Background: Blood sample hemolysis affects pre-analytical quality and may cause pseudohyperkalemia. We established a statistical model to estimate the corrected potassium (K+) in serum.
Methods: Serum K+ and H index were analyzed, and blood cell index was obtained from the examined Full Blood Examination (FBE) results. A linear-regression model was developed using hemolysis (H) index, K+ and covariates of blood cell index from 139 cell lysates of blood samples. The model was then validated against 26 in vitro physically hemolyzed serum samples.
Results: The final model selected H index, hemoglobin concentration (HGB), and hematocrit (HCT) as important predictors in estimating the K+ content. The model was validated against artificially hemolyzed serum samples, which returned a correlation of 0.942 between observed and predicted net K+ increase by hemolysis. The predictors H index, HCT, and HB contributed 93.7%, 3.5% and 2.8% to the model R2, respectively.
Conclusion: In vitro hemolysis induced pseudohyperkalemia could be accurately predicted and restored by our model for clinical application.
Keywords: H index; Hemolysis; Hyperkalemia; Linear regression.
Copyright © 2019 Elsevier B.V. All rights reserved.