Development of prediction model for chronic kidney disease in type 2 diabetics in Thailand

Abstract

Diabetes mellitus (DM) is a major contributing factor that leads to end stage renal disease (ESRD) with unsuitable clinical management. Chornic kidney disease (CKD) prediction model could prevent the progression of CKD to ESRD. However, current CKD prediction models in patients with type 2 DM were developed to predict ESRD. This study aimed to develop prediction model for CKD stage 3 in patients with type 2 DM in Thailand. This was a 10-year retrospective cohort study obtaining data of patients with type 2 DM from electronic database of Taksin hospital during 1 January 2008 to 31 December 2017. The outcome variable was the present of CKD stage 3 which was defined as estimated glomerular filtration rate (eGFR) <60 mL/min/1.73m2. The total patients of 2,178 were randomly assigned into training dataset for developing model (N=1,525) and validation dataset for model validation (N=653). The study used Cox proportional hazard regression for model development. For model performance, while model discrimination was conducted using Harrell’s C-statistic, model calibration was evaluated by Hosmer-Lemeshow chi-square test and survival probability curve. The results showed that the median follow-up time was 1.29 years (interquartile range, 0.5- 2.5 years) and 385 patients or 17.68% with CKD stage 3. Data analysis identified five CKD stage 3 predictors including age, female, urinary albumin to creatinine ratio, baseline eGFR, and Hemoglobin A1c. Based on these five CKD predictors, two CKD prediction models were developed, model 1 using laboratory testing data and model 2 using simplified or proxy data. Both models demonstrated good discrimination with C-statistic of 0.890 and 0.812, respectively, and accurate prediction. These two CKD prediction models are recommended for health providers to use as an input for decision making on clinical management which could prevent diabetic kidney disease and for raising patients’ awareness on health prevention.