Friday, January 23, 2009

Evidence-Based Medicine

I've long admired the attendings who really know EBM. I have aspirations to be one someday: have an office filled with articles to print out for residents, know which trial said what, who conducted the important trials, be able to read articles critically and pick up on subtle clues that the research was or wasn't great.

I initially said "always" instead of "long" in that first sentence, but I changed it. My first experiences with EBM were pretty pointless. In our first year of med school, we took a mini-class in statistics, where we had to memorize the "ABCD" tables and what sensitivity/specificity meant. It didn't make much sense to any of us, and everyone groused through the course. That summer, I had to do a project for my preceptorship that involved "PICO" questions. They're ridiculously easy to write, so I couldn't figure out why I had to do so many. I was told that I needed to learn how to search Pubmed properly, so I had to have a good question. Later, I had more lectures on how to search Pubmed well. Seriously, people, most med students now are pretty familiar with Google or other search engines and know how to conduct a quick internet-based search to get what they want. Sitting through stifling lectures about "boolean operators" and learning the difference between searching with AND and with OR... Shoot me.

I think med students may get turned off by the statistics and the uselessness of learning how to search the internet and lose sight of why EBM overall actually matters. It has very little to do with PICO questions, after all.

The way I see it, EBM has several points. First, doctors should know how to critically appraise an article, so when the drug rep hands you the article about linezolid vs vancomycin you aren't blindsided by the pretty graphics. Second, doctors should know how to search the literature to find answers to clinical questions--this is where PICO comes in, but it isn't always necessary to go through that whole process. Still, if you want to read the actual study that UpToDate based their guidelines on, you should have an idea of where to find it. Third, all of this critical appraisal *should* lead to evidence-based guidelines for treatment. I know many doctors get upset at the idea of "cookbook medicine", but I'm all for some standardization (with final discretion always with the doctor and the patient, of course). If the literature said Drug X is very good in diabetes, but not Drug Y, then I think a doctor who wants to prescribe Drug Y to a diabetic should have a really good reason for doing so.

So what are the downsides to EBM? Why isn't everyone doing it? I think there are several potential problems:

First, while the randomized controlled trial is the gold standard for testing therapies (new drugs, new imaging, new surgical techniques & devices, etc), not everything can feasibly have an RCT. Pregnant women and children are often overlooked for studies, because who wants to have their fetus or small child experimented upon? The elderly are often excluded from studies, as are the really ill patients. Thus, RCT's often ignore whole populations that may need a treatment, so then we have to try to extrapolate the results to an untested population (or, if you're a purist, just say "there's no evidence for X in pregnant women" and don't treat). Diseases with very small numbers of patients may be studied in fantastic trials that can't reach statistical significance due to lack of power. And sometimes, you just can't randomize people to have a certain condition--see the satirical "Parachute use to prevent death and major trauma related to gravitational challenge: systematic review of randomised controlled trials" from the BMJ.

Second, there's often a disconnect between study data and clinical practice guidelines. You don't usually base guidelines on the results of one study, but sometimes that happens--see the WHI hoopla. A treatment may become standard of care based on one study (like using steroids for spinal cord injury) and then even when later studies debunk it you can't change the standard of care (defensive medicine, anyone?). Other times, it takes years and multiple studies to "prove" something works or doesn't work, meaning at any given time the guideliness available are many years behind the evidence. Also, guidelines are written by "panels of experts", so sometimes it's difficult to tell what's truly EBM and what's "expert opinion".

Which brings me to my third problem: we in medicine haven't done a good job of selling the public on EBM. Patients don't really know about sensitivity and specificity, false positives and false negatives, statistical likelihood of disease, number needed to treat, etc. How many people would take Lipitor if they knew that between 100-250 people have to take it to prevent one MI? How many people would clamor for increased HIV testing in the ER if they were the patient with the false positive who had to go through the extra testing and fear of having HIV? Even worse, one may still be sued successfully for following EBM practices when it results in a poor outcome.

Fourth, there's a wide disparity of practices between patients with money and good insurance and patients who do not, or even between geographic areas. In a rural area, if you have stable angina, you're likely to get nitro; in an urban area, you're likely to get a cath. Do people in urban areas live longer? (I don't know, but I haven't seen the evidence). I feel like one goal of EBM *should* be to reduce some of these types of disparities, but it hasn't really happened. The growth of medical technology outpaces the body of literature.

Finally, what do we do when there is no evidence? Many of us are not comfortable doing nothing, and many patients are not comfortable doing nothing. For example, when a patient has viral bronchitis, we know the evidence says DO NOT GIVE ANTIBIOTICS. Yet, patients still come to the doctor with cough and runny nose. There's no evidence for giving cough syrup or inhalers, yet I think many of us do so, simply for the sake of doing something (and to get the patient off our backs about the freaking antibiotics, the answer was NO).

So why do I heart EBM? I feel that it's the best system out there for keeping abreast of the unbelievable amount of medical knowledge available. Sure, it's got flaws, and perhaps in the future a new system will come along and wipe EBM off the map, but until then we should use what we have.

You can wake up now, I'm done.

1 comment:

red rabbit said...

We have a great EBM tutor who loves shredding the drug reps. So much fun, it makes everyone want to go, me next, I want to torture a drug rep!

Also, I hate meta-analyses. Grr.