The Centers for Medicare & Medicaid Services (CMS) proposed changes in 2019 to the Clinical Laboratory Improvement Amendments of 1988 (CLIA) regulations for proficiency testing. It was proposed that HbA1c be a regulated analyte, i.e., any laboratory in the US using test methods that are not CLIA-waived for HbA1c would need to enroll in and successfully pass PT with acceptance limits (AL) of ± 10%. Since releasing the proposal, various professional groups have expressed concerns that ± 10% AL would be too wide to maintain already improved analytical accuracy and precision performance, therefore, the clinical utility of HbA1c testing in diabetes mellitus management and diagnosis. Then what would be an appropriate PT allowable limit for HbA1c?
The CMS-proposed AL of ± 10% was based upon simulated outcomes generated by CLIA-approved PT programs using various potential ALs. The goal was to identify an AL that provided overall unsatisfactory event score rates of less than 2%. An unsatisfactory event score occurs when more than one of five results are outside the AL. The simulation studies were performed retrospectively using PT results from previous challenges over a 2-year period. But, the proposed rule did not release detailed information on how the ± 10% AL was generated, which specific PT program providers were involved, what criteria or acceptance limits were applied, and how many challenge samples were used per event. With these questions, we carried out a study aimed to assess impact of various ALs on laboratories’ PT performance, therefore, to identify an appropriate AL as a PT evaluation criterion. The study was carried out in collaboration with Shanghai Center for Clinical Laboratory (SCCL). Ten PT samples prepared from de-identified left-over patient’s EDTA-whole blood specimens. The samples from patients with hemoglobinopathies, renal failure and pregnancy were removed. Prior to distribution, patient specimens were stored at -70 °C and thawed, grouped and pooled based on their HbA1c concentrations, mixed to homogeneity, and aliquoted into 0.5 mLs to produce PT samples. The PT samples were distributed at 2-8 °C to 318 and 336 participant laboratories for the 2018 and 2019 PT events, respectively, of the SCCL PT program. The results were reported to SCCL via a secured internet system. Homogeneity and stability of the PT samples were assessed by following the assessment procedure and found meeting the criteria by ISO13528:2015. Participant laboratories measured HbA1c concentrations using ion-exchange and boronate HPLC-based methods and Immunoturbidimetric-based test kits. The target values of HbA1c were determined at SCCL using 4 IFCC Secondary Reference Measurement Procedures (SRM).
We evaluated the results using the criteria of SRM-defined target mean ± AL (5%, 6%, 7%, 8%, 9% and 10%, respectively) for each sample and defined results outside the target ± AL as Failure Results. Laboratories with Failure Results ≥ 2 out of 5 samples for a PT event obtained Event Unsatisfactory Status according to CLIA ’88. The laboratories’ Event Unsatisfactory rates were calculated by dividing the number of laboratories with Event Unsatisfactory Status by the number of laboratories using the same method-system (or peer- group) for a PT event, generating Event Unsatisfactory Rates sorted based on method group, and Overall Laboratory Event Unsatisfactory Rates. We found that at AL of ± 5%, Overall Laboratory Event Unsatisfactory Rates were the highest, i.e., 11.3% and 12.2% in the events 2018 & 2019, respectively, and reduced by about a half (5.3%, 4.8%) at AL ± 6%, and by about another half and more (3.1%, 0.9%) at AL of ± 7%, then reduced to further (2.2%, 0.6%) at AL of ± 8%, and the lowest rate (1.4%, 0.6%) at both ± 9% and ± 10% ALs. These observations suggested the criteria of target ± 6% AL (same as CAP’s current criterion) or ± 7% AL for PT evaluation of HbA1c results would be reasonable. Using either of these criteria, about 95% of participant laboratories would obtain a passing score for a PT event. (Detailed contents are published in Clin Chem Lab Med. 2021 Jul 6. doi: 10.1515/cclm-2020-1311)