Comparison of measuring serum osmolality and equations estimating osmolality in peritoneal dialysis patients

The osmolar gap increases with kidney failure. A number of equations have been proposed to calculate serum osmolality, allowing determination of the osmolar gap by comparison with measured osmolality. As glucose and icodextrin absorption can potentially interfere with the laboratory measurement of serum sodium, a key component in equations calculating osmolality, we reviewed the performance of 14 equations used to calculate serum osmolality compared to the measurement of serum osmolality in 144 patients with peritoneal dialysis (PD); 81 (56.3%) males, 76 (52.5%) diabetics, mean age of 64.4 ± 16.3 years, 115 (79.9%) prescribed icodextrin and 38 (26.4%) 22.7 g/L glucose dialysates. Measured serum osmolality was 311 (304-320) mosmo/kg (mmol/kg), whereas calculated osmolality for the 14 equations ranged from a median of 274 (269-284) mosmo/kg to 307 (300-316) mosmo/kg. Bland-Altman mean bias showed that measured serum osmolality was greater than the calculated osmolality ranging from 4.0 mosmo/kg to 36.2 mosmo/kg between the 14 equations, with wide 95% limits of agreement (LoA) ranging from -27.1 mosmo/kg to 19.4 mosmo/kg and from -58.5 mosmo/kg to -13.8 mosmo/kg. Only 2 of the 14 equations gave a mean osmolar gap of <10 mosmo/kg and showed no systematic bias, median serum osmolality of 307 (300-316) and 303 (298-312) mosmo/kg, Spearman ρ of 0.57, 0.62, both p < 0.001, respectively. Our study would suggest that only 2 of the 14 equations we compared with measured serum osmolality showed no systematic bias, but still had too great a bias to be useful in clinical practice. As such we propose a new equation to calculate serum osmolality in patients with PD.