Establishment and validation of a predictive model for post-transplantation diabetes mellitus based on glucose status on day 3 after kidney transplantation

Abstract Background and objectives: Post-transplantation diabetes mellitus (PTDM) is a common complication following kidney transplantation that significantly impacts long-term clinical outcomes. The current study aimed to develop a practical model for predicting the risk of PTDM based on glucose status on day 3 after kidney transplantation.Methods: This study enrolled 300 kidney transplant recipients between October 2018 and July 2019. All recipients were randomly assigned to either development or validation cohorts in a 2:1 ratio. Univariate logistic regression was used for predictor selection. The nomogram was created using a multivariate logistic regression model and calibrated by the receiver operating characteristic curves (ROC), the calibration curves, and the decision-curve analysis. The validation model externally validated the model.Results: The cumulative probability of developing PTDM within one-year of a kidney transplant was 11%. Univariate and multivariate logistic regression analyses identified hyperglycemia on day 3 (OR 2.02, 95% CI: 1.34-3.07, p < 0.001) and recipient age (OR 1.09, 95% CI: 1.02-1.15, p = 0.007) as independent risk factors for PTDM. The calculated potential predictive risk for PTDM was found to be in perfect agreement with the observed prevalence of PTDM. When compared to the previously used glucose statute on day 5, our model had better predictive performance. Moreover, higher risk of PTDM was associated with a lower recovery rate of graft function.Conclusion: Our study confirmed that the prediction model based on day 3 blood glucose levels can provide an early and accurate prediction of PTDM. This model will assist our clinicians in identifying and intervening in high-risk patients on time to prevent PTDM.