OBJECTIVES:

To develop the risk-adjustment capitation model for diabetes mellitus (DM) patients, and to further simulate the financial impact on healthcare insurance and healthcare providers.

METHODS:

Data were obtained from Tianjin Basic Medical Insurance database (2015-2019). Adult patients who had continuous enrollment and primary diagnosis of DM outpatient specific disease claims both in 2018 and 2019 were included. Five sets of risk-adjusters in 2018 were considered, containing age, sex, 17 DM complications, 17 Charlson comorbidities, and 36 interactions among DM complications. Three types of econometric models, including ordinary least squares (OLS), log-transformed OLS, and generalized linear model (GLM), were used to predict individuals’ annual DM-related total costs and outpatient costs in 2019. In-sample and out-sample predictive performance were evaluated by adjusted R², mean absolute error (MAE), and predictive ratio (PR). Based on the best-fit model, the impacts of implementing different risk-adjusted capitation payment schemes on healthcare insurance and providers were estimated using the ratio of payment compared with fee-for-service (FFS) or unadjusted capitation payment.

RESULTS:

A total of 241,120 patients (53.3% male; mean age 63.6±11.1 years) were identified, with the mean annual DM-related total and outpatient cost of RMB 19,069.88, and RMB 14,758.50, respectively. The optimal independent variables included all five risk-adjusters, and the optimal econometric model was OLS (adjusted R²: 16.74% and 30.48% for total and outpatient cost). The blended payment scheme that combined risk adjustment and partial FFS generated fewer financial fluctuations, in which the ratio of payment was 0.99 and 1.20 for healthcare insurance, and ranged from 0.96 to 1.21 and 0.95 to 1.35 for different healthcare providers, compared with FFS and unadjusted capitation payment, respectively.

CONCLUSIONS:

A capitation payment that adjusts for age, sex, and complications, especially those related to DM, performs better in predicting future DM-related costs. Blended payment schemes are more conducive to healthcare insurance payment reform.