Validation and Evaluation of Eight Commercially Available Point of Care CRP Methods
Validation and Evaluation of Eight Commercially Available Point of Care CRP Methods
a r t i c l e i n f o a b s t r a c t
Article history: Background: There are several situations compelling to measure CRP with a point of care (POC) method. Assay
Received 2 April 2014 performance of various available POC CRP methods were evaluated as analytical quality is important and should
Received in revised form 1 October 2014 be known before clinical use.
Accepted 20 October 2014 Methods: We compared 2 semi-quantitative strips; Actim and Cleartest and 6 quantitative CRP tests; Afinion,
Available online 31 October 2014
QuikRead go, Smart, iChroma, Microsemi and AQT90 Flex to the Synchron CRP method, using the CLSI EP9 pro-
tocol. The coefficient of variance (CV) was determined. Various aspects of pre-analytical and analytical steps were
Keywords:
Point of care
evaluated.
C-reactive protein Results: CRP strips showed 50–60% concordance with the Synchron CRP. The linear regression lines (95% CI) of the
General practice quantitative POC CRP methods compared to the Synchron CRP method were: y = [0.96–1.04]x + [−4.7 to −2.04]
(Smart); y = [1.00–1.06]x + [1.05–4.99] (AQT90 Flex), y = [0.84–0.91]x + [−1.13 to 3.95] (Afinion); y = [0.83–
0.87]x + [0.25–1.5] (QuikRead go); y = [0.76–0.82]x + [−0.18 to 1.35] (iChroma) and y = [1.14–1.18]x + [−3.17
to −1.83] (Microsemi).
Conclusions: At best, the semi-quantitative CRP strips could be used to discriminate between normal and increased
levels of CRP. Of the quantitative methods, when combining analytical with practical evaluation, the Smart and
Afinion would be the preferred analyzers for POCT.
© 2014 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.cca.2014.10.028
0009-8981/© 2014 Elsevier B.V. All rights reserved.
196 N. Brouwer, J. van Pelt / Clinica Chimica Acta 439 (2015) 195–201
Fig. 1. Correlation of the POC CRP strips, interpreted by observer 1 after 5 minute incubation, with the Synchron CRP method. Each dot represents one sample. The gray boxes represent the
corresponding categories on the strips. A. Actim strips and B. Clear strips.
2) quantitative methods, i.e. analyzers using either capillary or venous performed with capillary blood, and an automated hematocrit (Ht)
blood. correction is applied.
• iChroma analyzer (Boditech, Med. Inc., Gangwon-do, Korea), CRP
2.1. Semi-quantitative methods method based on fluorescence sandwich immunoassay. The assay
can be performed with capillary blood. Since Ht correction is not pos-
• Actim® CRP strips (Medix Biochemica, Kauniainen, Finland), an sible, we did not correct for Ht during the evaluation either.
immunochromatographic assay producing a semi-quantitative result • Microsemi (Horiba Ltd., Kyoto, Japan), CRP method based on
subdivided into the following categories: b10, 10–40, 40–80, and immunoturbidimetric assay. The assay can be performed either with
N80 mg/L. capillary blood or with venous blood. CRP analysis can only be per-
• Cleartest® CRP strips (Servoprax, Wesel, Germany), an immuno- formed in combination with hematology parameters (measured:
chromatographic assay producing a semi-quantitative result sub- WBC, RBC, Hb, Ht, platelets, lymphocytes, monocytes, granulocytes.
divided into the following categories: b10, 10–40, 40–80, and Calculated: MCV, MCH, MCHC, RDW, PDW, MPV). Automated Ht cor-
N80 mg/L. rection is applied.
• AQT90 Flex (Radiometer Medical ApS, Brønshøj, Denmark), CRP
The results of these tests were interpreted by two independent ob- method based on solid phase sandwich immunoassay with time-
servers after 5 min and after 15 minute incubation. resolved fluorometric detection. The assay can only be performed
with venous blood. Automated Ht correction is applied.
2.2. Quantitative methods
2.3. Analytical validation
• QuikRead go (Orion Diagnostica, Espoo, Finland), CRP method based
on immunoturbidimetric assay. The assay can be performed with cap- For comparison, all the samples used for the POC CRP evaluation were
illary blood, and an automated Ht correction is applied. analyzed in the laboratory of Clinical Chemistry of the Medical Centre Alk-
• Smart analyzer (Eurolyser Diagnostica, Salzburg, Austria), CRP meth- maar by means of the routine biochemical CRP method, on the
od based on immunoturbidimetric assay. The assay can be performed Synchron® analyzer (Beckman Coulter, Krefeld, Germany). The analytical
with capillary blood, and Ht correction can be applied manually (only range of this method is between 1.0 and 500 mg/L, with an intra- and
with known Ht). We did not correct for Ht during this evaluation, be- inter-assay variation (CRP 80 mg/L) of 1.1% and 5.5%, respectively.
cause in practice the Ht will not be known either for most patients at The CLSI EP9 protocol was performed for each of the POC CRP
the moment the CRP assay is performed. methods, using 100 K2-EDTA blood samples for each method. All
• Afinion™ AS 100 analyzer (Axis Shield, Oslo, Norway), CRP method blood samples were from GPs' patients, aged N 18 years, with CRP
based on solid phase immunochemical assay. The assay can be concentrations ranging from 5 to 200 mg/L, determined with the
Table 1
Correlation of the CRP strips with the Synchron CRP method. Correlation is expressed with the kappa-index [11] as well as % discrepancy.
Observer 1 Observer 2
3. Results
Fig. 2. Correlation of the 6 POC CRP analyzers with the Synchron CRP method given as
Bland–Altman plots with absolute and percentage differences. Each dot represents one
sample. A: QuikRead go (n = 87), B: Smart (n = 86), C: Afinion (n = 98), D: iChroma
(n = 111), E: Microsemi (n = 106), and F: AQT90 Flex (n = 86).
198 N. Brouwer, J. van Pelt / Clinica Chimica Acta 439 (2015) 195–201
Table 2
Linear regression equation, coefficient of correlation and coefficient of variance of the 6 POC CRP methods compared to the Synchron CRP method.
QuikRead go 87 0.85 (0.83–0.87) 1.04 (0.25–1.5) 0.996 (0.994–0.997) 2.1 [74]; n.e.
Smart 86 1.00 (0.96–1.04) −3.26 (−4.70 to −2.04) 0.970 (0.954–0.980) 2.8 [79]; 7.6 [113]
Afinion 98 0.87 (0.84–0.91) 2.78 (1.13–3.95) 0.982 (0.973–0.988) 6.0 [82]; 2.6 [120]
iChroma 111 0.79 (0.76–0.82) 0.59 (−0.18–1.35) 0.967 (0.953–0.976) 18.7 [57]; 4.8 [77]
Microsemi 106 1.16 (1.14–1.18) −2.5 (−3.17 to −1.83) 0.997 (0.996–0.998) 3.0 [57]; 1.3 [113]
AQT90 Flex 86 1.03 (1.00–1.06) 2.68 (1.05–4.99) 0.992 (0.995–0.998) 3.5 [63]; 7.6 [120]
3.1.2. Quantitative methods manual of the test suggested (Fig. 3). This was demonstrated by the
The results of the comparisons between the six quantitative POC CRP low concordance between observer 1 and observer 2 at strip reading,
analyzers and the Synchron CRP method are graphically shown as especially for the Clear strips. Although both manufactures state that it
Bland–Altman plots in Fig. 2. Table 2 summarizes the 95% CI of the linear is in the nature of immunochromatographic assays to see an increase
regression lines, correlation coefficients and coefficients of variation in line intensity in time, anyone who does not perform these tests fre-
found for all six POC CRP analyzers. None of the analyzers met all the quently might find it difficult to judge whether the third blue line
correlation criteria determined beforehand. All analyzers had a (CRP N 80 mg/L) is visible or not.
r2 N 0.95. However, the 95% CI for the slope of the linear regression Moreover, it might happen that the test is evaluated after more than
equation compared to the Synchron CRP method only included 1.0 for 5 min. To investigate the effect of postponed interpretation of the strips,
the Smart analyzer and the AQT90 Flex analyzer. The linear regression we also analyzed the strips after 15 min. By this time, the discrepancy
equation of QuikRead go, Afinion and iChroma revealed an underesti- between the results of the strip and the Synchron CRP from the lab in-
mation of the CRP value compared to the Synchron CRP, while the creased by 10% for the Actim Strip and by 27% for the Clear strip. More-
Microsemi overestimated the CRP compared to the method used in over, the results seemed to indicate a higher CRP level than before. For
our laboratory. The iChroma analyzer was the only analyzer with an in- example, CRP in the category 40–80 mg/L after 5 min became category
tercept including 0.0 in the 95% CI. N80 mg/L after 15 min. This phenomenon was observed for all levels of
The coefficients of variation, measured in patient samples with CRP CRP.
values between 50 and 130 mg/L, varied among the analyzers During the practical evaluation we experienced a long hands-on
(Table 2). Only the iChroma did not meet the criterion of VC b 10%. time for the analysis (10–15 min) and difficult analytical handling for
persons who does not perform the test on a regular basis. In addition, re-
3.2. Practical evaluation sults are not automatically included in the patient record.
The main advantage of the strips is the low costs. Since no analyzer is
Evaluation of the various POC CRP methods with regard to relevant required, the costs of performing a semi-quantitative CRP are much
points for successful implementation of POC CRP revealed that Afinion lower than the quantitative CRP tests.
required the least pre-analytical handling in combination with a capil- In our opinion, the semi-quantitative CRP strips could at best be used
lary blood sample. The pre-analytical steps of both semi-quantitative to discriminate between no infection (CRP b 10 mg/L, only the control
strips, the QuikRead go and iChroma required 2–3 minute hands-on line is visible on the strip) and infection likely (CRP elevated, one or
time compared to 30–45 s for Afinion and Eurolyser. Both Microsemi more blue lines are visible on the strip in addition to the control line).
and AQT90 Flex did not required additional handling after venous If one or more of the blue test lines appear on the strip, the user should
blood sampling. The strict timing of interpretation after 5 min is a disad- be advised to measure CRP with a quantitative method.
vantage of the strips, while the analysis time of 13 min is the major
drawback for the AQT90 Flex. All other analyzers produced a result
within 5 min. Table 3 presents the results of all the items addressed in 4.2. Evaluation of the quantitative CRP analyzers
our evaluation.
The r2 of all quantitative POC CRP analyzers was N0.95. Most proba-
4. Discussion bly, the r2 of the Smart and iChroma CRP test was lower than the other
analyzers due to the absence of an automated Ht correction. However,
POC requires a good quality assurance program. This program should we did not manually correct for Ht, as this is in general not possible in
contain a validation of the CRP method, training procedure for applicants, POC situations. Several POC CRP analyzers have already been evaluated
certification, and re-certification after performing an e-learning training. by others [4–10] but all except one [10] in a one to one comparison with
Distribution and storage of cuvettes and reagents and internal control a central laboratory method. The correlation coefficients are in general
should follow the user manuals of the distributor. A contact person from N0.95 comparable with our findings.
the laboratory should be available for consultation. Only if all these condi- With regard to the correlation with the laboratory CRP test on the
tions are met, POC CRP can be successfully implemented. To our knowl- Synchron analyzer, there were some remarkable differences between
edge, this is the first study to compare both capillary and venous blood the various POC CRP analyzers. The slope of Smart and AQT90 Flex
CRP methods as well as semi-quantitative CRP strips regarding their ana- CRP was around 1.00, and therefore showed the best correlation with
lytical performance and their applicability for POC testing. the Synchron CRP method. The slope of QuikRead go and Afinion CRP
was 0.85, while the iChroma even underestimated the CRP value by
4.1. Evaluation of the semi-quantitative CRP strips more than 20%. On the other hand, the slope of the Microsemi CRP
was 1.15 compared to our laboratory method. These four methods failed
Actim and Clear semi-quantitative CRP strips showed an insufficient to have a 95% CI for the slope including 1.00. Almost never did the inter-
correlation with the Synchron CRP method used in the laboratory. Over- cept include 0.0 as defined in the correlation criteria. Instead, it varied
estimation as well as underestimation of the CRP value was observed for between −4.7 and 5.0 among the analyzers. Since the clinical relevant
about one third of the samples analyzed. The evaluation revealed that range is in the higher range, the deviation at the intercept had a relative-
interpretation of the lines was more difficult in practice than the user ly small impact and is therefore less relevant than the slope of the linear
Table 3
Practical aspects of the semi-quantitative strips and quantitative POC methods for the analysis of CRP.
Analyzer (producer) Actim (Medix) Cleartest (Servoprax) QuikRead go (Orion) Smart (Eurolyser) Afinion (Axis iChroma (Boditech) Microsemi (Horiba) AQT90 Flex
Shield) (Radiometer)
Material Capillary blood Capillary blood Capillary blood Capillary blood Capillary blood Capillary blood Capillary or venous Venous blood
blood
Method Immunochromatographic Immunochromatographic Immunoturbidimetric Immunoturbidimetric Solid phase Fluorescence sandwich Immunoturbidimetric Solid phase
assay assay assay assay immunochemical immunoassay assay sandwich
assay immunoassay
Minimum amount 10 μL 10 μL 20 μL 5 μL 1.5 μL 10 μL 18 μL + dead volume 2 mL
of material in the tube: 100 μL
Analytical range 0–80 mg/L 0–80 mg/L 5–200 mg/L 1–240 mg/L 8–200 mg/L 2.5–300 mg/L 2.0–230 mg/L 5–500 mg/L
(full blood)
199
200 N. Brouwer, J. van Pelt / Clinica Chimica Acta 439 (2015) 195–201
When comparing the different POC CRP analyzers with each other,
the Afinion CRP had the most simple pre-analytical handling, with the
Smart as second best. The AQT90 Flex CRP does not need any pre-
analytical handling at all, but it requires a venous puncture, which
makes it less suitable for POCT.
For QuikRead go and iChroma, the pre-analytical handling of the
samples is much more complicated than for the other analyzers
(Table 3, row pre-analytical handling). This might introduce variation
if the test is not performed on a regular basis. Besides variation, incor-
rect pre-analytical handling can generate errors during the analysis.
This might create spill of materials, having to use more than one cassette
or cuvette per patient sample to gain a result. Simple pre-analytical han-
dling is therefore an important matter in performing POC tests.
A drawback of the Afinion analyzer is that the sample volume is only
1.5 μL, which means that it is very important to place the cassette direct-
ly into the analyzer after adding the blood sample, otherwise an error is
generated because the material will clot. A disadvantage of the Smart
and iChroma CRP analyzers is the absence of automated Ht correction,
which can introduce incorrect results, since in most POC settings the
Fig. 3. Comparison of the lines on the strip (Actim) in practice (A) and on the manufac- actual Ht of the patient will not be known before CRP measurement.
tures insert (B). Line 1: CRP 10–40 mg/L, line 2: CRP 40–80 mg/L, line 3: CRP N80 mg/L, With an analytical time between 2 and 4 min, all analyzers are gen-
and line 4: control line. The strip was incubated with a full blood sample with a CRP con-
erating a result fast enough for POCT, with the exception of the AQT90
centration of 90 mg/L as determined by the Synchron method.
Flex, which has an analytical time of 13 min.
Taking into account both the analytical validation and the practical
regression line. The Smart analyzer showed the best correlation with evaluation, we consider the Afinion and Smart CRP to be the most suit-
the Synchron CRP method. able analyzers for POCT. Although the correlation coefficient of the
The CV-analytical of the various CRP methods varied, but only the Afinion was better than the Smart, Afinion underestimated the CRP con-
iChroma did not pass our criterion of CV b10%. This might be a direct centration by more than 10% and therefore our preferred method is the
consequence of the pre-analytical handling necessary for the iChroma Smart analyzer.
method, which is relatively complex. We experienced that it was almost The AQT90 Flex also is a reliable analyzer, but it is less suitable be-
impossible to replace the cap on the tube after filling it with the material cause of the necessary venous sampling, the amount of necessary
from the capillary without getting fluid on the outside of the cap, there- blood, the analysis time and the size and weight of this benchtop appa-
by inducing variation in the liquid volume. Besides, the presence of liq- ratus. However, it could work well at an emergency department or in a
uid containing human material on the outside of the vessel also is a smaller health care center.
negative aspect of this method with regard to hygiene and infection
risk. Remarkably others found better results for the performance of Competing interests
the iChroma [4,8].
In general the agreement between the laboratory method and the None declared.
POCT methods is not very good but also the outcomes among various
studies with the same POCT method seems to be rather variable. First
Funding
of all most POCT methods are not calibrated before measurements and
also no quality controls are available. Furthermore a lot-to-lot variation
All analyzers used in this research were provided free of charge for
can be present and POCT methods have mostly single cartridges for
the purpose of validation and evaluation only. No other forms of funding
every measurement. Incorrect pre-analytical handling of the samples
were received for this research.
can also introduce considerable variation although POC devices should
be user-friendly and full-proof. In our studies all pre-analytical handling
and measurements were carried out by the same technician so the Ethical approval
obtained CV's are completely comparable with each other. Our study
was conducted with fresh venous blood samples and differences be- Not required.
tween venous blood and capillary blood cannot be argued. For practical
reasons, we performed no manual Ht correction which possibly could Acknowledgments
lead to better results.
We would like to thank Erik Allon, Ria Baltus and the point of care
technicians, of the Laboratory of Clinical Chemistry, Hematology and
4.3. Practical evaluation of the quantitative CRP analyzers Immunology of the Medical Centre, Alkmaar, for their practical assis-
tance and their critical views on various analyzers.
Pre-analytical handling necessary for the various methods was one
of the items addressed in our practical evaluation of the suitability of References
the different POC CRP tests.
The advantages of the quantitative methods compared to the semi- [1] Joshi A, Perin DP, Gehle A, Nsiah-Kumi PA. Feasibility of using C-reactive protein for
point-of-care testing. Technol Health Care 2013;21:233–40.
quantitative strips are obvious: a quantitative result is obtained, there
[2] Cals JW, Chappin FH, Hopstaken RM, van Leeuwen ME, Hood K, Butler CC, et al. C-
is less hands-on time and, importantly, there is no critical time of reactive protein point-of-care testing for lower respiratory tract infections: a quali-
read-out, so the operator can walk away from the test after putting it tative evaluation of experiences by GPs. Fam Pract 2010;27:212–8.
into the analyzer. Most analyzers can be connected to the laboratory in- [3] Verheij TJM, Prins JM, Salomé PhL, Bindels PJ, Ponsioen BP, Sachs APE, et al. NHG-
Standaard Acuut hoesten. Huisarts Wet 2011;54:68–92.
formation system, thus facilitating automatic transfer of the results to [4] Oh SW, Moon JD, Park SY, Jang HJ, Kim JH, Nahm KB, et al. Evaluation of fluorescence
the patient's electronic lab dossier. hs-CRP immunoassay for point-of-care testing. Clin Chim Acta 2005;356:172–7.
N. Brouwer, J. van Pelt / Clinica Chimica Acta 439 (2015) 195–201 201
[5] Papaevangelou V, Papassotiriou I, Sakou I, Ferentinos G, Liapi G, Kyrka A, et al. Eval- [9] Kotani K, Minami T, Abe T, Sato J, Taniguchi N, Yamada T. Development of a new
uation of a quick test for C-reactive protein in a pediatric emergency department. point-of-care testing system for measuring white blood cell and C-reactive protein
Scand J Clin Lab Invest 2006;66:717–21. levels in whole blood samples. Clin Chim Acta 2014;433:145–9.
[6] Siebenhaar R, Mushholt PB, Forst T, Weber MM, Maurer R, Pfützner A. Laboratory [10] Minnaard MC, van de Pol AC, Broekhuizen BD, Verheij TJ, Hopstaken RM, van Delft S,
evaluation of a new lateral-flow-based point-of-care rapid test for assessment of et al. Analytical performance, agreement and user-friendliness of five C-reactive pro-
chronic systemic inflammation. J Diabetes Sci Technol 2010;4:717–22. tein point-of-care tests. Scand J Clin Lab Invest 2013;73:627–34.
[7] Verbakel JY, Aertgeerts B, Lemiengre M, Sutter AD, Bullens DM, Buntinx F. Analytical [11] Fleiss JL, Levin B, Paik MC. Statistical methods for rates and proportions. Wiley; 1881
accuracy and user-friendliness of the Afinion point-of-care CRP test. J Clin Pathol 229–32.
2014;67:83–6.
[8] Ciftci IH, Koroglu M, Karakece E. Comparison of novel and familiar commercial kits for
detection of C-reactive protein levels. World J Microbiol Biotechnol 2014;30:2295–8.