Evidence based medicine

Sounds like a good idea, it's the right way to go but, it ain't easy.

Here's a problem I've been working on. Suppose you wanted to search all the relevant scientific literature and do a systematic review and metanalysis to figure out the best way to diagnose acute appendicitis. (I'll leave the definitions of unfamiliar terms in the preceding sentence for later; suffice it to say we'll be applying the best available methods for combing information from all of the good quality studies that have been done on the subject.) Then you can write a report and tell doctors how to do this. Fewer people get sent home only to have their appendix rupture; fewer people get surgery unnecessarily. Yeah!

Oh my gosh is it not nearly that easy. First of all, why are we trying to diagnose acute appendicitis in the first place? There are a lot of different tests you can use, but before you use any of them, you have to decide who to use them on. The facile answer is, people who show up at the ER with pain in the lower right quadrant of the abdomen. Then there are various other more specific clinical indicators that will raise or lower your suspicion that this really is acute appendicitis. But are we trying to diagnose appendicitis, or figure out whatever the reason might be that they have this symptom? If we go straight to question 1, maybe we're missing something else we ought to be thinking about.

And is this person a child, a woman (or for that matter a 14 year old girl) of reproductive age not known to be pregnant, a woman known to be pregnant, an adult male of [insert age category], or a very old person? How long have they had the pain? Every one of these variables means you're looking at a whole different kettle of fish.

Does the person have insurance that will pay for a CT scan or will they have to pay out of pocket -- or maybe the hospital will have to eat the cost. This is a medical emergency, potentially, so they have to treat you whether or not you can pay, but what does "treating" you mean? If you do have appendicitis, the cheapest thing to do (in hindsight), and the best from your point of view as well, is to go straight to surgery. On the other hand the totally cheapest thing to do, and the best from your point of view if you don't have appendicitis, is to send you home with advice to come back if it doesn't get better. But in either case, the wrong decision is likely to cost more and result in a much worse outcome.

You can always diagnose appendicitis, with almost perfect sensitivity and specificity, by opening up the abdomen and removing the appendix and having a pathologist look at it. But that's not really the question after all, is it? The question is, should I open the abdomen of this person with lower right quadrant pain. Maybe I'll find out they don't have appendicitis but there is some very good reason why I should have opened their abdomen. On the other hand, maybe they have an ectopic pregnancy or an ovarian tumor or colon cancer, and the wrong kind of surgeon is looking at it, who doesn't know what to do.

You can start with ultrasound, which has essentially no risk, but it's often equivocal. Then you can go to a CT scan, which does irradiate you with potentially some small long term risk, unless you're in early pregnancy in which case there may be greater risk to the fetus; you might also be less inclined to do a CT scan on a young child. If your hospital has an MRI scanner, there's no ionizing radiation but it's more expensive. And the hospital might not have one.

I could go on but you get the idea. You might be particularly averse to post-surgical pain, or you might not care so much. You might have known someone who died of a ruptured appendix so you're really anxious about it, or you might be happier to go home and wait. Whatever.

So . . . Doctors don't generally "know" what they're doing. They're just muddling through, hopefully for the best. As are we all.