Currently listening to RANGE: Why Generalists Triumph in a Specialized World by David Epstein. In the first chapter he discusses the differences between wicked problems and kind problems. For example, chess is a kind problem. It has specified rules and clear outcomes. Because it is a kind problem, AI can successfully play the game.
Wicked problems do not have rules or even a single known solution. One cannot always determine outcomes because we have many variables and many dimensions to the outcomes. Internists face wicked problems regularly. Many of us chose internal medicine because we love the challenge of these wicked problems.
Diagnosis is a classic wicked problem. We deal with a large variety of inputs and are not restricted to one clear answer. Some patients have more than one diagnosis explaining their symptoms.
Often performance measures, experts and guidelines treat testing or management decisions as if they were kind problems. But clinicians quickly understand that each decision has nuances based upon the patient’s complexity. Even Evidence Based Medicine can suffer because the evidence base does not really fit the patient we are considering.
And that is the problem we confront. We want to measure quality, yet we are facing a wicked game. With some literary stretching, consider that once famous song “Wicked Game” by Chris Isaak – “What a wicked game you played to make me feel this way”. While he is talking about love, I will take the leap to say that trying to measure quality is actually a wicked game. We have too many variables for which we cannot account.
We can measure the kind (or tame) problems in medicine – central line infections, incorrect medications delivered to patients, wrong site surgery, etc. But what should we do with measures that we try to apply to complex patients.
We would love to measure diagnostic error, but no one has yet solved this wicked problem – and I suspect there is no solution, because diagnosis is so complex process. We rarely have a gold standard diagnosis. Sometimes we cannot make a diagnosis at the initial presentation. At what point do we declare a diagnostic error?
I would love your thoughts on the wicked problem concept. We are using kind problem tools to solve wicked problems. This will often not work and create frustration. We therefore have measures that can drive the wrong care.
