OBJECTIVES: A traditional statistical measure such as concordance(c) statistics has been used widely for comparison of risk adjustment models; however, c-statistic has been criticized for its insensitivity. To overcome the limitations of c-statistics, novel reclassification measures (reclassification calibration statistics, Net Reclassification Index (NRI) and Integrated Discrimination Index (IDI)) have recently been proposed. The objective was to compare Charlson Comorbidity Index (CCI), Chronic Disease Score (CDS) and CCI+CDS in predicting one year mortality in type-2 diabetes mellitus patients (T2DM) by applying novel reclassification measures. METHODS: The Clinical Practice Research Database, electronic medical record data from UK, was used for this retrospective longitudinal cohort study. Patients diagnosed with T2DM from January 1, 2006 to December 31, 2006 were included. Diagnosis and prescription information upto 1-year prior to the index date, i.e. first date of T2DM diagnosis, was used to create CCI and CDS, respectively. Patients were followed for 1 year from the index date to observe mortality. Descriptive statistics was used to describe the study cohort. In addition to traditional c-statistics from logistics regression, NRI and IDI were used to compare risk adjustment models. RESULTS: The cohort consisted of 26,191 patients with T2DM. The mean CCI and CDS were 0.24±0.67 and 1.58±1.06, respectively. The c-statistics values for CCI, CDS and CCI+CDS models were 0.791 (95%CI: 0.777-0.805), 0.788 (95%CI: 0.774-0.802) and 0.803 (95%CI: 0.789-0.817), respectively. The CDS and CCI+CDS reclassified 1.92% (p=0.238) and 6.50% (p=<0.001) patients into correct strata compared to the CCI. The IDI values for CDS and CCI+CDS were -0.64% (p=<0.001) and 0.43 (p=<0.001). This means that addition of CDS in CCI improved the prediction of mortality. CONCLUSIONS: Combined score (CCI + CDS) performed better than individual scores. In addition to c-statistics, reclassification measures such as NRI and IDI can be added to the armamentarium of risk adjustment model comparisons.