsee.

Soon, you can become the person yelling, Stop. And, if you do that enough, you meet those who have little patience for this approach.

I began my talk by urging the mostly young audience to first consider what a study is for: is it for marketing or is it to answer an important question.

Sometimes it is hard to sort this out, but it is always important to ask this question. It turns out that there are great examples of both.

The next chapter centered on how our attempts to infer causation from nonrandom comparisons in observational studies can result in serious harm.

I showed the young people the CAST trial, wherein we learned that the accepted practice of using drugs to suppress premature beats in patients who had an heart attack was actually killing people. (I think CAST is literally the most important medical trial of our generation.)

In my clinic at the old Wishard hospital, in Indianapolis, in the 1990s, I prescribed hormone replacement therapy to older women to prevent heart disease. But this also turned out to be harmful. I told the story of how the drug digoxin is unfairly maligned because of flawed observational studies.

Then I showed how results of a study can turn on how the data is analyzed. That sounds weird, doesn’t it? But it’s true in some cases.

Another chapter of my talk explored the choice of study endpoints. Cardiology studies used to measure death rates. Beta-blockers, ACE-inhibitors, spironolactone, defibrillators have all been shown in randomized trials to improve survival in patients with heart failure.

But now, it’s rare for trials to only measure death rates. Most new interventions reduce only surrogate endpoints—like hospitalizations due to heart failure. Progress creates headwinds.

I gave an example of how a flawed endpoint led to acceptance of a dubious measurement technique in the field of interventional cardiology.

The final two chapters dealt with the hardest task any clinician faces: translating evidence to the patient in the clinic.

Trials produce average effects in selected patients who are treated in the special environment of a clinical trial. (Trials often have frequent follow-up and research nurses to help care delivery run smoothly.)

If the average age of patients in the trial was 65 years old, how does that evidence apply to the 85-year-old in your clinic?

---

A nice example of the challenge of evidence translation came from a study published last week in JAMA-IM.

There are two strategies to treat patients with a certain kind of a heart attack. One is to take the patient immediately to the cath lab and the other is to use medicines first and go to the cath lab only if patients have recurrent symptoms.

Previous trials have shown a modest benefit for the immediate approach, but patients in those trials were relatively young.

The recent study, however, enrolled older patients with some degree of frailty. And they found no advantage to the immediate strategy and a strong trend in favor of the conservative approach.

This figure, from my colleague Andrew Foy, explains the complexity of how treatment effects can vary in different patients.

The simplest concept is that a treatment has benefit. But it also can have harm. These are opposing forces. Anticoagulant drugs reduce clots but increase bleeding, for example.

But that is not all. Also relevant is the risk of having the outcome. And the competing risks. Screening colonoscopy won’t increase survival in young adults because their risk of colon cancer is too low for early detection to make a difference. Screening also won’t work in ninety-year-olds, because there are too many competing risks of death.

---

These six chapters of evidence translation only skim the surface of what it takes to understand and apply evidence. But I hope you get the sense of why I’ve adopted a conservative approach to practice.

My final slide:

The learners asked many great questions. Perhaps in a coming post, I will tell you what they are, and how I answered them.