Circumstantial Evidence

Got a great comment yesterday on my evidence-based medicine post, but it's kind of buried down below so I'll reprint it here:


Is EBM an intervention, or is it an organizing principle about interventions? You can write a prescription for acyclovir, but not for EBM. The teacher may have been put out by the confusion of logical levels rather than by the proposal to examine EBM. Its proponents so far as I know are not so woebegone as to insist on RCTs for interventions with very large effects (like insulin for diabetic ketoacidosis or parachute use for skydiving), but for those with small to moderate effects, which are the rule in medicine. Growth by small steps happens much more often than by great leaps...


This comment, from someone who goes by Pontificator-In-Chief, raises a good point. Of course I'll acknowledge that Randomized Controlled Trials aren't always insisted upon, and that EBM as an organizing principle is really quite reasonable.

My point, however, is that lots of things are really quite reasonable. Antivirals for viral infections, for instance. The beauty of EBM is that it ignores mechanisms and looks at the bottom line: does the antiviral reduce the duration of illness? Severity of symptoms? Or, more broadly, for things like lipitor -- sure, cholesterol numbers fall, but does lipitor cut down the number of heart attacks? Extend life?

Evidence-based medicine should really examine its own bottom line. It hasn't been proven that the rigid application of evidence-based guidelines actually improves patient outcomes. It probably cuts costs, sure -- by reducing the number and choice of prescriptions. But using EBM brings some side effects, listed below. One problem I didn't dwell on previously is erosion of the doctor-patient relationship. By inhibiting patient autonomy and forcing adherence to certain guidelines, EBM might damage them to the point where patients are less inclined to go to the doctor, less likely to self-medicate, etc.

After a couple of sore throats in which no antibiotics are prescribed, will a patient become less likely to have that fever checked out? Or ask about that lump? I can't say. Either way, it's probably hard to assess. But the bottom line is worth examining, and EBM proponents should be finding ways to measure it. After all, PSA testing seems quite reasonable at first glance, but now no one balks at multi-decade, multi-million dollar trials of PSA testing on mortality.

And, by the way, lots of non-prescribable interventions are measured and assessed all the time. Check this one out:


Evidence-based physicians' dressing: a crossover trial.
Med J Aust. 2002 Dec 2-16;177(11-12):681-2.
Nair BR, Attia JR, Mears SR, Hitchcock KI.
Department of Geriatric Medicine, John Hunter Hospital, Hunter Region Mail Centre, NSW.
OBJECTIVE: To describe the effect of physicians' dress on patient confidence and trust. DESIGN: A prospective crossover trial involving physicians dressed in "respectable" versus "retro" attire. SETTING: A general medicine ward at a tertiary hospital. PARTICIPANTS: 12 male general physicians and 1680 patients. MAIN OUTCOME MEASURES: Patient trust and confidence as measured by a questionnaire mailed after hospital discharge. RESULTS: Formal attire was correlated with higher patient confidence and trust. Nose rings were particularly deleterious to patients' reported trust and confidence. A minimum threshold of two items of formal attire (dress pants, dress shirt, tie, or white coat) were necessary to inspire a reasonable amount of confidence; this is the NND (number needed to dress). CONCLUSIONS: We highlight the need for more research into the effects of physician dress, and coin the term "evidence-based dressing".


It's a small step (made during a silly walk) from this kind of research, to the kind of EBM vs. 'traditional medicine' comparison that I'm proposing. EBM's been around for over a decade -- why hasn't anyone done it?