Bibliography

This publication, published in 2026, compares sodium and potassium measurements performed in the emergency department using point-of-care testing (POCT) and those obtained from the central laboratory. It is a retrospective single-center study conducted over a one-year period. The authors emphasize that only a limited number of studies have evaluated discrepancies in electrolyte measurements, particularly sodium and potassium, despite the fact that these parameters are easily accessible on blood gas analyzers used for point-of-care testing. Nevertheless, these differences are already known. The study compared 6,404 sodium measurements and 5,622 potassium measurements performed: on the one hand, on whole blood samples in the emergency department using a Radiometer ABL90 blood gas analyzer; on the other hand, in the central laboratory on plasma samples using a Roche analyzer. Samples for point-of-care testing and laboratory analysis were collected less than one minute apart. Laboratory analyses were performed within two hours after the POCT measurement. The main finding demonstrated a systematic difference between values obtained by point-of-care testing and those obtained in the central laboratory, with POCT values being generally higher than laboratory values. However, across more than 5,500 measurements, this discrepancy did not result in any major overall clinical impact on patient management.
The authors observed that sodium measurements showed only moderate agreement with laboratory results, whereas potassium measurements demonstrated better concordance.

For sodium, the difference between the two methods exceeded 4 mmol/L in 8.5% of cases and exceeded 8 mmol/L in 0.5% of cases. They also observed that the discrepancy between methods tended to increase at higher sodium concentrations. For potassium, the difference exceeded 0.5 mmol/L in 4.5% of cases and exceeded 1 mmol/L in 1% of cases. These discrepancies may nevertheless have important consequences for certain clinical situations, which constitutes one of the central points of the discussion. In particular, in severe hyponatremia, management relies on the administration of hypertonic saline with repeated biological reassessments. Additional treatment is indicated if sodium levels fail to increase sufficiently, generally by approximately 5 mmol/L. The accuracy of sodium measurement is therefore essential in order to avoid undercorrection or overcorrection. Similar concerns apply to the management of hyperkalemia. The authors therefore strongly recommend maintaining the same analytical method for electrolyte monitoring throughout the patient’s management. Regarding the study limitations, the authors point out that this retrospective analysis does not allow precise identification of the causes of the observed discrepancies. Several hypotheses are discussed, including: differences related to the electrodes used by the different analytical techniques; the fact that point-of-care testing is performed on whole blood whereas laboratory measurements are performed on plasma; potential biases related to hematocrit, protein concentration, lipemia, or hemolysis, which were not specifically collected in this study. The authors also highlight the limited number of patients presenting with severe electrolyte disturbances, particularly severe hypernatremia.

The authors present a retrospective study involving 120 patients conducted over a three-month period in the emergency department of a hospital in Dubai. The aim was to compare, on a paired basis, sodium and potassium values obtained either by Point-of-Care Testing (POCT) using an Cobas b221 device or by the central laboratory on plasma, with a maximum sampling interval of 30 minutes between measurements. The authors emphasize that discrepancies in electrolyte measurements are a well-known and widely published issue, particularly for potassium and sodium. Ils rappellent également que, selon les recommandations du CLSI, les biais acceptables sont inférieurs ou égaux à 0,3 mmol/L pour le potassium et à 3 mmol/L pour le sodium. Dans cette étude, la concordance observée pour le potassium montre un r de 0,91 avec un biais de +0,12 mmol/L, tandis que pour le sodium, le coefficient de corrélation est de 0,88 avec un biais de -1,2 mmol/L. Les auteurs insistent sur l’essor actuel des méthodes de biologie délocalisée aux urgences ainsi que sur les différents facteurs pouvant influencer les résultats : calibration, méthodologie analytique, nature des électrodes, etc., notamment dans les situations d’hyperkaliémie ou d’hyponatrémie où l’interprétation clinique est particulièrement sensible. Leur conclusion met en évidence une légère surestimation du potassium, dans laquelle l’hémolyse pourrait jouer un rôle, tout comme la nature des électrodes utilisées. Ils soulignent néanmoins que 97,5 % des points comparés n’entraînent aucune incidence sur la prise en charge clinique. Concernant le sodium, la sous-estimation apparaît un peu plus importante, avec 9,2 % des sodiums normaux sous-estimés en biologie délocalisée. Il est cependant intéressant de noter que, malgré leur conclusion globalement favorable sur la concordance, l’observation du diagramme de Bland-Altman montre une dispersion relativement large des points, entre les limites de ±1,96 SD, allant d’environ -2 mmol/L à +6 mmol/L. Les auteurs reconnaissent plusieurs limites à cette étude : une cohorte relativement réduite de 120 patients adultes ; l’exclusion des populations pédiatriques, des femmes enceintes et des patients d’oncologie ; l’absence d’étude sur les conséquences cliniques potentielles en termes de traitement, de durée de séjour, d’effets secondaires ou de devenir des patients. Une petite analyse est également réalisée sur le différentiel de délai de rendu des résultats, qui apparaît relativement faible, de l’ordre de 10 minutes. However, it should be noted that the laboratory involved appeared to provide exceptionally rapid result reporting. The authors’ practical conclusion is that any extreme value — potassium above 5.5 mmol/L or sodium below 130 mmol/L — should systematically prompt confirmation by central laboratory testing. Finally, as observed in the other publication analyzed in this literature review, the authors do not truly address the distinction between direct and indirect potentiometry, despite mentioning methodological and electrode-related differences.