EXTERNAL VALIDATION OF A RISK PREDICTION TOOL IN DETERMINING EARLY DEATH AMONG ELDERLY PATIENTS INITIATED ON DIALYSIS
Rachelle Zolla Ciudadano-Manalaysay1 1Southern Philippines Medical Center, Davao City, Davao del Sur, Philippines
Abstract
External Validation of a Risk Prediction Tool in Determining Early Death
among Elderly Patients Initiated on Dialysis
Rachelle Zolla S. Ciudadano-Manalaysay, MD
Trisha Michelle V. Manalaysay, MD
Southern Philippines Medical Center, Philippines
Aim: This study aimed to validate if the Thamer Risk Scoring Tool is a significant predictor in determining risk of early mortality among elderly patients initiated on dialysis in a tertiary hospital in the Philippines
Background: The increase in the senior population is one of the factors that has contributed to the rise in the number of dialysis patients in the Philippines. The elderly has an increased risk for poor dialysis-associated outcomes, and studies have been conflicting if they have better survival with dialysis as compared to conservative management. The Thamer Risk Scoring tool uses simple variables to predict early mortality in these patients, however this has not been validated in the Philippines.
Methods: A retrospective cohort research design was used. Patients ≥ 60 years old, diagnosed with end-stage renal disease, and initiated on dialysis from 2018 to 2021 were subjected to the tool to predict 3-month, 6-month, and over-all mortality.
Results: Compared to the Thamer study with an AUROC = 0.691 for the tool in their validation cohort, this study showed an AUROC = 0.6245, 0.2847, and 0.5993 in the 3-month, 6-month, and over-all mortality groups respectively. As the score increased for the three groups, the tool became less sensitive and more specific with a concomitant increase in accuracy for predicting mortality.
Conclusion: Despite the poor performance of the tool in predicting mortality in terms of AUROC in this study, the sensitivity, and specificity had a similar performance in the Thamer study.
Biography
Bio to come