|

Received: 24 January 2017    Accepted: 3 April 2017

DOI: 10.1002/jcla.22249

RESEARCH ARTICLE

UriSed 3 and UX-­2000 automated urine sediment analyzers

vs manual microscopic method: A comparative performance
analysis

Sathima Laiwejpithaya | Preechaya Wongkrajang  | Kanit Reesukumal  | 

Chonticha Bucha  | Suriya Meepanya | Chanutchaya Pattanavin | 
Varanya Khejonnit | Achara Chuntarut

Department of Clinical Pathology, Faculty of

Medicine Siriraj Hospital, Mahidol University, Background: Fully automated urine analyzers now play an important role in routine
Bangkok, Thailand urinalysis in most laboratories. The recently introduced UriSed 3 has a new automated

Correspondence
Preechaya Wongkrajang, Department of
Clinical Pathology, Faculty of Medicine Siriraj
Hospital, Mahidol University, Bangkok,
Thailand.
Email: preechaya.wok@mahidol.ac.th
digital imaging urine sediment analyzer with a phase contrast feature. The aim of this
study was to compare the performance of the UriSed 3 and UX-­2000 automated urine
sediment analyzers with each other and with the results of the manual microscopic
method.
Methods: Two hundred seventy-­seven (277) samples of leftover fresh urine from our
hospital’s central laboratory were evaluated by two automated urine sediment analyz-
ers—UriSed 3 and UX-­2000. The results of urine sediment analysis were compared
between the two automated analyzers and against the results of the manual micro-
scopic method.
Results: Both devices demonstrated excellent agreement for quantitative measure-
ment of red blood cells and white blood cells. UX-­2000 had a lower coefficient corre-
lation and demonstrated slightly lower agreement for squamous epithelial cells.
Regarding semiquantitative analysis, both machines demonstrated very good concord-
ance, with all applicable rates within one grade difference of the other machine. UriSed
3 had higher sensitivity for small round cells, while UX-­2000 showed greater sensitiv-
ity for detecting bacteria and hyaline casts. UriSed 3 demonstrated slightly better
specificity, especially in the detection of hyaline and pathological casts.
Conclusions: Both instruments had nearly similar performance for red blood cells and
white blood cells measurement. UriSed 3 was more reliable for measuring squamous
epithelial cells and small round cells, while the UX-­2000 was more accurate for detect-
ing bacteria and hyaline casts.

KEYWORDS
automated urine analyzer, urinalysis, urine sediment, UriSed 3, UX-2000

1 | INTRODUCTION
Abbreviations: RBCs, red blood cells; WBCs, white blood cells; ECs, squamous epithelial cells;
AIEM, Auto Image Evaluation Module; r, coefficient correlation; CV, coefficient of variation;
ICC, intraclass correlation coefficient; HPF, high power field; PPV, positive predictive value;
Urinalysis is an essential test in clinical medicine that is used for screen-
NPV, negative predictive value. ing, diagnosis, and monitoring of diseases of the urinary system, and

J Clin Lab Anal. 2018;32:e22249. wileyonlinelibrary.com/journal/jcla © 2017 Wiley Periodicals, Inc.  |  1 of 10
https://doi.org/10.1002/jcla.22249
 |
2 of 10       LAIWEJPITHAYA et al.
diseases that are detectable via the urinary system. Urinalysis consists
of urine physical appearance, chemical analysis, and microscopic sed-
iment examination. Manual microscopic method for the examination
of urine is time consuming, labor intensive, and requires well-­trained
and experienced technicians. Moreover, and as a result of some inher-
ent subjectivity, variations among observers are not uncommon. As
a result, automated urine analyzers were developed to improve both
productivity and consistency of results in urinalysis.1-5
Automated urine sediment analysis is performed, using either
®
digital imaging or flow cytometry. In 1985, the Yellow IRIS auto-
mated urinalysis instrument (International Remote Imaging Systems,
Inc., Chatsworth, CA, USA) was introduced that uses a video camera
and an image analysis system. A few years later, flow cytometry was
™
developed and introduced in the UF-­100 fully automated urine cell
analyzer (Sysmex Corp., Kobe, Japan). Since that time, many manufac-
turers have launched new and more advanced analyzers, including the
IQ-­200 (IRIS, Chatsworth, CA, USA), UF-­1000i (Sysmex Corp., Kobe,
Japan), and FUS100 (Dirui Industrial Co. Ltd., Changchun, China).
Many studies have compared the performance of automated urine
sediment analyzers against each other and against manual sediment
microscopy.1,6-11
In 2009, a Hungarian company named 77 Elektronika Kft. de-
veloped and introduced a new automated urine sediment analyzer
and marketed it under the names UriSed or sediMAX in some coun-
tries. This instrument captures high-­power field-­like images from
urine centrifuged in a disposable cuvette via a digital camera that is
mounted in a bright-­field microscope at 400× magnification. Particles
are then identified and categorized with image processing software.
The automated process used in the UriSed analyzer is similar to the
process used in manual microscopic examination. The images are
then presented on a screen, which allows for visualization and iden-
tification of particles that can be rechecked and adjusted, as needed,
by a technician. Although the UriSed software can evaluate red blood
cells (RBCs), white blood cells (WBCs), and squamous epithelial cells
(ECs) correctly, and is able to differentiate other sediment particles
with acceptable results, but manual verification is still needed.5,12 As
a result, 77 Elektronika Kft. set forth to improve their technology and
software, and recently launched the UriSed 3 in 2016. The UriSed
3 uses technology similar to that used in their first UriSed analyzer,
but with some new features. New features include a built-­in micro-
scope with close to 100× magnification so that more native urine
is examined within 15 images, and a phase contrast microscope to
supplement the preexisting bright-­field microscope.13,14 In contrast,
the Sysmex UX-­2000 is an automated urine analyzer that uses flow
cytometry technology for sediment analysis. The advantage of the
UX-­2000 is that it can evaluate all three components of urinalysis in
one machine. However, the UX-­2000 is limited in its ability to cate-
gorize some type of particles, such as pathological casts and crystals,
with some results still needing to be confirmed by an experienced
15
technician.
The objective of this study was to compare the performance of the
UriSed3 and UX-­2000 automated urine sediment analyzers with each
other and with the results of manual sediment microscopy.
2 | MATERIALS AND METHODS
2.1 | Urine samples
This study was performed at the central laboratory of the Department
of Clinical Pathology, Faculty of Medicine, Siriraj Hospital, Mahidol
University during the March 2016 to May 2016 study period. Siriraj
Hospital is Thailand’s largest national tertiary referral center. Two hun-
dred and seventy-­seven (277) leftover urine specimens that were col-
lected from both inpatients and outpatients in clean, preservative-­free
containers were included. All leftover samples must have contained
at least 20 mL of uncontaminated urine. All samples were analyzed
within 2 hours after reaching our laboratory. After being processing by
the UX-­2000 according to routine working instructions, residual sam-
ples were then divided into two aliquots. One aliquot was sent to be
examined by manual microscopic method and the other was analyzed
by the UriSed 3 automated urine sediment instrument. The protocol
for this study was approved by the Siriraj Institutional Review Board
(SIRB) [069/2559 (EC3)].
2.2 | Automated urine analyzers
2.2.1 | UX-­2000
UX-­2000 is a single machine that can analyze all three components
of urinalysis, including physical, chemical, and microscopic sediment
examination. A sample must contain at least 5 mL of urine, of which
2 mL is used to evaluate all three components without centrifugation.
For physical examination, the transmission refractometry method,
reflectivity measurement method, and light-­scattering measurement
method are used for measurement of specific gravity, turbidity, and
color, respectively. The chemical examination involves the use of a test
strip that is measured by the dual-­wavelength reflectance method.
For microscopic analysis, the fluorescence flow cytometry principle
is used to measure particles and categorize them into the following
parameters: RBCs, WBCs, ECs, hyaline casts, or bacteria. Since the
UX-­2000 cannot categorize other parameters, such as small round
cells, yeast-­like cells, crystals, pathological casts, and sperm, they are
flagged for manual microscopic examination.16
2.2.2 | UriSed 3
The UriSed 3 analyzer requires a urine sample of least 2 mL, of which
200 μL is pipetted into an individual disposable cuvette without any
additional reagents. The filled cuvette is centrifuged at 2000 rpm for
10 seconds. An automated built-­in camera then takes 15 images of
the settled monolayer of urine particles. Images are recorded in three
types, including bright-­field, phase contrast, and composite. Images
are then evaluated by the Auto Image Evaluation Module (AIEM), the
automatic, real-­time, image processing software. Urine particles are
classified into the following parameters: RBCs, WBCs, white blood cell
clumps, hyaline casts, pathological casts, ECs, nonsquamous epithe-
lial cells (non-­ECs), bacteria, yeast, mucous, sperm, and crystals, which
LAIWEJPITHAYA et al. |
      3 of 10

are sub-­classified into calcium-­oxalate monohydrate, calcium-­oxalate was used for UriSed 3, and UF II CONTROL™ (Sysmex America, Inc.,
dehydrate, uric acid, or triple phosphate. After the automatic results Lincolnshire, IL, USA) was used for UX-­2000. For between-­run preci-
are obtained, a well-­trained technologist can reclassify or further sub-­ sion, both levels of quality control materials were run daily for a total
classify particles from the images displayed on the screen in order to of 20 days. We assessed the precision of each method by calculating
achieve more accurate results.13,14 the coefficient of variation (CV).
In this study, the results analyzed by AIEM were defined as ‘UriSed
3 pre-­classification’ results, while the results rechecked by technolo-
2.4.2 | Linearity study
gists were defined as ‘UriSed 3 postclassification’ results.
Linearity is the ability of a device to accurately count particles in vari-
ous concentrations of urine specimen relative to the actual concentra-
2.3 | Manual microscopic examination
tion. To estimate linearity, two urine samples were diluted with normal
For the manual method, 10 mL of each urine specimen was poured urine into a series of five concentrations. The first sample contained
into conical urine tubes and centrifuged at 2000 rpm (400 g) for high RBCs, and the second sample contained a high level of WBCs and
9
5 minutes according to the following equation : bacteria. The same specimens in each concentration were analyzed
in duplicate by both the UX-­2000 and the UriSed 3. The results were
RCF (g) = 1.118 × 10−5 × r × (rpm)2 , plotted on a linear regression graph and the correlation coefficient (r)
was calculated.
where RCF, relative centrifugal force; r, distance between the rotation
axis and the center of the sample tube in centimeter; rpm, revolutions
per minute. 2.4.3 | Carry-­over study
Nine milliliter of supernatant was discarded, using a tri-­
bulb.
A urine specimen with a high concentration of particles was aliquoted
After resuspending the remaining 1 mL of urine, the sample was
into three tubes (S1, S2, S3) and analyzed consecutively, followed by
poured into a Vetriplast slide with counting grid (Vacutest Kima,
the consecutive analysis of three successive aliquots of normal urine
Arzergrande, Italy). Two well-­trained laboratory technologists then
(N1, N2, N3) in both automated urine analyzers. Two urine samples
examined the slide under light microscope and separately recorded
were selected—one with high RBCs, and the other with high WBCs
their results on a case record form. If the findings of the two tech-
and bacteria. The carry-­
over was determined using the following
nologists were not in agreement for the same sample, a third labo-
formula18:
ratory technologist with 10 years of clinical microscopy experience
was asked to examine the sample. The finding made by of two of %carry- over = {(N1- N3)∕(S3- N3)}×100.
three technicians was then used as the reference result. The counting
grid in Vetriplast slides contained nine large squares of 1×1 mm for 2.4.4 | Accuracy evaluation
a volume of 0.1 μL each. Every large square was divided into nine
The component results of urine sediment analysis were categorized
smaller squares of 0.333×0.333 mm for a volume of 0.0111 μL each.
into two groups. The first group included RBCs, WBCs, and ECs re-
For cell counts, such as RBCs, WBCs, ECs, and small round cells, we
ported in cells/μL. We analyzed agreement between various combi-
randomly counted the number of elements presented in 10 different
nations of methods using intraclass correlation coefficient (ICC) and
small squares at 400× magnification. Hyaline casts and pathological
Bland–Altman plot. We also calculated the correlation coefficient (r)
casts were reviewed and counted in all nine large squares at 100×
to determine the strength of the linear relationship between meth-
magnification. The quantity of cells per 1 μL of urine was calculated
ods. In addition, these quantitative results were converted to counts/
using the following formula17:
HPF and categorized into five ranges: 0-­5, 5-­10, 10-­20, 20-­50, and
n >50 counts/HPF. The concordance rate within one grade difference
Cells∕μL =
VolChamber ×CF was then analyzed together with weighted kappa (κ). The following
remaining parameters were included in the second group: bacteria,
where n, number of cells counted in 10 small squares; VolChamber, vol- yeasts, small round cells, hyaline casts, pathological casts, and crystals.
ume of 10 small squares; CF, concentration factor. All parameters in the second group were reported as either positive or
negative. Sensitivity, specificity, positive predictive value (PPV), and
negative predictive value (NPV) were calculated for each parameter
2.4 | Method comparison design
result from each method, with manual microscopic examination being
used as the standard method.
2.4.1 | Precision study
Within-­run precision was evaluated by measuring RBCs and WBCs,
2.5 | Statistical analysis
using commercially available high-­level and low-­level quality control
materials a total of 20 times in 1 day. QuanTscopics® urinalysis mi- Data were analyzed, using PASW Statistics for Windows, version
croscopics control (Quantimetrix Corp., Redondo beach, CA, USA) 18.0 (SPSS, Inc., Chicago, IL, USA) and Microsoft Excel (Microsoft
4 of 10       | LAIWEJPITHAYA et al.

T A B L E   1   Within-­run and between-­run precision of microscopic analysis by UX-­2000 and UriSed 3

Within-­run precision Between-­run precision

Level 1 (Low) Level 2 (High) Level 1 (Low) Level 2 (High)

Analyzer Parameter (cells/μL) Mean±SD %CV Mean±SD %CV Mean±SD %CV Mean±SD %CV
a a a a
UX-­2000 RBCs 38.80±2.44 6.3 181.75±4.68 2.6 42.09±1.82 4.3 198.48±8.38 4.2
a a a a
WBCs 40.12±1.68 4.2 775.28±10.31 1.3 41.71±3.05 7.3 804.01±40.63 5.1
Uri Sed 3 RBCs 18.14±2.97 16.4 101.87±12.87 12.6 20.90±3.26 15.6 127.35±11.58 9.1
WBCs 11.04±1.79 16.2 43.11±4.84 11.2 12.28±1.97 16.1 49.91±4.25 8.5

%CV, percent coefficient of variation; RBCs, red blood cells; SD, standard deviation; WBCs, white blood cells.
a
Data derived from the study by Wesarachkitti, et al.1

3500 1400
Measured RBC count (cells/µl)

Measured WBC count (cells/µl)

3000 1200

2500 1000

2000 800
1500 600
1000 400
y = 0.944x + 140.268 y = 0.950x + 47.636
500 200
r = 0.997 r = 1.000
0 0
0 1000 2000 3000 0 500 1000 1500
(A) Calculated RBC count (cells/µl) (B) Calculated WBC count (cells/µl)

3500 1400
Measured WBC count (cells/µl)
Measured RBC count (cells/µl)

3000 1200
2500 1000
2000 800
1500 600
1000 400
500 y = 0.947x + 293.529 200 y = 0.929x + 92.345
r = 0.993 r = 0.999
0 0
0 1000 2000 3000 0 500 1000 1500
Calculated RBC count (cells/µl) Calculated WBC count (cells/µl)
(C) (D)
F I G U R E   1   Linearity of red blood cells and white blood cells counted by UX-­2000 (A, B) and UriSed 3 (C, D)

Corporation, Redmond, WA, USA). Data are presented as percent-

age (%), mean±standard deviation (SD), correlation coefficient (r),
coefficient of variation (%CV), intraclass correlation coefficient
(ICC) with 95% confidence interval (95% CI), concordance rate, and
weighted kappa (κ). ICC and κ values were interpreted according to
Landis and Koch criteria, as follows: 0.00-­0.20 for slight agreement;
0.21-­0.40 for fair agreement; 0.41-­0.60 for moderate agreement;
0.61-­0.80 for substantial agreement; and, 0.80-­1.00 for almost per-
fect agreement.19 A P-­value <.05 was considered to be statistically
significant.
3 | RESULTS
3.1 | Precision and linearity study
Within-­run and between-­run precision of microscopic analysis for both
the UX-­2000 and UriSed 3 are shown in Table 1. For linearity analy-
sis, RBC counts ranged from 2586.6 to 398.4 cells/μL and 3038.9 to
551.2 cells/μL for the Ux-­2000 and UriSed 3, respectively. For WBC
analysis, concentrations ranged from 1139.9 to 183.7 cells/μL and
1306.3 to 226.4 cells/μL for the UX-­2000 and UriSed 3, respectively.
LAIWEJPITHAYA et al. |
      5 of 10

T A B L E   2   Carry-­over analysis of UX-­2000 and UriSed 3

UX-­2000 UriSed 3

Sample RBCs (cells/μL) WBCs (cells/μL) Bacteria (cells/μL) RBCs (cells/μL) WBCs (cells/μL) Bacteria (cells/μL)

S1 3488.30 923.50 2565.20 4062.08 842.16 360.36

S2 3435.50 862.70 2214.10 4160.64 889.46 315.92
S3 3529.20 825.30 2152.50 4045.07 843.04 335.43
N1 5.60 2.40 0.90 0.29 1.10 44.44
N2 4.60 1.60 0.90 0.59 0.44 41.80
N3 5.20 1.40 0.00 0.59 0.66 42.53
%Carry-­over 0.01% 0.12% 0.04% −0.01% 0.05% 0.65%

RBCs, red blood cells; WBCs, white blood cells.

T A B L E   3   Comparison of intraclass
Paired method Parameter r ICC 95% CI P-­value
correlation coefficients between various
combinations of methods UriSed 3 preclassification RBCs .851 0.904 (0.878-­0.924) <.05
vs Manual WBCs .884 0.918 (0.896-­0.935) <.05
ECs .914 0.934 (0.917-­0.948) <.05
UriSed 3 postclassifica- RBCs .861 0.909 (0.885-­0.928) <.05
tion vs Manual WBCs .884 0.918 (0.896-­0.935) <.05
ECs .915 0.935 (0.918-­0.949) <.05
UX-­2000 vs Manual RBCs .923 0.922 (0.901-­0.938) <.05
WBCs .778 0.861 (0.824-­0.890) <.05
ECs .683 0.755 (0.690-­0.807) <.05
UX-­2000 vs UriSed 3 RBCs .847 0.912 (0.889-­0.931) <.05
preclassification WBCs .838 0.847 (0.806-­0.879) <.05
ECs .671 0.672 (0.584-­0.741) <.05
UX-­2000 vs UriSed 3 RBCs .860 0.920 (0.899-­0.937) <.05
postclassification WBCs .838 0.847 (0.807-­0.879) <.05
ECs .672 0.673 (0.585-­0.742) <.05
UriSed 3 preclassification RBCs .998 0.999 (0.999-­0.999) <.05
vs UriSed 3 WBCs 1.000 1.000 (1.000-­1.000) <.05
postclassification
ECs .998 0.999 (0.999-­0.999) <.05

RBCs, red blood cells; WBCs, white blood cells; ECs, squamous epithelial cells; r, correlation coefficient;
ICC, intraclass correlation coefficient; CI, confidence interval.
P-­value <.05 indicates statistical significance.

Linearity graphs and correlation coefficients (r) for the UX-­2000 and
UriSed 3 are shown in Figure 1.
3.2 | Carry-­over study
The carry-­over (%) was calculated in RBCs, WBCs, and bacteria on
both the UX-­2000 and UriSed 3, as shown in Table 2. A carry-­over of
<1% was observed on both devices for all three parameters.
3.3 | Comparison of urine sediment
Quantitative counting of RBCs, WBCs, and ECs by all three meth-
ods was analyzed in pairs for correlation coefficient (r) and intraclass
correlation coefficient (ICC), as summarized in Table 3. The ICCs of
both UriSed 3 preclassification and postclassification compared with
the manual method for all three parameters (RBCs, WBCs, and ECs)
were >0.8. For UX-­2000 compared with the manual method, the ICC
values were similar to those of the UriSed 3, except for the measure-
ment of ECs, which ranged from 0.61 to 0.80.
Figures 2-4 show Bland–Altman plots of all methods. Very
little difference was observed among systems for RBC and WBC
counts. Regarding ECs and as compared to the manual method,
lower ­results were obtained for both preclassification and post-
classification UriSed 3 analysis. Conversely and as compared to the
manual method, the UX-­2000 tended to report higher EC counts
(Figure 4).
6 of 10       | LAIWEJPITHAYA et al.

Difference between Manual and UriSed 3 Pre Difference between Manual and UriSed 3 Post Difference between Manual and UX-2000
2000 2000 2000
1500 1500 1500

Difference (cells/µL)

Difference (cells/µL)
Difference (cells/µL) 1000 1000 1000
500 x̄ = -2.17 +1.96SD = 272.39 500 x̄ = 1.32 +1.96SD = 269.17 500 x̄ = 37.41 +1.96SD = 311.01
0 0 0
-500 -1.96SD = -276.73 -500 -1.96SD = -266.54 -500 -1.96SD = -236.19
-1000 -1000 -1000
-1500 -1500 -1500
-2000 -2000 -2000
0 500 1000 1500 2000 2500 3000 0 500 1000 1500 2000 2500 3000 0 500 1000 1500 2000 2500 3000
Average (cells/µL) Average (cells/µL) Average (cells/µL)
(A) (B) (C)
Difference between UriSed 3 Pre and UX-2000 Difference between UriSed 3 Post and UX-2000 Difference between UriSed 3 Pre and UriSed 3 Post
2000 2000 250
1500 1500

Difference (cells/µL)

Difference (cells/µL)
Difference (cells/µL)

150
1000 1000
x̄ = 39.58 +1.96SD = 356.36 x̄ = 36.09 +1.96SD = 339.60 x̄ = -3.49 +1.96SD = 31.63
500 500 50
0 0
-500 -500 -1.96SD = -267.42 -50 -1.96SD = -38.60
-1.96SD = -277.20
-1000 -1000
-150
-1500 -1500
-2000 -2000 -250
0 500 1000 1500 2000 2500 3000 0 500 1000 1500 2000 2500 3000 0 500 1000 1500 2000 2500 3000
Average (cells/µL) Average (cells/µL) Average (cells/µL)
(D) (E) (F)

F I G U R E   2   (A) Bland-­Altman plot for red blood cells counted by manual and UriSed 3 preclassification assays; (B) Manual and UriSed 3
postclassification assays; (C) Manual and UX-­2000; (D) UX-2000 and UriSed 3 pre-classification assays; (E) UX-­2000 and UriSed 3 postclassification
assays and, (F) UriSed 3 preclassification and UriSed 3 postclassification assays

Difference between Manual and UriSed 3 Pre Difference between Manual and UriSed 3 Post Difference between Manual and UX-2000
4000 4000 4000
3000 3000 3000
Difference (cells/µL)
Difference (cells/µL)

Difference (cells/µL)
2000 2000 2000
1000 +1.96SD = 517.33 1000 x̄ = -15.78 x̄ = 92.23 +1.96SD = 963.89
x̄ = -14.80 +1.96SD = 515.99 1000
0 0 0
-1000 -1000 -1000
-2000 -1.96SD = -546.93 -2000 -1.96SD = -547.56 -1.96SD = -779.43
-2000
-3000 -3000 -3000
-4000 -4000 -4000
0 1000 2000 3000 4000 0 1000 2000 3000 4000 0 1000 2000 3000 4000 5000 6000
Average (cells/µL) Average (cells/µL) Average (cells/µL)
(A) (B) (C)
Difference between UriSed 3 Pre and UX-2000 Difference between UriSed 3 Post and UX-2000 Difference between UriSed 3 Pre and UriSed 3 Post
4000 4000 200
3000 3000 150
Difference (cells/µL)
Difference (cells/µL)
Difference (cells/µL)

2000 2000 x̄ = 108.02 +1.96SD = 937.39 100

x̄ = 107.03 +1.96SD = 937.06
1000 1000 50 x̄ = 0.98 +1.96SD = 23.86
0 0 0
-1000 -1000 -50 -1.96SD = -21.89
-1.96SD = -722.99 -1.96SD = -721.35
-2000 -2000 -100
-3000 -3000 -150
-4000 -4000 -200
0 1000 2000 3000 4000 5000 6000 0 1000 2000 3000 4000 5000 6000 0 1000 2000 3000 4000
Average (cells/µL) Average (cells/µL) Average (cells/µL)
(D) (E) (F)

F I G U R E   3   (A) Bland-­Altman plot for white blood cells counted by manual and UriSed 3 preclassification assays; (B) Manual and UriSed
3 postclassification assays; (C) Manual and UX-­2000 assays; (D) UX-­2000 and UriSed 3 preclassification assays; (E) UX-­2000 and UriSed 3
postclassification assays; and, (F) UriSed 3 preclassification and UriSed 3 postclassification assays

For the semiquantitative grading, almost all pairwise concordance In this study, we compared the performance of the UriSed 3 and
rates of RBCs, WBCs, and ECs were higher than 90%. The weighted UX-­2000 automated urine sediment analyzers with each other and
kappas of RBCs and WBCs of the automated methods compared with with the results of the manual microscopic method. For an overall
the manual method were between 0.61-­0.80, while the κ value of ECs finding, we observed that a higher concentration of sediment was
ranged from 0.41 to 0.60 (Table 4). associated with a lower %CV, which corresponded with the findings
The diagnostic accuracy of both devices compared with manual of other studies.12,20 UX-­2000 test data from this study was similar to
microscopic examination is given in Table 5. UriSed 3 postclassifica- UX-­2000 test data reported in a previous study.1 Our UX-­2000 test
tion had the highest specificity of all parameters, followed by UriSed 3 data was also similar to Sysmex UF-­1000i test data reported by Lee
preclassification. However, the UX-­2000 demonstrated higher sensi- et al.21 Both machines use flow cytometry technique for sediment
tivity for detecting bacteria, hyaline casts, and pathological casts. evaluation.
However, Jiang et al.,22 reported slightly lower between-­run pre-
cision results than found in the present study. For UriSed 3, the %CV
4 |  DISCUSSION was very similar to %CV results reported from a study in UriSed 3
performance that was conducted in a central laboratory in Budapest,
Automated urine analyzers were developed to improve laboratory Hungary.14 Moreover, the overall precision of the UriSed 3 observed in
productivity and to reduce interobserver variability. However, auto- this study was consistent with the findings reported in previous stud-
mated analyzers vary by analytic method and performance. ies of the UriSed and sediMAX analyzers.12,23
LAIWEJPITHAYA et al. |
      7 of 10

Difference between Manual and UriSed 3 Pre Difference between Manual and UriSed 3 Post Difference between Manual and UX-2000
70 70 70
50 50 50

Difference (cells/µL)

Difference (cells/µL)
Difference (cells/µL) +1.96SD = 55.63
30 +1.96SD = 13.58 30 30
+1.96SD = 13.50 x̄ = 10.23
10 x̄ = -2.92 10 x̄ = -2.92 10
-10 -10 -10
-30 -1.96SD = -19.43 -30 -1.96SD = -19.34 -30 -1.96SD = -35.16
-50 -50 -50
-70 -70 -70
0 50 100 150 200 250 300 0 50 100 150 200 250 300 0 50 100 150 200 250 300
Average (cells/µL) Average (cells/µL) Average (cells/µL)
(A) (B) (C)
Difference between UriSed 3 Pre and UX-2000 Difference between Urised 3 Post and UX-2000 Difference between Urised 3 Pre and Urised 3 Post
70 70 20
50 +1.96SD = 60.89 50 +1.96SD = 60.84 15

Difference (cells/µL)

Difference (cells/µL)
Difference (cells/µL)

30 30 10
x̄ = 13.16 x̄ = 13.16 5 x̄ = 0.00 +1.96SD = 1.80
10 10
0
-10 -10 -1.96SD = -1.80
-5
-30 -1.96SD = -34.58 -30 -1.96SD = -34.53
-10
-50 -50 -15
-70 -70 -20
0 50 100 150 200 250 300 0 50 100 150 200 250 300 0 50 100 150 200 250 300
Average (cells/µL) Average (cells/µL) Average (cells/µL)
(D) (E) (F)

F I G U R E   4   (A) Bland-­Altman plot for squamous epithelial cells counted by manual and UriSed 3 preclassification assays; (B) Manual and
UriSed 3 postclassification assays; (C) Manual and UX-­2000 assays; (D) UX-­2000 and UriSed 3 preclassification assays; (E) UX-­2000 and UriSed
3 postclassification assays; and, (F) UriSed 3 preclassification and UriSed 3 postclassification assays

T A B L E   4   Comparison of concordance
Within one grade
rates and kappa coefficients between
Paired Method Parameter concordance rate (%) κ 95% CI
various combinations of methods
UriSed 3 preclassifica- RBCs 95.31 0.675 (0.569-­0.781)
tion vs Manual WBCs 94.95 0.759 (0.706-­0.812)
ECs 100.00 0.558 (0.426-­0.690)
UriSed 3 postclassifica- RBCs 94.95 0.708 (0.616-­0.800)
tion vs Manual WBCs 95.67 0.765 (0.715-­0.816)
ECs 99.64 0.559 (0.422-­0.696)
UX-­2000 vs Manual RBCs 96.03 0.721 (0.616-­0.826)
WBCs 96.75 0.803 (0.741-­0.865)
ECs 98.56 0.452 (0.264-­0.640)
UX-­2000 vs UriSed 3 RBCs 94.95 0.607 (0.494-­0.720)
preclassification WBCs 89.17 0.634 (0.565-­0.703)
ECs 98.19 0.458 (0.289-­0.626)
UX-­2000 vs UriSed 3 RBCs 94.95 0.623 (0.512-­0.733)
postclassification WBCs 89.89 0.644 (0.575-­0.713)
ECs 98.19 0.459 (0.294-­0.624)
UriSed 3 preclassifica- RBCs 98.56 0.903 (0.849-­0.956)
tion vs UriSed 3 WBCs 99.28 0.971 (0.946-­0.996)
postclassification
ECs 99.28 0.965 (0.916-­1.000)

RBCs, red blood cells; WBCs, white blood cells; ECs, squamous epithelial cells; κ, weighted kappa coef-
ficient; CI, confidence interval.

Overall, the UX-­

2000 demonstrated better within-­
run and For quantitative results of RBCs, WBCs, and ECs, we used ICC and
between-­run precision than the UriSed 3, with all CVs <10%. However, Bland–Altman plot to evaluate correlation between methods. UriSed
both instruments had good precision when compared with the preci- 3 preclassification showed excellent agreement for all three elements,
sion data of the manual microscopic method from studies by Chien, which was consistent with data from a study by Bottini et al.23 In addi-
20,22
et al. and Jiang et al. tion, most of data points in Bland–Altman plot were within ±1.96s (stan-
Linear regression graphs from both devices showed excellent correla- dard deviation) of mean difference, which indicates good agreement.
tion, with r>.99 for RBCs and WBCs. Slight differences in measured con- Moreover, Zaman et al.12 showed that after editing, the diagnostic ac-
centration of RBCs and WBCs between instruments, even though they curacy of all three parameters tended to increase, which was similar to
were evaluating the same specimen, may be due to differences in analytic our UriSed 3 postclassification data. For semiquantitative analysis of
method. For carry-­over study, percentages of carry-­over were <2%.24 all three parameters, the results demonstrated very good concordance
 |
8 of 10       LAIWEJPITHAYA et al.

T A B L E   5   Diagnostic accuracy of urine sediment analysis by various methods compared to manual microscopy

Sediment analysis Method Sensitivity (%) Specificity (%) PPV (%) NPV (%)

Bacteria UX-­2000 74.76 90.23 81.91 85.79

UriSed 3 preclassification 59.22 95.40 88.41 79.81
UriSed 3 postclassification 59.22 95.40 88.41 79.81
Yeasts UX-­2000 37.78 96.55 68.00 88.89
UriSed 3 preclassification 40.00 97.84 78.26 89.37
UriSed 3 postclassification 51.11 99.57 95.83 91.30
Small round cells UX-­2000 26.67 84.35 8.89 95.26
UriSed 3 preclassification 80.00 89.31 30.00 98.73
UriSed 3 postclassification 66.67 95.80 47.62 98.05
Hyaline casts UX-­2000 83.33 77.36 14.29 99.03
UriSed 3 preclassification 41.67 99.62 83.33 97.42
UriSed 3 postclassification 66.67 99.25 80.00 98.50
Pathological casts UX-­2000 75.00 80.38 14.75 98.61
UriSed 3 preclassification 75.00 92.08 30.00 98.79
UriSed 3 postclassification 58.33 99.25 77.78 98.13
All types of crystals UX-­2000 42.50 97.97 89.47 80.75
UriSed 3 preclassification 57.50 97.46 90.20 84.96
UriSed 3 postclassification 55.00 98.48 93.62 84.35
Calcium oxalate crystals UX-­2000 N/A N/A N/A N/A
UriSed 3 preclassification 46.03 100.00 100.00 86.23
UriSed 3 postclassification 55.56 100.00 100.00 88.38
Uric acid crystals UX-­2000 N/A N/A N/A N/A
UriSed 3 preclassification 9.09 100.00 100.00 96.38
UriSed 3 postclassification 18.18 100.00 100.00 96.73
Triple phosphate crystals UX-­2000 N/A N/A N/A N/A
UriSed 3 preclassification 33.33 100.00 100.00 98.55
UriSed 3 postclassification 33.33 100.00 100.00 98.55

PPV, positive predictive value; NPV, negative predictive value; N/A, not applicable.

(95%-­100%), which corresponded to outcomes reported by Erzsébet
(90%-­98%).14 However, weighted kappa data from Zaman et al.12 in-
dicated substantial agreement for all RBC, WBC, and EC parameters,
which was similar to our data, except we found only moderate agree-
ment for ECs. In contrast and relative to the qualitative parameters
evaluated by UriSed 3, only non-­ECs and pathological casts showed
sensitivity and specificity similar to data supplied by the manufacturer.
Other parameters, including bacteria, yeasts, hyaline casts, and crys-
14
tals, had similarly high specificity (>85%), but lower sensitivity. These
disparities may be attributable to differences in the microscopes used
in the manual method. Our study used bright-­field microscope, but
the manufacturer used phase-­contrast microscope, which is more ac-
curate. Manual editing resulted in a slight increase in specificity for
almost all parameters. Manual editing also slightly increased the sen-
sitivity for detecting yeasts and hyaline casts. However, the sensitivity
for detecting small round cells, pathological casts, and crystals tended
12
to decrease. Zaman et al. reported that after manually reviewing
three out of 15 images, there was a substantial increase in sensitiv-
ity for all elements, except for pathological casts. These differences
in findings between our study and theirs may be due to differences
in microscope magnification. Zaman, et al. evaluated the performance
of the sediMAX analyzer, which captured images at a magnification
of 400×, but our UriSed 3 images were captured at 100×. Although
15 images at 100× should facilitate a more comprehensive evaluation
of native urine, the quality of the images was lower. The low magnifi-
cation and low resolution observed in UriSed 3 images made particle
identification by the technologist difficult. Even though UriSed 3 pre-
classification and postclassification results were only slightly different,
reanalysis by a well-­trained technologist is recommended, especially
for pathological urine specimen.
Several studies in the UF-­1000i and UF-­100, both of which use
flow cytometry (the same method that is used by the UX-­2000), found
good correlation for quantitative RBCs, WBCs, and ECs between each
of the two urine analyzers and the manual examination method.20-22,25
These findings corresponded with our data, which showed excellent
agreement for RBCs and WBCs, and substantial agreement for ECs be-
tween the automated and manual methods. In contrast, Wesarachkitti
et al.1 reported that UX-­2000 demonstrated only moderate agreement
LAIWEJPITHAYA et al.
for all three types of cells. From a study by Sanchez-­Mora et al., UX-­
2000 had a weighted kappa of 0.573 and 0.819 for categorical analysis
of RBCs and WBCs, respectively. The weighted kappa for WBCs in that
study was similar to our data (κ=0.803); however, we found greater
agreement for RBCs (κ=0.721).26 As for concordance rate within one
grade difference, our results were higher than those from a study of the
UF-­1000i by Lee, et al.—especially for RBCs and ECs (RBCs, 96.1% vs
83.7%; WBCs 96.8% vs 93.8%; ECs 96.6% vs 81.3%). This difference
in findings may be explained by different ranges of grading between
21
the two studies. For the other qualitative sediments, most of our
results had higher sensitivity, except for small round cells, when com-
pared with the study by Wesarachkitti, et al. These differences may be
due to differences in manual microscopic method procedures between
studies. Our study used concordant results obtained by two technolo-
gists. If the technologists were not in agreement, the specimen would
be checked by another highly experienced expert. Wesarachkitti et al.1
used manual examination data obtained by only one well-­experienced
technologist.
When comparing between the UriSed 3 and UX-­2000 against
the manual method, the correlations for quantitative RBC and WBC
counting were in excellent agreement for both devices, but the UX-­
2000 tended to report higher cell counts for ECs, and expressed only
substantial agreement. In terms of semiquantitative analysis, both
machines had very good concordance rates and presented similar κ
values, with substantial agreement for RBCs and WBCs, and moder-
ate agreement for ECs. UriSed 3 revealed higher sensitivity for the
detection of small round cells, while the UX-­2000 showed greater
sensitivity for detecting bacteria and hyaline casts. For specificity,
UriSed 3 was slightly better—especially for hyaline casts and patho-
logical casts.
From our findings, each of the two instruments had its own dis-
tinct advantages, but also some identifiable inferiorities. UriSed 3 re-
quires only a small amount of urine and does not require any reagents.
Moreover, the digital imaging with editing function on the UriSed
3 reduces the time needed for the manual review process, and the
specimen preparation time is also shorter than the preparation time
needed for the UX-­2000. Moreover, the microscopic images from the
UriSed 3 can be documented and can be used for technician training.
The UriSed 3 also has a dual microscope system that includes both a
bright-­field microscope and a phase-­contrast microscope.
European Urinalysis Guidelines recommend using a phase-­contrast
microscope for routine particle identification, and this corresponds
with Clinical and Laboratory Standards Institute (CLSI) approval of
phase-­contrast microscopy as a method for identifying and measuring
urine sediment.2,27 It was expected that the special optical properties
of phase-­contrast technology would yield higher accuracy in the identi-
fication of particles.14 Although the additional phase-­contrast function
may be helpful in differentiating some particles, such as dysmorphic
RBCs, yeasts, and hyaline casts, the 100× magnification resulted in
low-­resolution images, which made particle identification difficult. In
the opinion of the authors, combining a 400× magnification bright-­
field microscope with the phase-­contrast feature would likely result in
better and easier particle identification.
|
      9 of 10
The notable features of the UX-­2000 are that it can analyze all
three parts of urinalysis and its small physical size, which makes it suit-
able for a laboratory with limited space. The UX-­2000 demonstrated
higher sensitivity for bacteria and hyaline casts, but was not able to
distinguish any type of crystals or pathological casts. Other associated
inconveniences and shortcomings include a fluorescent reagent re-
quirement and the lack of an image reviewing function. Moreover, and
according to data reported by Khejonnit et al.,15 approximately 55% of
specimens analyzed by the UX-­2000 required additional examination
by manual microscopic method.
Limitations of this study include a relatively small number of urine
specimens, with limited pathological samples—especially small round
cells, casts, and yeasts. In addition, our performance evaluation be-
tween and among methods was based on statistical correlations, not
clinical correlations. Larger studies that include a larger sample size
and a more sufficiently and statistically quantifiable proportion of all
measured parameters are warranted.
In conclusion, UriSed 3 and UX-­2000 had nearly similar perfor-
mance for the measurement of RBCs and WBCs. However, UriSed
3 was more reliable for measuring ECs and small round cells, while
UX-­2000 was more accurate for identifying bacteria and hyaline
casts.
AC KNOW L ED G M ENTS
Research authors Preechaya Wongkrajang and Kanit Reesukumal
were supported by a “Chalermphrakiat Grant”, Faculty of Medicine
Siriraj Hospital, Mahidol University. The research authors would like
to gratefully acknowledge Miss Julaporn Pooliam for assistance with
statistical analysis.
REFERENCES
1. Wesarachkitti B, Khejonnit V, Pratumvinit B, et al. Performance evalu-
ation and comparison of the fully automated urinalysis analyzers UX-­
2000 and Cobas 6500. Lab Med. 2016;47:124‐133.
2. European Confederation of Laboratory Medicine. European urinalysis
guidelines: summary. Scand J Clin Lab Invest. 2000;60:1‐96.
3. Du J, Xu J, Wang F, et al. Establishment and development of the
personalized criteria for microscopic review following multiple auto-
mated routine urinalysis systems. Clin Chim Acta. 2015;444:221‐228.
4. Winkel P, Statland BE, Jorgensen K. Urine microscopy: an ­ill-­defined
method examined by a multifactorial technique. Clin Chem.
1974;20:436‐439.
5. Block DR, Lieske JC. Automated urinalysis in the clinical lab. MLO
Med Lab Obs. 2012;44:8‐10, 12; quiz 14.
6. Roe CE, Carlson DA, Daigneault RW, Statland BE. Evaluation of the
yellow IRIS®: an automated method for urinalysis. Am J Clin Pathol.
1986;86:661‐665.
7. Deindoerfer FH, Gangwer JR, Laird CW, Ringold RR. The Yellow IRIS”
urinalysis workstation–the first commercial application of “automated
intelligent microscopy. Clin Chem. 1985;31:1491‐1499.
8. Yüksel H, Kiliç E, Ekinci A, Evliyaoğlu O. Comparison of fully auto-
mated urine sediment analyzers H800-­FUS100 and labumat-­urised
with manual microscopy. J Clin Lab Anal. 2013;27:312‐316.
9. Mayo S, Acevedo D, Quinones-Torrelo C, Canos I, Sancho M. Clinical
laboratory automated urinalysis: comparison among automated mi-
croscopy, flow cytometry, two test strips analyzers, and manual
 |
10 of 10       LAIWEJPITHAYA et al.
microscopic examination of the urine sediments. J Clin Lab Anal. 2008;22:
262‐270.
10. Lamchiagdhase P, Preechaborisutkul K, Lomsomboon P, et al. Urine
sediment examination: a comparison between the manual method
and the iQ200 automated urine microscopy analyzer. Clin Chim Acta.
2005;358:167‐174.
11. Altekin E, Kadicesme O, Akan P, et al. New generation IQ-­200 auto-
mated urine microscopy analyzer compared with KOVA cell chamber.
J Clin Lab Anal. 2010;24:67‐71.
12. Zaman Z, Fogazzi GB, Garigali G, Croci MD, Bayer G, Kránicz T. Urine
sediment analysis: analytical and diagnostic performance of sedi-
MAX® — A new automated microscopy image-­based urine sediment
analyser. Clin Chim Acta. 2010;411:147‐154.
13. UriSed 3 Fully automated urine sediment analyzer user manual sw
3.1. 77 Elektronika Kft. 2016.
14. Erzsébet N. Diagnostic Performance of the UriSed 3 Automated Urine
Sediment Analyzer Instrument: Clinical Study Made in the Polyclinic of the
Hospitaller Brothers of St. John of God in Budapest, Central Laboratory.
Budapest; 2015.
15. Khejonnit V, Pratumvinit B, Reesukumal K, Meepanya S, Pattanavin C,
Wongkrajang P. Optimal criteria for microscopic review of urinalysis fol-
lowing use of automated urine analyzer. Clin Chim Acta. 2015;439:1‐4.
16. Fully Automated Integrated Urine Analyzer UX-2000 Instructions for use.
Sysmex Corporation; 2010-2011.
17. Plastic Counting Chambers. VACUTEST KIMA.
18. Jintasuthanont P, Khejonnit V, Opaskiattikul N, Chinswangwatanakul
W, Gonggetyai V. Evaluation of the performance of the automated
urine sediment analyzer “Urised” compared with the manual method.
Siriraj Med J. 2010;62:152‐156.
19. Landis JR, Koch GG. The measurement of observer agreement for cat-
egorical data. Biometrics. 1977;33:159‐174.
20. Chien T-I, Kao J-T, Liu H-L, et al. Urine sediment examination: a com-
parison of automated urinalysis systems and manual microscopy. Clin
Chim Acta. 2007;384:28‐34.
21. Lee W, Ha J-S, Ryoo N-H. Comparison of the automated cobas u 701
urine microscopy and UF-­1000i flow cytometry systems and man-
ual microscopy in the examination of urine sediments. J Clin Lab Anal.
2016;30:663‐671.
22. Jiang T, Chen P, Ouyang J, Zhang S, Cai D. Urine particles analysis:
performance evaluation of Sysmex UF-­1000i and comparison among
urine flow cytometer, dipstick, and visual microscopic examination.
Scand J Clin Lab Invest. 2011;71:30‐37.
23. Bottini PV, Martinez MH, Garlipp CR. Urinalysis: comparison between
microscopic analysis and a new automated microscopy image-­based
urine sediment instrument. Clin Lab. 2014;60:693‐697.
24. Broughton PM, Gowenlock AH, McCormack JJ, Neill DW. A revised
scheme for the evaluation of automatic instruments for use in clinical
chemistry. Ann Clin Biochem. 1974;11:207‐218.
25. Manoni F, Tinello A, Fornasiero L, et  al. Urine particle evaluation: a
comparison between the UF-­1000i and quantitative microscopy. Clin
Chem Lab Med. 2010;48:1107‐1111.
26. Sanchez-Mora C, Acevedo D, Porres MA, et al. Comparison of auto-
mated devices UX-­2000 and SediMAX/AutionMax for urine samples
screening: a multicenter Spanish study. Clin Biochem. 2017; (in press).
http://doi.org/10.1016/j.clinbiochem.2017.02.005.
27. Clinical and Laboratory Standard Institute (CLSI). Urinalysis: Approved
guideline, 3rd edn. CLSI document GP16-A3. Wayne, PA, USA: CLSI;
2009.
How to cite this article: Laiwejpithaya S, Wongkrajang P,
Reesukumal K, et al. UriSed 3 and UX-­2000 automated urine
sediment analyzers vs manual microscopic method: A
comparative performance analysis. J Clin Lab Anal.
2018;32:e22249. https://doi.org/10.1002/jcla.22249