Monday, January 22, 2007

How People..I mean, Doctors Think

The New Yorker recently posted an essay about ED (emergency department) doctor cum experimental psychologist Dr. Pat Croskerry concerning the flawed heuristics (manner of approaching an investigation in this context) doctors bring into the doubt-riddled and often misled diagnosis process (please don't ever lie to your doctor about just how hard you work to dick yourself up). It's an easy read, telling stories of the atypical patients that are misdiagnosed before waxing theoretical, and I strongly suggest you read it.

The essay brings up three particular heuristics:
1. "Representativeness" a.k.a. expecting what's true most of the time to be true
2. "Availability" a.k.a. following trends in recently admitted patients
3. "Affective Error" a.k.a. liking your patient too much (WTF?)

Now, I don't approach the essay as a polemic, but it is critical in tone and, like many critiques, lacking in suggestions for remedial action. The first two heuristics are simply good logic, and warping the physician's approach to diagnosis in an attempt to prevent the percentage of errors it causes would undoubtedly be extremely expensive. The majority of patients that create a sense of "representativeness" or "availability" would undergo additional testing to rule out "unrepresentativeness" and, well, that'd be fucking awful considering the number of uninsured patients and universally accepted bloat of American healthcare.
It's a classic question: Do we punish the majority for the sake of the minority? As always, it becomes a matter of weighing advantages and disadvantages. Until better and cheaper diagnostic tools become available, error is inevitable in the attempt to run an efficient (and in this case that means life-saving) operation. I purposefully omit often-proffered solutions, such as colleague oversight, because they don't apply to the premise of the essay. In no way does this indemnify substandard doctoring (as is seen in the second two examples of the essay), but believing your doctor is malfeasant in expecting the norm absent of other risk factors can only raise the expected standard of care to a level the current technological and legal environment (think: $$) cannot support. You don't see the bill for a patient of House M.D.

I do think the technology is in the pipeline, so despair not, but that's another post.

Anyways, as always, I invite contrary opinions. I'm just beginning to wrestle with this issues, and I recognize the limitation of my understanding and the benefit of a little intellectual adversity. Also, Nelly Furtado gives me an ambiguous-ethnicity-beauty chubby.

Labels:

4 Comments:

At 12:30 PM, January 23, 2007, Blogger sdvknsdvkn said...

At no time in your incoherent ramblings did I feel entertained. I want more 80's music posts. Your veiwing public demands it.

 
At 8:44 AM, January 24, 2007, Blogger Sir Cody said...

Thank you for bringing up that stuff about House. I swear shows like that are a hypochondriacs wet dream because it gives them more reasons to think they're dying of some exotic disease when its just allergies. Also, does that affective error thing come into play a lot? They put that problem on Izzy a bunch in Grey's Anatomy. Oh man, I can feel my manhood slipping. Penis, weiner, boobs!

 
At 3:12 PM, January 24, 2007, Blogger Brian said...

heh heh, heh..he said 'chubby'

your intellect is lost on us.

 
At 9:58 AM, January 25, 2007, Anonymous Anonymous said...

Sorry I'm just getting around to this, Richard.

A generalization of the problem that Groopman presents:

Any attempt to analyze an unfathomably complex universe is severely limited by the finite amount of resources at our disposal (time, and the processing capacity of the mind, most importantly).

Heuristic methods, therefore, are not the problem but are evolved solutions. As you say, representativess and availability are simply good logic.

Still, any heuristic involves a certain amount of uncertainty. That the stakes of medicinal diagnoses are life and death is the reason that doctors must every second be aware of their cognitive biases - their instincts; a tall order for any human. I think Groopman was not criticizing the implementation of heuristics, only explaining their inherent limitations and pointing out the potential recipe disaster when said limitations are ignored.

As far as future technology is concerned, as an uber-conservative I am always wary of the "potential benefits" any new technology has to offer, especially with respect to the inevitable added costs (both economic and social). Also, any technology is always ultimately subservient to human interpretation and human nature, which will never change. But you are right, that is for another post.

All this makes me want to re-read Pinker. To the Batcave.

 

Post a Comment

<< Home