78. Do Different Casemix Measures Affect Judgments of VISN Performance?
SA Loveland, Center for Health Quality, Outcomes and Economic Research; AK Rosen, Center for Health Quality, Outcomes and Economic Research; JJ Anderson, Center for Health Quality, Outcomes and Economic Research; C Rakovski, Center for Health Quality, Outcomes and Economic Research; D Berlowitz, Center for Health Quality, Outcomes and Economic Research
Objectives: Diagnosis-based case-mix measures are increasingly being used for provider profiling. This may lead to case-mix adjusted comparisons of VISN performance in the VA. We examined whether use of two such measures, Adjusted Clinical Groups (ACGs) and Diagnostic Cost Groups (DCGs), results in different assessments of performance across VISNs.
Methods: A 40% random sample of veteran patients, excluding individuals with telephone or dental encounters only, was obtained from VA inpatient and outpatient databases. The resulting sample consisted of 1,046,803 veterans who received acute, long-term, or outpatient care during FY’97. This sample was cross-validated with an additional 20% random sample of FY'97 patients (N=524,461). We developed two weighted concurrent risk adjustment models using a reparameterized model from each of the ACG and DCG systems. To age and gender categories, we added 136 Hierarchical Condition Categories (for the DCG/HCC model) and 32 Adjusted Diagnostic Groups (for the ACG/ADG model) to explain FY'97 utilization (the annualized sum of inpatient and outpatient days). Three profiling indicators were used: 1) an unadjusted utilization ratio (VISN mean utilization days/VA mean utilization days); 2) morbidity ratios (VISN mean predicted days based on DCG or ACG parameters/VA mean actual days); and 3) efficiency ratios (VISN mean actual days/VISN mean predicted days based on ACGs or DCGs). The first ratio is a measure of relative utilization per VISN; the second ratio adjusts this for case-mix; and the third is a measure of the over- or under-utilization of each VISN adjusted by its case-mix.
Results: Based on unadjusted utilization ratios, we identified the three VISNs with the highest relative utilization (1.20 to 1.35) and the three VISNs with the lowest relative utilization (0.80 to 0.88). Both the ACG-based and DCG-based morbidity ratios identified two of these top three VISNs as also being the highest with respect to case-mix burden; for the bottom three, the ACG-based ratio classified all three as lowest in case-mix burden while the DCG-based ratio identified two. There was slightly more disagreement between the ACGs and DCGs on morbidity ratios, in that the DCG adjustment rated more VISNs as having higher case-mix than did the ACG adjustment. Although more VISNs scored below 1.00 (i.e. efficient VISNs) based on the ACG efficiency ratio, the DCG efficiency ratio was lower (more efficient) than the corresponding ACG efficiency ratio for thirteen of the twenty-two VISNs. The degree of efficiency was related to whether the VISN was considered low or high in morbidity.
Conclusions: There was some consistency in profiling high and low users by the ACG and DCG risk-adjustment methodologies. However, depending upon the measure and model selected, a VISN will not consistently rank among the highest or lowest performers on all indicators.
Impact: Development of accurate tools for making fair comparisons across providers (i.e., VISNs) is critical. Since use of different case-mix measures affects judgments of performance, caution should be used in interpreting VISN-level results so that "worse-than average" VISNs are not penalized unfairly for inefficiency, and similarly, "better-than-average" VISNs are not rewarded prematurely.