Predictors of early mortality in dialysis patients

The mortality among end-stage renal failure (ESRF) patients undergoing rena l replacement therapy (RRT) remains high. An important proportion of these patients die shortly after the initiation of RRT. The present study aims to determine the best predictors for the early mortality in a group of 140 ES RF patients who initiated RRT between october 96 and december 99. The mean age of the study group was 61 +/- 13 years, and the mean follow-up time was 20 +/- 12 months. Diabetic nephropathy was the most prevalent etiology of renal failure (30%). The following data, collected immediately before the i nitiation of RRT, were included as independent variables: demographic and c linical characteristics, including the nutritional status established by th e Subjective Global Assessment (SGA) follow-up time in the predialysis clin ic (less or longer than 3 months) EPO therapy, vascular access, renal funct ion (creatinine and urea clearances, and Kt/V urea), hematological and bioc hemical data including serum albumin, bicarbonate, transferrin, PTH and C-R eactive protein, as well as the protein catabolic rate and the percent of l ean body mass normalized for ideal body weight, calculated from the 24 h to tal urine excretion of nitrogen and creatinine. The Cox proportional hazard regression model, stratified for an age over or less than 65 year, was uti lized to determine the best predictors for the mortality during the study p eriod. Sixty percent of patients had at least one comorbid condition, and 35% had cardiovascular diseases. Mild-moderate or severe malnutrition was observed in 48% of patients. The creatinine clearance and Kt/V urea before the initi ation of RRT were: 9.50 +/- 2.64 ml/min/1.73 m(2) and 1.47 +/- 0.44, respec tively. Forty-one patients died during the study period (annual death rate: 17%). The best predictor of mortality was the nutritional status assessed by the SGA (OR: 2.32, IC 95% 1.54-3.48, p < 0,0001). In a second analysis i n which the SGA was removed from the model, the previous history of cardiov ascular diseases (OR: 2.07 CI 95%: 1.06-4.06, p = 0.032), and the percent o f lean body mass/ideal weight (OR: 0.96; IC 95%: 0.93-0.99; p = 0.042), pro ved to be the best predictor of mortality. In conclusion, nutritional indices prior to the initiation of RRT, and the previous history of cardiovascular diseases were the best predictors of the early mortality in this unselected population on dialysis. Because nutriti onal status appeared to be a marker of the severity of the comorbid conditi ons, a better control of the number and severity of these comorbid conditio ns may be the best way for reducing the mortality in patients on RRT.