© Copyright 2007 Health Grades, Inc. All rights reserved.
Hospital Report CardsTM Methodology 3
May not be reprinted or reproduced without permission from Health Grades, Inc.
contains 427.31 Atrial Fibrillation, that code is considered a risk if it occurs by itself and a complication if
there is a corresponding 997.1 Cardiac Complications, NEC code also present in the patient record.
Some diagnosis codes were merged together (e.g., primary and secondary pulmonary hypertension) to
minimize the impact of coding variations.
Outcomes were binary, with documented major/minor complications either present or not, and patients
recorded as either alive or expired. In cohorts where the quality measure is major complications, mortality is
considered a complication. See Appendix B for a list of complications included in the quality measure "Major
Complications."
Risk-Adjustment Methodology
Fair and valid comparisons between hospital providers can be made only to the extent that the risk-
adjustment methodology considers important differences in patient demographic and clinical characteristics.
The risk-adjustment methodology used by HealthGrades defines risk factors as those clinical and
demographic variables that influence patient outcomes in significant and systematic ways. Risk factors may
include age, gender, specific procedure performed, and comorbid conditions such as hypertension, chronic
renal failure, heart failure, and diabetes. The methodology is disease-specific and outcome-specific. This
means that individual risk models are constructed and tailored for each clinical condition or procedure, and
also for each outcome.
Developing the HealthGrades ratings involved four steps for each cohort (e.g., coronary bypass surgery)
and quality measure (e.g., inhospital mortality or complications).
·
First, the predicted value (e.g., predicted mortality or complications) was obtained using logistic
regression models discussed in the next section.
·
Second, the predicted value was compared with the actual, or observed, value (e.g., actual
mortality or complications). Only hospitals with at least 30 cases across three years of data and at
least five cases in the most current year were included.
·
Third, a test was conducted to determine whether the difference between the predicted and actual
values was statistically significant. This test was performed to make sure that differences were very
unlikely to be caused by chance alone.
·
Fourth, a star rating was assigned based upon the outcome of the statistical test.
Statistical Models
Unique statistical models were developed for each patient cohort and each outcome using logistic
regression.
Comorbid diagnoses (e.g., hypertension, chronic renal failure, anemia, diabetes), demographic
characteristics (e.g., age and gender), and specific procedures (where clinically relevant) were classified as
potential risk factors. We used logistic regression to determine which of these were actually risk factors and
to what extent they were correlated with the quality measure (e.g., mortality). A risk factor stayed in the
model if it had an odds ratio greater than one (except that clinically relevant procedures, cohort defining
principal diagnoses, and some protective factors as documented in the medical literature were allowed to
have an odds ratio less than one) and was also statistically significant (p<0.05).
Exceptions to this rule should be noted for the cardiac service line (specifically CABG, PCI and AMI) where
cardiogenic shock, anoxic brain injury, and cardiac arrest were excluded from the final model as risk factors.
In a few cases, risk factors that have odds ratios less than one are included in the models if the risk has
been previously accepted in the medical literature. Complications were not counted as risk factors as they
were considered a result of care received during the admission. See Appendix C for a list of the top five risk
factors for each procedure or diagnosis.
