Statistical correction of potassium in haemolysed samples

In response to the letter of Adel Ismail

We thank Dr Ismail for his interest in our publication and agree with his comments that caution is required with application of any method. Our basic premise is that mathematical corrections have uncertainties that can be modelled and rational decisions taken on the basis of the outcomes. A model for serum potassium could include the uncertainty in the ratio of red cell intracellular potassium to intracellular haemoglobin and make decisions about correction including that uncertainty. We would note, however, that deciding not to correct a result does not avoid this issue, as this uncertainty should be considered in setting haemolysis limits for release of the potassium. We also agree that in vitro haemolysis can void the correction and care needs to be taken with interpretation in this situation. It is then a balance between withholding very many valid potassium results caused by in vitro haemolysis to avoid a much smaller number of potentially misleading results which are incorrectly corrected. In one of our laboratories where potassium results corrected for haemolysis are released (GJ), they are always accompanied by a comment to be aware of intravascular haemolysis. The issue with delayed separation, and indeed the many other causes of artefactual hyper- and hypokalaemia, occurs with uncorrected as well as corrected results. In our paper, potassium was given as an example of the process of including a rational assessment of correction for interferences. While agreeing that care is required, we also believe that the use of a valid model of the uncertainties on results both with and without corrections allows a rational consideration of the different approaches.