Clinical Biochemistry 71 (2019) 38–45 Contents lists available at ScienceDirect Clinical Biochemistry journal homepage: www.elsevier.com/locate/clinbiochem A multicenter evaluation of a point of care CRP Test T a,⁎ b c c c Volkher Scharnhorst , Peter G. Noordzij , Andrea Lutz , Uwe Graser , Daniel Püntener , Aitor Alquézar-Arbéd a Clinical Laboratory, Catharina Hospital and Technical University Eindhoven, Eindhoven, The Netherlands b Department of Anesthesiology, Intensive Care and Pain Medicine, St. Antonius Hospital, Nieuwegein, The Netherlands c Roche Diagnostics International Ltd, Rotkreuz, Switzerland d Servicio de Urgencias, Hospital de la Santa Creu i Sant Pau, Barcelona, Spain ARTICLE INFO ABSTRACT Keywords: Background: Point-of-care (POC) C-reactive protein (CRP) testing in the primary healthcare setting is a cost- Antibiotic use effective approach for reducing antibiotic prescriptions, but has yet to be widely adopted. POC CRP Test Methods: Analytical performance of the cobas CRP Test on the cobas b 101 system was evaluated at three POC C-reactive protein sites and one reference laboratory. Within-run (repeatability), within-laboratory (intermediate precision), and Infection between-laboratory precision (reproducibility) were assessed. Method comparison (reference test: CRPNX re- Inflammation agent [cobas c 501 module]) and matrix/lot-to-lot comparison experiments were conducted using prospectively Point-of-care testing Diagnostic equipment collected blood samples from 217 adults (apparently healthy or with clinically relevant conditions). Usability Laboratory medicine and reliability were assessed by questionnaire and error reporting. Results: Coefficients of variation (CV) for repeatability and intermediate precision ranged from 1.7%–4.0% and 1.9%–4.5%, respectively, for human serum pools containing CRP 4.7–350.7 mg/L; repeatability in clinical samples ranged from 1.6%–5.9% (3.3–360.3 mg/L). CVs for reproducibility ranged from 2.5%–4.0% (4.7–344.3 mg/L). CRP concentrations were comparable for capillary whole blood, serum, Li-heparin whole blood/plasma, K2 and K3 EDTA whole blood/plasma (Pearson's r ≥ 0.996), and among three CRP Test lots (r ≥ 0.993). Clinically relevant CRP concentrations measured with the CRP Test showed good agreement with those measured by CRPNX reagent (serum, weighted Deming regression y = 0.97× + 0.11; Pearson's r ≥ 0.996). The overall mean usability score was 4.18/5 and the error rate across 9378 tests was 1.00%. Conclusions: The cobas CRP Test on the cobas b 101 system demonstrates robust analytic performance when used by healthcare professionals in the POC setting. 1. Introduction presenting with symptoms of lower respiratory tract infection in pri- mary care, if pneumonia has not been diagnosed and it is unclear C-reactive protein (CRP) is a key mediator of the acute-phase re- whether antibiotics should be prescribed [13]. Results of the CRP test sponse, with blood levels of CRP increasing rapidly after an in- can be used to guide antibiotic prescribing in people without a clinical flammatory stimulus [1–3]. Therefore, changes in serum levels of CRP diagnosis of pneumonia [13]. However, despite the availability of POC are a clinically useful marker of infection, inflammation, and tissue devices, POC CRP testing has not yet been widely adopted in primary injury [4]. CRP testing in the primary setting (i.e. point-of-care [POC]) care clinics [13–16]. can help reduce diagnostic uncertainty by differentiating between The cobas b 101 POC system (Roche Diagnostics International Ltd., bacterial and viral infections [5–7] and has been shown to be cost-ef- Rotkreuz, Switzerland) provides glycated hemoglobin (HbA1c) and lipid fective for reducing inappropriate antibiotic prescriptions [8–12]. panel tests (measurement of cholesterol, triglyceride, and high-density UK National Institute for Health and Care Excellence (NICE) lipoprotein; calculation of low-density lipoprotein) for managing dia- guidelines recommend consideration of POC CRP testing for people betes and dyslipidemia at point-of-need. The test options on the cobas b Abbreviations: CRP, C-reactive protein; CV, coefficient of variation; HSP, human serum pool; NICE, National Institute for Health and Care Excellence; POC, point-of- care ⁎ Corresponding author at: Clinical Laboratory, Catharina Hospital, PO Box 1350, 5602 ZA Eindhoven, The Netherlands. E-mail addresses: volkher.scharnhorst@catharinaziekenhuis.nl (V. Scharnhorst), p.noordzij@antoniusziekenhuis.nl (P.G. Noordzij), andrea.lutz@roche.com (A. Lutz), uwe.graser@roche.com (U. Graser), daniel.puentener@roche.com (D. Püntener), aalquezar@santpau.cat (A. Alquézar-Arbé). https://doi.org/10.1016/j.clinbiochem.2019.06.009 Received 17 January 2019; Received in revised form 11 June 2019 Available online 21 June 2019 0009-9120/ © 2019 Roche Diagnostics International Ltd. Published by Elsevier Inc. on behalf of The Canadian Society of Clinical Chemists. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/). V. Scharnhorst, et al. Clinical Biochemistry 71 (2019) 38–45 101 system have recently been expanded to include POC testing for present in the sample binds with the CRP antibody–latex conjugate. The CRP. We evaluated the analytical performance, usability and reliability concentration of CRP is determined by photometric measurement of the of the cobas POC CRP Test (Roche Diagnostics International Ltd., latex agglutination reaction using wavelengths of 525 and 625 nm. Rotkreuz, Switzerland) and conducted a method comparison versus a Blood flow within the CRP test is illustrated in Supplementary Fig. 1. reference comparator. All experiments with the POC CRP Test in the present study were performed according to the manufacturer's instructions. 2. Methods 2.3. Precision in clinical samples, whole-blood and plasma samples 2.1. Study design Repeatability of the POC CRP Test was evaluated at a single site A multicenter evaluation of the cobas POC CRP Test was performed using EDTA K3 whole-blood and plasma samples. For each of the three at three hospitals (Hospital Sant Pau, Barcelona, Spain; Catharina reagent lots, one sample from each of the following CRP concentration Ziekenhuis, Eindhoven, The Netherlands; St Antonius Ziekenhuis, categories was tested: healthy (< 5 mg/L); cutoff (~10 mg/L); decision Nieuwegein, The Netherlands), and one reference laboratory in Japan (~40 mg/L); acute (~100 mg/L); acute high (~350 mg/L). For each between May and November 2017. Each healthcare professional sample, a series of 21 sequential measurements was performed to assess (nurses and physicians with limited POC device experience and no within-run precision. technical laboratory background) received a short training on how to obtain the sample, use the disk, and operate the device. A total of three CRP Test lots and three CRP Control lots were evaluated; each site was 2.4. Precision according to CLSI EP05-A3 guidelines assigned two POC CRP Test lots and two CRP Control lots. Within-run precision (repeatability), within-laboratory precision Repeatability and intermediate precision of the POC CRP Test were (intermediate precision), and between-laboratory precision (reprodu- also assessed according to Clinical and Laboratory Institute (CLSI) cibility) were assessed using six human serum pools (HSPs; Biomex EP05-A3 guidelines [18]. Samples comprised six HSPs and two CRP GmbH [D-Heidelberg] and Kentucky Clinical Trials Laboratory Controls (levels 1 and 2). For each lot (n = 2 lots per site), samples were [Louisville, USA]; HSP 1, < 5 mg/L [healthy]; HSP 2, ~10 mg/L measured four times daily for 21 days (two replicates of each sample for [cutoff]; HSP 3, ~40 mg/L [decision]; HSP 4, ~100 mg/L [acute]; HSP each of two runs per day). Samples were analyzed in random order per 5, ~280 mg/L [acute high]; HSP 6, spiked with recombinant CRP to day for 21 days. ~350 mg/L [acute high]) and two cobas CRP Control pools (level 1, Reproducibility of the POC CRP Test was assessed at three sites over representing a low-range sample; level 2, representing a high-range 5 days using HSPs 1–6 and CRP Controls (levels 1 and 2). Five replicates sample). Method comparison, matrix comparison, lot-to-lot comparison of each sample were analyzed per day over 5 days. experiments, and an additional assessment of repeatability were per- formed using samples collected prospectively from adults aged 2.5. Sample matrix comparison ≥18 years who were either apparently healthy or who had clinically relevant conditions including inflammatory disorder and associated Test performance in different matrices was compared using matched diseases (e.g. Crohn's disease, acute asthma without infection), infec- serum and capillary blood samples, and matched serum and whole- tion (e.g. strong evidence of infection in the clinical record, sepsis) or blood/plasma samples containing EDTA K2, EDTA K3, and Li-heparin tissue injury (e.g. contusion, fracture, open wound, surgery). Exclusion as anticoagulants. Matched pairs were analyzed in parallel, in a single criteria were pregnancy or breastfeeding, and any individual whom the run for each of two POC CRP Test lots, at a single site. attending clinician deemed to be clinically unstable or whose condition could be compromised by blood draw. Participants at each POC site 2.6. Lot-to-lot comparison provided two capillary blood samples (one per lot tested) and a venous whole-blood sample for measurement of serum CRP; samples were not POC CRP Test lot-to-lot comparison was performed using capillary treated with an anticoagulant. Participants enrolled at POC site 1 pro- whole-blood/serum samples (at three sites with a total of three disk vided additional venous whole-blood samples for measurement of CRP lots) or EDTA K2, EDTA K3, and Li-heparin whole-blood/plasma sam- in EDTA (K2 and K3) and lithium (Li)-heparin whole blood/plasma. ples (at a single site with two disk lots). Samples were tested in a single Samples were processed and tested immediately at each site; serum run per reagent lot. Multiple samples were taken from the same subject samples were split into three 1.5 mL aliquots, frozen at –70°C or colder, and run in parallel on two different instruments, using a different disc and shipped on dry ice for method comparison experiments performed lot for each sample. at the reference laboratory. The study protocol was approved by the relevant Institutional Review Board/Independent Ethics Committee at each study site prior to 2.7. Method comparison study initiation and the study was conducted in accordance with the Declaration of Helsinki [17]. Written informed consent was provided by Healthcare professionals determined CRP concentrations in capil- all participants. The study was sponsored by Roche Diagnostics Inter- lary whole blood/serum and EDTA K2, EDTA K3, and Li-heparin in national Ltd. whole blood/plasma using the cobas POC CRP Test (cobas b 101 system; three sites); some users had limited experience of using the 2.2. Point-of-care CRP Test system previously, and some were naïve to the system. For comparison, CRP values from matched serum samples were obtained by laboratory The cobas POC CRP Test is an in vitro diagnostic test for the professionals using the cobas c 501 module (Roche Diagnostics quantitative determination of CRP in human capillary whole blood and International Ltd., Rotkreuz, Switzerland) with the CRPNX reagent serum, and EDTA (K2/K3) and Li-heparin anticoagulated whole blood (reference test; Denka Seiken, Tokyo, Japan). The reference test has a and plasma. The capillary whole-blood, serum or plasma sample is measuring range of 0.1~320 mg/L and a limit of detection of 0.05 mg/ applied to the disk, which is inserted into the cobas b 101 instrument. L; the CV for within-run precision ranged from 0.62–0.93% across Once the system is initiated, the liquid component of the sample is control samples. No interference of the test is shown with ascorbic acid separated by centrifugation (i.e. separation from the blood cells), the up to concentration of 50 mg/dL; hemoglobin up to 500 mg/dL; bilir- isolated plasma or serum is diluted with dilution buffer, and CRP ubin up to 30 mg/dL and intralipid up to 5 mg/dL. 39 V. Scharnhorst, et al. Clinical Biochemistry 71 (2019) 38–45 2.8. Usability and reliability 3.3. Precision according to CLSI EP05-A3 guidelines Usability of the POC CRP Test on the cobas b 101 system was Repeatability and intermediate precision are presented in Table 1 evaluated by questionnaires completed by all operators at each of the (overall) and Supplemental Table 3 (by lot and site) and were consistent three sites. Questions related to six individual domains: general; general across sites and lots. CVs for repeatability in HSPs ranged from 1.7% to aspects of the software; processing of the sample; CRP disk; quality 4.0% (site 1, 2.0%–4.0% [mean CRP range, 4.8–350.7 mg/L]; site 2, control; and cleaning/disinfection. The usability questionnaire em- 1.7%–3.6% [mean CRP range, 4.7–344.2 mg/L]; site 3, 1.7%–3.9% ployed a 5-point scoring system (1, very poor; 2, poor; 3, average; 4, [mean CRP range, 4.7–345.3 mg/L]); CVs for intermediate precision in good; 5, excellent), with mean scores calculated for each domain and HSPs ranged from 1.9% to 4.5% (site 1, 2.1%–4.1%; site 2, 1.9%–4.5%; overall. Reliability of the system was assessed by capturing CRP Test site 3, 2.3%–4.2%). and instrument errors. An overall error rate was calculated based on The SD for reproducibility was 0.2 mg/L for the low-concentration reported CRP Test and instrument errors. HSP 1 sample (mean CRP concentration 4.7 mg/L). CVs for reproduci- bility ranged from 2.5% to 4.0% for HSPs 2–6 (mean CRP concentra- tions 10.4–344.3 mg/L; Table 2). 2.9. Data analysis and statistical methods 3.4. Matrix comparison The target sample size was ≥60 adults per site to provide samples covering the measuring range of the POC CRP Test (3–400 mg/L) in the CRP concentrations measured using the different sample types following prespecified distribution: < 20 mg/L (> 40% of partici- correlated well with CRP concentrations measured in serum (Pearson's pants); 20–100 mg/L, 100–200 mg/L, and 200–400 mg/L (each > 10% r ≥ 0.996, intercept ≤0.32, slope ≥ 0.96; Supplemental Table 4); of participants). All data were evaluated and analyzed with validated across lots and matrices, mean bias was low, ranging from 0.02% to statistical tools (WinCAEv and Medrio, Mannheim, Germany; SAS 0.34% for low concentration samples (0–5 mg/L), from −2.18% to software, SAS Institute Inc., Cary, NC). 0.90% for mid-range samples (5–200 mg/L), and from −5.53% to For precision experiments, the standard deviation (SD; for low −0.74% for higher-concentration samples (Supplemental Table 5). concentration samples) and coefficient of variation (CV) were used as There was good agreement for CRP concentrations measured in serum the measure of assay imprecision. Matrix and lot-to-lot comparisons with Li-heparin whole-blood samples (Pearson's r, 0.998; Fig. 1A) and were evaluated using Passing–Bablok analysis. Agreement between EDTA K3 whole blood (Pearson's r, 0.998; Fig. 1B). methods was evaluated using weighted Deming regression analysis. Bias was examined using Bland-Altman analysis, with relative differ- ence calculated as (Y–X)*100/X (where X = CRPNX test [serum] and 3.5. Lot-to-lot comparison Y = POC CRP Test). Mean bias between CRP values on the different systems was calculated and the relative bias at medical decision points Correlations among different test lots were very good for whole- was determined. blood, plasma, and serum samples, with all Pearson's r values ≥ 0.993, intercepts ≤ 0.30, and slopes ≥ 0.96 (Fig. 2; Supplemental Table 6). Mean bias with different lots was ≤2.37%. 3. Results 3.6. Method comparison 3.1. Samples There was very good correlation between CRP values measured with A total of 217 adult patients were recruited from semi-intensive the cobas POC CRP Test on the cobas b 101 POC system (all sample care, cardiovascular care, ambulatory care or the emergency depart- matrices) and values measured in matched serum samples with the ment, and comprised 158 males (72.8%) and 59 females (27.2%). CRPNX test on the cobas c 501 laboratory module (Pearson's r ≥ 0.994, Median age (range) was 68 (range, 22–88) years. Participants with slope ≥ 0.93, intercept ≤0.47; Supplemental Table 7). Across matrix valid measurements (n = 201) were categorized according to clinical types, lots, and sites, bias at the lower (5.0 mg/L) and upper (10.0 mg/ relevance for CRP, as follows: tissue injury (n = 116 [57.7%]), infection L) medical decision points was ≤6.07%; results for capillary whole- (n = 54 [26.9%]), inflammatory disorders and associated diseases blood and serum samples pooled by site and for anticoagulated whole- (n = 28 [13.9%]) or CRP ≤5 mg/L (n = 3 [1.5%]). Samples (n = 201) blood and plasma samples at a single site are presented in Table 3. were categorized by CRP concentration range (Supplemental Table 1). Biases for the comparison of the POC CRP Test versus the CRPNX test (serum) are presented in Table 3. For the CRP range 3–200 mg/L, mean bias (95% CI) was −2.05% (−10.54, 6.43) for serum and − 2.05% 3.2. Precision in clinical samples, whole-blood and plasma samples (−15.64, 11.54) for capillary whole blood. Example Bland Altman plots are shown for serum in Fig. 3A and capillary whole blood in Repeatability results for EDTA K3 whole-blood samples (mean CRP Fig. 3B. When CRP was measured in serum samples with the POC CRP concentration 3.3–360.3 mg/L) and plasma samples (3.3–356.5 mg/L) Test and CRPNX (pooled site; lot 1; n = 130 samples), weighted Deming are presented in Supplemental Table 2. Across sample types, CVs for regression analysis yielded a Pearson's r of 0.996 (Supplemental repeatability ranged from 1.6% to 5.9%. Fig. 2A). Weighted Deming regression analysis of the POC CRP Test Table 1 Precision of the POC CRP Test, as assessed according to CLSI EP05-A3 guidelines. Data are presented as a range across individual lots and sites; each range is based on n = 504 measurements (n = 84 replicate measurements per sample, per lot, across three sites). See Supplementary Table 1 for results by site and by lot. HSP 1 HSP 2 HSP 3 HSP 4 HSP 5 HSP 6 Control level 1 Control level 2 Mean CRP, mg/L 4.7–4.8 10.3–10.6 42.3–43.2 101.5–104.1 224.0–231.5 342.0–350.7 9.6–10.1 44.1–45.8 Intermediate precision, CV (%) 3.3–4.5 2.5–3.7 2.9–4.4 1.9–3.1 2.7–3.5 2.6–3.5 3.3–5.0 3.1–4.1 Repeatability, CV (%) 3.1–4.0 1.8–3.1 2.5–3.4 1.7–2.2 2.2–3.2 2.0–2.9 2.6–4.7 2.4–3.8 CRP, C-reactive protein; HSP, human serum pool. 40 V. Scharnhorst, et al. Clinical Biochemistry 71 (2019) 38–45 Table 2 Reproducibility of the POC CRP Test. Data shown are combined results across all three sites. HSP 1 HSP 2 HSP 3 HSP 4 HSP 5 HSP 6 Control level 1 Control level 2 Mean CRP, mg/L 4.7 10.4 42.5 102.1 225.6 344.3 9.7 44.7 Reproducibility, CV (%) 0.2 3.4 3.2 2.5 4.0 3.9 4.7 4.0 Site-to-site, CV (%) 0.1 1.7 0.7 0.8 0.0 0.8 0.0 0.7 Lot-to-lot, CV (%) 0.0 0.2 0.4 0.0 1.9 1.5 3.4 2.1 Day-to-day, CV (%) 0.1 0.4 1.2 0.9 0.0 1.3 0.6 1.6 Repeatability, CV (%) 0.2 2.9 2.9 2.1 3.6 3.3 3.2 2.9 CRP, C-reactive protein; CV, coefficient of variation; HSP, human serum sample pools; SD, standard deviation. (whole capillary blood; lot 1) versus CRPNX test (serum) results, based CRP testing may facilitate greater use by clinicians. In the current on pooled site data for 136 samples, yielded a Pearson's r of 0.994 analyses, we show that the cobas POC CRP Test delivered precise and (Supplemental Fig. 2B). repeatable CRP values with clinical specimens (including samples with very high CRP values) and with HSPs measured according to CLSI EP05- 3.7. Operator-reported usability A3 criteria. Notably, CVs for imprecision were well below the specifi- cations reported in the desirable biological variation database, which Overall mean score for operator-reported usability assessed across state a CV for imprecision of 21.1% [19]. CRP values measured using six domains was 4.18 out of 5 (i.e. between “good” and “excellent”; different sample types correlated well, with low mean bias. Correlations n = 7 completed questionnaires across the three sites). Mean scores by between different test lots were very good, indicating that different lots individual domain are presented in Supplemental Fig. 3. Of interest, all of the POC CRP Test perform equally well. Moreover, CRP values sites noted the rapid turnaround time for results and the ease of use of measured in a range of sample types with the POC CRP Test demon- the instrument, but also the missing battery-operated modus, meaning strated very good correlation with CRP values measured in serum that it was cumbersome to plug in the instruments at patient beds. samples with the CRPNX test on the cobas c 501 module; these results A total of 9378 POC CRP Test disks were tested during this per- indicate that the POC CRP Test delivers comparable performance to formance evaluation and 94 errors were reported (~1% error rate). laboratory CRP testing. Finally, usability of the cobas b 101 system was Most errors (46%) were reaction errors due to overfilling the disk where rated as “good” to “excellent” by operators, and reported error rates the sample is able to move out of the chamber due to backflow. The were low, indicating convenience of the system for use in the POC second most common error (24%) was large or insufficient sample, and environment. most cases occurred with serum or control solution (only one error was POC CRP testing offers several clinical and health-economic ad- reported with capillary whole blood). None of the reported errors were vantages. Compared with CRP testing, procalcitonin testing is more associated with wrong measurement results. expensive in both POC and automated settings [20,21]. Notably, POC CRP testing has been shown to be cost-effective for reducing in- appropriate antibiotic prescriptions in patients presenting with sus- 4. Discussion pected respiratory tract infection [13,22]. A recent systematic review of randomized controlled trials and observational studies reported that Use of CRP testing at POC can help to reduce inappropriate anti- POC CRP testing was associated with a significant 25% reduction in biotic prescriptions [8–12]; however, widespread use in primary care antibiotic prescribing at the index consultation relative to no POC CRP settings has not yet materialized. Availability of easy-to-use and reliable Fig. 1. Matrix comparison: Passing–Bablok regression analysis of CRP values measured with the POC CRP Test using serum (reference) compared with (A) Li-heparin whole blood and (B) EDTA K3 whole blood (test). Data are shown for Lot 1 analyzed at POC site 1. CRP, C-reactive protein; POC, point-of-care; WB, whole blood. 41 V. Scharnhorst, et al. Clinical Biochemistry 71 (2019) 38–45 Fig. 2. Lot-to-lot comparison: Passing–Bablok regression analysis of CRP values measured with different lots of the POC CRP Test using serum (A and B) and capillary whole blood (C and D). CRP, C-reactive protein; WB, whole blood. testing (risk ratio 0.75; 95% confidence interval [CI], 0.67–0.83) [8]. therapy should not be routinely offered if the CRP concentration is < Reductions in unnecessary antibiotic prescriptions are desirable given 20 mg/L, that a delayed antibiotic prescription (i.e. for use at a later the growing problem of antibiotic resistance [23]. Moreover, over half date, if symptoms worsen) should be considered if the CRP concentra- of patients presenting to primary care with lower respiratory tract in- tion is 20–100 mg/L, and that antibiotic therapy should be offered if the fections are prescribed antibiotics [24], yet there is limited evidence CRP concentration is ≥100 mg/L. that the marginal benefits of these treatments outweigh the risk for Clinical experience with existing POC CRP tests supports their use in harm [25,26]. Although CRP tests are available through local hospital primary care [28]. For example, Matheeussen and colleagues [28] com- pathology services at a lower cost, POC testing allows the findings to be pared a CRP QuikRead POC testing device (Orion Diagnostica, Espoo, discussed between patient and primary care practitioner during the Finland) with a central laboratory method for CRP measurement in over consultation and may be helpful in persuading patients to agree to a 2900 serum samples from adult patients presenting to primary care with non-antibiotic approach [13,27]. symptoms of lower respiratory infection. The POC device had a sensitivity UK NICE guidelines recommend that POC CRP testing is considered of 92.2% and specificity of 99.4% at a CRP cutoff of ≥30 mg and was for people presenting to primary care with symptoms of lower re- deemed a good candidate for POC CRP testing. Interestingly, the slope of spiratory tract infection. The guidelines recommend that antibiotic the Passing–Bablok analysis suggested a 6% underestimation of CRP 42 Table 3 Method comparison: mean bias for CRP measured with the cobas POC CRP Test (cobas b 101 POC system) using (A) capillary whole blood and serum (data from all sites combined), (B) EDTA K2/K3 and Li-heparin in whole blood and plasma samples (data from POC site 1 only), on the CRPNX (cobas c 501 laboratory module). A V. Scharnhorst, et al. Lot CRP range/*MDP (mg/L) Capillary whole blood Serum n Mean bias, % (95% CI) n Mean bias, % (95% CI) 1 3–200 121 −2.05 (−15.64, 11.54) 115 −2.05 (−10.54, 6.43) 200–400 15 −2.41 (−16.50, 11.68) 15 −0.67 (−17.59, 16.25) 5* 136 2.88 (0.32, 5.44) 130 −0.32 (−2.66, 2.02) 10* 136 −0.64 (−1.90, 0.62) 130 −1.47 (−2.58, −0.36) 2 3–200 119 −2.39 (−15.27, 10.48) 116 −1.44 (−12.19, 9.32) 200–400 17 −4.43 (−11.27, 2.40) 16 −1.34 (−11.53, 8.85) 5* 136 2.44 (−0.16, 5.04) 132 2.89 (0.61, 5.18) 10* 136 −0.99 (−2.24, 0.27) 132 −0.04 (−1.13, 1.05) 3 3–200 116 −3.26 (−13.06, 6.54) 112 −1.43 (−10.86, 8.01) 200–400 12 −6.38 (−18.63, 5.88) 11 −2.63 (−8.29, 3.04) 5* 128 −1.38 (−3.88, 1.12) 123 2.26 (0.31, 4.21) 10* 128 −2.81 (−4.03, −1.58) 123 −0.22 (−1.16, 0.72) B Lot CRP range/*MDP Whole blood Plasma (mg/L) EDTA K2 EDTA K3 Li-heparin EDTA K2 EDTA K2 Li-heparin 43 n Mean bias, 95% CI (%) n Mean bias, 95% CI (%) n Mean bias, 95% CI (%) n Mean bias, 95% CI (%) n Mean bias, 95% CI (%) n Mean bias, 95% CI (%) 1 3–200 62 −1.79 (−14.11, 10.54) 62 −4.05 (−13.47, 5.37) 62 −2.58 (−12.28, 7.11) 58 −3.24 (−12.72, 6.24) 56 −4.68 (−14.04, 4.68) 53 −2.72 (−14.55, 9.10) 200–400 10 −3.02 (−10.43, 4.40) 10 −5.22 (−14.43, 3.99) 10 −1.55 (−9.99, 6.88) 10 −2.96 (−17.75, 11.83) 10 −5.14 (−13.07, 2.79) 10 −2.87 (−8.33, 2.59) 5* 72 1.55 (−2.67, 5.77) 72 −0.89 (−4.88, 3.11) 72 0.30 (−2.42, 3.02) 68 −0.97 (−5.01, 3.08) 66 −2.33 (−6.16, 1.49) 63 1.23 (−3.95, 6.41) 10* 72 −1.02 (−2.95, 0.92) 72 −3.32 (−5.09, −1.55) 72 −1.72 (−3.15, −0.29) 68 −2.64 (−4.48, −0.80) 66 −4.09 (−5.78, −2.39) 63 −1.61 (−3.94, 0.72) 2 3–200 62 −2.50 (−14.06, 9.06) 62 −4.38 (−17.06, 8.30) 62 −2.30 (−15.59, 11.00) 58 −3.17 (−13.92, 7.59) 57 −4.85 (−16.08, 6.38) 53 −2.14 (−11.89, 7.61) 200–400 10 −3.24 (−11.02, 4.54) 10 −4.38 (−12.65, 3.90) 10 −3.63 (−15.66, 8.40) 10 −5.31 (−15.03, 4.42) 10 −3.31 (−10.16, 3.54) 10 −3.51 (−9.75, 2.73) 5* 72 1.83 (−1.77, 5.44) 72 −0.24 (−3.91, 3.44) 72 2.98 (0.04, 5.91) 68 2.97 (−0.95, 6.88) 67 −0.15 (−3.77, 3.47) 63 1.36 (−3.10, 5.82) 10* 72 −1.40 (−3.09, 0.28) 72 −3.26 (−4.96, −1.56) 72 −1.00 (−2.51, 0.52) 68 −1.76 (−3.55, 0.03) 67 −3.40 (−5.07, −1.73) 63 −1.29 (−3.27, 0.70) CI, confidence interval; CRP, C-reactive protein; MDP, medical decision point; PL, plasma; WB, whole blood. Clinical Biochemistry 71 (2019) 38–45 V. Scharnhorst, et al. Clinical Biochemistry 71 (2019) 38–45 Fig. 3. Method comparison: Bland-Altman analysis of CRP values measured with the cobas POC CRP Test (Lot 1) on the cobas b 101 system using (A) serum (B) capillary whole blood, compared with values measured using matched serum samples on the cobas c 501 laboratory module with the CRPNX test. Data are for all sites combined. Relative difference (%) calculated as (Y–X)*100/X, where X = CRPNX test (serum) and Y = POC CRP Test. CRP, C-reactive protein. levels by the POC device (Passing–Bablok slope 0.94; 95% CI, 0.93–0.95) collection, analysis, and interpretation of data; in the writing of the [28]. For comparison, in the present study there was strong agreement report; or in the decision to submit the report for publication. All au- between the POC CRP Test and the central laboratory reference test thors have accepted responsibility for the entire content of this sub- (CRPNX reagent); for serum samples, the slope of the weighted Deming mitted manuscript and approved submission. regression was 0.97 and the upper 95% CI ranged from 0.98 to 0.99 Andrea Lutz, Uwe Graser and Daniel Püntener are employees of, and across lots, indicating a 3% underestimation of CRP concentrations. own stocks/shares, in Roche Diagnostics International Ltd. Dr. Strengths of the present study include the use of clinical samples Alquézar-Arbé received an honorarium from Roche Diagnostics for his derived from prospectively enrolled individuals, and that precision was participation as investigator of the study. Dr. Noordzij received an measured according to CLSI EP05-A3 guidelines [18]. Importantly, the honorarium from Roche for an IL-6 Advisory Board on September method comparison was performed by healthcare professionals in the 20–21, 2016 in Frankfurt, Germany. POC environment and laboratory operators at a reference laboratory (i.e. relevant user groups), and used a relevant reference test (CRPNX). Appendix A. Supplementary data A reference range for CRP specific to the cobas POC CRP Test was not determined in the present study; therefore, further studies are war- Supplementary data to this article can be found online at https:// ranted to define a reference range and explore the utility of this POC doi.org/10.1016/j.clinbiochem.2019.06.009. CRP Test in clinical populations. As this study was conducted in hos- pital settings, additional studies conducted in a primary care population References would be beneficial to validate our findings and further examine the utility of the POC CRP Test in clinical populations. Use of the POC CRP [1] S. Black, I. Kushner, D. Samols, C-reactive protein, J. Biol. Chem. 279 (2004) 48487–48490. Test would be expected to have equivalent benefit in clinical popula- [2] W. Ansar, S. Ghosh, C-reactive protein and the biology of disease, Immunol. Res. 56 tions, particularly since all healthcare professionals are able to learn to (2013) 131–142. use the device after a short training session. [3] C. Gabay, I. Kushner, Acute-phase proteins and other systemic responses to in- flammation, N. Engl. J. Med. 340 (1999) 448–454. In conclusion, these findings indicate that healthcare professionals [4] P. Póvoa, L. Coelho, E. Almeida, A. Fernandes, R. Mealha, P. Moreira, et al., C- can obtain precise and reproducible CRP values with the cobas POC reactive protein as a marker of infection in critically ill patients, Clin. Microbiol. CRP Test that show very good correlation with laboratory measure- Infect. 11 (2005) 101–108. [5] J.P. Haran, F.L. Beaudoin, S. Suner, S. Lu, C-reactive protein as predictor of bac- ments. Importantly, operators considered the system convenient for use terial infection among patients with an influenza-like illness, Am. J. Emerg. Med. 31 in the POC environment. (2013) 137–144. [6] R.M. Hopstaken, J.W. Muris, J.A. Knottnerus, A.D. Kester, P.E. Rinkens, G.J. Dinant, Contributions of symptoms, signs, erythrocyte sedimentation rate, and C-reactive protein to a diagnosis of pneumonia in acute lower respiratory tract infection, Br. J. Acknowledgments Gen. Pract. 53 (2003) 358–364. [7] S.F. van Vugt, B.D.L. Broekhuizen, C. Lammens, N.P.A. Zuithoff, P.A. de Jong, S. Coenen, et al., Use of serum C reactive protein and procalcitonin concentrations This study was funded by Roche Diagnostics International Ltd. in addition to symptoms and signs to predict pneumonia in patients presenting to (Rotkreuz, Switzerland). Medical writing assistance was provided by primary care with acute cough: diagnostic study, BMJ 346 (2013) f2450. [8] Y. Huang, R. Chen, T. Wu, X. Wei, A. Guo, Association between point-of-care CRP David Evans, PhD, and Louise Kelly, BSc, (Gardiner-Caldwell testing and antibiotic prescribing in respiratory tract infections: a systematic review Communications, Macclesfield, UK), and was funded by Roche and meta-analysis of primary care studies, Br. J. Gen. Pract. 63 (2013) e787–e794. Diagnostics International Ltd. (Rotkreuz, Switzerland). [9] R. Aabenhus, J.U. Jensen, K.J. Jørgensen, A. Hróbjartsson, L. Bjerrum, Biomarkers as point-of-care tests to guide prescription of antibiotics in patients with acute The funding organization played no role in the study design; in the 44 V. Scharnhorst, et al. Clinical Biochemistry 71 (2019) 38–45 respiratory infections in primary care, Cochrane Database Syst. Rev. 11 (2014) Updated 2014, Available at: https://www.westgard.com/biodatabase1.htm , CD010130. Accessed date: 10 October 2018. [10] P. Little, B. Stuart, N. Francis, E. Douglas, S. Tonkin-Crine, S. Anthierens, et al., [20] F. Khorvash, F. Abdi, K. Dialami, Can serum procalcitonin and C-reactive protein as Effects of internet-based training on antibiotic prescribing rates for acute re- nosocomial infection markers in hospitalized patients without localizing signs? J. spiratory-tract infections: a multinational, cluster, randomized, factorial, controlled Res. Med. Sci. 16 (2011) 1280–1285. trial, Lancet 82 (2013) 1175–1182. [21] B. Andreola, S. Bressan, S. Callegaro, Procalcitonin and C-reactive protein as di- [11] R. Oppong, M. Jit, R.D. Smith, C.C. Butler, H. Melbye, S. Mölstad, et al., Cost- agnostic markers of severe bacterial infections in febrile infants and children in the effectiveness of point-of-care C-reactive protein testing to inform antibiotic pre- emergency department, Pediatr. Infect. Dis. J. 26 (2007) 672–677. scribing decisions, Br. J. Gen. Pract. 63 (2013) e465–e471. [22] R. Hunter, Cost-effectiveness of point-of-care C-reactive protein tests for respiratory [12] J.W.L. Cals, A.J. Ament, K. Hood, C.C. Butler, R.M. Hopstaken, G.F. Wassink, et al., tract infection in primary care in England, Adv. Ther. 32 (2015) 69–85. C-reactive protein point-of-care testing and physician communication skills training [23] European Centre for Disease Prevention and Control (ECDC). Surveillance of anti- for lower respiratory tract infections in general practice: economic evaluation of a microbial resistance in Europe 2016, Annual Report of the European Antimicrobial cluster randomized trial, J. Eval. Clin. Pract. 17 (2011) 1059–1069. Resistance Surveillance Network (EARS-Net), ECDC, Stockholm, 2017. [13] National Clinical Guideline Centre, Pneumonia: Diagnosis and management of [24] C.C. Butler, K. Hood, T. Verheij, P. Little, H. Melbye, J. Nuttall, et al., Variation in community- and hospital-acquired pneumonia in adults, Clin. Guidel. 191 (2014) 3 antibiotic prescribing and its impact on recovery in patients with acute cough in December. Available at: https://www.nice.org.uk/guidance/cg191/evidence/full- primary care: prospective study in 13 countries, BMJ 338 (2009) b2242. guideline-193389085 , Accessed date: 10 October 2018. [25] S.M. Smith, J. Smucny, T. Fahey, Antibiotics for acute bronchitis, JAMA 312 (2014) [14] M.C. Minnaard, A.C. van de Pol, B.D. Broekhuizen, T.J. Verheij, R.M. Hopstaken, 2678–2679. S. van Delft, Analytical performance, agreement and user-friendliness of five C- [26] P. Little, B. Stuart, M. Moore, S. Coenen, C.C. Butler, M. Godycki-Cwirko, et al., reactive protein point-of-care tests, Clin. Lab. Invest. 73 (2013) 627–634. GRACE consortium. Amoxicillin for acute lower-respiratory-tract infection in pri- [15] N. Brouwer, J. van Pelt, Validation and evaluation of eight commercially available mary care when pneumonia is not suspected: a 12-country, randomised, placebo- point-of-care CRP methods, Clin. Chim. Acta 439 (2015) 195–201. controlled trial, Lancet Infect. Dis. 13 (2013) 123–129. [16] [No authors listed], Point-of-care CRP testing in the diagnosis of pneumonia in [27] J.R. Huddy, M.Z. Ni, J. Barlow, A. Majeed, G.B. Hanna, Point-of-care C reactive adults, Drug Ther. Bull. 54 (2016) 117–120. protein for the diagnosis of lower respiratory tract infection in NHS primary care: a [17] World Medical Association, WMA Declaration of Helsinki – Ethical Principles for qualitative study of barriers and facilitators to adoption, BMJ Open 6 (2016) Medical Research Involving Human Subjects. 9 July 2018, Available at: https:// e009959. www.wma.net/policies-post/wma-declaration-of-helsinki-ethical-principles-for- [28] V. Matheeussen, V. Van Hoof, K. Loens, C. Lammens, A. Vanderstraeten, S. Coenen, medical-research-involving-human-subjects/ , Accessed date: 10 October 2018. et al., GRACE consortium. Analytical performance of a platform for point-of-care [18] Clinical and Laboratory Standards Institute, CLSI EP05-A3. Evaluation of Precision CRP testing in adults consulting for lower respiratory tract infection in primary of Quantitative Measurement Procedures, 3rd ed, NCCLS, Wayne, PA, 2014. care, Eur. J. Clin. Microbiol. Infect. Dis. (2018), https://doi.org/10.1007/s10096- [19] J. Westgard, Quality Requirements: Desirable Specifications for Total Error, 018-3253-3 May 10. (Epub ahead of print). Imprecision and Bias, Derived from Intra- and Inter-Individual Biologic Variation. 45