Author + information
- Frederick A. Masoudi, MD, MSPH* (, )
- John C. Messenger, MD,
- W. Douglas Weaver, MD and
- Ralph G. Brindis, MD, MPH
- ↵*Denver Health Medical Center and University of Colorado, Division of Cardiology, MC0960, Denver Health Medical Center, 777 Bannock Street, Denver, Colorado 80204
As contributors to the NCDR (National Cardiovascular Data Registry), we read the letter by Ellis and colleagues with interest. The NCDR has long been dedicated to the measurement and improvement in care quality. Undoubtedly, these goals are best promoted when the measures used to characterize quality are as valid as possible.
We agree regarding the distinction between measures used for the purposes of quality improvement and those intended for the purposes of external accountability. Indeed, the American College of Cardiology/American Heart Association (ACC/AHA) Performance Measures Task Force methodology explicitly acknowledges this dichotomy (1). Measures used for the purposes of accountability must rise to a particularly high standard with respect to validity, burden of data collection, and susceptibility to “gaming.”
The specific issue Ellis and colleagues raise is the exclusion in the reperfusion measure for patient-centered reasons for delaying therapy in the NCDR CathPCI Registry. Clinically appropriate reasons for delays in reperfusion therapy are numerous. Indeed, enumerating them completely is impractical. The NCDR CathPCI Registry reperfusion measures allow for exclusions for patient-centered reasons for delay (e.g., the need for a decision-altering diagnostic test prior to possible primary percutaneous coronary intervention). Although the measures may include examples, they do not include specific lists of these reasons. Such exclusions were integrated into the measure to acknowledge the fact that high-quality clinicians providing the best care will, on occasion, face situations where their delivery of reperfusion therapy is delayed for clinically appropriate reasons. As Ellis and colleagues point out, the flexibility intrinsic to this exclusion may create opportunities for gaming.
Although we believe that the high standards of medical professionals protect the integrity of the measures to some extent, we are not so naive to assume that professional integrity alone will eliminate gaming. Unfortunately, however, addressing this issue by removing the exclusion as proposed by Ellis and colleagues raises substantial problems of its own by undermining the clinical face validity of the measure. The absence of such exclusions creates other compelling arguments—namely, that centers that care for particularly complex patients, where clinically reasonable delays are more common—are disproportionately penalized. Indeed, before the Department of Health and Human Services' Centers for Medicare and Medicaid Services/Joint Commission measure incorporated this exclusion, such complaints were among the most common causes for objection to the reperfusion measures (Jo DeBuhr, Colorado Foundation for Medical Care, personal communication, July 2010).
This dilemma, among the many complex issues surrounding measuring reperfusion quality, was addressed explicitly by an ACC/AHA Writing Group comprised of experienced clinicians and experts in performance measurement (2). This writing group concluded that this exclusion is important, despite its limitations. This opinion is reflected in the current ACC/AHA performance measures for acute myocardial infarction (3).
Further, the Writing Group recommended: 1) surveillance for the proportion of cases where exclusions are noted, including the distribution of the exclusions by institution; and 2) audit of the clinical appropriateness of exclusions both in a targeted manner (i.e., among institutions with the highest numbers of excluded cases) as well as randomly.
To this point, NCDR metrics have been used predominantly for quality improvement. Although some of the metrics reported to registry participants are not intended for accountability purposes, others—including the time-to–primary percutaneous coronary intervention metric in question—might reasonably be viewed as useful in this regard. As this occurs, we agree with Ellis and colleagues that greater scrutiny of exclusions, consistent with the recommendations by the ACC/AHA Writing Group is warranted.
- American College of Cardiology Foundation
- Bonow R.O.,
- Masoudi F.A.,
- Rumsfeld J.S.,
- et al.
- Masoudi F.A.,
- Bonow R.O.,
- Brindis R.G.,
- et al.
- Krumholz H.M.,
- Anderson J.L.,
- Bachelder B.L.,
- et al.