Abstract: SA-PO396
Novel Risk Score to Predict 3-Year Mortality in Patients on Hemodialysis
Session Information
- Hemodialysis and Frequent Dialysis: CV and Risk Prediction
November 05, 2022 | Location: Exhibit Hall, Orange County Convention Center‚ West Building
Abstract Time: 10:00 AM - 12:00 PM
Category: Dialysis
- 701 Dialysis: Hemodialysis and Frequent Dialysis
Authors
- Okubo, Aiko, Iryo Hojin Ichiyokai Harada Byoin, Hiroshima, Japan
- Doi, Toshiki, Iryo Hojin Ichiyokai Harada Byoin, Hiroshima, Japan
- Morii, Kenichi, Iryo Hojin Ichiyokai Harada Byoin, Hiroshima, Japan
- Nishizawa, Yoshiko, Iryo Hojin Ichiyokai Harada Byoin, Hiroshima, Japan
- Yamashita, Kazuomi, Iryo Hojin Ichiyokai Harada Byoin, Hiroshima, Japan
- Shigemoto, Kenichiro, Iryo Hojin Ichiyokai Harada Byoin, Hiroshima, Japan
- Mizuiri, Sonoo, Iryo Hojin Ichiyokai Harada Byoin, Hiroshima, Japan
- Masaki, Takao, Hiroshima Daigaku Byoin, Hiroshima, Hiroshima, Japan
Background
The risk of death in patients on hemodialysis (HD) is high. Therefore, identifying patients at a high risk of death at the induction of HD is essential in clinical practice. However, few reports have examined the models predicting all-cause mortality in patients on HD using clinical factors, including electrocardiographic findings. This study aimed to investigate the clinical factors that have the most effect on the mid-term prognosis in patients on HD and to evaluate a novel risk model using these risk factors for predicting all-cause death.
Methods
We analyzed 385 patients with initiated HD at our four facilities from November 2008 to February 2019. We investigated demographics, medical history, laboratory data, and electrocardiographic findings at the initiation of HD therapy. We used the logistic regression model to predict 3-year all-cause mortality and evaluated it by cross-validation.
Results
During the 3-year follow-up, 86 (24.2%) patients died. Age (P<0.01), prior stroke (P<0.01), and the corrected QT interval (P=0.03) were identified as independent predictors of all-cause death by a multivariate logistic regression analysis. The predictive model was constructed using all these parameters with good discrimination of all-cause death (area under the curve: 0.80, with 80.1% sensitivity and 76.7% specificity). The area under the curve based on the 10-fold cross-validation was 0.82, with 78.2% sensitivity and 70.9% specificity, which suggested a good model. Patients above the cut-off value of the risk model were more likely to have a significantly higher risk of all-cause death than those below the cut-off value (hazard ratio: 9.13, 95% confidence interval: 5.16–16.16, P<0.001).
Conclusion
This novel risk model composed of age, prior stroke, and the corrected QT interval can stratify high-risk patients and be useful in predicting 3-year all-cause death in patients on HD. Our new risk score is easily measured at the bedside and provides good prediction of the mortality risk of patients on HD.