Accepted quality measures don't measure quality

Today's issue of JAMA has a research article (press release here) that sounds a little dry, but is actually pretty important. It examines the ability of five accepted quality-of-care measures for heart failure to predict rehospitalization and mortality. The five are process measures that are used by the Joint Commission on Accreditation of Healthcare Organizations (JCAHO) to accredit hospitals.

The study, by UCLA researchers, found that the five measures (provision of discharge instructions, evaluation of left ventricular systolic function, administration of ACE/ARB for left ventricular systolic dysfunction, smoking cessation advice/counseling, and anticoagulant at discharge for patients with atrial fibrillation) were by and large not strongly or statistically signficantly predictive of either 60- to 90-day mortality or of rehospitalization rate. Another measure not in standard use, though - administration of beta-blocker - was strongly associated with improved outcomes.

Kudos to JAMA for publishing this article. The hospitals, etc., who are investing a lot of money measuring quality are going to be upset, throw up their hands, and say, "What are we doing?" However, there is the saving grace of the one non-measure - beta-blockers - that was strongly related to the outcomes. That means that there is, in fact, the potential to measure actual quality via process measures; we just need to pick the right ones. Everyone, now, should be primed to demand better validation of quality measures, and, at the very least, transparency in the validation process which has to this point been quite obscure.