Verification of automatic analysers Roller 20PN and iSED for measuring erythrocyte sedimentation rate

Introduction Automated erythrocyte sedimentation rate (ESR) analysers are based on different methodology than Westergren method. It is questionable whether ESR values obtained from those analysers are comparable with determined values with Westergren method. The aim was verification of the precision, method comparison and accuracy of automated ESR analysers: Roller 20PN (Alifax S.p.A., Polverara, Italy) and iSED (Alcor Scientific, Smithfield, USA). Materials and methods Blood samples (N = 752 for Roller 20PN and N = 213 for iSED) were sampled into K2EDTA (Kima, Italy) tubes for automated and 3.8% Na-citrate tubes (Kima, Italy) for Westergren method. The data was divided into three groups according to the ESR values obtained with the Westergren method: Group Low (L) (ESR ≤ 20 mm), Group Medium (M) (ESR 21-60 mm), and Group High (H) (ESR ≥ 61 mm). Method agreement was assessed by Bland-Altman analysis and Passing-Bablok regression. Results Analyser iSED has shown better comparability with Westergren method (bias 0.0 (95%Cl -1.4 to 1.5) range than Roller 20 PN (bias = - 6.4 (95%Cl - 7.1 to -5.7) in the whole measuring. For Roller 20 PN, Passing-Bablok regression has shown constant and proportional difference for Groups L and M, and for iSED only for Group H. Roller 20 PN had lower sensitivity (0.51 (95%Cl: 0.45-0.57) than iSED (0.72 (95%Cl: 0.59-0.80) while they had comparable specificity (> 0.90) and accuracy (≥ 0.80) in comparison with the Westergren method. Conclusion Both analysers are not comparable with the Westergren method and should not be used interchangeably.


Introduction
Erythrocyte sedimentation rate (ESR) is a widely used assay which is still among the top 25 most common laboratory tests (1). Because of its simplicity and use for monitoring inflammatory, autoimmune and malignant diseases, it still has clinical use. Many physiological and pathophysiological causes can increase the ESR values which make this test nonspecific, e.g., in pregnancy, for women who have menstruation, at an older age, arthritis and inflammatory bowel disease (2). The "gold standard" for the determination of ESR is the Westergren method which is standardized but timeconsuming (3). Westergren method uses whole blood sample that is diluted with a liquid citrate anticoagulant (4:1) and the value of ESR is determined after one hour in a vertically placed tube. Furthermore, for performing ERS with the Westergren method larger sample volume is needed. In order to shorten the time of the ESR measurement, there are modified Westergren methods that have certain modifications (e.g., shorter turnaround time, use of non-diluted samples) and alternate methods that are based on a different methodology. The survey conducted by International Council for Standardization in Haematology (ICSH) indicate that 2/3 of all laboratories worldwide use modified or alternate ESR test method (3). These alternate ESR methods can be based on photometric rheology which measures Rouleaux formation (transmitted or reflexed light intensity depending on the duration of erythrocyte aggregation) or centrifugation (3). The obtained results by the alternate method are mathematically transformed into values comparable with the Westergren method (3,4).
Not only does the alternate methods reduce the turnaround time to 30 minutes or less for measurement of ESR but they also have more advantages, e.g. reducing possibility of a human error (the possibility of misreading ESR values or wrong transcript on the laboratory report). Furthermore, they reduce the cost of the blood collection devices and a volume of blood required for analysis, because the ESR analysers use ethylenediaminetetraacetic acid (EDTA) whole blood. The use of EDTA blood also reduces the possibility of errors, due to the dilution of the sample with citrate anticoagulants and decrease the risk of exposure of laboratory technician to possible infectious blood pathogens, during sample handling (2). Several automated or semi-automated methods are available for estimating ESR values and consequently, various ESR analysers have been developed (3). Over the last 20 years there were numerous verification and comparison studies of several automated analysers with Westergren method. Several studies have already investigated analytical performances of analysers iSED (Alcor Scientific, Smithfield, SAD) and TEST1 (Alifax S.p.A, Polverara, Italy) (5-7). Analyser TEST1 is widely used for the measurement of ESR, while the newly developed Roller 20PN (Alifax S.p.A, Polverara, Italy) adheres to similar technology but compared to the earlier models has slightly different performance of measuring ESR (8). Moreover, because of the altered methodology of auto-mated ESR analysers that are increasingly used in routine practice instead of the Westergren method, the verification of these methodologies is highly recommended by ICSH. Additionally, it is questionable whether the obtained ESR results of these automated ESR analysers are even comparable to the Westergren method. Furthermore, the performance and comparison of two automated ESR analysers, Roller 20 PN and iSED, have not yet been investigated.
The aim of this study was therefore to perform the verification of the precision and accuracy of two automated ESR analysers: Roller 20 PN and iSED and to compare these two automated analysers with Westergren method and with one another.

Subjects
This verification study was done from September to December 2017, in the Department of medical laboratory diagnostics in University Hospital "Sveti Duh". All patients for whom tests for ESR and complete blood count have been ordered were included in this study and their leftover blood samples were used for analysis. The exclusion criteria were not used because we wanted to include as many participants available to cover a wider range of ESR values. Furthermore, all included patients represent a representative population of our laboratory. For each patient two blood tubes were used: one 1,6 mL tube with 3,8% sodium citrate (Kima, Piove di Sacco, Italy) and other 3 mL tube with K 2 EDTA (Kima, Piove di Sacco, Italy). Samples were analysed within 4 hours from arrival to the laboratory. The study was done with the approval of the hospital Ethics Committee for using leftover blood samples. . Between-run precision was determined in triplicate for 5 consecutive days. For within-run, control samples and leftover blood samples (6 samples for Roller 20PN and 7 samples for iSED) were analysed 20 times in a row in one day. Between-and within-run precision was expressed as coefficient of variation in percentage (CV%). For Roller 20PN, the obtained CVs% were compared to declared CVs% by manufacturer. For analyser iSED the manufacturer did not declare the criteria for within-and between run CV%. The criteria for within and between-run, for iSED, were calculated from 1 standard deviation (SD) and the mean which were declared by the manufacturer for control samples.

Results
The obtained CVs% for between-and within-run precision on control samples for both analysers are presented in   Westergren method the mean bias was not statistically significant in the whole measuring range (Table 3). Moreover, Passing-Bablok regression for Roller 20PN showed a constant and proportional difference from the Westergren method for groups L and M, but for analyser iSED there was a constant and proportional difference only for group H ( Table 3). Scatter graphs provide insight into the relationship of data measured with different methods (Figure 1-3).     Roller 20 PN had lower sensitivity compared to iSED, while they had comparable specificity and accuracy (Table 4).     Mean Roller 20PN and iSED showed significant bias despite the similar methodology.

Roller 20PN iSED
Verification of automated analyser iSED was performed by several studies (6,7,12 All three studies showed similar results for comparison of iSED with Westergren method, with the exception of the negative bias for lower ESR values, which was observed in our study and not in others (6,7,12). This disagreement may be due to some patient characteristics and sample diversity used for comparison. were obtained by comparison of two analysers (iSED and TEST1). The difference between analysers that are based on a similar methodology may be because of different time frame of pre-analysis sample mixing. Despite the similarity of methods, the whole analysis process is not yet harmonized completely (7).
The limitation of this study is that it included only a minor proportion of the samples with ESR values higher than 60 mm. Further studies are needed to confirm our findings in the range of elevated ESR values. Moreover, we are aware of potential interference of low haematocrit and concentrations of positive acute phase proteins (fibrinogen) and potential physiological and pathophysiological conditions of patients that may affect analysis of ESR values. The more extensive verification which would include the investigation of potential interferences in order to upgrade the quality measurement of ESR is highly warranted. Another limitation of this study was that the number of included samples for verification of Roller 20PN and iSED was disproportional.
Implementation of automated measurement of ESR have many advantages, such as the use of the control samples for better monitoring of measurement quality, reduction of preanalytical and analytical errors of manual measurement and many more which were mentioned previously in this study. Before replacing a test used in the routine with a test based on different methodology, the clinicians need to be informed about the potential effect on the test results and about interpretation on which future clinical decisions are going to be based.

Conclusion
Both analysers, Roller 20PN and iSED, are not comparable with the Westergren method and should not be used interchangeably. Overall, the disagreement with Westergren method is less pronounced for iSED analyser.

Potential conflict of interest
None declared.