Proficiency testing matters.

Far too many laboratories consider proficiency testing just a necessary evil, little more than periodic pass–fail exercises we perform solely to meet regulatory requirements. In addition, too many of us belittle point-ofcare (POC) testing as a passing fad, a technology so inferior to what we use in our own laboratories that it hardly warrants our attention. Clearly, a report that combines these 2 topics, such as the one in this issue of Clinical Chemistry (1 ), runs the risk of commanding little attention. That would be a very unfortunate mistake, because it has important lessons for all of us who practice laboratory medicine in our efforts to improve patient care. As Stavelin et al. (1 ) suggest, POC testing represents an important and growing segment of laboratory medicine. Many important clinical decisions are based on these technologies. Even if the CVs of POC testing, and indeed their accuracies, are not as good as centrallaboratory techniques, POC testing modalities can sometimes have a more positive impact on certain aspects of healthcare. A POC creatinine test that predicts a patient’s risk for contrast-induced nephropathy and is carried out in 10 min in the emergency room prior to radiologic imaging trumps (as long as it is “good enough”) a central-laboratory test that takes 30 min (or longer) to reach the caregivers (2 ). Even in nonurgent situations, the availability of such POC tests can be extremely valuable in preventing contrast-induced nephropathy (3 ). Among the more common, and more important, decisions that clinicians make on the basis of laboratory test results are those related to achieving therapeutic international normalized ratios (INRs) in the management of oral anticoagulants (specifically, vitamin K antagonists such as warfarin). In general, data indicate that POC devices can perform reasonably well analytically (4, 5 ). In addition, the use of POC devices in ambulatory settings can enhance patient care by allowing immediate feedback between physicians and patients (6, 7 ). Even more recently, the use of POC devices by patients themselves, analogous to how diabetic patients use glucose meters, has been shown to be effective (8, 9 ). But how do we know, in an ongoing way, whether POC devices are actually performing well (or poorly)? As is the case for the central laboratory, proficiency testing (PT) can shed some light. Doing measurements on blinded samples and then comparing them with the results of others should reveal, at a minimum, how close one’s results are to those of one’s peers. Although participating in PT may not be an absolute requirement for POC devices, many authorities, if not most, believe that PT in this setting is especially important (10 ). Even for central-laboratory techniques, traditional PT suffers from “matrix effects,” in that samples used for testing often react differently from native patient samples. Therefore, comparisons must be made only to peer groups, rather than to the “true value” (11 ). What if the peer group as a whole is wrong (12 )? The problem of matrix effects becomes even more complicated for POC devices, in which the sample typically used for patient testing is fresh whole blood, a sample type especially difficult to simulate in PT programs. In their report, Stavelin et al. (1 ) propose a novel method for performing PT for INRs with POC devices. In brief, for each of 3 different POC methods, they asked a relatively small number (approximately 20) of “expert” centers to measure 5 patient samples with their POC device in parallel with their centrallaboratory method. They then used these data to calculate a mean percentage bias for each POC method. At the same time, these centers and all the other participants in the program (approximately 1500), performed PT testing with standard, noncommutable PT material. The medians of the values from the “expert” centers became the POC method–specific target values. Stavelin et al. then plotted each participant’s performance on each (noncommutable) challenge on a graph in which the x axis represented the percent bias established by the “expert” laboratories for the POC method itself compared with the central laboratory method and the y axis represented the percent bias of the participant’s value compared with the peer group target value. The authors used this graph to define a set of regions representing different levels of performance. 1 Department of Pathology, Beth Israel Deaconess Medical Center, Boston, MA. * Address correspondence to the author at: Beth Israel Deaconess Medical Center, 330 Brookline Ave., Boston, MA 02215. Fax 617-667-4533; e-mail [email protected]. Received November 19, 2012; accepted November 20, 2012. Previously published online at DOI: 10.1373/clinchem.2012.197574 2 Nonstandard abbreviations: POC, point-of-care (testing); INR, international normalized ratio; PT, proficiency testing. Clinical Chemistry 59:2 335–337 (2013) Editorials