Abstract: SA-PO739
Validation of Indices Predicting 6 Month Mortality for Incident Dialysis Patients Over 75
Session Information
- Geriatric Nephrology
October 27, 2018 | Location: Exhibit Hall, San Diego Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Geriatric Nephrology
- 1100 Geriatric Nephrology
Authors
- Thorsteinsdottir, Bjoerg, Mayo Clinic , Rochester, Minnesota, United States
- Hickson, LaTonya J., Mayo Clinic , Rochester, Minnesota, United States
- Giblon, Rachel, Mayo Clinic , Rochester, Minnesota, United States
- Connell, Natalie B., Mayo Clinic , Rochester, Minnesota, United States
- Zand, Ladan, Mayo Clinic , Rochester, Minnesota, United States
- Tangri, Navdeep, Seven Oaks General Hospital, Winnipeg, Manitoba, Canada
- Albright, Robert C., Mayo Clinic , Rochester, Minnesota, United States
Background
Prognosis is one of the pinnacles of shared decision making. Much research has been done in the development of valid prognostic tools for mortality in patients on dialysis, however the performance and clinical applications of these tools in elderly incident dialysis patients remains to be determined. We aimed to find the best performing prognostic tool for short term mortality in elderly incident dialysis patients.
Methods
Six validated indices (Foley, NCI, REIN, newREIN, Thamer, and Wick) were identified in a systematic review for short term mortality prediction in incident dialysis patients. We applied these indices to an established cohort of 349 incident dialysis patients who were >75y old and used each to predict 6 month mortality. Variables were extracted from the medical record, and missing variables were imputed. Models were compared using the c-statistic, i.e. area under the receiver operating characteristic curve (ROC). Sensitivity and specificity were also calculated.
Results
The average age of our population was 81.5 years, 65.9% male, median survival was 351 days. The ROC ranged from 0.57 to 0.73, with the Foley and Wick indices performing the best, see Figure 1. The following cutoff scores yielded a specificity of >50% and >90% respectively in predicting mortality: Foley 7 and 10, NCI 4 and 10, REIN 4.2 and 8.2, REIN_new 12.1 and 16.4, Thamer 4.2 and 6.5, Wick 6 and 10. Sensitivity ranged from 45.8-76.4 for 50% mortality prediction with Wick having highest sensitivity and 9.0-22.0 for 90% mortality with Foley having highest sensitivity.
Conclusion
Most of the indices can predict which patients are more likely than not to die within 6 months of dialysis start with acceptable sensitivity and specificity. Higher cutoff scores can be used to predict death with high specificity (>90%) but at the considerable loss of sensitivity.The Wick and Foley indices performed best.
Funding
- NIDDK Support