Lena M Chen1, E John Orav, Arnold M Epstein. 1. Veterans Affairs Center for Clinical Management Research, VA Ann Arbor Healthcare System, Ann Arbor, MI, USA. lenac@umich.edu
Abstract
BACKGROUND: Since the advent of public reporting on risk-adjusted mortality for coronary artery bypass graft surgery, public reporting on outcomes has expanded to include a variety of dissimilar conditions and procedures. We have little evidence to support such broad-based efforts. METHODS AND RESULTS: We examined the quality performance of 351 cardiologists at 48 hospitals in New York State, using publicly reported risk-adjusted mortality rates (RAMRs) for nonemergent percutaneous coronary interventions between 1998 and 2007. In the year after report release, we examined the following: (1) average RAMR for hospitals, (2) change in market share for hospitals and cardiologists, and (3) proportion of physicians leaving practice. We found that patients who picked a hospital that performed significantly better than expected in prior years had lower RAMRs (0.47, 0.61, and 0.72 for patients choosing hospitals whose prior reports were better than, as, and worse than expected; P=0.02). However, choosing a hospital in the top quartile (or decile) of performance in prior years did not decrease a patient's chance of dying (P=0.29, or P=0.27). Performance ranking was not associated with a change in market share for hospitals or for physicians, or with leaving practice (all P>0.05). CONCLUSIONS: Public reporting on nonemergent percutaneous coronary interventions in New York State identifies very high and low performers but provides insufficient information to differentiate between most hospitals. It appears to have had no effect on market share or physicians' decisions to leave practice. The utility of public reporting on RAMRs may differ for different conditions and procedures.
BACKGROUND: Since the advent of public reporting on risk-adjusted mortality for coronary artery bypass graft surgery, public reporting on outcomes has expanded to include a variety of dissimilar conditions and procedures. We have little evidence to support such broad-based efforts. METHODS AND RESULTS: We examined the quality performance of 351 cardiologists at 48 hospitals in New York State, using publicly reported risk-adjusted mortality rates (RAMRs) for nonemergent percutaneous coronary interventions between 1998 and 2007. In the year after report release, we examined the following: (1) average RAMR for hospitals, (2) change in market share for hospitals and cardiologists, and (3) proportion of physicians leaving practice. We found that patients who picked a hospital that performed significantly better than expected in prior years had lower RAMRs (0.47, 0.61, and 0.72 for patients choosing hospitals whose prior reports were better than, as, and worse than expected; P=0.02). However, choosing a hospital in the top quartile (or decile) of performance in prior years did not decrease a patient's chance of dying (P=0.29, or P=0.27). Performance ranking was not associated with a change in market share for hospitals or for physicians, or with leaving practice (all P>0.05). CONCLUSIONS: Public reporting on nonemergent percutaneous coronary interventions in New York State identifies very high and low performers but provides insufficient information to differentiate between most hospitals. It appears to have had no effect on market share or physicians' decisions to leave practice. The utility of public reporting on RAMRs may differ for different conditions and procedures.