Wednesday, August 02, 2006

Real medicine's "ill structured problems" versus guideline's well structured ones

I have been thinking lately about the type of medical problems that physicians face. Mathematicians and cognitive scientists talk about well-structured problems and ill-structured problems.

It seems that much of the challenge of patient care falls under the heading of ill-structured problems (ISP). Well structured problems (WSP) are those for which there is a known algorithm.

ISPs have these characteristics:
1.inadequate information form the outset
2.lack of defining guidelines to evaluate the problem
3.mutability of the problem-things changes as you go alone
4.lack of assurance that the problem has been solved

These are complicated problems without a clear cut solution and for which there may not be one right answer. There is no back-of-the-answer to compare with your analysis.

A recent post by Aggravated docsurg gives some great examples of ISPs that a general surgeon faced.Internists have equally demanding cases as well in addition to the simpler, quasi-no brainers.

An ill -structured problem is , by definition, one for which there is no algorithm. Much of the formal education I received in college physics and chemistry and calculus involved the mastery of WSPs of the following type. If a rock drops into a well and the splash is heard 4 seconds later how deep is the well? Physicians do not seem to do much of that type of thinking in their offices.

The folks who paint medicine as mostly a series of WSPs solable by algorithms and auditable for quality and reimbursable of the basis of obedience to those guidelines are either ignorant of or choose to ignore the reality of just how complex and ill structured the issues are that physicians face.

Clinical decisions in these ISPs will require all the knowledge, expertise and judgment the physician can bring to bear factoring in the values and wishes of the patient to try and find the particular "clinical truth" for the circumstances at hand. The quality gurus have no generic algorithm for that process.

1 comment:

Sid Schwab said...

It's a problem that most physicians, I assume, understand at some level but don't typically address. We deal with data that are often incomplete and innately innaccurate (how reliable, really, are manometrics, for example, measured via an endoscope?) Is "gut instinct" the x-factor in such cases? Is it that we in fact know way less than we think we know about human pathology? As a surgeon, when faced with a sick person in whom the diagnosis was unclear, I often (not to my regret, that I can recall) reminded myself that the immediate question was a relatively simple one: to operate, or not. After that, things had a way of clarifying.