Select your timezone:

P15.07 Risk prediction model for immunosuppressive medication nonadherence in kidney transplantation recipients

Lei Dong, People's Republic of China

Central South University

Abstract

Risk prediction model for immunosuppressive medication nonadherence in kidney transplantation recipients

Lei Dong1, Jia Liu1,2,3, Xiao Zhu2, Hao Li2.

1XiangYa School of Nursing, Central South University, Changsha, People's Republic of China; 2Transplantation Department, The Third Xiangya Hospital, Central South University, Changsha, People's Republic of China; 3 Nursing department, The Third Xiangya Hospital, Central South University, Changsha, People's Republic of China

Objective: To identify the risk factors associated with immunosuppressive medication (IM) nonadherence in kidney transplantation recipients and to develop a simple and effective personalized risk prediction model of it.

Methods: A total of 1011 kidney transplantation recipients in a tertiary hospital in China were recruited from March 2020 to April 2021. Least absolute shrinkage and selection operator (Lasso) regression analysis was used to screen the optimized variables. Multivariate Logistic regression analysis was applied to develop a risk prediction model for IM nonadherence in kidney transplantation recipients, and the relevant nomogram was drawn. The receiver operating characteristic curve (ROC), calibration curves, and Hosmer-Lemeshow test were used to validate and evaluate the discrimination and calibration of the model, and Bootstrap method for internal verification.

Results: The multivariate regression analysis showed that age, perceived barriers to adherence score, marital status, family income, medical insurance, preoperative drinking history and gender were the risk factors for IM nonadherence in kidney transplantation recipients (P<0.05). The nomogram model demonstrated good discrimination, with the area under the ROC curve of 0.775. The Hosmer-Lemeshow showed perfect fitting degree (P=0.395, P>0.05), and the calibration curve approached to the ideal curve. The C-index of 0.782 was reached in internal verification.

Conclusion: The model developed in this study with 7 predictors has satisfactory predictive efficacy, with good discrimination and calibration, which can provide references for medical staff to early identify high-risk kidney transplantation recipients with IM nonadherence and to formulate relevant intervention measures.

National Natural Science Foundation of China, 71904209.

Social Media Promotion Image

right-click to download

© 2022 TTS 2022