A-145 Are EDTA and Citrate Collection Devices Interchangeable for Estimating Blood Platelets Values using Sysmex XN™Analyzers ?

Abstract

Abstract Background In the Sentara Healthcare System patient blood platelets are estimated using Sysmex XN™ analyzers. The blood specimens are collected in devices with EDTA anticoagulant. In the Emergency Departments, upon the first assessment of a patient a number of collection devices, including one with Na citrate anticoagulant, are obtained in case repeated testing is required. To prevent the collection of an additional EDTA device we investigated whether devices with Na citrate anticoagulant could be used interchangeably with EDTA devices for platelets values estimation. Methods Instruments: Sixteen Sysmex XN analyzers (Siemens) located in several institutions. Specimens: Seven-hundred nighty-eight patient blood specimens were obtained with both EDTA and citrate BD™ collection devices. The specimens, platelets numbers between 5000 and 800,000 platelets/mcL, were assayed in parallel within two minutes, after receipt in the laboratory, and then at intervals to reach 20 hours. The data were transferred electronically to Minitab® (version 21, Minitab Inc) statistical software. The paired data were analyzed with orthogonal, ordinary least squares (OLS), weighted polynomial ordinary least squares (WPOLS) and locally weighted scatterplot smoother (lowess) regression models and their graphic representations. The CLIA’s total error criterion (target value ± 25%) was used for evaluating acceptable variability. Results The parallel box plot of the differences between specimens as obtained with EDTA and those as obtained with Na citrate, showed that the distribution of the differences was similar between instruments and within the allowable total error of ± 25%. The orthogonal and the OLS models showed similar regression lines (Orthogonal: Y=3.3+0.94X: OLS: Y=-1+0.93X) and increase in variance for increasing platelets values. Consequently, a WPOLS model was employed for additional analyses. The ANOVA table of the WPOLS showed that there were no probabilistically significant differences between regression lines for either platelet’s values (P = 0.5) or specimen age (P = 0.1). Tridimensional surface plots clearly illustrated that the differences were distributed within -12.5% and the zero lines. The aptness of the WPOLS model was demonstrated by the standardized residual (std. res.) analysis showing quasi-normal distribution, with a few outliers (std. res. > |3|) and two influential observations for (Hi > 0.5) which were for platelets values less than 10,000 platelets/mcL. Conclusions The differences between EDTA and citrated specimens were within zero and -12.5%, this represented half of the CLIA’s criterion. The weighted polynomial regression model was employed due to the heteroscedasticity of the variance, The simultaneous evaluation by platelets values and the age of the specimen did not detect probabilistically significant differences by either platelets values or age of the specimen. This limited study suggested that EDTA and citrate anticoagulated specimens may be used interchangeably for patient care for specimens with platelets values between 10,000 and 800,000/mcL and within twenty hours from collection. However, the limited number of patients precluded to evaluate whether the detected outliers represented group(s) of patients with a particular medical condition and or pharmacotherapeutic treatment. Finally, the electronic collection and transfer of specimens to statistical software, such as Minitab, was crucial for performing this study.