| 저자 |
Su Hee Kim* 1, Sukyung Lee1, Hee-Yeon Jung1, Ji-Young Choi1, Jang-Hee Cho1, Chan-Duck Kim1, Sun-Hee Park1, Yong-Lim Kim1 |
| 초록 |
Background: Modern epidemiologic approaches using multiple markers for predicting adverse outcomes may improve risk assessment in the general population. However, there have been limited evidences in dialysis population. We evaluated conventional risk factors and multiple inflammatory markers (uric acid, ferritin, WBC, CRP and albumin) as predictors of all cause, cardiovascular and infection-related mortalities in multicenter prospective observational cohort in patients on dialysis.
Methods: We enrolled 3,309 ESRD patients who were in dialysis between May 1, 2009 and December 31, 2013 at the 31 centers of the Clinical Research Center for ESRD in Korea. Cox proportional hazards regression methods and time dependent ROC curves were constructed and net reclassification index (NRI) and integrated discrimination improvement (IDI) were calculated for precise comparison of predictive power between multiple inflammatory markers with conventional risk factors and conventional risk factors alone.
Results: Each of three predictors (WBC, CRP and albumin) plus conventional risk factors had more predictive power than conventional risk factors alone in all cause and infection-related mortality and each of two inflammatory markers (CRP and albumin) in cardiovascular mortality. The best prognostic marker was albumin in all cause, cardiovascular and infection-related mortality. Addition of all combination of multiple inflammatory markers to conventional risk factors resulted in improving predictive ability in all cause and infection related mortality except in cardiovascular mortality.
Conclusion: Albumin was the best predictive marker and inflammatory markers such as WBC, CRP and albumin had more predictive ability beyond conventional risk factors. Multi-marker approaches using multiple inflammatory markers provided higher predictive power in total dialysis patients.
Table:
Table 1. Predictive power of multiple biomarker models for all cause, cardiovascular and infection related mortality using NRI and IDI
* CR; conventional risk factor, NRI; net reclassification index, IDI; integrated discrimination improvement.
Figures: |