Mortality predictors after 10 years of dialysis: a prospective study of Japanese hemodialysis patients.

Background: This work aimed to examine the predictive value for death of various clinical variables after long-term hemodialysis (HD).
Design, Setting, Participants, and Measurements: A total of 947 patients (597 men and 350 women, aged 21 to 93 yr) who were undergoing maintenance HD in Niigata, Japan, were stratified into two cohorts: Those with >10 yr of prior HD at study enrollment (n = 391) and those with < or =10 yr of previous therapy (n = 556). The survival of patients was examined for up to 40 mo (1999 to 2003) with the Cox proportional hazards model. Baseline clinical and dialysis data and serum biochemistries were used as independent variables. For adjustment for bias in patient selection, patient survival in either cohort was analyzed separately.
Results: In patients with >10 yr of HD, high pulse pressure, cerebrovascular disease, low serum creatinine, and low Kt/V values were the mortality risk predictors, whereas for those with < or =10 yr of HD, age and cerebrovascular disease were independent risk predictors for death. Diabetes, coronary artery disease, serum albumin, and C-reactive protein were NS predictors in those with long-term HD.
Conclusions: Providing adequate dosage of dialysis and achieving a better control of pulse pressure may further improve survival in selected patients who had undergone HD for >10 yr.