Casemix adjustment of managed care claims data using the clinical classification for health policy research method

The objective of this study was to explore the use of the Clinical Classification for Health Policy Research (CCHPR) as a casemix adjustment method for examining physician practice patterns.

The data source was 2 years of administrative claims from an 86,000 member health maintenance organization in southeastern Michigan. The CCHPR version 2 algorithm, which is in the public domain, was used to assign each claim to one of 260 clinical categories. CCHPR and age-sex categories were used as explanatory variables in multiple linear regression models with approved claims payments in dollars as the outcome variable. Regressions were performed retrospectively for 1994 and 1995, and with 1994 claims' history to predict 1995 utilization. Similar regressions were performed with age-sex categories alone, and also with the ambulatory diagnostic groups.

The adjusted R2 value of the retrospective regression models for total approved dollars was 0.42 for both study years when CCHPR categories were used. In contrast, age-sex explanatory variables alone achieved an R2 of 0.02.

The CCHPR method appears to be a promising tool to understand variability in physician resource utilization in managed care.