Influence of protein on the determination of sodium, potassium and chloride in serum by Ektachem DT 60 with the DTE module; evaluation with special attention to a possible protein error by flame atomic emission spectrometry and ion-selective electrodes; proposals to their calibration.

The reliability of the Ektachem DT 60 with the DTE module was evaluated. The precision of the determination of sodium, potassium and chloride in serum was adequate. The relative standard deviation for precision between days was Na+ 0.7%, K+ 1.5% and Cl- 1.0%. The means of the Ektachem results for 7 control sera differed from those of the reference method values by 0.9% (Na+), -0.9% (K+) and +4.4% (Cl-). Similar results were obtained for the analysis of patient sera. The influence of protein was investigated, using sera of increasing protein concentration prepared by ultracentrifugation. The results from the Ektachem corresponded to the values obtained by flame atomic emission spectrometry, even at high protein concentrations, although Ektachem measurements are performed by ion-selective electrodes without predilution. In paraproteinaemia, the Ektachem and flame atomic emission spectrometry results disagreed. Chloride determinations by Ektachem distinctly differed from measurements of the chloride concentration in total serum. It is proposed that ion-selective electrodes should be calibrated and linearized with respect to sodium chloride, in order to obtain an accurate value for the concentration of electrolyte in serum water. Concentrations are easier to interpret than "activities" for therapeutical purposes, and they can be used to define protein- and lipid-independent reference intervals for these electrolytes. With this calibration procedure, the results from ion-selective electrodes are never lower than values obtained by flame atomic emission spectrometry. The accuracy of ion-selective electrode measurements should be evaluated by applying reference methods for sodium, potassium and chloride to the ultracentrifugation supernatant of the corresponding serum. This approach can help to settle the dispute concerning the influence of protein on the residual liquid junction potential.