Developing a Combined Measure of Rehospitalization and Mortality to Asses Hospital Quality

Wong ES, Liu C, Rinne S, Perkins M, Hebert PL. Developing a Combined Measure of Rehospitalization and Mortality to Asses Hospital Quality. Poster session presented at: AcademyHealth Annual Research Meeting; 2014 Jun 10; San Diego, CA.

Related HSR&D Project(s)

• IIR 09-354 – Organizational Factors Related to Hospital Readmissions

Developing a Combined Measure of Readmission and Survival to Assess Hospital Quality Edwin S. Wong, PhD1,2; Chuan-Fen Liu, PhD MPH1,2; Seppo T. Rinne, MD, PhD1,3; Mark Perkins, PharmD1; Paul L. Hebert, PhD1,2 1Center of Innovation for Veteran-Centered and Value-Driven Care, VA Puget Sound Health Care System, Seattle, WA; 2Department of Health Services, University of Washington, Seattle, WA; 3Department of Medicine, University of Washington, Seattle, WA Objective: Assessment of hospital quality often uses indirect measures such as risk-adjusted mortality and readmission because quality of care is difficult to measure directly and comprehensively. To the extent that both mortality and readmission reflect the underlying hospital quality of care, both measures should be of interest to patients and administrators. However, the Centers for Medicare and Medicaid Services currently uses only readmission rates to assess financial penalties for poor hospital performance through the Hospital Readmissions Reduction Program. Patients who die within-hospital or within 30 days of discharge are excluded from these models, so mortality or survival does not inform the measure of hospital quality. In this study, we develop a new utility-based metric for assessment of hospital quality by combining readmission and survival outcomes and apply this metric to comparing hospitals in the Veterans Affairs Healthcare System (VA). Study Design: Using VA and Medicare administrative data, we identified all patients hospitalized for congestive heart failure (CHF) in VA hospitals and tracked patients for 30 days following their discharge. The 30-days following discharge were partitioned into three mutually exclusive categories: days hospitalized, days deceased and hospital-free days alive. For each patient, we measured utility as a weighted average of days hospitalized and hospital-free days alive. Prior studies suggest utility weights of 0.6 and 0.4, respectively. We risk-adjusted days hospitalized and hospital-free days alive jointly using a fractional probit model. We then calculated VA hospital-specific utility scores as the average utility score of all patients discharged from a given hospital. We compared VA hospital rankings based on the combined utility metric with rankings based on the 30-day risk-adjusted all-cause readmission rate. Population Studied: 16,929 patients hospitalized in 127 VA hospitals for CHF in fiscal year 2007. Principal Findings: Over the 30-days following discharge, the average number of hospital-free alive, hospitalized and deceased days per patient was 27.2, 2.1 and 0.7 respectively. Rates of 30-day all-cause readmission and mortality were 25.2% and 4.7%, respectively. The mean combined metric score after risk adjustment was 17.10 (SD = 0.180) and ranged from 16.47 to 17.47 across hospitals. The correlation between the combined-metric ranking and the readmission-only ranking was 0.585. However, a substantial number of hospitals ranked highly on one metric, but ranked poorly on the other. The rank differed by more than 50 positions for 14 hospitals. Four hospitals ranked in the top 25 on the readmission-only metric were ranked among the bottom 25 on the combined metric. The difference in rankings across metrics ranged from -103 to 81. Conclusions: VA relative hospital rankings based on our combined utility metric differed substantially for some hospitals compared to rankings based on the 30-day risk-adjusted readmission measure. Implications for Policy, Delivery or Practice: The differences in hospital rankings identified suggest some hospitals are potentially misclassified as high or low performers. Assessment of hospital performance using outcomes following hospitalization should consider mortality or survival in conjunction with readmission.