Orthodoxies rust easily. Even a benign truth — like the desirability of evidence-based medicine or health information technology — will decay if it is repeated too often or invested with an aura of magical infallibility. So the world needs Jerome Groopman.

An M.D. and Harvard professor who writes for the New Yorker, Groopman is creating a widening public audience for a problem that internists have been worrying themselves about for decades: an apparent decline in physicians’ clinical skills, driven at least in part by increasing dependency on high-tech diagnostic tools and financial incentives to see more and more patients. Much of the evidence comes from studies based on autopsies that reveal morbidities not detected (or treated) in living patients. One such study found major diagnostic errors in 32 percent of patients who died during stays in intensive care. Estimates of the overall rate of diagnostic error fall in the 10-15 percent range.

Diagnostic error visits multiple implications on the quality-improvement movement. Most emerging practice in the domains of evidence-based medicine (EBM), performance measurement, and performance incentives focuses on treatment and presumes correct diagnosis as a point of departure. If such assumptions are incorrect 15 percent of the time, those schemes will need to be modified appropriately. Groopman’s analysis of the likely sources of diagnostic error also prompts him to be skeptical about transformational expectations for health information technology and EBM.

But for policy jocks who are already up to their eyebrows in EBM, IT, and P4P, a more interesting dimension of the literature on diagnostic error is the window it opens on physicians’ behavior, patient-physician interactions, and cognitive psychology. Readily perceived problems such as rushed patient encounters and imperious, overconfident practitioners are merely the most obvious symptoms of what can go wrong. The Institute of Medicine’s canonical analysis of medical error aptly notes that a culture of medical safety needs to begin with an environmental change that makes it professionally acceptable for doctors to admit uncertainty and to be candid about mistakes.

But the subtlety of the cognitive biases associated with diagnostic error in Groopman’s work and other studies suggests a daunting re-education challenge. Greater reciprocity in doctor-patient relations is a part of the answer that is consistent with many IT champions’ vision of patient-centered care. Other problems are not nearly as well recognized. Mark Graber and colleagues, cited above, found “premature closure,” or the inclination of physicians to settle on a diagnosis before they have gathered enough information or considered all the alternatives, to be the single most common source of cognitive failure. But there are a whole slew of other mental and behavioral weaknesses that may enter into the clinical encounter and drive it into a ditch.

The bottom line, according to Groopman, is that doctors often don’t ask the right questions and don’t listen carefully enough when the patient answers. Expect to hear more about this. The blogs are already buzzing.