Commutability of external quality assessment materials for point‐of‐care glucose testing using the Clinical and Laboratory Standards Institute and International Federation of Clinical Chemistry approaches

Abstract Objectives The aim of this study was to assess the commutability of three external quality assessment (EQA) materials for point‐of‐care (POC) glucose testing using two approaches, to identify suitable EQA materials to evaluate and monitor the quality of POC testing. Methods Commercial control materials (CCMs), pooled human serum samples (PHSs), and homemade human whole‐blood samples (HWBs) were measured along with 33 individual clinical samples using five POC instruments and a Hitachi 7600 analyzer. Data were analyzed by Deming regression analysis with a 95% prediction interval as described in Clinical and Laboratory Standards Institute (CLSI) EP30‐A, and by difference in bias analysis as described by the International Federation of Clinical Chemistry (IFCC) Working Group on Commutability. Results Using the CLSI approach, HWBs, CCMs, and PHSs were commutable with five, one, and two instruments, respectively. With the IFCC approach, HWBs were commutable with two instruments, while CCMs and PHSs were largely inconclusive or non‐commutable on five instruments. Conclusions HWBs were commutable on all instruments by the CLSI approach and may be a suitable EQA material for POC testing. Although some results differed between the IFCC and CLSI approaches, both indicated that HWBs were far superior to CCMs and PHSs in commutability.


| INTRODUC TI ON
Point-of-care testing (POCT) is a popular means of providing laboratory testing at or near the site of patient care. It has become an important component of laboratory medicine by virtue of its portability and ease of operation by non-laboratory personnel or patients themselves. [1][2][3][4][5] Point-of-care (POC) glucose testing plays an important role in the treatment and management of diabetes mellitus, enabling strict glycemic control and creates opportunities to increase the efficiency of clinical services to improve patient outcomes. 6,7 Most analytical methods use one of three enzymatic reactions to quantify glucose: glucose oxidase (GOD), glucose dehydrogenase (GDH), or hexokinase/glucose-6-phosphate dehydrogenase (HK). In these systems, enzymatic activity produces an electrical current or color change proportional to the glucose concentration. Isotope dilution gas chromatography-mass spectrometry (ID-GC/MS) serves as a higher-order procedure in reference laboratories, while the HK method is widely accepted for routine calibration and accuracy evaluation. 8 Stringent accuracy assessment criteria for both self-and hospital-based blood glucose monitoring have been proposed by many international organizations, including the International Standardization

Organization (ISO) and the Clinical Laboratory and Standards
Institute (CLSI). [9][10][11][12] However, in clinical application, the accuracy of POC glucose testing remains unsatisfactory. Several studies have described variability in measurements made by different POC glucose instruments or between these instruments and central laboratory analyzers, [13][14][15][16] mainly due to the lower specificity of the enzymes used (GOD and GDH), which make them susceptible to interference. 17,18 External quality assessment (EQA) is crucial to ensure the continuous high quality of medical laboratories. Commutability is required to be able to use EQA results to evaluate the performance of participating laboratories, as it enables measurement standardization. The International Vocabulary of Metrology defines the commutability of a reference material (RM) as close agreement between the measurements of a stated quantity of the material obtained by two different measurement procedures (MPs), as well as agreement between patient sample (PS) measurements. Miller et al have suggested an EQA scoring system with six categories, based on the ability of an EQA to evaluate participant and instrument performance. 19 Category I is the most desirable, as programs in this category use commutable samples with target values established by a reference system, and can evaluate both individual laboratories and MPs for reproducibility, calibration traceability, and uniformity between laboratories and between MPs. Particularly for EQAs, the lack of commutability of applied samples is internationally recognized as one of the major hurdles in achieving a Category I POC glucose testing, 6,19 as it often impedes interpretation. 20,21 Because evaluating the commutability of EQA materials requires consistent sample typology (capillary samples) and stringent requirements that are difficult to apply, a pragmatic evaluation approach is required to ensure the correct interpretation of results provided in POC EQA reports. The aim of this study was to assess the commutability of three types of EQA materials by two different approaches, and to define suitable EQA materials to evaluate and monitor the quality of POC glucose testing.

| Study design
As EQA materials, we evaluated commercial control material (CCM), pooled human serum (PHS), and homemade human whole blood (HWB), all at three concentrations (denoted 1-3), using five POC instruments and a laboratory-based analyzer.

| POC glucose instruments
Five different mainstream-brand POC glucose instruments were evaluated in this study (Table 1). Each POC instrument was operated and performed according to the specifications of its manufacturer. We performed one run with each instrument using one lot of strips and internal control materials, and these measurements were within the specified limits, indicating that all instruments were stable throughout the analysis period.

| Ethics statement
Because the study used anonymized leftover clinical samples, it did not require the consent of an ethical committee or review board.

| Measurements
PSs and the three EQA materials were measured with five POC instruments and the Hitachi 7600 analyzer on the same day. All samples were adequately mixed at room temperature before analysis and measured in triplicate; for the EQA materials, three replicates were performed on each instrument. Samples were evaluated by the instruments in a set order, and the elapsed time between the first and last measurements was <30 minutes. All measurements were performed in a laboratory setting with controlled room temperature (23 ± 5°C) and humidity, according to the manufacturers' specifications.

| Data analysis
Microsoft Excel 2013 (Microsoft) was used to process the data, using formulas provided in the CLSI EP30-A and IFCC WG on Commutability documents. Outlier values were excluded based on CLSI EP30-A section 6.3.5: exclusion of data and handling of outliers in Part 2 of the IFCC document. 22,24 Of the 33 PSs, 30 were suitable for statistical analysis.

| Precision and comparability of different instruments
To evaluate the precision of each POC instrument, within-run coefficients of variation (CVs) were calculated using triplicate measurements of PSs. Passing-Bablok regression analysis was used to estimate the slopes and intercepts of each of the POC instruments vs the Hitachi 7600 analyzer, and the Spearman rank correlation coefficient was also calculated.

| Commutability assessment
Two different approaches were used for commutability evaluation.
Difference plots were generated separately for comparisons between each POC instrument and the Hitachi 7600, and logarithm-transformations were determined if scattering increased with concentration.
1. According to CLSI EP30-A, the log 10 -transformed results of PSs were analyzed by Deming regression analysis. A 95% PI around this regression line was calculated using the formulas described in CLSI EP30-A Appendix C and was plotted along with the log 10 -transformed results of the three EQA materials. When the result of each EQA material fell within the 95% PI it was regarded as commutable; otherwise, it was considered non-commutable. 22 As the materials in this study are used as EQAs, we have defined results touching the PI as commutable.

| Precision and comparability of different instruments
As shown in Table 2, the median within-run CVs of the five POC instruments varied from 1.36% (the HORIBA LP-150C) to 4.13

(the StatStrip Xpress). Passing-Bablok slopes and intercepts and
Spearman rank correlation coefficients for each POC-Hitachi 7600 comparison are also shown in

| Commutability of the EQA materials according to the CLSI approach
Commutability assessments of the three EQA materials according to the CLSI approach are shown in Figure 1. CCM-1, -2, and -3 were commutable on 3/5, 2/5, and 4/5 instruments, respectively. PHS-1 -2, and -3 were commutable on 4/5, 3/5, and 5/5 instruments, respectively. HWBs at three concentrations were commutable on all five POC instruments, exhibiting the best performance among the three EQA materials by this approach.

| Commutability of the EQA materials according to the IFCC approach
Commutability assessments of the three EQA materials according to the IFCC approach are shown in Figure 2. HWB-1, -2, and -3 were commutable on 3/5, 4/5, and 3/5 instruments, respectively, while CCMs and PHSs were inconclusive or non-commutable on all five POC instruments. All three HWB concentrations were commutable on the ACCU-CHEK Performa and HORIBA LP-150C. Abbreviations: CI, confidence interval; CV, coefficient of variation; N/A, not applicable.

F I G U R E 1
Commutability of the three EQA materials using the CLSI approach. Commutability assessment of the three external quality assessment (EQA) materials (commercial control materials (CCMs), pooled human serum samples (PHSs), and homemade human whole-blood samples (HWBs) according to Clinical and Laboratory Standards Institute (CLSI) EP30-A. 22 The glucose levels of the EQA materials and patient samples (PSs) were measured with five point-of-care (POC) instruments and a Hitachi 7600 analyzer. The log-transformed results measured by the Hitachi 7600 and the POC instruments are plotted on the x-and y-axes, respectively. Solid and dashed lines represent the regression lines and the limits of the 95% PIs of Deming regressions, respectively. The black circles represent the log-transformed results of the PSs, and the blue squares, green triangles, and red circles represent the log-transformed results of the HWBs, CCMs, and PHSs, respectively instruments using the CLSI approach, but produced non-commutable results using the IFCC approach.

| D ISCUSS I ON
The use of POCT in laboratory medicine is evolving at an increasing rate, with progressively more medical treatment decisions made based on it. Therefore, it is crucial to conduct EQAs to assess the accuracy and clinical reliability of POCT. 29 If an EQA is category I, the consistency of results between different measuring systems can be assessed using a true value, which would improve the harmonization and standardization of POCT. However, a main issue for EQA organizers is the scarcity of commutable EQA materials that are compatible with different POC instruments. 21 This study aimed to assess the commutability of three EQA materials using five POC glucose instruments and a central laboratory platform through two different approaches, to identify EQA materials that are as similar to native PSs as possible.
Before assessing the commutability of the three EQA materi- (large sample-specific differences). 24 In addition, poor commutability characteristics can be caused by the nature of the analyte and its matrix, and the concentration can also affect the uncertainty. 35 Based on this study, we suggested increasing the number of replicate measurements of the EQA materials to reduce the uncertainty.
A major limitation of the study was using the Hitachi 7600 analyzer as the comparative method. According to the IFCC's latest recommendations on commutability assessment, the results of each routine method should be compared with those obtained using a higher-order reference method. Although the HK method is still listed as a reference method for glucose measurement, ID-GC/MS may provide a better reference point.
In conclusion, compared to CCMs and PHSs, HWBs had better commutability characteristics with mainstream POC glucose instruments by two different approaches, indicating that they are suitable EQA materials to evaluate and monitor the analytical quality of POC glucose testing. Furthermore, the results suggest that the IFCC approach for commutability evaluation should be used when selecting EQA materials for POCT.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.