Benchmarking in Understanding Resource Utilization and Disease Management — Are We Ready?

Abstract

Objective: The study was designed to examine 3 years of a healthplan’s claims data (1997, 1998 and 1999) in order to accomplish an in-depth analysis on the costs and use of services and pharmaceuticals in the treatment of diabetes. It was also intended to benchmark those findings against similar data for a large national population of managed-care lives, in order to compare the healthplan’s performance with regional and national averages. A larger goal was to begin exploring the value of mining large benchmarking databases, such as this, to support healthplan activities.

Study design: Retrospective claims data analysis.

Methods: Encrypted claims data from Blue Cross and Blue Shield of Minnesota (BCBSM) were employed to identify and categorize episodes of diabetes using the Episode Treatment Group™ (ETG™) methodology from Symmetry Health Data Systems of Phoenix, Arizona. A clinical disease model developed by Phar-Metrics, Inc. of Watertown, Massachusetts, was then superimposed on the ETG™ group to allow the analysis to focus on patients with diabetes, stratified according to type 1 versus type 2 diabetes mellitus, and by the presence or absence of comorbid conditions and complications. These data were analyzed as follows along several parameters, including the type of product: health-maintenance organization (primary care coordinated) and preferred-provider organization (open access or fee-for-service); region of the state; and by the ten largest providers in the state. Three major pathways of analysis focused on (i) the cost of an episode of diabetes, (ii) utilization of recommended care services for members with diabetes, and (iii) the impact of different drug treatment cohorts on the costs of an episode of diabetes. In addition to the internal analysis of BCBSM data, a similar analysis was used on the large Integrated Outcomes Database (about 16 million covered lives at the time of this analysis) developed by PharMetrics, Inc., in order to benchmark BCBSM performance against that of the Midwest and the US in general. Although the study included both type 1 and type 2 diabetes mellitus patients, this report focuses primarily on the results noted for those with type 2 diabetes.

Results: Analysis of the data showed that costs (based on charges) for an episode of diabetes were higher in the Midwest than in the rest of the US. Costs within BCBSM were comparable to those of the Midwest. Results also showed that members of BCBSM received the (minimum) recommended care services for diabetes at a higher rate than the rest of the country. Finally, the study showed considerable variation in the cost of an episode of diabetes based on the drug or drugs used to treat the illness, but did not indicate that any individual drug or combination of drugs was superior from a cost-effectiveness perspective.

Conclusions: The methodology and focus on episodes of care appear to be a good way to examine complex illnesses to try to understand the impact of variation in resource utilization. The lack of specific outcomes data limits the interpretation of the results. Finally, the value of benchmarking healthplan data against large, national databases has yet to be established. This study was a first step in assessing the value of such analyses. It is my opinion that having access to such databases can enhance the value of healthplan studies and that this study represents only a very preliminary approach to understanding and harnessing the value of such a resource.