No abstract

Serum biochemical measures suggestive of undernutrition have been reported to correlate with 1-yr mortality risk in prevalent groups of hemodialysis patients. The predictive power of these variables has not been reported in newly diagnosed patients or in patients whose dialysis prescription is guided by urea kinetics. The relationship of these predictors to mortality over periods of longer than 1 yr is also unreported. Therefore, the survival of 184 hemodialysis patients was examined for up to 44 months (1987 to 1991) with the Cox proportional hazards model. Baseline demographic, clinical, and biochemical parameters were used as independent variables. To adjust for bias in patient selection, the survival of patients with 12 months or less of prior dialysis at the time of enrollment ("new cases") was analyzed separately from that of patients with more than 1 yr of prior treatment ("long-standing cases"). Serum albumin was less than 3.5 g/dL in 31% of new cases and in 12% of long-standing cases. Adjusting for the other variables, low serum albumin was the strongest mortality risk predictor in both new and long-standing cases. Low serum cholesterol was an independent risk predictor in both groups. Diabetes and race were not significant predictors. Mean age at enrollment was nearly a decade higher for nonsurvivors than for survivors, in both new and long-standing groups. Yet, age was not an independent risk predictor in the Cox model for the new group because of an unexpectedly high death rate among young black men. Female gender, which was confounded by increased age, took the place of age in the model for the new group. For each model, there was good agreement between observed and predicted mortality for up to 24 months. To assess the influence of dialysis treatment time and dose (measured as pre-to-post treatment urea ratio) on risk, survival was examined in a subset of 139 patients monitored for up to 22 months, from 1989 to 1991, a period when the urea ratio was used routinely. Adjusting for the other variables, low serum albumin and cholesterol again independently increased risk. The urea ratio was also a significant independent predictor. The pattern of mortality by urea ratio was U shaped, with minimum risk for values between 2.5 and 3.4 Treatment time did not influence risk. It was concluded that baseline serum values of albumin and cholesterol strongly influence survival for up to 2 yr in new and long-standing hemodialysis patients.(ABSTRACT TRUNCATED AT 400 WORDS)

Urinary red blood cells (RBC) are usually small and morphologically abnormal in glomerular (GN) hematuria, and slightly enlarged and morphologically normal in nonglomerular (NG) hematuria. This study was performed to evaluate the diagnostic value of urinary red cell size and morphology and to investigate the mechanism of the alteration in cell size. In 34 consecutive patients with hematuria we examined the urinary RBC size distribution and mean corpuscular volume (MCV) by electronic sizing of suspensions of RBC in an isotonic medium and, in 28 cases, compared it with the presence of 50% or greater dysmorphia. In 15 consecutive cases, we correlated MCV values with urine chemistries. In two GN cases we recorded the urine MCV serially during a furosemide-induced diuresis. In vitro incubations of peripheral or urinary RBC in various electrolyte solutions prior to sizing were also performed. The MCVs were significantly lower in GN (p < 10∼⁶) and probable GN (p < 10–⁴) than NG hematuria. A cutoff of 72 fl completely separated GN and probable GN from NG cases. Fifty percent or more ‘dysmorphic’ RBC were seen in 12 of 13 GN, 3 of 4 probable GN but in no NG sediments. In patients with NG hematuria, the ratio of urinary to peripheral MCV tended to be greater than unity and correlated strongly with pH (r = -0.97; p < 0.002). The effect of pH was confirmed in vitro. Furosemide diuresis induced a partial correction of the microcytosis of GN RBC, which correlated with the changes in urine composition. Furosemide had no effect on GN cells in vitro. Incubation of venous RBC in saline ranging from 0.5 to 6% did not alter MCV. Saline of 0.45% or less caused partial to total hemolysis and the appearance of RBC fragments with extremely low ‘MCV values. We conclude that electronic RBC sizing is a highly accurate and objective method for differentiating GN from NG hematuria. Urinary pH is an important determinant of MCV for NG RBC. A hypothesis is proposed to explain the microcytosis of GN RBC.

No abstract

No abstract