BACKGROUND: It has been advocated that measuring the haemoglobin concentration in a haemolysed sample can be used to correct the potassium concentration. However, estimation of the degree of potassium adjustment has not been validated with patient-derived data. This study aimed to ascertain the relationship between haemolysis and potassium increment using patient samples. METHODS: A total of 35 hospital patients (16 male, 19 female; median age 73 years, inter-quartile range 45-83 years) for whom haemolysed samples were repeated within 24 h were included in the study. The difference between potassium in the haemolysed and non-haemolysed repeat sample was correlated with the serum haemolysis index measured on a Beckman Synchron LX20PRO. RESULTS: The linear relationship derived for patient samples showed a potassium increase of approximately 0.16 mmol/L for each increment in the haemolysis index. However, Bland-Altman statistics in the patient-derived model showed the variability (95% limit of agreement) in predicted potassium concentration to be +/-0.4 mmol/L. CONCLUSIONS: The magnitude of the relationship using authentic patient data with the Beckman LX20PRO haemolysis index is in keeping with previously described relationships. Use of the haemolysis index on the Beckman LX20PRO to predict potassium concentration in haemolysed samples is not sufficiently accurate to obviate the need for repeat analysis, but can be used to prioritise the need for urgent repeat analysis.
BACKGROUND: It has been advocated that measuring the haemoglobin concentration in a haemolysed sample can be used to correct the potassium concentration. However, estimation of the degree of potassium adjustment has not been validated with patient-derived data. This study aimed to ascertain the relationship between haemolysis and potassium increment using patient samples. METHODS: A total of 35 hospital patients (16 male, 19 female; median age 73 years, inter-quartile range 45-83 years) for whom haemolysed samples were repeated within 24 h were included in the study. The difference between potassium in the haemolysed and non-haemolysed repeat sample was correlated with the serum haemolysis index measured on a Beckman Synchron LX20PRO. RESULTS: The linear relationship derived for patient samples showed a potassium increase of approximately 0.16 mmol/L for each increment in the haemolysis index. However, Bland-Altman statistics in the patient-derived model showed the variability (95% limit of agreement) in predicted potassium concentration to be +/-0.4 mmol/L. CONCLUSIONS: The magnitude of the relationship using authentic patient data with the Beckman LX20PRO haemolysis index is in keeping with previously described relationships. Use of the haemolysis index on the Beckman LX20PRO to predict potassium concentration in haemolysed samples is not sufficiently accurate to obviate the need for repeat analysis, but can be used to prioritise the need for urgent repeat analysis.