Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
2016;62:49-56
©Copyright
ORIGINAL ARTICLE
SUMMARY
Background: Urinalysis is the third major test in clinical laboratory. Manual technique imprecision urges the need
for a rapid reliable automated test. We evaluated the H800-FUS100 automatic urine sediment analyzer and com-
pared it to the manual urinalysis technique to determine if it may be a competitive substitute in laboratories of
central hospitals.
Methods: 1000 urine samples were examined by the two methods in parallel. Agreement, precision, carry-over,
drift, sensitivity, specificity, and practicability criteria were tested.
Results: Agreement ranged from excellent to good for all urine semi-quantitative components (K > 0.4, p = 0.000),
except for granular casts (K = 0.317, p = 0.000). Specific gravity results correlated well between the two methods
(r = 0.884, p = 0.000). RBCS and WBCs showed moderate correlation (r = 0.42, p = 0.000) and (r = 0.44,
p = 0.000), respectively. The auto-analyzer's within-run precision was ≥ 75% for all semi-quantitative components
except for proteins (50% precision). This finding in addition to the granular casts poor agreement indicate the
necessity of operator interference at the critical cutoff values. As regards quantitative contents, RBCs showed a
mean of 69.8 ± 3.95, C.V. = 5.7, WBCs showed a mean of 38.9 ± 1.9, C.V. = 4.9). Specific gravity, pH, mi-
croalbumin, and creatinine also showed good precision results with C.Vs of 0.000, 2.6, 9.1, and 0.00 respectively.
In the between run precision, positive control showed good precision (C.V. = 2.9), while negative control's C.V.
was strikingly high (C.V. = 127). Carryover and drift studies were satisfactory. Manual examination of inter-
observer results showed major discrepancies (< 60% similar readings), while intra-observer's results correlated
well with each other (r = 0.99, p = 0.000).
Conclusions: Automation of urinalysis decreases observer-associated variation and offers prompt competitive re-
sults when standardized for screening away from the borderline cutoffs.
(Clin. Lab. 2016;62:49-56. DOI: 10.7754/Clin.Lab.2015.150520)
Table 1. Showing results of the manual and automated techniques for the tested parameters, with the corresponding method of
comparison results.
Manual Automated r, K *, †
Specific gravity 1019.8 (± 5.7) 1019 (± 7.2) r: 0.884,
[mean (± SD)] C.V. = 0.6 C.V. = 0.7 p = 0.00
pH Acidic Neutral Alkaline Acidic Neutral Alkaline
K: 0.914
(%) 89.9 2.0 8.1 89.9 3.6 6.5
Albumin Nil + ++ +++ ++++ Nil + ++ +++ ++++
K: 0.695
(%) 62.8 7.2 11.1 9.9 9.0 62.8 7.6 9.7 8.9 11.0
Glucose Nil + ++ +++ ++++ Nil + ++ +++ ++++
K: 0.78
(%) 88.0 3.5 4.1 4.2 0.2 87.8 3.2 2.4 6.2 0.4
Ketone Negative Positive Negative Positive
K:1
(%) 98.6 1.4 98.6 1.4
Nitrite Negative Positive Negative Positive
K:1
(%) 98.2 1.8 98.2 1.8
RBCs [median r: 0.417
1.5 (0.5 - 5) 4 (1.75 - 16)
(25th - 75th percentiles)] p = 0.00
WBCs [median r: 0.442,
1.5 (0.5 - 9) 2 (0.75 - 12)
(25th - 75th percentiles)] p = 0.00
Epithelial cells Nil Few + ++ +++ ++++ Nil Few + ++ +++ ++++
K: 0.765
(%) 59.0 12.0 17.4 7.9 2.2 1.5 58.0 15.3 14.8 7.6 2.8 1.5
Calcium Nil Few + ++ +++ ++++ Nil Few + ++ +++ ++++
K: 0.728
Oxalate 90.1 2.2 3.7 0.9 1.3 1.8 90.3 2.7 3.4 0.8 1.1 1.7
Crystals Nil Few + ++ +++ ++++ Nil Few + ++ +++ ++++
Uric acid K: 0.65
(%) 93.5 1.6 2.2 1.2 0.8 0.7 94.4 1.9 1.7 0.8 0.5 0.7
Triple Nil Few + ++ +++ ++++ Nil Few + ++ +++ ++++
K **
phosphate 99.6 0.2 0.0 0.1 0.0 0.1 99.7 0.1 0.1 0.0 0.0 0.1
Nil Few + ++ +++ ++++ Nil Few + ++ +++ ++++
Urate K: 0.71
Amorphous 97.8 0.1 1.0 0.5 0.2 0.4 98.0 0.2 0.6 0.4 0.1 0.7
(%) Nil Few + ++ +++ ++++ Nil Few + ++ +++ ++++
Phosphate K **
99.4 0.0 0.0 0.2 0.3 0.1 99.3 0.1 0.0 0.2 0.2 0.2
Nil Few + ++ Nil Few + ++
Hyaline K **
97.9 0.9 1.2 0.0 97.2 2.3 0.1 0.0
Casts Nil Few + Nil Few +
Granular K: 0.317
(%) 97.4 1.7 0.9 99.2 1.7 0.1
Nil Few Nil Few
Waxy K **
99.9 0.1 100.0 0.0
Bacteria Nil Few + ++ +++ ++++ Nil Few + ++ +++ ++++
K: 0.74
(%) 72.9 4.9 7.3 4.1 3.2 7.6 72.1 7.6 5.9 3.9 3.3 7.2
Fungi Nil Few + ++ +++ ++++ Nil Few + ++ +++ ++++
K: 0.57
(%) 92.1 2.6 2.4 0.9 1.0 1.0 88.2 6.5 2.4 0.9 1.3 0.7
Mucous Nil Few + ++ +++ ++++ Nil Few + ++ +++ ++++
K: 0.559
(%) 83.5 0.4 8.4 4.6 2.2 0.9 76.1 7.7 9.9 3.4 2.1 0.8
Sperm Nil Few + ++ +++ ++++ Nil Few + ++ +++ ++++
K **
(%) 99.3 0.4 0.2 0.0 0.0 0.1 99.1 0.5 0.1 0.2 0.0 0.1
* Agreement was calculated for qualitative variables expressed in K, correlation was calculated for quantitative variables expressed in r.
** Kappa statistics could not be performed as it requires a symmetric 2-way table in which values of the first valuable must match the values
of the second variable. † The p-value of the kappa test was < 0.05 for all shown parameters denoting sound statistical calculations.
Table 2. Performance characteristics of the H-800 versus the manual technique in diagnosis of hematuraia and pyuria.
Table 4. Showing the descriptive statistics of the semiquantitative variables of the inter-observer study.
Table 5. The descriptive statistics of the quantitative variables of the inter-observer study.
9.8%, general medicine 8.4%, dermatology 7.2%, neu- H800-FUS100 are shown in Table 2. Cutoff points were
rology 5.9%, ICUs collectively 6.6%, cardiology 3.7%. considered > 5 cells/HPF for the diagnosis of hematuria
The remaining 15% were patients of general and special [5] and ≥ 10 for leukocytes as it is more relevant to the
surgeries, psychiatry, and occupational health. Males reflex culture cutoff [6].
represented 48% of the cohort, while 52% were fe- A second cutoff for hematuria was searched. It was des-
males. ignated by the researchers to be ≥ 10/HPF for the cases
Agreement and correlation studies’ results are shown in tested by automation, and the specificity was recalculat-
Table 1. Figures 1 and 2 show how H800-FUS100 strat- ed. Specificity improved to 88.9%, sensitivity remained
ified the studied patients as regards the diagnosis of he- acceptable at 98.8%, with DOR at 1.3 (CI; 1.22 - 1.36).
maturia and pyuria in comparison to the manual meth- Within-run precision testing for the semi-quantitative
od. The derived performance characteristics of the variables by FUS-100/H-800 showed exactly the same
Figure 1. How the FUS100H800 stratified hematuria patients at cutoff more than 5HPF.
Figure 2. How the FUS100H800 stratified pyuria patients at cutoff equal or more than 10HPF.
results for 20 successive times for casts, bacteria, fungi, with a C.V. = 127%. Median negative control was 0.75
urobilinogen, bilirubin, ketone, occult blood, glucose, with (25th - 75th% = 0 - 2).
WBCs, and nitrite. Results of crystals, mucus, and epi- No carryover contaminated the three negative samples
thelial cells showed 95%, 80%, and 75% concordance run after three positive samples. No drift was observed.
respectively. Protein content precision was the least For the manual technique, Tables 4 and 5 show the de-
(50% ++, 45% +, and 5% nil). Quantitative variables of scriptive statistics of the semi-quantitative and quantita-
within run precision are shown in Table 3. tive variables of the inter-observer assay. In the intra-
In the between-run precision study, the positive control observer study, a strong positive correlation was found
showed a mean ± SD of 1050 ± 29.5, C.V. = 2.9%. between the two runs of samples read by the same ob-
Negative control repeats had a mean ± SD of 1.04 ± 1.3 server with r = 0.999 for both erythrocytes and leuco-
cytes and a p = 0.000. Practicability criteria of both further study relating the lab diagnosis to the clinical
methods are shown (Table 6). outcome is mandatory to resolve this debate, and per-
haps new cutoffs for hematuria and pyuria will be des-
ignated for the automated count. Furthermore, regarding
DISCUSSION the manual method, at the high level counts, erythro-
cytes and leucocytes are expressed as > 100/HPF. Auto-
Urine analysis is the third major diagnostic screening mated methods, on the other hand, can read figures up
test in the clinical laboratory, only preceded by serum/ to orders of thousands. This means that automated
plasma chemistry profiles and complete blood count methods in general will yield higher readings when
analysis [2]. A correct urine analysis result offers a di- compared to manual results especially at higher levels
rect indication of the state of the patient's renal and gen- of RBCs or WBCs, which is not necessarily clinically
itourinary system and a monitoring of other body sys- significant. This is comparable to previously shown re-
tems [5]. sults [4] on a different instrument (IQ200 analyzer) in
In this study, the H800-FUS100 was assessed as an al- which more RBCs and WBCs were obtained with the
ternative to the manual urine analysis testing procedure. automated than with the manual microscopy.
In contrast to other published studies; this study includ- The fully automated urine devices classify some cells as
ed a large sample size, with a single observer perform- “unclassified” and allow the trained operator to correct
ing both the manual microscopic examination and auto- the designations of the cells. The device accuracy in-
mated visual classification of unclassified particles. The creases when the operator examines these cells and
effectiveness of the manual technique as a standard makes the correct designations [5].
method was also explored. Other performance criteria for the H800-FUS100
The H800-FUS100 agreed well with the manual method showed a good within-run precision for either semi-
in ketone (K = 1), nitrite (K = 1), pH (k = 0.914), glu- quantitative or quantitative parameters except for the
cose (k = 0.78), epithelial cells (k = 0.765), calcium ox- protein content of the sample. Casts, bacteria, fungi,
alate crystals (k = 0.728), bacteria (k = 0.74), amor- urobilinogen, bilirubin, ketone, occult blood, glucose,
phous urates (k = 0.71), albumin (k = 0.695), uric acid and nitrite showed exactly the same results for 20 re-
crystals (k = 0.65), mucous (k = 0.559), and fungi peated times. Calcium oxalate crystals, epithelial cells,
(k = 0.57) with a p-value of (0.000) for all. Granular and mucus showed > 75% precision. Only the protein
cast was the only item that showed poor agreement be- content of the sample failed to show a good precision
tween the two methods (k = 0.317) (p = 0.0000). with 50% of the times (10 times ) showing two pluses
A good correlation was found between the two methods (++), 45% showing one plus (+), and one time showing
as regards the specific gravity (r = 0.884, p = 0.000). a negative result.
Also a moderate correlation was found for RBCs Red blood cells content showed a mean of 69.8 ± 3.95
(r = 0.42, p = 0.000) and WBCs (r = 0.44, p = 0.000). with a C.V. of 5.7. WBCs showed a mean of 38.9 ± 1.9,
Yuksel et al. [5] did a study on the same apparatus and C.V. = 4.9. Specific gravity, pH, microalbumin, and
found that the strip tests for SG and pH and the micro- creatinine also showed good precision results with C.Vs
scopic tests for leukocyte and epithelial cell counts of 0.000, 2.6, 9.1, and 0.00 respectively.
showed good correlations, while protein, glucose, ke- In the between-run precision, good results were found
tones, and erythrocytes showed moderate correlations. for level II (positive) control, with a mean of 1050
Due to the small number of positive specimens for uro- + 29.5, C.V. = 2.9. For level I (negative) control the
bilinogen and bilirubin, it was difficult to find concor- C.V. was strikingly high (C.V. = 127). However, this
dance either in our study or in theirs. However, nitrite can be explained by the very low values of the negative
positive cases were prevalent enough in the current control (0 - 4), which are below the critical medical de-
study and thus agreement could be established. cision level of “5 particles/HPF”.
The H800-FUS100 showed acceptable performance In Yuksel et al. [5] study, the between- and within-pre-
characteristics; however, there was a high false positive cision were lower for the specimens with fewer cells,
rate noted especially in erythrocytes, which means that while for the specimens containing a large number of
the H800-FUS100 diagnoses 41.2% more cases of he- cells, the reproducibility was much better, which is sim-
maturia than the manual method does. Another cutoff ilar to our study and to the previous studies [7].
was set for hematuria for the H800-FUS100 at ≥ 10HPF Also, Mayo et al. [3] stated that precision studies pro-
and as expected the specificity (88.9%) was increased vided adequate variation coefficient values except for
with no major effect on the sensitivity (98.8%). This the samples with poor cellularity.
means that the H800-FUS100 has more tendency to Lamchiagdhase et al. [8] also reported that between-run
classify results in the range of (> 5 - ≥ 10HPF) as a CVs of RBC for negative and positive control were
round figure, a range at which manual microscopic ob- 61.6% and 6.4%, respectively.
servers may designate the sample as normal. It is not The results of the carryover study were consistent with
certain whether this is due to automation overestimation all urine contents of the three negative samples remain-
or to manual technique underestimation (cellular loss ing the same, even when run after three positive sam-
during centrifugation or a manual counting error). A ples. This ensures the efficiency of the apparatus in ana-