Reply to Abildgaard et al.: lot variation and inter-device differences contribute to poor analytical performance of the DCA vantage™ HbA_1c POCT instrument in a true clinical setting

To the Editor,

We, Siemens Healthineers Point of Care, are writing this letter to address our concerns related to the article authored by Abilgaard et al. which was published in volume 60(1) page 127–34 of the Clinical Chemistry and Laboratory Medicine (CCLM). In this article, Abilgaard and team [1] provide a retrospective performance analysis of Siemens Healthineers DCA Vantage^® point-of-care device hemoglobin A_1c against the laboratory-based HPLC Tosoh G8 and G11 analyzers (Tosoh Bioscience, Tokyo, Japan). Unfortunately, we believe that the authors incorrectly conclude that the DCA Vantage Analyzer “does not comply with recommendations for analytical performance” and that variations from different reagent lots and instruments are the major contributors to their observation. They also recommend against the use of point-of-care testing unless no other method is available. We respectfully disagree with the conclusion that they reached because the utility of DCA HbA_1c test has been well-established at the POC clinical settings. Furthermore, as explained in detail below there are significant limitations to Abilgaard’s study design.

Most results analyzed in this study came from data collected retrospectively over approximately 2 years. The study compared data from two scenarios, one with a pair of central lab measurements and the other with a central lab and POCT measurement. Oddly, sample collection of the pairs occurred up to 48 h apart from each other. This time frame is uncommon, and while the authors have shown that time between samples collected on two lab devices is relatively stable from 0–2 days, this does not justify use of POCT and a lab device in this same manner. In our opinion, further validation of this method would have helped further clarify this point. Additionally, a single replicate of a fingerstick result is being used to estimate assay precision. This is well below the guidelines used to estimate assay precision of any kind and should be considered a serious limitation of this study. The statistical methods used here are also not standardized and have been put into practice on retrospective data that was not collected for this purpose. We also had questions about the use of the Dahlberg formula with this data, as the results collected are not true replicates for the POCT and lab devices. As stated in Kallner et al. [2], the Dahlberg formula is defined for measurements under true repeatability conditions that have not been met here. The non-canonical method of sample collection, along with unique statistical analysis, must be considered when reading this article and judging the author’s conclusions about assay performance.

Taking the above limitations into consideration, the results presented still do not show a clear-cut picture of assay performance. The original and adjusted CVs taken from the article tables are summarized below in Table 1. Data for the CVs estimated at 58 mmol/mol shows almost no change for any of the adjustments applied. In addition, the data from the Regional Hospital Central Jutland shows only a modest change when adjusted for instrument, but not for lot. Even though this analysis examined data from only three lots, if the impact of lot and instrument variation is as profound as the authors suggest, one would expect some improvement in CV. However, the difference is insignificant. Additionally, the smaller prospective study performed showed that in the clinically relevant range from 43–53 mmol/mol, a CV_A was estimated at 3.22% and calculated at 4.27%. This study was presumably done using a single lot of reagents and shows comparable values, especially for the calculated CV_A to the non-adjusted results presented in the study. This does not support the authors’ conclusions about lot-to-lot variation and does not support the use of their statistical method.

The authors state “the DCA Vantage method does not comply with recommendations for analytical performance.” We believe this is not supported by their own assessment that the evaluation was done in a non-standardized method that is not as robust as the 20 day precision studies recommended by the Clinical Laboratory Standards Institute. The DCA Vantage Analyzer meets all analytical performance specifications presented in the instructions for use and is cleared by the U.S. FDA and other regulatory bodies. Performance routinely meets the standards set by IFCC and NGSP and been peer-reviewed in over 600 publications. Among these, several studies have shown that the Siemens Healthineers DCA Vantage Analyzer meets HbA_1c performance requirements and has acceptable correlation with laboratory devices [3], [4], [5]. Several of these studies have shown that the use of POC devices in general help lower HbA_1c values in patients when compared to using laboratory methods [6, 7].

The authors of the article also argue that “a POCT result should only be requested in situations where no laboratory HbA_1c is attainable.” Multiple studies have shown the positive impact of point-of-care HbA_1c testing in diabetes care [8, 9]. This work shows POC testing improves multiple aspects of the clinical experience, including patient satisfaction resulting from spending less time in appointments and limited blood draw of fingerstick vs. venous sampling. It also improves patient adherence to treatment recommendations, as results are often available and discussed with the patient during the same appointment window. This is among many other attributes of point-of-care testing that have been disregarded by the authors, including increased access to care, as point-of-care devices allow for more opportunities to measure HbA_1c, especially in locations where access to clinics and central labs is limited [10].

In conclusion, we disagree with the evidence provided by Abildgaard et al. in suggesting the DCA Vantage HbA_1c assay has poor analytical performance, which is not supported by the literature. It is well-known that both lot and instrument, among other variables, will contribute to error, as is the case with any analytical device. The authors’ non-standardized assessment relies on data collected in a non-conventional manner, and we have verbalized our disagreement with their statistical analysis methods. We hope the authors will reconsider their adverse position on the performance of DCA Vantage Analyzer with the evidence presented in this letter. Point-of-care testing has improved significantly over the decades and has enhanced patient compliance, outcomes, and population management as well as healthcare systems overall.