Abstract Background Hemolysis is one of the important preanalytical factor that can influence the laboratory test results. As arterial blood gas analysis (ABGA) is performed with whole blood, it is difficult to visually check hemolysis state of the specimen, and even blood gas analyzers cannot detect hemolysis. However, there is not enough consensus or guideline on which parameters are influenced with hemolysis. This study was aimed to identify the influence of hemolysis on ABGA results and develop a predictive model for hemolysis interference. Methods A total of 142 residual arterial blood specimens were collected from a teritary hospital in South Korea. Samples were aliquoted into three groups for mechanical hemolysis. Hemolysis was induced via 16, 22, and 26 Gauge needles and measured using the Profile pHOx Ultra Blood Gas Analyzer (Nova Biomedical, Waltham, MA, USA). The remaining blood was centrifuged, and hemolysis index was identified using plasma. The differences among groups were evaluated by one-way analysis of variance (ANOVA). In addition, the test results were divided into two sets: the training set and the test set. R studio (version 4.3.1) was used for the statistical analysis and the development of the predictive model. To select the optimal combination of parameters, we employed the backward elimination method. Results Among the parameters, pH and K values increased whereas pCO2, Na, Ca2+, and HCO3- values decreased. Values of pCO2, Hb, K, and Ca2+ increased as the degree of hemolysis increased with % biases exceeding the desirable bias. When the predictive model developed with training set was tested, the model that included pCO2, Hb, K, Cl, Ca2+, Lac, and tCO2 as variables demonstrated the top-flight performance. This model achieved the highest area under the curve (AUC) among all tested combinations (AUC=0.905). It exhibited a sensitivity of 89.6% and a specificity of 92.1%. Conclusion This study confirms that hemolysis influences the values of pH, pCO2, and K significantly. Furthermore, a predictive model was validated for detection of hemolysis interference in whole blood specimens used for ABGA. Therefore, the predictive model to detect spurious hemolysis should be adopted in ABGA instruments in clinical laboratories and POCT environments to reduce the analytical errors caused by hemolysis.
Kang et al. (Wed,) studied this question.