LipidFieldEvalProtocolLabvsCCPABiasRegressionPrec012510.docVIP

Polymer Technology Systems, Inc. CardioChek?PA Comparison Study Evaluation Protocol Bias Study, Precision and Linear Regression For application with comparisons to the Reference Laboratory Standard Recommended Evaluation Protocol – Bias plus Precision Scope This protocol provides direction for comparative study of the Cardiochek lipid panel to a reference laboratory. Expected Results Bias Evaluation Study – Average Difference Using a side-by-side evaluation, which employs the same venous blood sample for both methods, the expectation is that the average deviation of CardioChek results will be within ±10% of results from the reference lab result for cholesterol, ±12% for HDL and ±15% for triglycerides. This value gives a general expectation of the expected average deviation of a population of studies, but does not predict any individual deviation. This type of evaluation is done using a regression model (see #2). Bias Evaluation Study – Regression Model Linear regression is the accepted statistical approach to analyzing comparative data and defining the relationship of one method to another. In a regression model, the key performance measures are the slope of the regression equation line, the y-intercept (i.e., that point at which the regression line crosses the y axis of the graph) and the correlation statistic (r) which represents the degree of variability for points around the regression line. To analyze the regression one must use the slope and y-intercept to determine the overall variance from one method to another. Typically this is done at clinical decision limits, e.g., for cholesterol this is 160, 200 and 240 mg/dL. This is explained in Data Analysis Section below, on page 5. The correlation coefficient expressed as (r) should be greater than 0.88. Precision Study: Using a side-by-side evaluation, which employs the same venous blood sample for both methods, the expectation is that the Coefficient of Variation (CV) of CardioChek