- Email: [email protected]
AutionMax for urine samples screening: A multicenter Spanish study
CLB-09471; No. of pages: 5; 4C: Clinical Biochemistry xxx (2017) xxx–xxx
Contents lists available at ScienceDirect
Clinical Biochemistry journal homepage: www.elsevier.com/locate/clinbiochem
Comparison of automated devices UX-2000 and SediMAX/AutionMax for urine samples screening: A multicenter Spanish study Catalina Sánchez-Mora a,⁎, Delia Acevedo b, Maria Amelia Porres c, Ana María Chaqués b, Javier Zapardiel d, Aurelia Gallego-Cabrera a, Jose María López b, Jose María Maesa a a
Clinical Biochemistry Department, Virgen Macarena University Hospital, Seville, Spain Clinical Biochemistry Department, Clinical Analysis Service, Dr. Peset University Hospital, Valencia, Spain Clinical Analysis Department, Jiménez Díaz Foundation University Hospital, Madrid, Spain d Microbiology Department, Jiménez Díaz Foundation University Hospital, Madrid, Spain b c
a r t i c l e
i n f o
Article history: Received 16 November 2016 Received in revised form 1 February 2017 Accepted 5 February 2017 Available online xxxx Keywords: Urine screening Test strips UX-2000 SediMAX Aution max Sediment
a b s t r a c t Objectives: In this study we aim to compare UX2000 (Sysmex Corp, Japan) and SediMAX/AutionMax (Arkray Factory Inc., Japan), totally automatized analyzers, against Fuchs-Rosenthal counting chamber, the gold standard technique for sediment analysis. Design and methods: Urine samples of 1454 patients from three Spanish hospitals were assessed for red and white blood cells (RBC; WBC) using three different techniques: flow cytometry, image-based method and FuchsRosenthal counting chamber. Test strip results were subjected to concordance evaluation. Agreement was assessed by Cohen's weighted kappa for multinomial results. Sensitivity (SE) and specificity (SP) were calculated. Results: The categorization of the results showed that UX-2000 had higher concordance over SediMAX for WBC (0.819 vs. 0.546) and similar for RBC (0.573 vs. 0.630). For RBC, UX-2000 had higher SE (92.7% vs. 80.3%) but lower SP (77.1% vs. 87.4%), and showed higher both SE (94.3% vs. 76.7%) and SP (94.7% vs. 88.2%) for WBC. Inter-devices test strip agreement was substantial (kappa N 0.600) for all variables except for bilirubin (kappa: 0.598). Intra-device test strip agreement was similar for UX2000 and SediMAX with regard to RBC (kappa: 0.553 vs. 0.482) but better for UX2000 with regard to WBC (0.688 vs. 0.465). Conclusions: Both analyzers studied are acceptable for daily routine lab work, even though SediMAX is easier to use in laboratories thanks to its lower maintenance procedure. UX-2000 has shown to have better concordance with the gold standard method. However, it needs some improvements such as an image module in order to decrease manual microscopy review for urine samples. © 2017 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.
1. Introduction At present, urine sample analysis is one of the most prescribed tests in both primary care and at the emergency room. Urine is an easily obtainable sample that provides a lot of information to physicians. It is a fundamental test for the detection and follow-up of various urinary tract and renal diseases as well as chronic diseases. This test includes physical and biochemical analysis such as density, color, turbidity, pH, glucose, ketones bodies, proteins, nitrites, leucocyte esterase, hemoglobin, bilirubin and urobilinogen; and urine particle analysis such as red and white blood cells (RBC, WBC), bacteria, yeasts, crystals, casts, epithelial cells, etc. [1,2]. Between these parameters, RBC and WBC are widely used for the determination of pathological urine. Screening is performed by test strip ⁎ Corresponding author at: Clinical Biochemistry Department, Virgen Macarena University Hospital, (Avenida Doctor.Fedriani n° 3 CP: 46017 Sevilla), Sevilla, Spain. E-mail address: [email protected] (C. Sánchez-Mora).
systems that yield positive or negative results, hence determining whether standard urinary sediment analysis should be performed. Sediment analysis could be performed either manually using standardized systems or by means of automated devices [3]. We focused on RBC and WBC as in Spanish Health System these are the variables considered relevant in the assessment of urines for Emergency Rooms and Clinical Biochemistry departments. Although bacteria count is important in Microbiology departments for urines, they use other methods to assess bacteriuria. Manual microscopy assessment is considered as the gold standard even though it is a burdensome and time-consuming technique for specialized staff, a fact that should be taken into account in terms of laboratory costs [4]. On the other hand, automated analysis provides lower subjectivity on results interpretation, thereby improving accuracy [5] and optimizing time. Regarding automatization, there are many devices for urinary sediment analysis based on different technologies. Digital optic microscopy includes a system of cameras that photographs the sample and features an automatic recognition of elements. Flow
http://dx.doi.org/10.1016/j.clinbiochem.2017.02.005 0009-9120/© 2017 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.
Please cite this article as: C. Sánchez-Mora, et al., Comparison of automated devices UX-2000 and SediMAX/AutionMax for urine samples screening: A multicenter Spanish study, Clin Biochem (2017), http://dx.doi.org/10.1016/j.clinbiochem.2017.02.005
2
C. Sánchez-Mora et al. / Clinical Biochemistry xxx (2017) xxx–xxx
Table 1 Patient's characteristics. n: number of patients; P25: percentile 25; P75: percentile 75; 95%CI: 95% confidence interval. Variable
Global
Center 1
Center2
Center3
Patients [n] Age [median P25–P75] % gender (95%CI) Female Male % procedence (95%CI) Primary care Secondary care In-hospital care Emergency room
1454 54.0 (38.0–70.0)
517 56.0 (43.0–70.0)
502 50.0 (34.0–69.0)
435 54.0 (37.0–72.0)
62.9 (60.4–65.3) 37.1 (34.7–39.7)
59.8 (55.5–63.9) 40.2 (36.1–44.5)
54.0 (49.6–58.3) 46.0 (41.7–50.4)
76.8 (72.6–80.5) 23.2 (19.5–27.4)
44.6 (42.0–47.1) 11.2 (9.7–12.9) 6.5 (5.3–7.9) 37.7 (35.3–40.2)
41.3 (37.0–45.5) 31.6 (27.7–35.7) 18.2 (15.1–21.8) 8.9 (6.8–11.7)
0.0 (0.0–0.8) 0.0 (0.0–0.8) 0.0 (0.0–0.8) 100.0 (99.2–100.0)
100.0 (99.1–100.0) 0.0 (0.0–0.9) 0.0 (0.0–0.9) 0.0 (0.0–0.9)
cytometry and fluorescence devices examine every particle and cell present in a laminar flow sample and accurately count and classify forming elements. Recently, several companies have developed completely automated analyzers integrating in a sole platform both physic-chemical testing and sediment particles analysis. These analyzers can remove the interobserver variability while optimizing laboratory workflow and improving accuracy and precision of the test [6]. Although manual sediment analysis cannot be completely excluded, they significantly reduce the number of samples to be reviewed. European guidelines [7] establish that in the new urine laboratory workflow strategy it is crucial to rationalize the combination of physicochemical, bacterial and sediment particle automated count and the preservation of the sample as done in non-centrifugation based methods. In addition, international recommendations encourage the use of count chambers when assessing the performance of automated devices, both based on either flow cytometry or imaging techniques [8]. The primary aim of this study was to assess the performance of the UX-2000 (flow citometry) and SediMAX/Aution Max systems (imagebased method), and the Fuchs-Rosenthal count chamber for urine RBC and WBC in routine clinical laboratory setting. The secondary aim was to assess correlation between test strips and urinary sediment. 2. Material and methods 2.1. Study design This is a multicenter, observational and prospective study of automatic analyzers for test strips and sediment analyses. The study was presented and approved by each Institution's Ethics Board Committee. Three Spanish centers participated in the study: Dr. Peset University Hospital (center 1), Virgen Macarena University Hospital (center 2) and Jiménez Díaz Foundation University Hospital (center 3). From August 2015 to February 2016, 1454 urine samples were randomly collected from those received at the department of Clinical Analysis/Clinical Biochemistry and analyzed within 2 h. Study samples had to be a minimum of 8 mL volume in order to be assessed by both devices. Nonhomogenous, aggregates or visible particle-containing samples were excluded in order to avoid erratic results. The samples were processed changing the order of the different techniques every ten to fifteen samples to avoid bias. The samples were analyzed by expert staff of each hospital. 2.2. Patient samples Patient median age was 54.0 (P25–P75;38.0–70.0) years, were usually women [62.9% (CI 95%: 60.4–65.3)] and came mostly from primary care [44.6% (42.0–47.1)] and the Emergency Room [37.7% (35.3–40.2)] and in a lower percentage from secondary care [11.2% (9.7–12.9)] and in-hospital care [6.5% (5.3–7.9)]. Center 1 assessed a total of 517 patient samples with a median age of 56 years and 59.8% were women. The 41.3% of urine samples came from primary care, 31.6% from secondary
care, 18.2% from inhospital care and 8.9% from the emergency room. Center 2 assessed a total of 502 patient samples with a median age of 50 years, 54.0% were women and 100% came from the emergency room. Center 3 analyzed a total of 435 patient samples with a median age of 54 years, 76.8% were women and 100% came from a primary care setting (Table 1). 2.3. Urines assessment Samples were assessed by Fuchs-Rosenthal count chamber (SKC Europe GMBH. Frankfurt am Main, Germany), UX-2000 (Sysmex, Kobe, Japan) and SediMAX (77 Elektronika, Hungary)/Aution Max (Arkray Factory Inc., Japan) according to provider instructions, for count of RBC and WBC. For manual particle counts, Fuchs-Rosenthal chambers were used as the gold standard method. Fifteen microliters of urine were applied on the chamber ensuring that the entire surface was properly impregnated avoiding air bubbles formation. Five big squares corresponding to 1 μL were assessed [9]. UX-2000 analyzes non-processed (fresh) samples by flow cytometry [10]. Briefly, it identifies particles and cells by light scatter and classifies them according to their size and cellular complexity. Particles are stained with a fluorescent compound that gives more detailed information about nucleic acids. SediMAX/AutionMax analyses pre-centrifuged samples [11,12]. They are analyzed by an integrated microscope that takes images and compares them to pre-defined libraries of identified particles. Output was modified from high/low power field to particles/μL. Both devices provided a test stripe analysis of semi-quantitative values for glucose, protein, bilirubin, urobilinogen, ketones, pH, RBC, WBC and nitrites. 2.4. Statistical analysis Data was retrospectively retrieved at completion of the study. Center 3 was unable to retrieve SediMAX data because it was recorded and
Table 2 Cell counts for the overall population and for each center. Results shown as median (percentile 25–percentile 75). RBC: red blood cells; WBC: white blood cells. Variable
Fuchs-Rosenthal
UX-2000
SediMAX
RBC (particles/μL) Global Center 1 Center 2 Center 3
2.0 (0.0–8.0) 2.0 (0.0–6.0) 6.0 (2.0–18.5) 0.0 (0.0–3.0)
9.2 (0.0–31.1) 7.8 (3.6–22.6) 25.9 (14.2–60.3) 0.0 (0.0–0.0)
2.9 (0.0–20.3) 1.3 (0.0–4.6) 11.6 (2.9–52.8) –
WBC (particles/μL) Global Center 1 Center 2 Center 3
5.0 (1.0–21.0) 4.0 (1.0–12.0) 9.0 (2.0–36.5) 4.0 (0.0–22.3)
5.8 (0.7–31.4) 5.0 (1.9–18.1) 14.9 (5.3–63.3) 0.0 (0.0–30.5)
8.7 (1.3–46.5) 2.6 (0.7–9.9) 31.9 (8.7–129.8) –
Please cite this article as: C. Sánchez-Mora, et al., Comparison of automated devices UX-2000 and SediMAX/AutionMax for urine samples screening: A multicenter Spanish study, Clin Biochem (2017), http://dx.doi.org/10.1016/j.clinbiochem.2017.02.005
C. Sánchez-Mora et al. / Clinical Biochemistry xxx (2017) xxx–xxx
3
Fig. 1. Patient distribution for red blood cells according to Fuchs-Rosenthal (white), UX-2000 (grey) and SediMAX (black) results in A) Center 1; B) Center 2; C) Center 3. RBC: red blood cells.
stored as a range value as opposed to a single unique value. For statistical purposes, RBC and WBC, in strips and sediment results, were categorized according to cut-offs of test strips given by provider and also dichotomized as positive/negative as follows: RBC categories (0–19; 20–59; 60–299; ≥ 300/μL) and RCB positives results for ≥ 20/μL; WBC categories (0–74; 75–249; 250–499; ≥ 500/μL) and WBC positives results for ≥75/μL. All study variables were subjected to descriptive analysis. UX-2000 and SediMAX/AutionMax concordance against FuchsRosenthal chamber was assessed by Cohen's weighted kappa coefficient for qualitative variables. Cohen's coefficient cutoffs were established as slight (0.00–0.20), fair (0.20–0.40), moderate (0.40–0.60), substantial (0.60–0.80) and almost perfect agreement (0.80–1.00). Diagnostic performance (sensitivity, specificity, positive and negative predictive values) were calculated for sediment RBC and WBC determination in both instruments. Statistical analyses were performed using NCSS 10 software (NCSS, LLC Utah, USA) and Cohen's weighted kappa coefficient using VassarStats (http://vassarstats.net/kappa.html). All tests were significant with an α = 0.050.
whereas UX-2000 showed significant higher concordance with FuchsRosenthal than SediMAX (0.819 vs. 0.546) for WBC. In the biochemical analysis of urine samples with test strips, all variables had a concordance over 0.60 (data not shown), showing a substantial agreement between both automatic analyzers. The only exception was bilirubin, which had a concordance of 0.598. Concordance between each automatic analyzer and own test strip system is depicted on Table 3. Both devices had a good concordance for RBC and WBC categorization analysis. Particularly, SediMAX had quite higher concordance than UX-2000 (0.553 vs. 0.482) for RBC and much lower than UX-2000 when assessing WBC (0.465 vs. 0.688). These results were maintained when center-specific sub-analysis was performed (Table 3). When RBC and WBC values were dichotomized into positive and negative results (Table 4), the UX-2000 shown higher sensitivity (92.7% vs. 80.3%) than SediMAX for RBC, but worse specificity (77.1% vs. 87.4%). For WBC count, UX2000 had higher sensitivity (94.3% vs. 76.7%) and specificity (94.7% vs. 88.2%) than SediMAX. 4. Discussion
3. Results For both analyzers and gold standard, center 2 had higher RBC and WBC values, while center 3 had the lowest counts (Table 2). These differences were greater when samples were analyzed using the UX2000 system (Figs. 1 and 2). In the categorical analysis of forming elements (Table 3), UX-2000 and SediMAX had similar concordance for RBC (0.573 vs. 0.630)
In urine sediment assessments, each laboratory must establish the most suitable procedure according to patient's clinical characteristics which can be classified by department receiving these samples [1]. At present, it is widely accepted that automatic analyzers may reduce the number of hours spent on manual review, as reported in several studies [13,14]. There are many comparative studies that evaluate currently available automatic analyzers [15–17]. Wesarachkitti et al. [15]
Fig. 2. Patient distribution for white blood cells according to Fuchs-Rosenthal (white), UX-2000 (grey) and SediMAX (black) results in A) Center 1; B) Center 2; C) Center 3. WBC: white blood cells.
Please cite this article as: C. Sánchez-Mora, et al., Comparison of automated devices UX-2000 and SediMAX/AutionMax for urine samples screening: A multicenter Spanish study, Clin Biochem (2017), http://dx.doi.org/10.1016/j.clinbiochem.2017.02.005
4
C. Sánchez-Mora et al. / Clinical Biochemistry xxx (2017) xxx–xxx
Table 3 Concordance for categorized RBC and WBC. κ:weighted kappa; 95%CI: 95% confidence interval; RBC: red blood cells; WBC: white blood cells. Variable
UX-2000 cytometry - UX-2000 test stripe
SediMAX image - SediMAX test stripe
UX-2000 cytometry - Fuchs-Rosenthal
SediMAX image - Fuchs-Rosenthal
RBC Global Center 1 Center 2 Center 3
0.482 (0.443–0.521) 0.557 (0.488–0.625) 0.333 (0.282–0.383) 0.465 (0.356–0.574)
0.553 (0.506–0.599) 0.682 (0.615–0.750) 0.430 (0.367–0.493) –
0.573 (0.528–0.618) 0.679 (0.605–0.752) 0.451 (0.388–0.514) 0.618 (0.498–0.739)
0.630 (0.577–0.683) 0.826 (0.759–0.893) 0.504 (0.433–0.575) –
WBC Global Center 1 Center 2 Center 3
0.688 (0.655–0.720) 0.639 (0.575–0.704) 0.682 (0.631–0.732) 0.707 (0.652–0.762)
0.465 (0.409–0.521) 0.454 (0.381–0.528) 0.459 (0.391–0.527) –
0.819 (0.783–0.856) 0.802 (0.723–0.881) 0.802 (0.746–0.859) 0.843 (0.784–0.902)
0.546 (0.479–0.613) 0.681 (0.550–0.811) 0.482 (0.404–0.561) –
compared UX-2000 and an image-based automatic analyzer (Cobas 6500) on 258 patients considering agreement as same result or ± 1 grade, showing a great performance for both devices. In assessing 214 samples by conventional method and with a flow cytometry-based device, Manoni et al. [16], that made a comparison between UF-1000i and quantitative microscopy, found a nearly perfect concordance between both methods. Finally, Demet et al. [17] compared two digital imagebased automatic analyzers (Iris iQ200 and Dirui FUS-200) in 209 samples concluding that RBC and WBC were similar to those on manual counting. Our multicenter study assessed a representative amount (n = 1454) of urine sample to gain deeper and more robust insight into the concordance between automatic analyzers and the gold standard technique, Fuchs-Rosenthal count chamber. We did not assess feasibility, linearity, or accuracy of UX-2000 and SediMAX as this was not our main objective, and it has been widely described in previous published studies [10,15, 18]. Particularly, Wesarachkitti et al. [15] concluded that UX-2000 improves reproducibility and reduced workload due to its precision and linearity, whereas from the study of 400 urine samples, Bottini et al. [18] showed that SediMAX could yield a better workflow due to its precision and accuracy. With regard to forming elements (WBC and RBC) assessed using the Fuchs-Rosenthal chamber, our results show that as sample complexity increases (presenting abnormal or pathological values), discrepancy widens. Consequently, particle counting is increased, leading to significant differences between chamber counts and automatic analyzers results. Overall, differences among hospital's samples were more significant for UX-2000 than for SediMAX or Fuchs-Rosenthal results, which may imply that UX-2000 has a higher discriminatory power, accurately counting and classifying samples. This may be due to inter-rate variability, sedimentation of formed elements, and non-homogeneous samples. The conversion factor from particle/field to particle/μL also decreases the accuracy of manual counting. When the unit is converted into another unit, a small counting deviation is amplified. Regarding the test strip, kappa coefficients show that there is substantial concordance between both automatic devices (UX-2000 test strips scanner and Aution Max), being lower (moderate) for bilirubin. This somewhat lower concordance may be due to the low prevalence of positive cases for bilirubin in the general population, which mathematically implies a lower kappa coefficient.
When assessing RBC results by test strips and sediment for each automatic analyzer, a moderate concordance was found for both devices. This could be explained by the fact that RBC hemolysis cannot be detected by automatic analyzers that count intact particles [19]. For WBC concordance has been shown to be substantial for UX-2000 and moderate for SediMAX/Aution Max. This may be explained by the lower sensitivity of SediMAX when counting WBC, leading to a higher discrepancy between counting and test strips. Negative predictive values for both UX-2000 and SediMAX are good for RBC (98.3% vs. 95.1%) and WBC (99.2% vs. 96.4%), as stated by other authors [20], an essential feature for avoiding misclassification of positive urine samples as negative. Positive predictive value of UX-2000 for RBC is quite low (42.6%). This may be explained by scatter plot analysis, where some small DNAless particles such as yeasts or small crystals could potentially be misclassified as RBC, even if this could be solved by changing wavelength [20,21]. For WBC, positive predictive value of SediMAX is low (47.7%). This could be due to sample centrifugation prior to analysis by this device that may form aggregates or even produce WBC lysis [22]. Our study has some limitations. Samples from different centers had different clinical and pathological features which, despite allowing testing device performance in a wide range of real conditions, make comparisons between centers quite difficult. Moreover, in one of the centers, SediMAX sediment data could not be retrieved, which lowers effective sample size leading to a lower statistical power. However, these limitations do not reduce the relevance to the study results. Regarding technical and practical issues, UX-2000 has proven to be a compact device that does work in fresh samples without the need for centrifugation, hence preserving all particles. However, this device needs a higher urine volume in automatic mode (5 mL vs. 2 mL for SediMAX). This, while not being a problem for daily practice, could pose some difficulties for pediatric urine samples, although this short coming can be simply resolved by switching to manual mode. For daily practice, it seemed that UX-2000 needed more exhaustive maintenance (automatic filter washes after hematic or pyuric samples) which if not met, could lead to workflow interruption and consequently length in response time to get back lab results. Altogether, UX-2000 showed higher performance for RBC, WBC and epithelial cells. SediMAX was superior for detecting crystals, yeast and cast (data not shown).
Table 4 Diagnosis performance of UX-2000 and SediMAX according to Fuchs-Rosenthal results, using RBC ≥20/μl or WBC ≥75/μl as positive results. Data shown as percentage (95% confidence interval). RBC: red blood cells; WBC: white blood cells. Variable
Sensitivity Specificity Positive predictive value Negative predictive value
RBC
WBC
UX-2000
SediMAX
UX-2000
SediMAX
92.7 (88.2–95.6) 77.1 (74.6–79.4) 42.6 (38.2–47.2) 98.3 (97.2–99.0)
80.3 (73.7–85.7) 87.4 (84.9–89.6) 59.5 (53.1–65.6) 95.1 (93.2–96.5)
94.3 (89.4–97.1) 94.7 (93.3–95.9) 71.7 (65.4–77.4) 99.2 (98.4–99.6)
76.7 (67.9–83.7) 88.2 (85.8–90.3) 47.7 (40.5–54.9) 96.4 (94.8–97.6)
Please cite this article as: C. Sánchez-Mora, et al., Comparison of automated devices UX-2000 and SediMAX/AutionMax for urine samples screening: A multicenter Spanish study, Clin Biochem (2017), http://dx.doi.org/10.1016/j.clinbiochem.2017.02.005
C. Sánchez-Mora et al. / Clinical Biochemistry xxx (2017) xxx–xxx
5. Conclusions Automatic devices for urine sediment examination improve reproducibility, accuracy and response time over manual microscopy. Although the pricing of automatic analyzers and their reagents is not low, reduction on analysis time and savings on direct personnel costs makes these devices more cost-effective than manual microscopy assessment of urine sediment. Both analyzers studied are acceptable for daily routine lab work, even though SediMAX is easier to use in laboratories thanks to its lower maintenance procedure. UX-2000 flow cytometry has shown to have a higher concordance with image-based devices and higher correlation with the gold standard method. However, it needs some improvements such as an image module in order to decrease manual microscopy review for urine samples. Discloser All authors disclose no conflict of interest. All authors have contributed to study design, sample analysis, manuscript conception and final approval of the article. This work has been carried out in accordance with Declaration of Helsinki and the International Committee of Medical Journal Editors. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Sysmex España S.L. financially supported the study providing reagents and UX2000 equipment. Acknowledgments We thank Sysmex España for their technical support. References [1] A. Rabinovitch, L. Arzoumanian, K.M. Curcio, B. Dougherty, A.B. Halim, Urinalysis; approved guideline, CLSI Doc GP16-A3, third ed.Clin Lab Stand Institue, Wayne PA, 2009. [2] R. McPherson, J. Ben-Ezra, Basic Examination of Urine, Henry's Clinical Diagnosis and Management by Laboratory Methods, Elsevier Saunders, Philadelphia 2011, pp. 445–479. [3] D. Carlson, B. Statland, Automated urianalysis, Clin. Lab. Med. 8 (1988) 449–461. [4] P. Winkel, B. Statland, J. Jorgenson, Urine microscopy: an illdefined method examined by a multifactorial technique, Clin. Chem. 20 (1974) 436–439. [5] A. Lun, R. Ziebig, F. Priem, P. Sinha, Routine workflow for use of urine strips and urine flow cytometer UF-100 in the hospital laboratory, 1, Clin. Chem. 45 (8) (1999) 1305–1307. [6] Richard Thompson, Andrew Gammie, Debbie Lewis, Rebecca Smith CE. Evidence review: Automated urine screening systems, Cent Evidence-based Purch (March) (2010) 1–46 (CEP10030).
5
[7] European Confederation of Laboratory Medicine, European urinalysis guidelines, Scand J Lab Invest. 60 (2000) 1–96. [8] T. Kouri, A. Gyory, R.M. Rowan, ISLH recommended reference procedure for the enumeration of particles in urine, Lab. Hematol. 9 (2) (2003) 58–63 Available from. http://www.ncbi.nlm.nih.gov/pubmed/12828299. [9] J. Neuendorf, Zellzählung in der Fuchs-Rosenthal-Zählkammer, Das Urinsediment: Mikroskopie, Präanalytik, Auswertung und Befundung, Springer Berlin Heidelberg, Berlin, Heidelberg 2015, pp. 49–52 Available from: http://dx.doi.org/10.1007/9783-662-46074-0_9. [10] V. Khejonnit, B. Pratumvinit, K. Reesukumal, S. Meepanya, C. Pattanavin, P. Wongkrajang, Optimal criteria for microscopic review of urinalysis following use of automated urine analyzer, Clin. Chim. Acta 439 (2015) 1–4 Available from: http://linkinghub.elsevier.com/retrieve/pii/S0009898114004239. [11] Z. Zaman, G.B. Fogazzi, G. Garigali, M.D. Croci, G. Bayer, T. Kránicz, Urine sediment analysis: analytical and diagnostic performance of sediMAX - a new automated microscopy image-based urine sediment analyser, Clin. Chim. Acta 411 (3–4) (2010) 147–154 Available from: 10.1016/j.cca.2009.10.018. [12] M. Miura, T. Kondo, M. Mizuno, T. Morikawa, The basic analytical performance of the fully automated integrated urine analyzer UX-2000 CHM unit, Sysmex J Int. 23 (2013). [13] J.A. Tworek, D.S. Wilkinson, M.K. Walsh, C. The, The rate of manual microscopic examination of urine sediment a College of American Pathologists Q-Probes Study of 11 243 urinalysis tests from 88 institutions, Arch. Pathol. Lab. Med. 132 (December) (2008) 1868–1873. [14] Z. Zaman, Automated urine screening devices make urine sediment microscopy in diagnostic laboratories economically viable, Clin. Chem. Lab. Med. 53 (2015) S1509–S1511. [15] B. Wesarachkitti, V. Khejonnit, B. Pratumvinit, K. Reesukumal, S. Meepanya, C. Pattanavin, et al., Performance evaluation and comparison of the fully automated urinalysis analyzers UX-2000 and Cobas 6500, Lab. Med. 47 (2) (2016 Apr 11) 124–133 Available from: https://labmed.oxfordjournals.org/content/47/2/124.full. [16] F. Manoni, A. Tinello, L. Fornasiero, P. Hoffer, V. Temporin, S. Valverde, et al., Urine particle evaluation: a comparison between the UF-1000i and quantitative microscopy, Clin. Chem. Lab. Med. 48 (8) (2010) 1107–1111. [17] F. Demet, H. Yaşar, M. Koseoğlu, N. Şimşek, H. Yalçına, M. Osman, The comparison of automated urine analyzers with manual microscopic examination for urinalysis automated urine analyzers and manual urinalysis, Pract Lab Med. 5 (2016) 14–20. [18] P.V. Bottini, M.H.M. Martinez, C.R. Garlipp, Urinalysis: comparison between microscopic analysis and a new automated microscopy image-based urine sediment instrument, Clin Lab. 60 (4) (2014) 693–697. [19] J. Jiménez García, G. Ruiz Martín, Estudio de los elementos formes de la orina, in: LABCAM (Asociación Castellano-Manchega de Análisis Clínicos) (Ed.), Estandarización del sedimento urinario 2010, pp. 1–96. [20] S. Mayo, D. Acevedo, C. Quiñones-Torrelo, I. Canós, M. Sancho, Clinical laboratory automated urinalysis: comparison among automated microscopy, flow cytometry, two test strips analyzers, and manual microscopic examination of the urine sediments, J. Clin. Lab. Anal. 22 (4) (2008) 262–270. [21] Y.U. Budak, K. Huysal, Comparison of three automated systems for urine chemistry and sediment analysis in routine laboratory practice, Clin. Lab. 57 (1–2) (2011) 47–52. [22] M. Íñigo, A. Coello, G. Fernández-Rivas, M. Carrasco, C. Marcó, A. Fernández, et al., Evaluation of the SediMax automated microscopy sediment analyzer and the Sysmex UF-1000i flow cytometer as screening tools to rule out negative urinary tract infections, Clin. Chim. Acta 456 (2016) 31–35 Available from: 10.1016/j.cca. 2016.02.016.
Please cite this article as: C. Sánchez-Mora, et al., Comparison of automated devices UX-2000 and SediMAX/AutionMax for urine samples screening: A multicenter Spanish study, Clin Biochem (2017), http://dx.doi.org/10.1016/j.clinbiochem.2017.02.005
Contents lists available at ScienceDirect
Clinical Biochemistry journal homepage: www.elsevier.com/locate/clinbiochem
Comparison of automated devices UX-2000 and SediMAX/AutionMax for urine samples screening: A multicenter Spanish study Catalina Sánchez-Mora a,⁎, Delia Acevedo b, Maria Amelia Porres c, Ana María Chaqués b, Javier Zapardiel d, Aurelia Gallego-Cabrera a, Jose María López b, Jose María Maesa a a
Clinical Biochemistry Department, Virgen Macarena University Hospital, Seville, Spain Clinical Biochemistry Department, Clinical Analysis Service, Dr. Peset University Hospital, Valencia, Spain Clinical Analysis Department, Jiménez Díaz Foundation University Hospital, Madrid, Spain d Microbiology Department, Jiménez Díaz Foundation University Hospital, Madrid, Spain b c
a r t i c l e
i n f o
Article history: Received 16 November 2016 Received in revised form 1 February 2017 Accepted 5 February 2017 Available online xxxx Keywords: Urine screening Test strips UX-2000 SediMAX Aution max Sediment
a b s t r a c t Objectives: In this study we aim to compare UX2000 (Sysmex Corp, Japan) and SediMAX/AutionMax (Arkray Factory Inc., Japan), totally automatized analyzers, against Fuchs-Rosenthal counting chamber, the gold standard technique for sediment analysis. Design and methods: Urine samples of 1454 patients from three Spanish hospitals were assessed for red and white blood cells (RBC; WBC) using three different techniques: flow cytometry, image-based method and FuchsRosenthal counting chamber. Test strip results were subjected to concordance evaluation. Agreement was assessed by Cohen's weighted kappa for multinomial results. Sensitivity (SE) and specificity (SP) were calculated. Results: The categorization of the results showed that UX-2000 had higher concordance over SediMAX for WBC (0.819 vs. 0.546) and similar for RBC (0.573 vs. 0.630). For RBC, UX-2000 had higher SE (92.7% vs. 80.3%) but lower SP (77.1% vs. 87.4%), and showed higher both SE (94.3% vs. 76.7%) and SP (94.7% vs. 88.2%) for WBC. Inter-devices test strip agreement was substantial (kappa N 0.600) for all variables except for bilirubin (kappa: 0.598). Intra-device test strip agreement was similar for UX2000 and SediMAX with regard to RBC (kappa: 0.553 vs. 0.482) but better for UX2000 with regard to WBC (0.688 vs. 0.465). Conclusions: Both analyzers studied are acceptable for daily routine lab work, even though SediMAX is easier to use in laboratories thanks to its lower maintenance procedure. UX-2000 has shown to have better concordance with the gold standard method. However, it needs some improvements such as an image module in order to decrease manual microscopy review for urine samples. © 2017 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.
1. Introduction At present, urine sample analysis is one of the most prescribed tests in both primary care and at the emergency room. Urine is an easily obtainable sample that provides a lot of information to physicians. It is a fundamental test for the detection and follow-up of various urinary tract and renal diseases as well as chronic diseases. This test includes physical and biochemical analysis such as density, color, turbidity, pH, glucose, ketones bodies, proteins, nitrites, leucocyte esterase, hemoglobin, bilirubin and urobilinogen; and urine particle analysis such as red and white blood cells (RBC, WBC), bacteria, yeasts, crystals, casts, epithelial cells, etc. [1,2]. Between these parameters, RBC and WBC are widely used for the determination of pathological urine. Screening is performed by test strip ⁎ Corresponding author at: Clinical Biochemistry Department, Virgen Macarena University Hospital, (Avenida Doctor.Fedriani n° 3 CP: 46017 Sevilla), Sevilla, Spain. E-mail address: [email protected] (C. Sánchez-Mora).
systems that yield positive or negative results, hence determining whether standard urinary sediment analysis should be performed. Sediment analysis could be performed either manually using standardized systems or by means of automated devices [3]. We focused on RBC and WBC as in Spanish Health System these are the variables considered relevant in the assessment of urines for Emergency Rooms and Clinical Biochemistry departments. Although bacteria count is important in Microbiology departments for urines, they use other methods to assess bacteriuria. Manual microscopy assessment is considered as the gold standard even though it is a burdensome and time-consuming technique for specialized staff, a fact that should be taken into account in terms of laboratory costs [4]. On the other hand, automated analysis provides lower subjectivity on results interpretation, thereby improving accuracy [5] and optimizing time. Regarding automatization, there are many devices for urinary sediment analysis based on different technologies. Digital optic microscopy includes a system of cameras that photographs the sample and features an automatic recognition of elements. Flow
http://dx.doi.org/10.1016/j.clinbiochem.2017.02.005 0009-9120/© 2017 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.
Please cite this article as: C. Sánchez-Mora, et al., Comparison of automated devices UX-2000 and SediMAX/AutionMax for urine samples screening: A multicenter Spanish study, Clin Biochem (2017), http://dx.doi.org/10.1016/j.clinbiochem.2017.02.005
2
C. Sánchez-Mora et al. / Clinical Biochemistry xxx (2017) xxx–xxx
Table 1 Patient's characteristics. n: number of patients; P25: percentile 25; P75: percentile 75; 95%CI: 95% confidence interval. Variable
Global
Center 1
Center2
Center3
Patients [n] Age [median P25–P75] % gender (95%CI) Female Male % procedence (95%CI) Primary care Secondary care In-hospital care Emergency room
1454 54.0 (38.0–70.0)
517 56.0 (43.0–70.0)
502 50.0 (34.0–69.0)
435 54.0 (37.0–72.0)
62.9 (60.4–65.3) 37.1 (34.7–39.7)
59.8 (55.5–63.9) 40.2 (36.1–44.5)
54.0 (49.6–58.3) 46.0 (41.7–50.4)
76.8 (72.6–80.5) 23.2 (19.5–27.4)
44.6 (42.0–47.1) 11.2 (9.7–12.9) 6.5 (5.3–7.9) 37.7 (35.3–40.2)
41.3 (37.0–45.5) 31.6 (27.7–35.7) 18.2 (15.1–21.8) 8.9 (6.8–11.7)
0.0 (0.0–0.8) 0.0 (0.0–0.8) 0.0 (0.0–0.8) 100.0 (99.2–100.0)
100.0 (99.1–100.0) 0.0 (0.0–0.9) 0.0 (0.0–0.9) 0.0 (0.0–0.9)
cytometry and fluorescence devices examine every particle and cell present in a laminar flow sample and accurately count and classify forming elements. Recently, several companies have developed completely automated analyzers integrating in a sole platform both physic-chemical testing and sediment particles analysis. These analyzers can remove the interobserver variability while optimizing laboratory workflow and improving accuracy and precision of the test [6]. Although manual sediment analysis cannot be completely excluded, they significantly reduce the number of samples to be reviewed. European guidelines [7] establish that in the new urine laboratory workflow strategy it is crucial to rationalize the combination of physicochemical, bacterial and sediment particle automated count and the preservation of the sample as done in non-centrifugation based methods. In addition, international recommendations encourage the use of count chambers when assessing the performance of automated devices, both based on either flow cytometry or imaging techniques [8]. The primary aim of this study was to assess the performance of the UX-2000 (flow citometry) and SediMAX/Aution Max systems (imagebased method), and the Fuchs-Rosenthal count chamber for urine RBC and WBC in routine clinical laboratory setting. The secondary aim was to assess correlation between test strips and urinary sediment. 2. Material and methods 2.1. Study design This is a multicenter, observational and prospective study of automatic analyzers for test strips and sediment analyses. The study was presented and approved by each Institution's Ethics Board Committee. Three Spanish centers participated in the study: Dr. Peset University Hospital (center 1), Virgen Macarena University Hospital (center 2) and Jiménez Díaz Foundation University Hospital (center 3). From August 2015 to February 2016, 1454 urine samples were randomly collected from those received at the department of Clinical Analysis/Clinical Biochemistry and analyzed within 2 h. Study samples had to be a minimum of 8 mL volume in order to be assessed by both devices. Nonhomogenous, aggregates or visible particle-containing samples were excluded in order to avoid erratic results. The samples were processed changing the order of the different techniques every ten to fifteen samples to avoid bias. The samples were analyzed by expert staff of each hospital. 2.2. Patient samples Patient median age was 54.0 (P25–P75;38.0–70.0) years, were usually women [62.9% (CI 95%: 60.4–65.3)] and came mostly from primary care [44.6% (42.0–47.1)] and the Emergency Room [37.7% (35.3–40.2)] and in a lower percentage from secondary care [11.2% (9.7–12.9)] and in-hospital care [6.5% (5.3–7.9)]. Center 1 assessed a total of 517 patient samples with a median age of 56 years and 59.8% were women. The 41.3% of urine samples came from primary care, 31.6% from secondary
care, 18.2% from inhospital care and 8.9% from the emergency room. Center 2 assessed a total of 502 patient samples with a median age of 50 years, 54.0% were women and 100% came from the emergency room. Center 3 analyzed a total of 435 patient samples with a median age of 54 years, 76.8% were women and 100% came from a primary care setting (Table 1). 2.3. Urines assessment Samples were assessed by Fuchs-Rosenthal count chamber (SKC Europe GMBH. Frankfurt am Main, Germany), UX-2000 (Sysmex, Kobe, Japan) and SediMAX (77 Elektronika, Hungary)/Aution Max (Arkray Factory Inc., Japan) according to provider instructions, for count of RBC and WBC. For manual particle counts, Fuchs-Rosenthal chambers were used as the gold standard method. Fifteen microliters of urine were applied on the chamber ensuring that the entire surface was properly impregnated avoiding air bubbles formation. Five big squares corresponding to 1 μL were assessed [9]. UX-2000 analyzes non-processed (fresh) samples by flow cytometry [10]. Briefly, it identifies particles and cells by light scatter and classifies them according to their size and cellular complexity. Particles are stained with a fluorescent compound that gives more detailed information about nucleic acids. SediMAX/AutionMax analyses pre-centrifuged samples [11,12]. They are analyzed by an integrated microscope that takes images and compares them to pre-defined libraries of identified particles. Output was modified from high/low power field to particles/μL. Both devices provided a test stripe analysis of semi-quantitative values for glucose, protein, bilirubin, urobilinogen, ketones, pH, RBC, WBC and nitrites. 2.4. Statistical analysis Data was retrospectively retrieved at completion of the study. Center 3 was unable to retrieve SediMAX data because it was recorded and
Table 2 Cell counts for the overall population and for each center. Results shown as median (percentile 25–percentile 75). RBC: red blood cells; WBC: white blood cells. Variable
Fuchs-Rosenthal
UX-2000
SediMAX
RBC (particles/μL) Global Center 1 Center 2 Center 3
2.0 (0.0–8.0) 2.0 (0.0–6.0) 6.0 (2.0–18.5) 0.0 (0.0–3.0)
9.2 (0.0–31.1) 7.8 (3.6–22.6) 25.9 (14.2–60.3) 0.0 (0.0–0.0)
2.9 (0.0–20.3) 1.3 (0.0–4.6) 11.6 (2.9–52.8) –
WBC (particles/μL) Global Center 1 Center 2 Center 3
5.0 (1.0–21.0) 4.0 (1.0–12.0) 9.0 (2.0–36.5) 4.0 (0.0–22.3)
5.8 (0.7–31.4) 5.0 (1.9–18.1) 14.9 (5.3–63.3) 0.0 (0.0–30.5)
8.7 (1.3–46.5) 2.6 (0.7–9.9) 31.9 (8.7–129.8) –
Please cite this article as: C. Sánchez-Mora, et al., Comparison of automated devices UX-2000 and SediMAX/AutionMax for urine samples screening: A multicenter Spanish study, Clin Biochem (2017), http://dx.doi.org/10.1016/j.clinbiochem.2017.02.005
C. Sánchez-Mora et al. / Clinical Biochemistry xxx (2017) xxx–xxx
3
Fig. 1. Patient distribution for red blood cells according to Fuchs-Rosenthal (white), UX-2000 (grey) and SediMAX (black) results in A) Center 1; B) Center 2; C) Center 3. RBC: red blood cells.
stored as a range value as opposed to a single unique value. For statistical purposes, RBC and WBC, in strips and sediment results, were categorized according to cut-offs of test strips given by provider and also dichotomized as positive/negative as follows: RBC categories (0–19; 20–59; 60–299; ≥ 300/μL) and RCB positives results for ≥ 20/μL; WBC categories (0–74; 75–249; 250–499; ≥ 500/μL) and WBC positives results for ≥75/μL. All study variables were subjected to descriptive analysis. UX-2000 and SediMAX/AutionMax concordance against FuchsRosenthal chamber was assessed by Cohen's weighted kappa coefficient for qualitative variables. Cohen's coefficient cutoffs were established as slight (0.00–0.20), fair (0.20–0.40), moderate (0.40–0.60), substantial (0.60–0.80) and almost perfect agreement (0.80–1.00). Diagnostic performance (sensitivity, specificity, positive and negative predictive values) were calculated for sediment RBC and WBC determination in both instruments. Statistical analyses were performed using NCSS 10 software (NCSS, LLC Utah, USA) and Cohen's weighted kappa coefficient using VassarStats (http://vassarstats.net/kappa.html). All tests were significant with an α = 0.050.
whereas UX-2000 showed significant higher concordance with FuchsRosenthal than SediMAX (0.819 vs. 0.546) for WBC. In the biochemical analysis of urine samples with test strips, all variables had a concordance over 0.60 (data not shown), showing a substantial agreement between both automatic analyzers. The only exception was bilirubin, which had a concordance of 0.598. Concordance between each automatic analyzer and own test strip system is depicted on Table 3. Both devices had a good concordance for RBC and WBC categorization analysis. Particularly, SediMAX had quite higher concordance than UX-2000 (0.553 vs. 0.482) for RBC and much lower than UX-2000 when assessing WBC (0.465 vs. 0.688). These results were maintained when center-specific sub-analysis was performed (Table 3). When RBC and WBC values were dichotomized into positive and negative results (Table 4), the UX-2000 shown higher sensitivity (92.7% vs. 80.3%) than SediMAX for RBC, but worse specificity (77.1% vs. 87.4%). For WBC count, UX2000 had higher sensitivity (94.3% vs. 76.7%) and specificity (94.7% vs. 88.2%) than SediMAX. 4. Discussion
3. Results For both analyzers and gold standard, center 2 had higher RBC and WBC values, while center 3 had the lowest counts (Table 2). These differences were greater when samples were analyzed using the UX2000 system (Figs. 1 and 2). In the categorical analysis of forming elements (Table 3), UX-2000 and SediMAX had similar concordance for RBC (0.573 vs. 0.630)
In urine sediment assessments, each laboratory must establish the most suitable procedure according to patient's clinical characteristics which can be classified by department receiving these samples [1]. At present, it is widely accepted that automatic analyzers may reduce the number of hours spent on manual review, as reported in several studies [13,14]. There are many comparative studies that evaluate currently available automatic analyzers [15–17]. Wesarachkitti et al. [15]
Fig. 2. Patient distribution for white blood cells according to Fuchs-Rosenthal (white), UX-2000 (grey) and SediMAX (black) results in A) Center 1; B) Center 2; C) Center 3. WBC: white blood cells.
Please cite this article as: C. Sánchez-Mora, et al., Comparison of automated devices UX-2000 and SediMAX/AutionMax for urine samples screening: A multicenter Spanish study, Clin Biochem (2017), http://dx.doi.org/10.1016/j.clinbiochem.2017.02.005
4
C. Sánchez-Mora et al. / Clinical Biochemistry xxx (2017) xxx–xxx
Table 3 Concordance for categorized RBC and WBC. κ:weighted kappa; 95%CI: 95% confidence interval; RBC: red blood cells; WBC: white blood cells. Variable
UX-2000 cytometry - UX-2000 test stripe
SediMAX image - SediMAX test stripe
UX-2000 cytometry - Fuchs-Rosenthal
SediMAX image - Fuchs-Rosenthal
RBC Global Center 1 Center 2 Center 3
0.482 (0.443–0.521) 0.557 (0.488–0.625) 0.333 (0.282–0.383) 0.465 (0.356–0.574)
0.553 (0.506–0.599) 0.682 (0.615–0.750) 0.430 (0.367–0.493) –
0.573 (0.528–0.618) 0.679 (0.605–0.752) 0.451 (0.388–0.514) 0.618 (0.498–0.739)
0.630 (0.577–0.683) 0.826 (0.759–0.893) 0.504 (0.433–0.575) –
WBC Global Center 1 Center 2 Center 3
0.688 (0.655–0.720) 0.639 (0.575–0.704) 0.682 (0.631–0.732) 0.707 (0.652–0.762)
0.465 (0.409–0.521) 0.454 (0.381–0.528) 0.459 (0.391–0.527) –
0.819 (0.783–0.856) 0.802 (0.723–0.881) 0.802 (0.746–0.859) 0.843 (0.784–0.902)
0.546 (0.479–0.613) 0.681 (0.550–0.811) 0.482 (0.404–0.561) –
compared UX-2000 and an image-based automatic analyzer (Cobas 6500) on 258 patients considering agreement as same result or ± 1 grade, showing a great performance for both devices. In assessing 214 samples by conventional method and with a flow cytometry-based device, Manoni et al. [16], that made a comparison between UF-1000i and quantitative microscopy, found a nearly perfect concordance between both methods. Finally, Demet et al. [17] compared two digital imagebased automatic analyzers (Iris iQ200 and Dirui FUS-200) in 209 samples concluding that RBC and WBC were similar to those on manual counting. Our multicenter study assessed a representative amount (n = 1454) of urine sample to gain deeper and more robust insight into the concordance between automatic analyzers and the gold standard technique, Fuchs-Rosenthal count chamber. We did not assess feasibility, linearity, or accuracy of UX-2000 and SediMAX as this was not our main objective, and it has been widely described in previous published studies [10,15, 18]. Particularly, Wesarachkitti et al. [15] concluded that UX-2000 improves reproducibility and reduced workload due to its precision and linearity, whereas from the study of 400 urine samples, Bottini et al. [18] showed that SediMAX could yield a better workflow due to its precision and accuracy. With regard to forming elements (WBC and RBC) assessed using the Fuchs-Rosenthal chamber, our results show that as sample complexity increases (presenting abnormal or pathological values), discrepancy widens. Consequently, particle counting is increased, leading to significant differences between chamber counts and automatic analyzers results. Overall, differences among hospital's samples were more significant for UX-2000 than for SediMAX or Fuchs-Rosenthal results, which may imply that UX-2000 has a higher discriminatory power, accurately counting and classifying samples. This may be due to inter-rate variability, sedimentation of formed elements, and non-homogeneous samples. The conversion factor from particle/field to particle/μL also decreases the accuracy of manual counting. When the unit is converted into another unit, a small counting deviation is amplified. Regarding the test strip, kappa coefficients show that there is substantial concordance between both automatic devices (UX-2000 test strips scanner and Aution Max), being lower (moderate) for bilirubin. This somewhat lower concordance may be due to the low prevalence of positive cases for bilirubin in the general population, which mathematically implies a lower kappa coefficient.
When assessing RBC results by test strips and sediment for each automatic analyzer, a moderate concordance was found for both devices. This could be explained by the fact that RBC hemolysis cannot be detected by automatic analyzers that count intact particles [19]. For WBC concordance has been shown to be substantial for UX-2000 and moderate for SediMAX/Aution Max. This may be explained by the lower sensitivity of SediMAX when counting WBC, leading to a higher discrepancy between counting and test strips. Negative predictive values for both UX-2000 and SediMAX are good for RBC (98.3% vs. 95.1%) and WBC (99.2% vs. 96.4%), as stated by other authors [20], an essential feature for avoiding misclassification of positive urine samples as negative. Positive predictive value of UX-2000 for RBC is quite low (42.6%). This may be explained by scatter plot analysis, where some small DNAless particles such as yeasts or small crystals could potentially be misclassified as RBC, even if this could be solved by changing wavelength [20,21]. For WBC, positive predictive value of SediMAX is low (47.7%). This could be due to sample centrifugation prior to analysis by this device that may form aggregates or even produce WBC lysis [22]. Our study has some limitations. Samples from different centers had different clinical and pathological features which, despite allowing testing device performance in a wide range of real conditions, make comparisons between centers quite difficult. Moreover, in one of the centers, SediMAX sediment data could not be retrieved, which lowers effective sample size leading to a lower statistical power. However, these limitations do not reduce the relevance to the study results. Regarding technical and practical issues, UX-2000 has proven to be a compact device that does work in fresh samples without the need for centrifugation, hence preserving all particles. However, this device needs a higher urine volume in automatic mode (5 mL vs. 2 mL for SediMAX). This, while not being a problem for daily practice, could pose some difficulties for pediatric urine samples, although this short coming can be simply resolved by switching to manual mode. For daily practice, it seemed that UX-2000 needed more exhaustive maintenance (automatic filter washes after hematic or pyuric samples) which if not met, could lead to workflow interruption and consequently length in response time to get back lab results. Altogether, UX-2000 showed higher performance for RBC, WBC and epithelial cells. SediMAX was superior for detecting crystals, yeast and cast (data not shown).
Table 4 Diagnosis performance of UX-2000 and SediMAX according to Fuchs-Rosenthal results, using RBC ≥20/μl or WBC ≥75/μl as positive results. Data shown as percentage (95% confidence interval). RBC: red blood cells; WBC: white blood cells. Variable
Sensitivity Specificity Positive predictive value Negative predictive value
RBC
WBC
UX-2000
SediMAX
UX-2000
SediMAX
92.7 (88.2–95.6) 77.1 (74.6–79.4) 42.6 (38.2–47.2) 98.3 (97.2–99.0)
80.3 (73.7–85.7) 87.4 (84.9–89.6) 59.5 (53.1–65.6) 95.1 (93.2–96.5)
94.3 (89.4–97.1) 94.7 (93.3–95.9) 71.7 (65.4–77.4) 99.2 (98.4–99.6)
76.7 (67.9–83.7) 88.2 (85.8–90.3) 47.7 (40.5–54.9) 96.4 (94.8–97.6)
Please cite this article as: C. Sánchez-Mora, et al., Comparison of automated devices UX-2000 and SediMAX/AutionMax for urine samples screening: A multicenter Spanish study, Clin Biochem (2017), http://dx.doi.org/10.1016/j.clinbiochem.2017.02.005
C. Sánchez-Mora et al. / Clinical Biochemistry xxx (2017) xxx–xxx
5. Conclusions Automatic devices for urine sediment examination improve reproducibility, accuracy and response time over manual microscopy. Although the pricing of automatic analyzers and their reagents is not low, reduction on analysis time and savings on direct personnel costs makes these devices more cost-effective than manual microscopy assessment of urine sediment. Both analyzers studied are acceptable for daily routine lab work, even though SediMAX is easier to use in laboratories thanks to its lower maintenance procedure. UX-2000 flow cytometry has shown to have a higher concordance with image-based devices and higher correlation with the gold standard method. However, it needs some improvements such as an image module in order to decrease manual microscopy review for urine samples. Discloser All authors disclose no conflict of interest. All authors have contributed to study design, sample analysis, manuscript conception and final approval of the article. This work has been carried out in accordance with Declaration of Helsinki and the International Committee of Medical Journal Editors. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Sysmex España S.L. financially supported the study providing reagents and UX2000 equipment. Acknowledgments We thank Sysmex España for their technical support. References [1] A. Rabinovitch, L. Arzoumanian, K.M. Curcio, B. Dougherty, A.B. Halim, Urinalysis; approved guideline, CLSI Doc GP16-A3, third ed.Clin Lab Stand Institue, Wayne PA, 2009. [2] R. McPherson, J. Ben-Ezra, Basic Examination of Urine, Henry's Clinical Diagnosis and Management by Laboratory Methods, Elsevier Saunders, Philadelphia 2011, pp. 445–479. [3] D. Carlson, B. Statland, Automated urianalysis, Clin. Lab. Med. 8 (1988) 449–461. [4] P. Winkel, B. Statland, J. Jorgenson, Urine microscopy: an illdefined method examined by a multifactorial technique, Clin. Chem. 20 (1974) 436–439. [5] A. Lun, R. Ziebig, F. Priem, P. Sinha, Routine workflow for use of urine strips and urine flow cytometer UF-100 in the hospital laboratory, 1, Clin. Chem. 45 (8) (1999) 1305–1307. [6] Richard Thompson, Andrew Gammie, Debbie Lewis, Rebecca Smith CE. Evidence review: Automated urine screening systems, Cent Evidence-based Purch (March) (2010) 1–46 (CEP10030).
5
[7] European Confederation of Laboratory Medicine, European urinalysis guidelines, Scand J Lab Invest. 60 (2000) 1–96. [8] T. Kouri, A. Gyory, R.M. Rowan, ISLH recommended reference procedure for the enumeration of particles in urine, Lab. Hematol. 9 (2) (2003) 58–63 Available from. http://www.ncbi.nlm.nih.gov/pubmed/12828299. [9] J. Neuendorf, Zellzählung in der Fuchs-Rosenthal-Zählkammer, Das Urinsediment: Mikroskopie, Präanalytik, Auswertung und Befundung, Springer Berlin Heidelberg, Berlin, Heidelberg 2015, pp. 49–52 Available from: http://dx.doi.org/10.1007/9783-662-46074-0_9. [10] V. Khejonnit, B. Pratumvinit, K. Reesukumal, S. Meepanya, C. Pattanavin, P. Wongkrajang, Optimal criteria for microscopic review of urinalysis following use of automated urine analyzer, Clin. Chim. Acta 439 (2015) 1–4 Available from: http://linkinghub.elsevier.com/retrieve/pii/S0009898114004239. [11] Z. Zaman, G.B. Fogazzi, G. Garigali, M.D. Croci, G. Bayer, T. Kránicz, Urine sediment analysis: analytical and diagnostic performance of sediMAX - a new automated microscopy image-based urine sediment analyser, Clin. Chim. Acta 411 (3–4) (2010) 147–154 Available from: 10.1016/j.cca.2009.10.018. [12] M. Miura, T. Kondo, M. Mizuno, T. Morikawa, The basic analytical performance of the fully automated integrated urine analyzer UX-2000 CHM unit, Sysmex J Int. 23 (2013). [13] J.A. Tworek, D.S. Wilkinson, M.K. Walsh, C. The, The rate of manual microscopic examination of urine sediment a College of American Pathologists Q-Probes Study of 11 243 urinalysis tests from 88 institutions, Arch. Pathol. Lab. Med. 132 (December) (2008) 1868–1873. [14] Z. Zaman, Automated urine screening devices make urine sediment microscopy in diagnostic laboratories economically viable, Clin. Chem. Lab. Med. 53 (2015) S1509–S1511. [15] B. Wesarachkitti, V. Khejonnit, B. Pratumvinit, K. Reesukumal, S. Meepanya, C. Pattanavin, et al., Performance evaluation and comparison of the fully automated urinalysis analyzers UX-2000 and Cobas 6500, Lab. Med. 47 (2) (2016 Apr 11) 124–133 Available from: https://labmed.oxfordjournals.org/content/47/2/124.full. [16] F. Manoni, A. Tinello, L. Fornasiero, P. Hoffer, V. Temporin, S. Valverde, et al., Urine particle evaluation: a comparison between the UF-1000i and quantitative microscopy, Clin. Chem. Lab. Med. 48 (8) (2010) 1107–1111. [17] F. Demet, H. Yaşar, M. Koseoğlu, N. Şimşek, H. Yalçına, M. Osman, The comparison of automated urine analyzers with manual microscopic examination for urinalysis automated urine analyzers and manual urinalysis, Pract Lab Med. 5 (2016) 14–20. [18] P.V. Bottini, M.H.M. Martinez, C.R. Garlipp, Urinalysis: comparison between microscopic analysis and a new automated microscopy image-based urine sediment instrument, Clin Lab. 60 (4) (2014) 693–697. [19] J. Jiménez García, G. Ruiz Martín, Estudio de los elementos formes de la orina, in: LABCAM (Asociación Castellano-Manchega de Análisis Clínicos) (Ed.), Estandarización del sedimento urinario 2010, pp. 1–96. [20] S. Mayo, D. Acevedo, C. Quiñones-Torrelo, I. Canós, M. Sancho, Clinical laboratory automated urinalysis: comparison among automated microscopy, flow cytometry, two test strips analyzers, and manual microscopic examination of the urine sediments, J. Clin. Lab. Anal. 22 (4) (2008) 262–270. [21] Y.U. Budak, K. Huysal, Comparison of three automated systems for urine chemistry and sediment analysis in routine laboratory practice, Clin. Lab. 57 (1–2) (2011) 47–52. [22] M. Íñigo, A. Coello, G. Fernández-Rivas, M. Carrasco, C. Marcó, A. Fernández, et al., Evaluation of the SediMax automated microscopy sediment analyzer and the Sysmex UF-1000i flow cytometer as screening tools to rule out negative urinary tract infections, Clin. Chim. Acta 456 (2016) 31–35 Available from: 10.1016/j.cca. 2016.02.016.
Please cite this article as: C. Sánchez-Mora, et al., Comparison of automated devices UX-2000 and SediMAX/AutionMax for urine samples screening: A multicenter Spanish study, Clin Biochem (2017), http://dx.doi.org/10.1016/j.clinbiochem.2017.02.005