Relation between serum urea and mortality of hemodialysis patients.

Background. It is well known that serum urea concentration is not a good predictor of mortality in hemodialysis patients. On the other hand, urea kinetic modeling has been very successfully used to measure dialysis dose by the Kt/V index, which was found to he a good predictor of mortality. Could there be a relation between urea and mortality, but in some more complex way? Methods. This is a post-hoc analysis of a single center observation study that included 242 patients and an 11-year observation period. Mortality rates between the quartiles of serum urea levels were examined by a 2 x 4 table with the chi(2) test. Both univariate and multivariate survival analyses were performed with standard and segmented extended Cox regression. Results. The relation between mean urea in the baseline period and mortality showed an irregular U-shaped curve. The lowest mortality was observed in the third quartile (28 to 31 mmol/L). The relation between mean urea in the whole observation period and mortality was a J-shaped curve. The lowest mortality was in the second quartile (25-27 mmol/L). Urea was not a predictor of mortality in the whole cohort, but low-urea (binary) and high-urea (binary) were independent predictors of mortality in the corresponding models using standard or extended Cox regression. Conclusion. This study revealed a complex relationship between urea and mortality in hemodialysis patients. Patients with low or high urea levels exhibited higher mortality than those with medium levels, while both low and high levels of urea were independent predictors of all-cause mortality.