Medical Devices Costs Groups: A risk adjuster for capitation payments based on the prior use of medical devices

Presenter: Richard van Kleef, Erasmus University, Rotterdam

Abstract

This paper introduces Medical Devices Cost Groups (MDCG’s) as a new risk adjuster for capitation payments. The essence is to classify individuals into MDCG’s based on the use of medical devices in year t-1 and to include the MDCG-classification as a risk adjuster in the capitation model for year t. In empirical and theoretical analyses we examine for 143 categories of medical devices, their feasibility to be included in this classification.

The data are from 7 Dutch insurers and include individual-level (N = 5,5 million) information on health care expenses in 2005, the current risk adjusters in the Dutch capitation system for insurers (i.e. age/gender, Pharmacy Cost Groups, Diagnostic Cost Groups, region, source of income and social-economic status) and provisions of medical devices in 2003-2005.

The methodology consists of the following steps:

1. we estimate the Dutch capitation model with the actual expenses in 2005 as the dependent variable and the risk adjusters mentioned above as explanatory variables;

2. for the users of medical devices in 2004 we combine actual and expected expenses in 2005 in order to calculate the average residual for each category. Categories with substantial residuals (which indicate substantial predictive power) proceed to the third step;

3. together with a team of medical experts we examine for each category its feasibility to be included in the MDCG-classification, using the following three criteria: measurability (e.g. to what extent is the use of the device recorded uniformly across insurers?), validity (e.g. to what extent is the use of the device related to long-term health problems?) and incentives (e.g. to what extent may the inclusion of the device in the MDCG-classification provoke undesired behavior from insurers and providers of care?). In addition, the medical experts cluster the selected categories into clinically homogeneous groups;

4. we re-estimate the Dutch capitation model with these clusters added as separate risk classes. Individuals who used multiple medical devices in 2004 are classified into the category with the highest residual (as calculated in the second step).

The results show that 12.8 percent of all individuals in the 2005-data used at least one medical device in 2004. For this group of users, the average residual in 2005 equals € 1,131. Of all 143 categories of medical devices, only 22 fulfill the criteria of measurability, validity and right incentives. These 22 categories can be grouped in 14 clinically homogeneous clusters. Inclusion of the 14 clusters as separate risk classes in the Dutch capitation model leads to an increase of the R-square from 23.7% to 24.7%, with only 1% of all individuals categorized in one of these classes. Starting from an age/gender-model the R-square would increase from 5.3% to 7.8%. Starting from an empty model the R-square would increase from 0 to 3.2%.

We conclude that MDCG’s can improve capitation schemes significantly. However, only a minority of medical devices fulfills the criteria of measurability, validity and right incentives.