Author + information
- Matthew W. Sherwood, MD, MHS∗ ( and )
- Eric D. Peterson, MD, MPH
- ↵∗Duke University Medical Center, Division of Cardiovascular Medicine, Duke Hospital Room 7400, 2301 Erwin Road, Durham, North Carolina 27710
We appreciate the perspective of the comments by Drs. Miner and Nield (1) regarding our study (2). As noted by Drs. Miner and Nield (1), there are many limitations to any form of observational risk-adjusted outcomes comparisons. Certainly cardiac arrest or shock patients have a gradient of risk, and certainly providers may have a gradient in who they take to the lab or how they classify and report “high risk.” In individual cases, a provider could overcall shock in a lower-risk case and thereby have better observed results than expected. However, the American College of Cardiology–NCDR (National Cardiovascular Data Registry) risk models were developed from real-world data. As such, provider-related factors would have already been incorporated into the models. Thus, such variation in community practice is unlikely to explain why, in aggregate, providers who take on more high-risk cases do better. More importantly, in our analyses of the “concentrated risk year,” we used the individual providers themselves as their own control group. We found in such high-risk scenarios, providers’ “risk-adjusted” outcome performance was as good or better in high-risk cases than when the provider faced normal-risk or low-risk groups. So, we believe our paper provides compelling evidence that, in aggregate, the NCDR percutaneous coronary intervention risk models adequately assess and compensate providers for taking high-risk cases to the lab.
However, Miner and Nield (1) also raise an important point regarding whether or not public reporting itself is harmful or helpful. To be clear, our paper should not be seen as an endorsement of public reporting, and we agree the assessment of the total impact of public reporting is complex. On the one hand, public reporting does provide consumers with information on provider outcomes as well as give providers an incentive to monitor and hopefully improve their procedural outcomes. Although there is much debate whether consumer choice is improved via public reporting, there has been consistent evidence supporting the value of performance measurement and subsequent provider-led quality improvement, including door-to-balloon times, as well as with the outcomes of acute myocardial infarction, heart failure (3), and stroke (4). On the other hand, public outcome reporting could make certain providers “gun shy” and unwilling to take high-risk cases to the lab, even in situations where revascularization may be beneficial (such as ST-segment elevation myocardial infarction or shock). Previous studies have indicated that states with public reporting use PCI less and perhaps have worse outcomes than do states without (5). However, these studies were the exact motivation for our paper. Risk-averse clinician behavior likely represents the provider’s fear that taking on high-risk cases will “hurt” their performance ratings relative to peers. Our data demonstrate that such fears appear unfounded.
In conclusion, although one can debate the impacts of public reporting, our study should be interpreted to say that if it is undertaken, current modeling methods are generally adequate to capture and adjust for case mix and risk and thereby avoid penalizing clinicians who take on high-risk patients. We hope such information encourages providers to think more about the outcomes of their high-risk patients than about the impact of these on their procedural report card results.
- American College of Cardiology Foundation
- Miner S.,
- Nield L.
- Sherwood M.W.,
- Brennan J.M.,
- Ho K.K.,
- et al.
- Fonarow G.C.,
- Stough W.G.,
- Abraham W.T.,
- et al.