46. Using the Rx Risk as a Case-mix Descriptor - Test-Retest Reliability
KL Sloan, VA Puget Sound Health Care System; AE Sales, VA Puget Sound Health Care System; CF Liu, VA Puget Sound Health Care System; P Fishman, Center for Health Studies, Group Health Cooperative of Puget Sound; ND Sharp, VA Puget Sound Health Care System
Objectives: The RxRisk, formerly the Chronic Disease Score, is a pharmacy-based case-mix measure for profiling and predicting health care resource utilization. Pharmacy data provide a consistently coded, direct measure of actively treated comorbidity and disease severity. The RxRisk uses outpatient pharmacy data to classify patients into 29 disease conditions. The VA is an ideal environment for using a pharmacy-based case-mix adjustment method because VA enrollees have financial incentives to obtain all their medications from the VA and pharmacy data are available in the VA national administrative databases. This study examines the test-retest reliability of the RxRisk categories by comparing Fiscal Year (FY) 1998 and FY1999 RxRisk classifications of veterans seen in the VHA Northwest Network (VISN 20).
Methods: Using data obtained from the VISN 20 data warehouse, we targeted all veteran users from facilities in VISN 20 who were classified into at least one RxRisk category during FY1998 and who had at least one prescription written during FY1999. We judged the test-retest reliability of the RxRisk categories by comparing the FY1998 and FY1999 patient classification on each of the 29 RxRisk categories with Cohen’s Kappa.
Results: There were 74,808 veterans in VISN 20 who met study inclusion criteria. These individuals averaged 14.7 (SD 14.0) and 16.0 (SD 15.2) prescriptions in FY1998 and 1999, respectively. They were classified into an average of 3.35 (SD 2.04) RxRisk categories in FY1998 and 3.56 (SD 2.22) RxRisk categories in FY1999. Examining the Kappa statistics revealed that 22 of the 29 RxRisk categories had Kappas >0.6 with HIV (K=0.883), Diabetes (K=0.876), Thyroid Disorder (K=0.867) and Transplant (K=0.806) all having Kappas >.80. The only two categories with Kappa <0.40 were Tuberculosis (K=0.251) and Renal Disease (K=0.209). Those categories defined by medications that are used in a more time-limited fashion [e.g., Rheumatoid Arthritis (prednisone, K=0.465), Pain (opiate-containing pain medications, K=0.427) and Liver Failure (lactulose, K=0.405)] tended to have lower Kappas while categories driven by medications that are rarely discontinued [e.g., Glaucoma (intraocular anti-glaucoma medications, K=0.797), Psychotic Illness (antipsychotics, K=0.785) and Gout (allopurinol, colchicine and probenecid, K=0.774] had higher Kappas.
Conclusions: In addition to its use as a casemix adjuster, the RxRisk may also be used to describe treated prevalence of chronic conditions with very reasonable confidence in the year-to-year test-retest reliability of the category classifications. Although the year-to-year stability of the classes varies, most Kappas are >0.60 which indicates quite acceptable reliability.
Impact: Administrators should consider using this non-diagnosis based casemix instrument to describe and understand better the actively treated chronic disease burden of their clinic and medical center populations.