You should go to the doctor

And other randomized trials that don't exist

Recently, a family member was sick— fever, cough, runny nose— and another dispensed the usual advice: “You should probably get seen in urgent care.”

But the ‘patient’ objected, “The last thing I want to do is sit in that rigid plastic chair for 4 hours, in a drafty waiting room, only for them to recommend some over the counter medications. I’ll probably get sicker!”

Eventually someone asked my opinion.

Of course, medicine is full of rules of thumb. When should someone with a upper respiratory tract symptoms seek attention? If the fever gets better, then worsens, if the cough lasts for longer than 1 week, if you have sharp pain in any one location, if you have trouble staying awake, or, if you are older than 65, immunocompromised… the list goes on and on, and there are variations.

Essentially, all of this guidance is fabricated. It is bio-plausible at best. We have no credible evidence that you should seek care for these symptoms or risk factors but that you should not seek care for lesser ones.

Don’t get me wrong: I think it is highly unlikely that anything we do in western medicine is worth the discomfort of the plastic chair for the vast majority of people with upper respiratory tract infections (URIs) or even influenza like illness (ILI), but I also think that it is unlikely we do anything of value for even people who have cough more than 1 week, or who are older than 65.

It’s amazing to me that a profession that hemorrhages trillions of dollars in US GDP has not gotten around to the most basic questions: who should see a doctor and when? There is a simple way to design the trial. A pragmatic, telephone based study. Everyone is asked to call in if they feel sick, and they are randomized to being told to come in, or tough it out. (They can call in again a week later).

With a huge sample size, we can pre-specify interaction terms. In other words, do our policies work in people with very high fevers (>103 or 104) but not lesser fevers, do they work in older >65 vs younger folks.

And what if symptoms last for more than a week, should you go in then? A second randomization can answer that. Until any of this happens, a doctor’s opinion is hardly better than a mechanic’s. And a patient’s intuition: well it depends on how busy the office is and how hard that plastic chair is— might be better than any experts.

The trial I propose is intensely simple, but it doesn’t exist. Why?

That’s the part that baffles me. Medicine is a juggernaut of spending, but we are unwilling to invest even a little bit in the most basic questions. Primary care is an important field in my opinion, but obviously does not recruit well. Money and prestige are surely factors, but I think less discussed is this simple fact: we just know so little about optimum management strategies.

Primary care researchers are far more committed to creating new useless benchmarks— make sure your mammogram percentage is above 65%— rather than trying to actually be useful. More people study the impact of wearables than optimal referral strategies for joint pain or rash. Everyone wants to study social media use, but who wants to study fever and chills. Yet, patients desperately want basic guidance.

Should my family member, a healthy middle age person with 5 days of flu like symptoms go see the doctor? (PS: COVID neg per home test— but whether they should have tested with this also lacks RCT data).

The answer is: Who the hell knows? You could read every paper in the biomedical literature, understand every statistical test, but your opinion is no better than they guy down the street. So I gave the universal right answer: I don’t know. What your mom say?