2007 HSR&D National Meeting Abstract
3054 — New Approaches for Measuring Quality of Care and the Health of Veterans with Diabetes
Litaker DG (Cleveland VAMC) , Rajan M
(East Orange VAMC), Miller DR
(Bedford VAMC), Aron DC
(Cleveland VAMC), Pogach L
(East Orange VAMC)
Current measures of the quality of diabetes care use dichotomous thresholds to indicate whether optimal control has been achieved. This approach, however, fails to reflect the relative contribution of different risk factors to mortality, adjust for differences across facilities in patient characteristics, and provides little information on the relationship between treatment and veterans’ perceived health status. We compared the current approach with ones that addressed these limitations.
A 10% random sample of facilities in the VA’s Diabetes Epidemiology Cohort (DEpiC) in FY 2000 was used to determine facility level performance in the combined control (yes/no) of three risk factors (A1c <7%; cholesterol <200 mg/dl; systolic blood pressure <130 mm/Hg). Using a simulation model, the United Kingdom Progression of Diabetes Study (UKPDS) risk engine, we generated measures that accounted for relative risk factor importance and patient characteristics (e.g., age, comorbid conditions, and pre-existing diabetes-related complications) and that expressed values in terms of Quality Adjusted Life Expectancy (QALE). New measures represented 1) the proportion of veterans’ QALE “captured” (observed control of these three values/values resulting from optimal control) and 2) the sum of differences in QALE (optimal-observed) for each facility. Pair-wise correlations in facility rankings and change in rankings using each measure were assessed.
Mean control of all three risk factors for 11,846 veterans with diabetes seen at 12 facilities was 19%. The mean proportion of QALEs captured was 95% (range: 93.8-96.0) and translated into a missed opportunity to capture an average of 227 quality-adjusted life years (range: 84-375 years). Correlations between facility rankings using current and each new approach ranged from 0.39 to 0.74. Rankings changed by >= 2 in either direction at 4/12 (33%) of facilities when based on the current approach vs. one combining information from all three.
Tools such as the UKPDS risk engine combined with readily available administrative data offer quality measures that better reflect the complex epidemiology of diabetes and yield important information on the health status of veterans.
Measurement approaches such as this may better target quality improvement efforts to reduce persisting and substantial variation in the quality of diabetes care in the VA.