© Copyright 2007 Health Grades, Inc. All rights reserved.
Hospital Report CardsTM Methodology 4
May not be reprinted or reproduced without permission from Health Grades, Inc.
The statistical models were checked for validity and finalized. All of the models were highly significant, with
C-statistics ranging from ~ 0.6 to ~ 0.9. These cohort and outcome-specific models were then used to
estimate the probability of the outcome for each patient in the cohort. Patients were then aggregated for
each hospital to obtain the predicted outcome for each hospital.
Statistical significance tests were performed for each patient cohort to identify, by hospital, whether the
actual and predicted rates were significantly different. We used binomial distribution to establish an
approximate 90% confidence interval.
Assignment of Star Ratings
The following rating system was applied to the data for all procedures and diagnoses:
Actual performance was better than predicted and the difference
was statistically significant.
Actual performance was not significantly different from what
was predicted.
Actual performance was worse than predicted and the difference
was statistically significant.
In general, 70 percent to 80 percent of hospitals in each procedure/diagnosis were classified as three stars,
with actual results not significantly different from predicted results. Approximately 10 percent to 15 percent
were one-star hospitals and 10 percent to 15 percent were five-star hospitals.
APR-DRG-Based Ratings
For respiratory failure and gastrointestinal procedures and surgeries, the risk adjustment
was based upon APR-DRGs, a methodology developed by 3MTM Corporation. APR-DRGs
are an enhanced extension of the basic DRG (diagnosis related group) concept developed
by 3MTM's Clinical Research Group, the National Association of Children's Hospitals and
Research Institutes (NACHRI), and several physician groups.
While DRGs focus on the Medicare population, APR-DRGs describe a complete cross-section of acute care
patients and are specifically designed to adjust data for severity of illness (How sick is the patient?) and risk
of mortality (How likely is it that the patient will die?).
The fundamental principle of APR-DRGs is that the severity of illness and risk of mortality are both
dependent on the patient's underlying condition. High severity of illness and risk of mortality are
characterized by multiple serious diseases and the interactions between the disorders.
The 3M
TM
APR-DRG methodology is the most widely used severity-of-illness and risk-of-mortality adjustment
tool available today. It has become the standard for adjusting large volumes of data to account for
differences related to the individual's severity of illness or risk of mortality. As a result, the focus can be on
the differences in clinical care, thus providing equitable comparisons of quality and cost of care. APR-DRGs
are also recognized as the tool of choice by commissions, state agencies, and others who disseminate
comparative performance data to regulators, payers and the general public.
Data Analysis
The output from the APR-DRG software was twofold:
·
It told us how many patients had respiratory failure or gastrointestinal procedures or surgeries in
each hospital.
·
It identified each patient as being in one of four subclasses of mortality risk:
·
Minor
·
Moderate
·
Major
·
Extreme
