If medicine does not want to get better at generating evidence, we are lost

Enough excuses

Over the last week, a debate has simmered between the pages of my Substack, Sensible Medicine, and the Sensible Medicine podcast, which you can watch here.

Mandrola and Prasad are back

Briefly put the argument is:

Yes, of course, medicine would benefit from more large, well done randomized studies, and better epidemiologic evidence, but this is expensive, costly, requires coordination, is time consuming, thankless, and ergo, it is too much to ask. For many questions, we will have to settle for suboptimal evidence.

I completely agree with the commenters that, in the current system, running well done studies is difficult. Most randomized trials have been hijacked by conflicted parties — they are often not reliable, and there is little reward or assistance to clarify medical science. I know it is hard.

But what troubles me is the fatalistic attitude that we cannot do better. If medicine wants to make recommendations, it must do better at generating evidence. Otherwise, we are lost.

Screen time

Pediatricians advise parents how much screen time to give to kids. This entire body of literature is hopeless. That’s because the types of parents who don’t need to give their kids any screen time — as their houses are full of activities and nannies — are different than the types of parents who give their kids tons of screen-time. In some studies, parents report more than 8 hrs a day of screen time. Imagine the type of parent that does that for a young child!

How can you control for parents and find the independent effect of screen time? How can you arrive at the truth? Is it good or bad for cognitive growth to give 30 mins, or 2 hours or 4 hours of screen time to a 1 year old — all else being equal? There are no good data to advise parents, and yet, the American Academy of Pediatrics does.

For kids under 18 months they say no screentime, except for Facetime — a hilarious carve out just to sate the grandma contingency.

The truth is they do not know.

There are hundreds of good studies one could perform to improve the understanding of this question (from small and cheap to large and expensive), but none of them are ongoing. Medicine is happy to just keep talking without knowing, and to remain as dumb as we ever were.

HLH or HITT

Every week I am on service, there is a consult for hemophagocytic lymphohistiocytosis or heparin induced thrombocytopenia and thrombosis. When I started medical school 18 years ago, these diagnoses were seldom evoked, and now they are evoked constantly.

The former leverages data from kids — who, by the way, have a different disorder entirely — and the latter is a mix of low levels of data pieced together into a narrative.

In both cases, there is an urgency. Do we give etoposide? Should we stop heparin? Give argatroban?

And in both cases, no one knows what they are doing. There are ZERO credible studies to drive testing, interpretation, or treatment. It is a sea of panic + weak data + pathophysiologic storytelling.

Shouldn’t we feel embarrassed?

I feel like the doctor who drained George Washington’s blood when faced with these consults. Worse, when someone tries to explain the diagnoses to me — as if I don’t understand — I find it patronizing or worse.

What evidence supports the work up? Etoposide? It is an evidence based disaster. For HITT, we wrote up our concerns. And if a heme onc fellow is reading this and wants to write up HLH, email me please.

Conclusion

Medicine cannot be satisfied not knowing — unless that also means not commenting. But since we cannot help the latter, we must improve the former. If we don’t aspire to learn more about optimal screen time, or don’t wish to run randomized trials in adult HLH or HITT, then we are hopelessly lost — a brilliant species doomed to be as ignorant as our ancestors.