There are plenty of good reasons why thoughtful physicians are often unhappy with the current approach to measuring the quality of care they provide. Some, of course, object to the whole notion of quality measurement, but I believe they are in a shrinking minority clinging to an anachronistic mental model in which each physician defines for himself what constitutes high quality care. I have addressed this previously. But even those, who like me, believe it is essential (and possible) to measure quality, can point to legitimate shortcomings in the way it is done.
Among these shortcomings is the imperfect process by which individual physicians’ “results” are “adjusted” to account for differences in the patients they care for. In the simplest case, when the quality of care is judged by looking at patient outcomes, this risk-adjustment is meant to reflect the fact that clinical outcomes reflect both the baseline characteristics of the patients being treated and the treatment they get. For example, if one were to use in-hospital mortality rates to assess the quality of care for acute myocardial infarction, it would be essential to know “how sick” the patients, on average, were on presentation. A 50 year-old man with a small inferior wall MI is likely to live even in the absence of good care (or any care for that matter), whereas a 90 year-old woman with cardiogenic shock from an anterior wall MI is likely to die even with state-of-the art care. Any attempt to assess the quality of care for a population of MI patients must take this into consideration.
There is a more subtle way in which patient characteristics play into quality measurement schemes, even when the measurements are about processes of care instead of patient outcomes. In this construct, providers are assessed by how often patients eligible for some service or intervention actually get it. Did the patients with diabetes get fundoscopic exams? Did the women in their 50s get mammograms? Are the patients with coronary heart disease all on aspirin? Here it is easy to prospectively define exclusion criteria, which are meant to mimic reasonable clinical decision-making, and shield the provider from a “grade” that really reflects unmeasured differences in patient populations. For example, it would not be reasonable (or be indicative of high quality care!) to give aspirin to a patient with an aspirin allergy, so patients with aspirin allergy are excluded from the denominator, and the provider is not judged harshly for a “failure” to prescribe it. So far, so good. This gets a whole lot trickier, however, when trying to figure out how to handle instances where care is recommended, but not done. What happens if the patient is advised that she should have that mammogram, but doesn’t get it?
Assuming for a moment that it is possible to accurately distinguish between a failure (on the part of the physician) to recommend and a failure (on the part of the patient) to adhere to the physician’s recommendation, who is responsible for the latter? On the one hand, it seems pretty straightforward: the “right” care was recommended, and the patient failed to take good advice, so this can’t possibly be used to judge the care the doctor provided, right? Well, maybe so, but maybe the patient didn’t take the recommended course of action because the doctor failed to explain it in a way the patient understood, or because the patient couldn’t access the recommended service, or because the patient experienced a side-effect that the physician did not elicit. In these instances, accountability is shared by patients and their physicians.
This idea of shared accountability was recently addressed by a joint committee of the American College of Cardiology and the American Heart Association. These organizations have been collaborating for years to produce clinical practice guidelines. The guidelines, in turn, have been used as the basis for a wide range of performance measures, which have been used to assess the quality of cardiovascular care. The published “statement” is a thoughtful consideration of how to balance the interdependent responsibilities of clinicians, patients and systems of care. The salient figure from the paper is reproduced below, but I urge you to read the whole thing. Once you do, let me know what you think.