Multiple Calibrator Measurements Improve Accuracy and Stability Estimates of Automated Assays
Calibration curves are used to convert measured signals—such as light, electrical current, fluorescence, or radioactivity—to concentrations of substances in blood samples. Prior to the development of automated chemistry analyzers, calibration curves were measured with every manual assay using a series of standards with well-established concentrations.
Today, calibration curves are established initially, and then automated analyzers are used to calculate the concentrations of hundreds or thousands of subsequent blood specimens from patients. Any variations in the establishment of the stored calibration curve are propagated into the numerous patient results derived from that curve. It has been hypothesized that extra care in the establishment of the stored calibration curve can help improve the patient results and that performing multiple calibrator measurements can improve the overall accuracy of an individual assay.
In a recent study, Mayo Clinic researchers investigated the effects of combining multiple calibrations on assay accuracy and measurement of calibration stability. The study was published in the Scandinavian Journal of Clinical and Laboratory Investigation.
“By performing multiple measurements of the reference standards over several days, and using the average of these measurements for the stored calibration curve, we can improve the accuracy of individual assay results and the accuracy of composite measurements such as assay stability estimates. Our study sought to confirm this,” said George Klee, M.D., Ph.D., Emeritus Consultant in Mayo’s Department of Laboratory Medicine and Pathology.
Researchers assessed calibrations for total triiodothyronine (TT3), vitamin B12, and luteinizing hormone (LH) using Beckman Coulter’s Access 2 analyzer. These assays were chosen based on how their signal measurements differ.
“The mathematical relationship between the measured signals and the concentrations are different for different types of assays. In some assays, like those used for LH, the signal increases progressively with concentration; whereas for other types of assays, like those used for TT3 and vitamin B12, the signal decreases with higher concentrations. We wanted to compare the effects of multiple calibration measurements on both types of assays,” said Dr. Klee.
The automated analyzer in the study used six standards of specified concentrations to establish a stored calibration curve. The manufacturer’s standard calibration procedure measures these calibrator levels in duplicate on the same day and uses the average for each level to establish the calibration curve, which was referenced as CC1 during the study. Researchers also evaluated two other calibration procedures: utilizing multiple measurements over two days (CC2) and over three days (CC3).
The study results show that all three assays were more accurate at time zero when using CC3 and more consistent when measuring their change over time including the competitive assays (TT3 and vitamin B12) which had positive time-regression slopes, and the LH assay which had a negative slope.
“These results indicate that the investment of a little extra care in the establishment of the stored calibration curve for automated analyzers utilizing replicate measurements over three days can improve the accuracy of patient laboratory test results and the accuracy of composite measurements such as stability estimates,” stated Dr. Klee.
The study also found that using the intercept of the stability regression of concentration versus time for the time zero baseline concentration increases the accuracy of stability estimates.
According to Dr. Klee, there are two ways to increase the accuracy of stability estimates. First, the best possible estimate of the value at time zero can be used as the baseline for determining the change over time. In this study, an accurate estimate was determined through the use of the CC3 calibration procedure. Alternatively, a regression fit can be made from all of the values obtained during the stability study. The resulting estimate of the actual result at time zero is accurately determined as the regression’s intercept. When the intercept is used as the baseline for determining the change over time, the stability measurement reverts to the measurement of percent change over time. The increased accuracy in the stability estimate however is not caused by measuring percent change but rather by using the regression intercept as a more accurate estimate of the time zero value.
“This specific finding could influence the way all in vitro diagnostic reagent stability studies are analyzed,” added Dr. Klee.
According to Dr. Klee, this study has potential applications for improving many laboratory procedures. However, since the instrument manufacturers generally prescribe specific calibration procedures in their FDA approved protocols, the real application of these techniques generally would depend on their incorporation into the manufacturer’s calibration protocols for patient care measurements.