Hospital Report Cards TM Maternity Care and Women's Health Methodology 2007-2008 6
© Copyright 2007 Health Grades, Inc. All rights reserved.
May not be reprinted or reproduced without permission from Health Grades, Inc.
Statistical Models for Predicting Mortality
1. For each patient cohort, unique statistical, female only models were developed using logistic
regression. Cohorts were defined by developing a list of specific diagnoses and procedures to be
included in the cohort. A list of the codes used to identify patients in the six cohorts can be found in
Exhibit A.
2. Outcomes were binary, with patients recorded as either alive or expired at hospital discharge.
3. Comorbid diagnoses (e.g., hypertension, chronic renal failure, anemia, diabetes), demographic
characteristics (e.g., age), and specific procedures were classified as possible risk factors. Some
diagnosis codes were merged together (e.g., primary and secondary pulmonary hypertension) to
minimize the impact of coding differences. HealthGrades used logistic regression to determine which of
these were actually risk factors and to what extent they were correlated with mortality. A risk factor
stayed in the model if it had a positive odds ratio and was also statistically significant in explaining
variation. Potential risk factors with odds ratios less than one are removed from the model except in a
few cases. Complications were not considered as potential risk factors predicting mortality.
4. The statistical models were checked for validity and finalized. All of the models were highly significant,
with p values not greater than 0.0001. These cohort specific models were then used to estimate the
probability of death for each patient in the cohort.
5. Patients were then aggregated for each hospital to obtain the predicted outcome for each hospital.
Assignment of Ratings for Cardiac/Stroke Services for Women
For each hospital, the actual mortality was summed for all of the six patient cohorts and the predicted
mortality (risk adjusted) was summed for all of the six patient cohorts. The predicted mortality rate was
compared to the actual mortality rate for each hospital and tested for statistical significance at 90 percent
(using a z-score and a two-tailed test). Percentile scores were calculated based on the z-score.
The following rating system was applied to the comparison of the actual mortality for all six patient cohorts
and the predicted mortality rate for all six patient cohorts.
·
Better than expected Actual performance was better than predicted and the difference was
statistically significant, limited to the top 15 percent of hospitals (by z-score).
·
As expected The middle 70 percent of hospitals (by z-score).
·
Worse than expected Actual performance was worse than predicted and the difference was
statistically significant, limited to the bottom 15 percent of hospitals (by z-score).
To be included in the study, a hospital must have had all of the following:
·
An open heart program in 2005.
·
Over the three years, a minimum of 30 female discharges in coronary bypass surgery, 30 female
discharges in stroke, and 30 female discharges in any three of the remaining four cardiac cohorts
(valve replacement surgery, interventional cardiology procedures, acute myocardial infraction, or
heart failure) for a minimum of 150 discharges total.
·
For the most recent year, a minimum of 5 female discharges in coronary bypass surgery, 5 female
discharges in stroke, and 5 female discharges in each of the three cohorts for which they met the
30 discharge criterion above.
