Institutional members access full text with Ovid®

Share this article on:

Challenges and Benefits of Adding Laboratory Data to a Mortality Risk Adjustment Method

McCullough, Elizabeth MS; Sullivan, Christopher PhD; Banning, Pamela BS; Goldfield, Norbert MD; Hughes, John MD

doi: 10.1097/QMH.0b013e318231cf4f
Original Articles

Background: There is increased interest in improving the clinical acceptance and statistical performance of hospital mortality rate comparisons. This study assessed the feasibility of linking separate electronic feeds of laboratory data and claims-based information and, if successful, to identify laboratory data elements that significantly improved mortality rate predictions for All-Patient Refined Diagnosis Related Groups (APRDRGs), a risk of mortality (ROM) classification tool in regular use for public reporting purposes.

Methods: The Florida Agency for Health Care Administration recruited 15 hospitals to supply computerized administrative and laboratory information that could be linked at the patient level. The hospitals, standardized computer code terminology, which was then merged with administrative data. We evaluated the ability of the merged data to improve APRDRG ROM predictions.

Results: We describe the procedures that the laboratory information systems used to link the electronic laboratory data with standard claims data. The addition of 11 clinical laboratory test results increased the C statistic by 0.574% and R2 by 4.53%.

Conclusions: This study supports the feasibility of linking laboratory data elements with claims-based administrative data to enhance ROM assessments. This linkage resulted in modest statistical improvement in a commonly used ROM model.

3M Health Information Systems, Inc, Wallingford, Connecticut (Ms McCullough and Banning, and Dr Goldfield); Image Research, Tallahassee, Florida (Dr Sullivan); and Yale University School of Medicine, New Haven, Connecticut (Dr Hughes).

Correspondence: Norbert Goldfield, MD, 3M Health Information Systems, 100 Barnes Rd, Wallingford, CT 06492 (

This study was funded by the Adding Clinical Data to Statewide Administrative Data AHRQ Contract #07–10042.

The authors (except for Christopher Sullivan) are clinical developers of all classification tools at 3M Health Information Systems Inc.

©2011Lippincott Williams & Wilkins, Inc.