Meeting Brochure and registration form      SMDM Homepage

Tuesday, October 23, 2007
P3-18

THE DEVIL IS IN THE DETAILS: COMPARING VA DATA REPOSITORIES FOR QUALITY SURVEILLANCE

Kathleen A. McGinnis, MS1, Melissa Skanderson, MSW1, Forrest Levin2, Cynthia A. Brandt, MD, MPH3, Joseph J. Erdos, MD2, and Amy C. Justice, MD, PhD4. (1) VA Pittsburgh Healthcare System, Pittsburgh, PA, (2) VA CT Healthcare System, West Haven, CT, (3) Yale University, New Haven, CT, (4) Yale University / VA Connecticut Healthcare Center, West Haven, CT

Purpose: VA Healthcare System is the largest US provider of care to those with HIV and benefits from one of the most highly developed electronic medical information systems in the world. Immunology Case Registry (ICR) is an automated electronic data repository designed to monitor the costs and quality of HIV care. Decision Support Systems (DSS) was developed separately to monitor utilization and costs on all veterans in care. Because these systems use different methods for collecting overlapping data, they provide an opportunity to compare the accuracy and completeness of each repository. Methods: We compared data collected in DSS and ICR for 22,647 HIV+ veterans with an in or outpatient visit in FY 2002 for nine laboratory tests (Table 1). For each test, out of all individuals with a lab from either source, we calculated the percent of individuals with a result from each source. For individuals with results in both repositories, we calculated Pearson's correlation coefficients to quantify the correlation between overlapping lab values. Results: For six of the nine laboratory tests, DSS provided laboratory values on more individuals. ICR provided laboratory values on slightly more individuals for alanine aminotransferase and on substantially more individuals for HIVRNA viral load (VL) and CD4 (Table 1). Overlapping results were nearly perfectly correlated with the exception of CD4. Conclusions:  Despite drawing data from the same electronic record system, ICR and DSS each miss individuals with lab results in the other repository. This is likely due to discrepancies in the downloading process. The result of performance measures that count the number of CD4 tests done within a year or the proportion of patients achieving undetectable viral load may depend upon the repository used. Validation of data repositories should be mandatory before they are used to measure healthcare quality.

 

Table 1. Summary of Lab Values                                    

 

# Individuals with Value from DSS or ICR

% with Value from DSS

% with Value from ICR

Correlation Coefficient for Overlapping Values

Hemoglobin

19,848

91

87

.997

Aspartate Aminotransferase

18,985

99

86

.997

Alanine Aminotransferase

19,725

86

88

.999

Glycosylated Hemoglobin

4,008

95

80

.991

Creatinine

19,470

92

86

.990

White Blood Count

19,901

91

87

.961

Glucose

19,535

96

85

.940

HIVRNA VL

16,654

63

98

.989

CD4

16,381

64

98

.875