Correcting potassium concentration of hemolyzed samples

Hemolysis, the breakdown of erythrocyte (red-cell) membranes, can occur in vivo as a result of disease, or more commonly in vitro when it is most usually a consequence of poor blood collection or sample handling technique. 

Hemolysis is of significance for laboratory staff because it results in the release of substances normally held within the red cell (e.g. hemoglobin) to plasma, and thereby a range of spurious effect on the results of plasma/serum measurements.

Of all routine blood tests plasma/serum potassium measurement is one of the most sensitive to the effect of hemolysis because red-cell potassium concentration is so much higher than that of plasma (approximately 20 times higher); hemolysis causes a spuriously high plasma potassium concentration. 

Serum/plasma samples submitted for potassium measurement are quite frequently rejected for analysis because they show the telltale signal of hemolysis, i.e. a reddish discoloration due to the presence of hemoglobin. Modern automated analyzers have the capacity to assess the hemoglobin content of plasma, thereby providing an objective measure of the degree of hemolysis (expressed as the hemolytic index, HI) for all plasma/serum samples. 

This new capacity has provoked studies aimed at investigating whether it is feasible to use HI to predict the "true" potassium concentration of hemolyzed samples. The most recent of these was conducted at a clinical laboratory in Plymouth UK. Correlation between potassium concentration and HI (µmol/L) was determined for 613 adult samples and for 523 neonatal samples. 

The regression equations were similar for both adults and neonates and allowed the conclusion that on average potassium of both adult and neonatal plasma/serum samples increases by 0.01 mmol/l for every HI unit increase. The derived equation used to compensate for potassium release from hemolyzed samples is:

In separate experiments it was demonstrated that detection of hemolysis by visual inspection is much less reliable for icteric samples (i.e. most neonatal samples) than for non-icteric samples.