Statistical correction of potassium in haemolysed samples

I read with interest the suggestion by Jones and Hawkins of applying a statistical model to quantitate and correct, using Microsoft Excel, the effect of interference of haemolysis (haemoglobin concentration) on potassium measurements. ¹ The authors suggest that correcting this form of interference would allow the release of more potassium results. They also emphasise that this approach is limited to the situation where the concentration of the interferent is measurable, with a linear relationship between concentration of interferent and its effect on the analyte of interest. The proposed mathematical model is well thought and properly constructed.

However, red blood cell indices such as mean corpuscular haemoglobin (MCH) and mean corpuscular haemoglobin concentration (MCHC) (per cell) are also relevant to this model. These parameters reflect the extent to which red blood cell intracellular haemoglobin content varies in normal individuals and in patients with anaemia. Lack of concordance between intracellular potassium content and haemoglobin in such cases is likely to manifest as different degrees of haemoglobin concentration in haemolysed samples, making potassium correction potentially inappropriate.

There are two further issues of concern. First, correcting potassium when haemolysis is in vivo as opposed to in vitro would not only be inappropriate but clinically misleading as well. The possibility of in vivo haemolysis should be suspected when haemolysis persists in repeated samples despite precautions during and after venepuncture. ² Second, delays in separating serum/plasma from cells can cause significant inaccuracy in potassium measurement without any visible haemolysis. When potassium leaks out of the cell without haemoglobin, measured potassium may be normal when in vivo it is low or may be high when it is normal. ³ These examples illustrate the importance of applying such a statistical approach with caution to individual samples.

Analytically, potassium can now be measured with accuracy of ±0.1 mmol/L. A robust policy focusing on eliminating all sources of pre-analytical error would help to secure the veracity of reported potassium concentrations.