Correlation of plasmatic sodium determined by the laboratory and that determined by the dialysis machine

Changes in plasma sodium concentration (_pNa, expressed in mEq/L) are common in hemodialysis (HD) patients. Hemodialysis monitors can estimate _pNa by using an internal algorithm based on ion dialysance measurements. The present study studies the accuracy of the correlation between the _pNa estimated by the dialysis monitor and that measured by the biochemistry laboratory at our center.

A single-centre prospective observational study in patients on a chronic HD program with the 6008 CAREsystem monitor and standard sodium (138mmol/L) and bicarbonate (32mmol/L) prescriptions. Venous blood samples were drawn from each patient before and after each HD session to ensure inter- and intra-individual validity. The _pNa was measured in the biochemistry laboratory using indirect potentiometry and simultaneously the estimated _pNa by the HD monitor was recorded at the beginning and at the end of the HD session. For statistical analysis, a scatterplot was made, and Spearman's correlation quotient was calculated. In addition, the differences between both methods were represented as Bland-Altman diagrams.

The pre-dialysis _pNa measured in the laboratory was 137.49±3.3, and that of the monitor, 137.96±2.91, with a correlation with R² value of 0.683 (p<0.001). The post-dialysis _pNa measured in the laboratory was 137.08±2.23, and that of the monitor was 138.87±1.88, with an R² of 0.442 (p<0.001). On the Bland-Altman plots, the pre-dialysis _pNa has a systematic error of 0.49, in favor of the monitor-estimated _pNa, with a 95% confidence interval (CI) of (-3.24 to a 4.22). In the post-dialysis _pNa, a systematic error of 1.79 with a 95% CI of (-1.64 to 5.22) was obtained.

The correlation between the _pNa estimated by Fresnius 6008 CAREsystem HD monitor and that measured by the laboratory is good, especially pre-dialysis measurements. Further studies should verify the external validity of these results.