Quality improvement is a misnomer

Bravo, now can you try to get the medical profession to abandon “reimbursement” to describe a fee or payment!

There’s more than a bit of a straw-man argument here. I have yet to meet a serious “quality improvement” researcher who doesn’t readily acknowledge the limitations you outline.

Most of us are exceedingly well-versed in logical argument. I would submit that what we need is agreement on terminology and an acknowledgment that we already agree on most of the issues you discuss. I once had a debate teacher who required me to state my opponent’s position–to my opponent’s satisfaction–before launching my argument. This is good advice.

You define “quality” as “better outcomes.” Fine. That is one definition, but there are others. I’m sure you’ve read Donobedian’s work, so I don’t need to review his structure-process-outcome framework for you. Donobedian doesn’t demand that improvement on a process measure always leads to better outcomes. His argument is population-level: process measure performance must be positively correlated with better health (i.e., on average produce better outcomes, however you define these). We all like outcome measures, but these are rarely available…especially in the world of chronic disease, where processes may produce measurable outcomes for decades.

When a researcher in QI studies “quality measures,” these can be anywhere in the structure-process-outcome chain. It is incumbent on the researcher to specify the level of measurement and to acknowledge that all measures (even measures of outcomes) suffer from serious limitations. There is no way around this. The best we can do is to continually test the relationships between “quality measures” at all levels and the final results we desire. Waiting for perfect measures is unrealistic.

I have to ask: if you don’t think guidelines should be followed in the populations for whom they are designed (which is how we identify the denominator populations for Bernoulli-distributed process measures), with whom are you really arguing?

I suppose the alternative is to reject empiricism entirely and adopt the framework of Cartesian doubt. After all, even outcomes-based studies don’t measure every possible outcome, and no outcomes measure is without flaws (Death? Death over what time period? Before death, are you absolutely sure you’ve measured quality of life correctly? How do you know you haven’t missed a death or two? In light of the “engage with grace” campaign, how do you know what percentage of the deaths are actually bad outcomes?)

So let’s not throw out the entire body of empirical work on QI. Trying to deduce from first principles of logic whether, say, lowering average A1c will improve “ultimate quality” will not be a productive use of time. I don’t think anybody wants to live in a world where all physicians are assumed (based on first principles, or maybe a one-time licensing exam) to deliver care of perfect quality.

An emotional reluctance to accept quality measurement is understandable. Nobody likes being measured, and the initial response of a prideful professional is to attack the measures by holding them to impossible standards…or to overinterpret them, suggesting that the measures imply an expectation of perfect care. We see the same thing every time we measure a new aspect of care, or a new subspecialty has its care examined. First, it’s “No information is better than imperfect information.” Then there’s anger, bargaining, …and I think you know the rest.