Date of Award

January 2013

Document Type

Open Access Thesis

Degree Name

Master of Public Health (MPH)


School of Public Health

First Advisor

Haiqun Lin


In the effort of comparing health care quality across hospitals and profiling hospitals, the Centers for Medicare and Medicaid Services (CMS) utilize risk-standardized outcome in which the hospital-level outcome is standarized adjusting for patient-covariates using hierarchical model for patients outcome clustered within hospital. Even though standardized measure is useful for comparing hospitals, we believe the analysis of observed hospital-level outcome without standarization can reveal additional information at the hospital-level. The implication of using both types of mortality rates was studied in the context of analyzing 30-day mortality trend of Acute Myocardial Infarction (AMI) patients. Therefore the objective of this study was 1) describe trend in 30-day all cause mortality rates of AMI patients, 2) identify groups of hospitals with distinct trends and determine patient and hospital characteristics associated with group membership, and 3) examine how the risk-standardization approach affects the trajectory shape.

During 2005-2010, the 30-day mortality trend of AMI patients showed a decreasing trend. The median observed mortality decreased by 4.3% from 18.8% to 14.5%, but the between-hospital variation remained unchanged. Five distinct groups of hospitals were identified based on their patterns of mortality rate over time. Trajectories using observed rates showed varying trend and level of mortality, whereas all trajectories of risk-standardization rates show decreasing trend that is rather smooth with different, levels and slope of mortality slightly differed. When using observed rates, 9 patient characteristics were associated with group membership including age, history of heart failure, chronic atherosclerosis, valvular heart disease, hypertension, pneumonia, functional disability, metastatic cancer, and chronic liver disease. For hospital characteristics, cumulative AMI volume was the only factor significantly associated (p < 0.0001). When using risk-standardized rates, 8 patient characteristics associated with group membership including age, history of heart failure, history of AMI, chronic atherosclerosis, valvular heart disease, COPD, and peripheral vascular disease. For hospital characteristics, four covariates including cumulative AMI volume, urban/rural classification, proportion of Medicaid patients, and region, were significantly associated (p < 0.0001). Lastly, we found that the change in trajectory shape between observed and risk-standardized rates was mainly driven by hospital covariates, and not patient covariates.


This is an Open Access Thesis.