Wednesday, June 12, 2013

The origins of second-best medicine


Medicine is not a science, but an art empowered by science. So it is little surprise to find that the same social factors that shape other workplaces - habit, hierarchy and deference to colleagues personal sensitivities - shape the antibiotic prescribing practices of physicians. Who likes to stand up and tell an esteemed colleague that they are making a mistake?  Well, some people do, but they soon get labeled as habitual contrarians to be ignored or suppressed. Not a career path for success. 

Esmita Charani and colleagues at The National Centre for Infection Prevention and Management in London have put a bit of science behind the suspicion that doctors behave in their workplace pretty much like everyone else does in their workplace. In a series of structured interviews with doctors, nurses and pharmacists, they found that evidence-based guidelines and policies counted for much less than personal experience, intuition and personal authority when it came to actual antibiotic prescribing practices. 

How big a problem is this?  After all, doctors care about giving the best care possible, and want to see their patients get better.  If poor prescribing practices were leading to bad outcomes, surely they would stop those practices - right?

The problem is not that doctors don't care, it's that they come equipped with human brains. And human brains are notoriously unreliable when it comes to evaluating success rates. We are good at remembering a few spectacular successes, thinking that they validate our competence. We forget the equally spectacular failures, believing that they are due to an unforeseeable combination of circumstances and bad luck. Most of all we are just plain bad at distinguishing natural fluctuations in success rates from pure chance.  

Baseball offers a great example of this phenomenon. Over the course of a week of baseball games, a star player will typically get 7 or 8 base hits. A journeyman would get 6, and a scrub would get 5. The difference between best and worst is a single event every few days.  No one could possibly tell the difference just by watching. That's why baseball teams keep score -  so that they always put the best players out on the field, rather than trust to recent experience, reputation or intuition. 

For critical diseases, such as Staph aureus bacteremias, the differences in outcomes between the best prescribing practices and the second-best are about the same as the differences between a star player and a scrub. MSSA patients that are prescribed general-purpose broad spectrum antibiotics have  roughly a 25% chance of dying; those whose prescriptions are evidence-based, and are specifically targeted have their risk of death reduced by half.

An individual doctor, who might see only 25 or fewer MSSA bacteremia cases per year, is not likely to notice the extra death or two that results from suboptimal prescribing practices. No doubt that patient was weaker to bnegin with, and had other complications. Nor will they be aware that the survivors spent more time in the ICU and more time in the hospital than necessary: they have treated their patients with a therapy that was helpful and most have survived. Why question what seems to be working?

Individual human brains are basically incapable of getting this right; we can't distinguish best from second best outcomes that are separated by frequencies of 10 or 20%. We need well designed and well controlled studies to identify these differences, and then we need to pay attention to them. But this can't happen if we defer to authority, trust intuition over published best practices, or remain quiet for fear of treading on a colleague's turf. 

About 5000 people are killed by MSSA infections each year in the US, and it is likely that the majority received suboptimal antibiotics. So it is safe to say that the behavior described by Charani et al are responsible for up to a thousand excess deaths per - for just one bad bug. That seems a pretty high price to pay to manage doctor's egos. 

No comments:

Post a Comment