Jennifer Schwartz1, Yongfei Wang2, Li Qin2, Lee H Schwamm2, Gregg C Fonarow2, Nicole Cormier2, Karen Dorsey2, Robert L McNamara2, Lisa G Suter2, Harlan M Krumholz2, Susannah M Bernheim2. 1. From the Center for Outcomes Research and Evaluation, Yale-New Haven Hospital, New Haven, CT (J.S., Y.W., L.Q., N.C., K.D., R.L.M., L.G.S., H.M.K.); Section of Cardiovascular Medicine, Department of Internal Medicine (J.S., Y.W., R.L.M., H.M.K.), Section of Rheumatology, Department of Medicine (L.G.S.), and Section of General Internal Medicine, Department of Internal Medicine (S.M.B.), Yale University School of Medicine, New Haven, CT; Department of Neurology, Massachusetts General Hospital, Harvard Medical School, Boston (L.H.S.); Division of Cardiology, Department of Medicine, Geffen School of Medicine at UCLA (G.C.F.); VA Connecticut Healthcare System, West Haven (L.G.S.); and Department of Health Policy and Management, Yale University School of Public Health, New Haven, CT (H.M.K.). Jennifer.schwartz@yale.edu. 2. From the Center for Outcomes Research and Evaluation, Yale-New Haven Hospital, New Haven, CT (J.S., Y.W., L.Q., N.C., K.D., R.L.M., L.G.S., H.M.K.); Section of Cardiovascular Medicine, Department of Internal Medicine (J.S., Y.W., R.L.M., H.M.K.), Section of Rheumatology, Department of Medicine (L.G.S.), and Section of General Internal Medicine, Department of Internal Medicine (S.M.B.), Yale University School of Medicine, New Haven, CT; Department of Neurology, Massachusetts General Hospital, Harvard Medical School, Boston (L.H.S.); Division of Cardiology, Department of Medicine, Geffen School of Medicine at UCLA (G.C.F.); VA Connecticut Healthcare System, West Haven (L.G.S.); and Department of Health Policy and Management, Yale University School of Public Health, New Haven, CT (H.M.K.).
Abstract
BACKGROUND AND PURPOSE: The Centers for Medicare & Medicaid Services publicly reports a hospital-level stroke mortality measure that lacks stroke severity risk adjustment. Our objective was to describe novel measures of stroke mortality suitable for public reporting that incorporate stroke severity into risk adjustment. METHODS: We linked data from the American Heart Association/American Stroke Association Get With The Guidelines-Stroke registry with Medicare fee-for-service claims data to develop the measures. We used logistic regression for variable selection in risk model development. We developed 3 risk-standardized mortality models for patients with acute ischemic stroke, all of which include the National Institutes of Health Stroke Scale score: one that includes other risk variables derived only from claims data (claims model); one that includes other risk variables derived from claims and clinical variables that could be obtained from electronic health record data (hybrid model); and one that includes other risk variables that could be derived only from electronic health record data (electronic health record model). RESULTS: The cohort used to develop and validate the risk models consisted of 188 975 hospital admissions at 1511 hospitals. The claims, hybrid, and electronic health record risk models included 20, 21, and 9 risk-adjustment variables, respectively; the C statistics were 0.81, 0.82, and 0.79, respectively (as compared with the current publicly reported model C statistic of 0.75); the risk-standardized mortality rates ranged from 10.7% to 19.0%, 10.7% to 19.1%, and 10.8% to 20.3%, respectively; the median risk-standardized mortality rate was 14.5% for all measures; and the odds of mortality for a high-mortality hospital (+1 SD) were 1.51, 1.52, and 1.52 times those for a low-mortality hospital (-1 SD), respectively. CONCLUSIONS: We developed 3 quality measures that demonstrate better discrimination than the Centers for Medicare & Medicaid Services' existing stroke mortality measure, adjust for stroke severity, and could be implemented in a variety of settings.
BACKGROUND AND PURPOSE: The Centers for Medicare & Medicaid Services publicly reports a hospital-level stroke mortality measure that lacks stroke severity risk adjustment. Our objective was to describe novel measures of stroke mortality suitable for public reporting that incorporate stroke severity into risk adjustment. METHODS: We linked data from the American Heart Association/American Stroke Association Get With The Guidelines-Stroke registry with Medicare fee-for-service claims data to develop the measures. We used logistic regression for variable selection in risk model development. We developed 3 risk-standardized mortality models for patients with acute ischemic stroke, all of which include the National Institutes of Health Stroke Scale score: one that includes other risk variables derived only from claims data (claims model); one that includes other risk variables derived from claims and clinical variables that could be obtained from electronic health record data (hybrid model); and one that includes other risk variables that could be derived only from electronic health record data (electronic health record model). RESULTS: The cohort used to develop and validate the risk models consisted of 188 975 hospital admissions at 1511 hospitals. The claims, hybrid, and electronic health record risk models included 20, 21, and 9 risk-adjustment variables, respectively; the C statistics were 0.81, 0.82, and 0.79, respectively (as compared with the current publicly reported model C statistic of 0.75); the risk-standardized mortality rates ranged from 10.7% to 19.0%, 10.7% to 19.1%, and 10.8% to 20.3%, respectively; the median risk-standardized mortality rate was 14.5% for all measures; and the odds of mortality for a high-mortality hospital (+1 SD) were 1.51, 1.52, and 1.52 times those for a low-mortality hospital (-1 SD), respectively. CONCLUSIONS: We developed 3 quality measures that demonstrate better discrimination than the Centers for Medicare & Medicaid Services' existing stroke mortality measure, adjust for stroke severity, and could be implemented in a variety of settings.
Authors: Chloe E Hill; Chun Chieh Lin; James F Burke; Kevin A Kerber; Lesli E Skolarus; Gregory J Esper; Brandon Magliocco; Brian C Callaghan Journal: Neurology Date: 2019-01-23 Impact factor: 9.910
Authors: Amit Kumar; Indrakshi Roy; Pamela R Bosch; Corey R Fehnel; Nicholas Garnica; Jon Cook; Meghan Warren; Amol M Karmarkar Journal: J Gen Intern Med Date: 2021-10-26 Impact factor: 6.473