A-331 Monitoring Operators Proficiency at Point of Care is Crucial for Reliable Patient Values. A Practical Example with Patient Blood Chloride, Potassium and Sodium Measurands Assayed with i-STAT®

Abstract

Abstract Background While i-STAT® cartridges offer the Physician a spectrum of analytical methods at the point of care for prompt diagnosis and interventions, these methods have to be harmonized with the laboratory methods in order to reliably detect shifts and trends. With this practical example we illustrate the importance of a program of routine comparisons between the i-STAT® and laboratory methods to detect analytical differences exceeding the CLIA’s criterion and potentially having adverse clinical implications. Methods Patient blood chloride (Cl), potassium (K) and sodium (Na) values as obtained by nurses with i-STAT Chem 8 ™ cartridges (Abbott Laboratories) at the point of care, where daily compared with the values as obtained with the Laboratory method (two cobas 6000®, Roche Diagnostics) using green top BD® lithium heparin tubes (Becton Dickinson) collected within thirty minutes of the i-STAT assay. The data were electronically stored in RALS™ (RALS Informatics) and transferred to Minitab® (Version 21, Minitab Inc.) statistical software. The two analytical methods were compared using the orthogonal and the polynomial ordinary least squares (POLS) regression models and their graphic representations. The aptness of the methods was evaluated with the standardized residuals diagnostics for normality, independence, outliers (Standardized residual >|3|) and influential observations (Hi>0.5). For acceptance of the differences the CLIA’s total error criterion was employed. (CI: target value ± 5%; K: target value ± 0.5 mmol/L; Na: target value ± 4 mmol/L). Results In the first four months of the study the number of differences exceeding the CLIA’s criterion was not acceptable. [Cl: 10% (165/1564); K: 7% (109/1593); Na: 4% (66/1586)]. These differences were occurring throughout the analytical range (AMR). Consequently, the operator’s technique was suspected and the POCT senior technologists (CA, FA) adopted operator specific interventions. This strategy was effective in the following month and the improvements were maintained for seven consecutive months. Monthly differences exceeding the CLIA’s criterion decreased significantly [Cl:1.9% (52/2777): K:0.9% (25/2868): Na:0.5% (15/2896)]. This was clearly illustrated with the dot plot by date and the parallel box plots by month. Regression analysis showed that the orthogonal and the OLS models were equivalent. Cl: Orthogonal y=1.7+0.99x, OLS y=12+0.88x. K; Orthogonal y=0.2+1.2x, OLS y=0.1+0.96x. Na: orthogonal y=3+0.98x, OLS y=14+0.9x. Additionally, the POLS model showed a linear relationship within the AMR intervals with few outliers (Cl=11; K=7; Na=9), no influential observations (Hi<0.5) and equality of monthly regression lines (P>0.05). Conclusions The implementation of daily comparisons between patient values as obtained with the i-STAT method and those as obtained by the Laboratory method in the interval of thirty minutes, allowed the identification of individual operators requiring additional training. This strategy ensured reliable performance for eight consecutive months. This study clearly showed that the addition of continuous supervision and training of operators to quality control and quality assurance practices, ensured the interchangeability of patient values as obtained with i-STAT and those as obtained with the Laboratory method. Electronic collection of data and analysis with appropriate statistical software, such as Minitab, was crucial for the implementation of this program.