Date of Award

January 2014

Document Type


Degree Name

Medical Doctor (MD)



First Advisor

Jeptha Curtis

Subject Area(s)

Medicine, Public policy



Philip W. Chui, Craig S. Parzynski and Jeptha P. Curtis. Center for Outcomes Research and Evaluation, Yale University, New Haven, CT.

Over the past decade, there has been an explosion of interest in measuring and reporting the quality of care delivered to patients undergoing percutaneous coronary intervention (PCI), with the newest performance metrics just released in December 2013. One aspect of quality measurement that remains poorly understood is the nature of the relationship between individual PCI process measures as well as between process and outcomes measures such as risk-standardized 30-day mortality (RSMR) and readmission rates (RSRR). While process and outcomes measures represent potentially important metrics of hospital quality, no study has examined whether these two are measuring distinct or overlapping domains of hospital quality or whether there is an association between the newly issued and previously existing set of process measures.

We performed a cross-sectional analysis using data from the National Cardiovascular Data Registry's (NCDR) CathPCI registry. We identified 1,219,544 patients across 1,331 catheterization centers that had a procedure from January 1, 2010 to December 30, 2011. We utilized generalized linear modeling to estimate hospital performance on individual and composite process measures and generated pair-wide correlations between each set of process metrics. We also performed correlation analyses to calculate the association between hospital performance on process metrics and hospital performance on outcome metrics. We found strong correlations between medication-specific process measures (p < 0.01), and the overall composite measure was significantly correlated with most other process measures, particularly referral to cardiac rehab (p < 0.01). Hospital performance on the new emerging process measures was lower than on existing ones, and there was little correlation in hospital performance between these two sets. The composite measure was correlated with hospital-specific 30-day RSRR for all patients and RSMR for NSTEMI patients (p value < 0.05), but the association was modest, explaining at most 6.6% of the variation in hospital-level outcome metrics. These results suggest that hospital-level performance on both individual and composite PCI process measures explain only a small percentage of the hospital variations in outcomes for PCI patients. Additional efforts are needed to better characterize how hospitals can utilize these two perhaps distinct markers of quality to improve hospital performance.