Clinical and Research Implications of Serum Versus Plasma Potassium Measurements

Potassium is a very commonly measured electrolyte across many fields of medicine, including cardiology. The normal range of serum potassium is typically cited as 3.5–5.0 mmol/L. However, several recent Danish publications suggest that the optimal range of serum potassium is much narrower than the normal range in several conditions including acute heart failure after myocardial infarction1 and chronic heart failure.2 These papers — which use data through 2012 — report serum potassium; however, beginning in 2010, some Danish laboratories changed from measuring serum potassium to plasma potassium, thus there was likely a mixture of serum potassium and plasma potassium values in the dataset used by the researchers in Denmark. This introduces some uncertainty regarding the potassium value for the nadir of risk in the U-shaped relationship between serum/plasma potassium and risk. Similarly, in the healthcare system of Stockholm region some centres measure potassium in serum while others measure potassium in plasma, and some centres have changed from one to the other in recent years.3 Having both values appear in clinical and research databases may impact research in this area and has implications for clinical care. This would be a trivial matter if not for the often overlooked fact that the optimal range differs depending on whether the measurement is derived from a serum or plasma sample. Serum measurements require blood clotting before analysis while plasma measurements are done directly after drawing the sample. Potassium is continuously released from cells during clotting. Therefore, imperfect sampling conditions, such as traumatic venipuncture, temperature variation, and prolonged time from collection to processing, can result in higher serum potassium levels, but would have less effect on plasma potassium levels. However, haemolysis and stasis also affect plasma potassium. This realization potentially diminishes the impact of any results or conclusions drawn from datasets which use clinically-derived potassium values, particularly datasets which have a combination of both serum and plasma samples. The Nordic Reference Interval Project (NORIP) reported in the year 2000 that the normal reference range for serum potassium was 3.6–4.6 mmol/L and the normal reference range for plasma potassium was 3.5–4.4 mmol/L.4 NORIP investigators performed additional analyses in which they excluded samples that deviated from results from the same individual, and found that the reference values remained unchanged, suggesting haemolysis did not substantially impact these results.5 These optimal values for serum and plasma potassium are similar, but NORIP was considered to have optimally collected and processed samples which minimized clotting effects on potassium, and excluded patients from whom it was difficult to draw blood samples. Samples were collected according to strict standards and instructions, as documented previously.6 Very early work reported that in real-world clinical conditions, potassium concentration determined from a coagulated blood sample can be 0.1 to 0.4 mmol/L higher than that measured in an anticoagulated plasma sample.7 Laboratories can quantify haemolysis but cannot correct for this and other factors that impact sample integrity.8 In the field of heart failure, potassium regulation within a normal range is particularly important, as both high and low potassium levels are associated with worse outcomes.9 In addition, treatments for heart failure, including diuretics and renin–angiotensin–aldosterone system (RAAS) inhibitors, and co-morbidities in heart failure, such as chronic kidney disease, all affect potassium levels. Importantly, treatments are often altered in response to potassium levels, such as replenishing potassium, discontinuing RAAS inhibitors, or using potassium binders. Thus, failing to consider the source (plasma or serum) of potassium determination may result in unnecessary or even inappropriate treatment decisions. Even with a solid understanding of the difference between serum and plasma potassium levels, and clinician awareness of this issue, there is still difficulty interpreting potassium values, as these values cannot simply be converted between serum and plasma. To demonstrate the average difference between serum and plasma values, we examined blood samples from 720 individuals in the NORIP database who had potassium measured in both serum and plasma, and plotted their potassium levels. We found a difference in variability across the spectrum of potassium values, with more variability with higher potassium values (Figure 1). However, there is limited utility of correlating between values. This correlation was derived from the NORIP database, where samples were collected under ideal conditions; the correlation would likely differ if calculated from samples obtained under imperfect conditions. Further, given the variability in sample collection conditions which are not routinely quantifiable, correction factors would not be reliable in routine clinical practice. Potassium values differ between serum and plasma samples, and our findings suggest that serum values are typically slightly higher. While accurate potassium values are important in both clinical practice and clinical research, converting between values is difficult and has limited utility. Moving forward, clinicians should be aware that there may be a difference between serum and plasma levels, and local practices may differ with respect to the use of serum or plasma for laboratory assessments. Furthermore, researchers may want to consider analysing blood samples from either plasma or serum, not both, to maintain consistency of results and maximize the impact of their research. Conflict of interest: L.B.C. reports receiving research funding from Abbott. G.S. reports receiving research grants from Boehringer Ingelheim, Merck Sharp & Dohme, Swedish Heart and Lung Foundation, and Italian Society of Cardiology. J.J.C. reports receiving research grants from AstraZeneca and Vifor Pharma. U.D. reports receiving speaking and consulting honoraria from Novartis and grant support from AstraZeneca. L.H.L. reports receiving research grants from Relypsa; consulting fees from Relypsa, Vifor Pharma, and AstraZeneca; and speaking for Vifor Pharma and AstraZeneca. All other authors have no conflicts of interest to declare.