Health Grades DHAPatient Safety Methodology (Page 2)

Distinguished Hospital Award Patient SafetyTM Methodology 2007 - 2

May not be reprinted or reproduced without permission from Health Grades, Inc.

HealthGrades used the QI Windows Software, version 3.0a, developed by the Agency for Healthcare
Research and Quality (AHRQ) and downloaded from

www.qualityindicators.ahrq.gov/winqi_download.htm

Major changes to the indicators in this most recent version of software are outlined in Appendix B.

Following all AHRQ guidelines for using PSI software, HealthGrades applied it to all short-term acute care
hospitals in the MedPAR file for three years (2003 through 2005).

Given that this data set applies mostly to patients over the age of 65, HealthGrades excluded the following
PSIs from the analysis:

Birth trauma injury to neonate

Obstetric trauma cesarean delivery

Obstetric trauma vaginal delivery with instrument

Obstetric trauma vaginal delivery without instrument

Due to coding variation in the use of E codes, HealthGrades excluded three additional indicators:

Complications of anesthesia

Accidental puncture or laceration

Transfusion reaction

Data Exclusions

HealthGrades modified the "Failure to rescue" patient group by excluding cancer patientspatients having
any ICD-9 code between 140.0 and 208.9 or between 230.0 and 239.9.

HealthGrades also removed hospitals in the U.S. territories and Puerto Rico from the data set.

Overall Patient Safety Score

To determine the overall patient safety score by hospital, HealthGrades performed the following steps.

1. AHRQ software calculated observed and expected rates for each hospital and PSI, provided that the

PSI had at least three cases. (HealthGrades used a stratification process to calculate expected rates for
those PSIs where AHRQ software only provided observed rates.)

2. HealthGrades identified significant bias in the expected rates for larger hospitals (which had

consistently higher observed rates than expected). Therefore, HealthGrades performed further risk
adjustment using the Medicare Case Mix Index (CMI). The case mix index adjustment compensates for
the fact that within a given DRG the most severely ill will probably be clustered at larger hospitals.

To perform the case mix index adjustment and remove the bias, HealthGrades:

a) Stratified hospitals by their CMI category. This was done separately for each of the three years
(20032005) using the corresponding year's CMI. Therefore, it is possible that a hospital could be in
different CMI strata from year to year. (See Appendix A for definitions.) CMI < 1.25 was the first level,
and so on.

Health Grades - DHAPatient Safety Methodology (Page 2)