| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Briefs |
1 Servei de Bioquímica Clínica, Ciutat Sanitària i Universitària de Bellvitge, L’Hospitalet de Llobregat, 08907 Catalonia, Spain
aauthor for correspondence: xfa{at}csub.scs.es
Frequently it is assumed that the precision of a measurement system observed using commercial control materials is the same as would be observed using human samples without additives. However, warnings that this assumption may be erroneous were published almost 20 years ago (1).
To provide more information regarding this obscure point of metrology applied to clinical laboratory science, we have studied the interchangeability of day-to-day imprecision for commercial control materials and serum pools (Table 1
).
|
Measurements of hormones and troponin T were made with the Elecsys 2010 analyzer (Roche Diagnostics), and the remaining measurements were performed with the Hitachi 747 analyzer (Roche Diagnostics), according the manufacturer’s protocols in both cases.
For each measurement procedure, day-to-day imprecisions were estimated simultaneously with Unassayed Chemistry Control 1 and Unassayed Chemistry Control 2 (unassayed reference materials 731 and 732, respectively; Bio-Rad) as control materials and two serum pools from routine patient samples having values near the values obtained in the control materials. The final values of control materials and pools are given in Table 1.
Each control material was reconstituted previously, and each serum pool was divided into 20 aliquots and stored at -20 °C until measurement. Daily, over 20 working days, one measurement of each analyte was carried out in each of the control materials and serum pools. When 20 replicated results for each analyte were obtained, the corresponding variances and CVs representing day-to-day imprecision were estimated. All CVs are shown in Table 1.
To determine the interchangeability of the day-to-day imprecisions determined for the two types of materials, each pair of variances (corresponding to results obtained with the control material and the serum pool having similar values) was compared by the F-test. P <0.05 was regarded as significant. Results of these comparisons are shown in Table 1. These data suggest that the variances for the control materials and serum pools used in this study were interchangeable for measurements of cholesterol, phosphate, thyrotropin, thyroxine, free thyroxine, triglyceride, triiodothyronine, and urea; however, these variances across control materials and patient serum pools were not interchangeable for measurements of alanine aminotransferase, bilirubin, calcium, creatine kinase, creatininium, ferritin, glucose, or troponin T. In the case of measurements of aspartate aminotransferase, pancreatic 
-amylase, and protein, interchangeability was dependent on concentration. For those tests that did not demonstrate interchangeability, day-to-day imprecision, in general, was lower with control materials. Hohnadel et al. (1) reported similar results for cholesterol, glucose, and urea.
The lack of interchangeability between commercial control materials and serum pools regarding day-to-day imprecision may have adverse consequences in monitoring patient results. In fact, the detection of a significant change between two consecutive results of an analyte in a patient requires knowledge of the day-to-day imprecision associated with patient results (2). Because estimations of day-to-day imprecision are usually made with control materials, a lack of interchangeability in these cases diminishes the reliability of criteria developed to detect significant changes. In addition, a lack of this type of interchangeability can lead to misestimation of the uncertainty of measurement of a patient’s results (3).
With regard to quality control, noninterchangeability does not appear to be as important as in the above cases. However, if this noninterchangeability were found among control materials from different manufacturers or, worse still, among different lots of the same control material, monitoring day-to-day imprecision during long periods would be very difficult. Consequently, it is very important for the in vitro diagnostic industry to make an effort to produce control materials that have imprecision interchangeable with human samples and to give documented evidence of it.
Footnotes
Editor’s Note: A consultant reminds us that frozen serum pools may give misleadingly high CVs, e.g., if aliquots are not thoroughly mixed after thawing or if the chemical being measured is not stable in the pool.
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
