Implicationsof ProbabilityAnalysisfor

CLIN. CHEM. 35/8, 1663-1668 (1989)
Implicationsof ProbabilityAnalysis for Interpreting Results of Leukocyte Esterase and Nitrite
Test Strips
Bjam
J. Bolann,’ Sverre Sandberg,’ and AsbJ.rn DIgranes2
We examined 288 urine samples, using test strips, sediment
microscopy, and culture. The ability of the leukocyte esterase
and nitrite test strips to detect or exclude urinaty tract
infection,as defined by a positive culture, was evaluated by
probability analysis. We found that the diagnostic efficiency
of the esterase-nitrite combination was similar to that of
sediment microscopy. Moreover, once the strip test results
had been obtained, little additional information was given by
microscopy. The importance of estimating the prevalence, or
pre-test probability, of infection before the test result is
evaluated is emphasized. We conclude that, for detecting or
excluding urinary-tract infection, microscopy can be replaced
by the esterase and nitrite test strips. If the probability of
infection predicted by the test strips is not high (or low)
enough compared with medical decision limits, the samples
should be cultured.
AddItIonal
urine
Keyphrases:
microscopy and culture compared
jflf1JQf7
.
screening
The laboratory
diagnosis
of urinary-tract
infection is
made by a quantitative determination of bacteria and (or)
leukocytes in the urine. This can be done by microscopy
and
(or) culturing the urine.
In recent years, test strips have been developed for detecting components from bacteria and leukocytes in urine. The
use of test strips instead of sediment microscopy would save
time and expense. However, despite numerous studies evaluating the strips, the role of test strips in clinicaland
laboratory decision making remains controversial (1).
We performed the present work to evaluate the test
strips for detecting leukocytes and bacteria in urine, with
an emphasis on how their results can form a basis for
decision making in the diagnosis and treatment of urinary
tract infections. The evaluation of the strips includes the
following steps:
and specificity of the test strips and of
were calculated, defining
urinary
tract infection by the result of urine culturing.
(b) The sensitivity and specificity of sediment microscopy
were recalculated in the subgroups of esterase-positive and
esterase-negative urines.
(c) The predictive values of the test results were calculated as a function of the pre-test probability of urinary
tract infection.
(d) Using medical decision limits, we calculated how
microscopy,
when performed subsequent to the use of test
strips, can influence the medical decision.
Finally, we discuss what tests should be done in clinical
practice, and how the test results should be interpreted.
(a) The sensitivity
sediment
microscopy
1Laboratory
of Clinical Biochemistry, and 2Depar’ent
of
Microbiology and Immunology, the Gade Institute, University of
Bergen, Haukeland Hospital, N-5021 Bergen, Norway.
Received January 19, 1989; acceptedMay 15, 1989.
MaterIals and Methods
Specimens were clean-voided midstream urine, obtained
as first morning collections from patients of both sexes,
hospitalized at Haukeland Hospital. The 288 urine samples
obtained were examined 1 to 3 h after collection, by use of
test strips for leukocyte esterase and nitrite, by routine
microscopy, and by culturing.
The test strips, Multistix#{174}7
(Ames Division,Miles Laboratories,
Inc., Elkhart,
IN) and
Nephur-Test#{176}+Leuco (Boehringer
Mannheim
GmbH,
Mannheim, F.R.G.), were used according to the manufacturers’ instructions. The strips for leukocyte esterase have
a color scale from 0 to 3. Results 1 were generally
considered positive, but cutoff points at 2 and 3 were also
evaluated. The nitrite field was recorded as either positive
(1) or negative (0).
For microscopy, 10 mL of urine was centrifuged at 180 X
g (1000 rpm) for 5 mm. The sediment was resuspended in
two drops of Sternheinier-Malbin
stain (2). Leukocytes
and
bacteria were counted per high-power field, defined as the
field of vision at 400 x magnification. Microscopy was
considered positive when the number of leukocytes was >5
per high-power field.
Urine specimens were cultured by application of 1 pL
with calibrated platinum loops onto bromthymol lactose
agar and blood agar plates. Cultures were examined after
incubation overnight at 37 #{176}C
in air. All isolates were
identified by routine bacteriological methods. For the purpose of this study, the growth of> iO bacteria per milliliter
of urine was considered to indicate infection.
Statistical methods. Sensitivity and specificity were calculated as described elsewhere (3,4). The predictive values
are defined as
pos. predictive value
true positive test results
all positive test results
=
SexP
=
Se x P + (1
Sp)(1
-
where Se = sensitivity, Sp = specificity, and P
lence of the disease in question, and
neg. predictive value
true negative test results
all negative test results
Sp(1
-
preva-
=
Sp(1
=
=
P)
-
-
P)
P) + P(1
-
Se)
Resu Its
Correlation between Sample Cultures and Test Strips
Cultures were positive for 26.4% of the samples. The
ability of the test strips and microscopy to detect or exclude
infection (as defined by a positive culture) can be expressed
as the sensitivity and specificity of the tests (Table 1).
The Ames and Boehringer nitrite test strips had respective sensitivities of 0.38 and 0.40, and respective specificiCLINICAL CHEMISTRY, Vol. 35, No. 8, 1989 1663
ties of 0.97 and 0.98. The esterase strips from these suppliers had sensitivities of 0.77 and 0.75, and specificities of
0.79 and 0.81, respectively. If the esterase strip was considered positive only when the observed value was 2, the
sensitivity of Boehringer’s esterase strip was better than
that of Ames. This was the only significant difference
between the strips (Table 1). Therefore, only results from
the Boehringer strips are presented in the following.
Using a combination of various tests can improve the
sensitivity or the specificity of an examination. If one
requires that both tests must be positive for the combination to be considered positive (series approach), then the
specificity will be higher, but the sensitivity will be lower
than that of either test alone, If, on the other hand, the
combination is considered positive when one or both tests
are positive, and negative only when both tests are negative (parallel approach), the sensitivity will be higher, but
the specificity will be lower than that of either alone (3, 5).
Table 2 shows the results of some test combinations.
Microscopy alone (Table 1) had either a lower sensitivity,
or a lower specificity, than the parallel combination of
esterase and nitrite, depending on the cutoff point chosen.
When the samples were divided into esterase-positive
and esterase-negative urines, the sensitivity and specificity
of microscopy in these subgroups was different (Table 3).
Especially, microscopy had a low sensitivity in esterasenegative samples and a low specificity in esterase-positive
samples.
Correlation between Test Strips and Microscopy
Esterase test vs leuhocyte count. Of samples with 5
leukocytes per high-power field, 92% were esterase negative; of samples with >5 leukocytes per high-power field,
69% were esterase positive. Conversely, of the esterasenegative urines, 81% had s5 leukocytes per high-power
field, and of the esterase-positive urines, 86% had >5
leukocytes per high-power field. Table 4 gives the correlation between results from the esterase test strips and
microscopy.
Nitrite test vs bacterial count. Of samples with 50
bacteria per high-power field, 32% were positive for nitrite.
Table 1..SensItivity and SpecificIty of Test Strips and
of Sediment Microscopy: SIngle Tests
Test
Nitrite
Crttsrlon for
positivity
1
Sensitivity
0.38
Ames
Boehringer
1
Esterase
Ames
3
Boehnnger
3
Microscopy
Leukocytes per HPF
>5
>15
>30
Bacteria per HPF
>10
>50
Specificity
0.40
0.98
0.97
0.77
0.43
0.20
0.75
0.55
0.30
0.79
0.93
0.97
0.81
0.84
0.68
0.47
0.79
0.65
0.73
0.84
0.90
0.54
0.80
HPF, high-powerfield.
1664 CLINICAL CHEMISTRY, Vol. 35, No. 8, 1989
0.92
0.96
Table 2. SensitIvity and Specificity
Combinations of Tests
Tests
Nitrite,
Criteria for positivity
N 1 and (or) E1
esterase
of Some
Sensitivity
0.82
0.69
N 1 and (or) E2
N 1 and (or) E 3
N 1 and E1
Esterase,
E1 and (or) >5 leuk/HPF
microscopy E2 and (or) >5 leuk/HPF
E1 and >5 leuk/HPF
Nitrite,
N 1 and (or) E1 and (or)
>5 Ieuk/HPF
esterase,
Specificity
0.33
0.89
0.87
0.68
0.91
0.80
0.90
0.94
0.98
0.70
0.74
0.86
0.69
0.91
0.63
0.58
0.90
0.59
microscopy
Microscopy
>5 leuk and (or) >50
bact/HPF
>5 leuk and >50 bactJHPF
N, nitrite;
E, esterase;
leuk,leukocytes;bact,bacteria;HPF, high-power
field.
Table 3. SensitIvity and SpecIficity of Microscopy
Results, AccordIng to Results of the Esterase Test
Strip
Microscopy result,
Ieukocytes/
field
high-power
Sensitivity
Specificity
0.60
0.35
0.10
0.86
0.95
0.97
Esterase test was negative
>5
>15
>30
Esterase test waspositive
>5
>15
0.93
0.23
0.79
0.36
>30
0.59
0.59
a Nottobe confusedwiththe sensitivity
and specificity of the combinationof
esterase
and
microscopy.
Table 4. Esterase Test Strip and Sediment Microscopy
Results Compared
Esterase
rsuit
0
1
2
3
Total
No.00 Ieukocyt.s
156k
10
3
0
169
per high-power field
6-15
16-30
>30
21
10
3
0
34
9
10
6
4
29
7
8
14
27
56
Total
31
26
38
193
288
Number of samples.
Of samples with <50 bacteria
were negative for nitrite.
per high-power field, 96%
Predictive Values
The predictive value of a test is a function of the prevalence of the condition being looked for. The positive predictive value is the probability that a positive test result
reflects the presence of the disease in question, and the
negative predictive value is the probability that a negative
test result reflects the absence of the disease (3,4).
The a priori probability
(pre-test probability
before any
examination has been performed) that a person has a given
disease is equal to the prevalence of the disease in the
tested group, as illustrated by the 45#{176}
line in Figure la. If
a relevant test result is positive, then the probability that
the patient has the disease exceeds the a priori probability.
This post-test probability equals the positive predictive
value of the test (Figure la, upper left curves). If, on the
other hand, the test result is negative, the probability of
disease is decreased, and the post-test probability now
equals 1 minus negative predictive value (lower right
curves). Thus, at a given prevalence, the vertical distance
from the a priori probability line to the post-test probability
curve illustrates the information gain given by the test.
Plotting results for several tests in the same diagram, as in
Figure la, more easily demonstrates which test gives the
most information.
The highest gain of positive information (confirming
infection) was given by a positive nitrite test result, followed by a positive esterase test result. The highest gain of
negative information (contradicting infection) was given by
a negative microscopy result (5 leukocytes per highpower field). Of the three tests, microscopy had the lowest
positive information gain when the cutoff value for positive
was >5 leukocytes
per high-power field; increasing the
cutoff to >15 leukocytes per high-power field increased the
informativeness of microscopy to that for the esterase test
(not shown in the figure).
Figure lb shows the probability
of infection
given by
some parallel combinations of tests, as compared with
microscopy alone. The highest gain of positive information
was given by the parallel combination of esterase and
nitrite. This combination also gave about the same amount
of negative information as did the observation of s5 leukocytes per high-power field by microscopy.
Discussion
In the traditional evaluation of the leukocyte esterase
test strip, the test result is compared with the number of
leukocytes determined
per microliter
of uncentrifuged
urine in a counting chamber. Both the original 15-mn
leukocyte esterase test strip (6-9), and later the 1-to 2-mn
version
(10, 11), which is represented in this study, have
been found to give positive results for 80% to 95% of the
samples with 10 leukocytes per microliter
of urine.
When the test strip result was compared with the number of leukocytes determined per high-power field in the
urinary
sediment (12-16), 70-95% of the samples with
leukocyturia
were esterase positive, compared with 69% in
the present study. However, in the previous studies the
sediments were prepared in different ways, and the criteria
for leukocyturia varied.
Beyond a general prediction of leukocyturia,
we found
that the esterase test had a limited ability to predict the
number of leukocytes found in the sediment (Table 4).
Similar observations have been made by others (12).
Still controversial is the number of bacteria per milliliter
of urine that must be present to reliably indicate infection
(17, 18). Consequently, the ability of the various tests to
detect infection is evaluated differently in the literature.
The sensitivity and specificity of the various tests in this
study (Table 1) were similar to the observations of other
authors who used similar criteria for infection (15, 19-24).
Sensitivity is often lower when the diagnosis of infection is
based on other criteria, e.g., low-count bacteriuria, other
abnormalities in the urine, or clinical symptoms (15,22,23,
25,26).
When >5 leukocytes per high-power field was used as a
cutoff point for sediment microscopy, we found that the
sensitivity of microscopy to diagnose infection was greater,
but the specificity was less, than that of the esterase test
strip (Table 1). Others have found microscopy to be rather
similar to esterase in this respect (19). To our knowledge,
no authors have found sediment microscopy to be markedly
better than the test strips for detection of bacteriuria.
The nitrite test had a low sensitivity, but it also had the
highest specificity of any of the tests we examined. Demonstration of bacteria by microscopy had either a low
sensitivity or a low specificity, depending on the cutoff
point chosen (Table 1).
When test strips are used, results from two or more tests
are obtained at the same time, and the result from a
1
c
Test po.
C
0
0
.4
.4
4)
U
S
C)
C
C
0
.4
0
-5
>‘
4)
>‘
$3.
.4
-I
.0
a
.0
a
.0
0
L
.0
0
L.
a-
a-
Test
nsg
0
0
0
.5
0
PrQvo
1 QflCQ
-
PrQva
1 nca
Fig. 1. Correlation between prevalenceand probabilityof infection,aftervarioustest results
Cutoff point for positive esterase: 1. A. Singletests:nitrite (- - -), esterase (---),
microscopyat >5 leukocytes per high-powerfield(-),
microscopy at
>50 bacteria per high-power field (
). B. Parallelcombinationsof tests: esterase-nitrite
(- - -),
esterase-microscopy
at >5 leukocytes per high-power field
-),
combinationof >5 leukocytesand (or) >50 bacteriaper high-powerfield (
), comparedwith microscopy at >5 leukocytes
per high-powerfieldalone
(-)
CLINICAL CHEMISTRY, Vol. 35, No. 8, 1989
1665
combination
of tests may give more information than
either test alone. Many studies have combined the esterase
and nitrite strips in a parallel approach, yielding a sensitivity that varied between 0.78 and 0.92 and a specificity
between 0.60 and 0.98 (15, 19, 20, 22, 23, 2 7-29). These
values are similar to our results (Table 2).
series of tests, one may finally achieve a post-test probability that is high (or low) enough to make a clinical decision.
The sensitivity and specificity of the combination of two
independent tests can be calculated from the corresponding
values of either test (3). Conversely, whether or not two
tests are independent can be calculated when the sensitivity and specificity of the tests and of their combination is
Predictive Values
known.
The pre-test probability of a disease’s occurring is of
fundamental importance for the predictive value of the test
result. This probability may be the same as the prevalence
of the disease in the population studied, or it may be the
result of preceding investigations, such as a clinical examination. From the formulas for predictive values (see Statistical methods) it follows that if the pre-test probability of
infection is very low, the post-test probability must also be
low. Correspondingly, at a very high pre-test probability, a
high post-test probability of infection may be present despite a negative test result.
To illustrate, suppose that urinalysis
is performed in a
population where the prevalence of urinary tract infection
is 0.05. Then the pre-test probability for an individual to
have an infection is 0.05. If a positive nitrite test result is
found, the probability of infection will be equal to the
positive predictive value, which is
If the esterase and nitrite strips were independent of
each other, the parallel combination
would give
1
-
Se’N/E
(1
where Se
-
(1
=
0.40)(1
-
SeN)(l
-
0.75)
-
=
SeE)
=
0.15,
sensitivity,N
= nitrite
test, E = esterase test
the parallel combination of the
two, and Se is the theoretical sensitivity for independent
(cutoff
=
point
l),
N/E
=
tests.
Thus, 1 = 0.15, and the theoretical sensitivity
would be 0.85, compared with the 0.82 we obtained empirically (Table 2).
Similarly, with regard to specificity (Sp)
SP’NIE = SPN X
SPE = 0.97 x 0.81
=
0.79,
SexP
Se x P + (1
-
Sp)(1
-
P)
0.4 x 0.05
0.4 x 0.05 + (1
-
0.97)(1
-
0.05)
=
0.41
as also illustrated in Figure la. Thus, the presence of
infection is considered more likely than it was before the
test was done.
If sediment microscopy was chosen instead of the nitrite
test, a corresponding calculation shows that the probability
of infection after a positive result would be only 0.14. This
means that the nitrite test gave much more information
than did microscopy. Regardless of which test was chosen,
however, infection would not be very likely, even after a
positive test, because of the low prevalence. Additional
investigations would have to be done.
Therefore, to obtain a clinical diagnosis, two or more
tests are often performed in succession. The post-test probability of disease, obtained from the first test, can then be
used as the pre-test probability of the next, presupposing
that the tests are independent of each other (30). After a
whereas 0.80 was obtained empirically.
Thus, the two test strips seem to be independent of each
other. This was to be expected, as their respective means of
detecting infection are quite different (10, 31).
On the other hand, if the esterase strip and microscopy
were independent of each other, one should expect a sensitivity of [1 (1
0.84)(1
0.75)] = 0.96 and a specificity
of (0.73 x 0.81) = 0.59, values far from the empirical values
of 0.89 and 0.70 (Table 2). Consequently, these tests cannot
be independent. This is reasonable, because they both
detect increased numbers
of leukocytes. However, the tests
are not completely dependent on each other, because they
use different principles to detect the leukocytes (10).
Because the esterase and nitrite tests are independent of
each other, the predictive values of the various combinations of test results can be calculated from the sensitivity
and specificity of either test. Table 5 lists the post-test
probability of infection, as evaluated after some test results, for various pre-test probabilities.
Microscopy is not independent of esterase, so that the
obtained post-test probability from esterase and nitrite
cannot be used directly as pre-test probability for micros-
-
-
Table 5. Post-Test Probability of infection, Based on Test Resuits for Esterase and Nitrite Test Strips,
Compared with Pre-Test Probability
Pre-testprobabIlity
0.01
0.05
0.10
0.26
0.002k
0.01
0.04
0.04
0.20
0.01
0.02
0.11
0.06
0.16
0.31
0.28
0.53
0.72
0.32
0.60
0.81
0.91
0.83
0.59
0.89
0.97
0.80
0.96
0.989
0.91
0.98
0.48
0.31
0.73
0.91
0.50
0.70
TestResult
Nitrite 0 and esterase0
Nitrite0 and esterase 1
Nitrite
0 and
Nitrite
1 and
Nitrite 1 and
Nitrite
1 and
esterase2
esterase0
esterase1
esterase2
0.06
0.18
0.18
0.56
5Post-testprobability.Boldnumbersindicatethe absence(upper left), or presence (lowerright) of infection.
1666 CLINICAL CHEMISTRY, Vol. 35, No. 8, 1989
0.995
copy. However, using the sensitivity and specificity of
microscopy in the subgroups of esterase-positive and
-negative samples (Table 3) allows calculation of new
predictive values (Table 6). The post-test probability obtained from esterase (with or without nitrite)
can then be
used as the pre-test probability for microscopy. Thus, Table
6 illustrates what additional information can be obtained
from sediment microscopy when the strip tests have been
performed.
Obviously,
the final post-test probability is independent
of the order in which the tests are performed. Microscopy
might be chosen as the first test, and the additional
information obtained from the subsequent use of test strips
could be calculated. However, microscopy is much more
time-consuming than the test strips, and it is reasonable to
start with the least expensive and time-consuming test.
Two common conditions under which urine is analyzed
are (a) screening and (b) evaluation of symptomatic patients. If we assume a threshold probability of >0.90 to
prescribe treatment for urinary
tract infection, and a
threshold probabilityof <0.10 to rule out an infection, the
following situations will appear.
(a) In screening symptomless, healthy persons, the prevalence of bacteriuria is generally <0.05 (32). If esterase 1
and nitrite = 0, then the post-test probability will be 0.06
(Table 5), and the patient can be considered not infected. If
esterase 2 and nitrite is 1, the post-test probability of
infection is 0.83,
which is close to the threshold for
deciding to treat the patient. For this pair of results, the
prevalence must be 0.09 to get the post-test probability
0.90, and a subsequent microscopy is not of much help:
even if >15 leukocytes per high-power fieldwere found, the
prevalence would still have to be 0.08 to get a final
post-test probability 0.90 (Tables 5 and 6). In this situation, no other combination of test-strip results leads to any
decision, and it is not possible to improve the decision
making by doing microscopy.
(b) In symptomatic patients the prevalence of bacteriuria
is 0.65 to 0.90 (17,33). Assuming a pre-test probability of
0.70, the finding of nitrite = 1 and (or) esterase 2 gives a
post-test probability of infection >0.90; i.e., the patient is
considered infected and will receive treatment. If, on the
other hand, nitrite = 0 and esterase 1, the the post-test
probability is 0.31 to 0.72 and cannot be increased to >0.90
or decreased to <0.10 by microscopy (Tables 5 and 6). Thus,
in these cases the use of microscopy does not add any
Table 6. Post-Test Probability of infection Based on
Various Results from Microscopy, with the Post-Test
Probability Obtained from Esterase Tests (with or
without Nitrite Tests) as the Pre-Test Probability
Post-test probablifty from esterase test
Leukocytesper
hIgh-power field
0.10
0.26
0.50
0.70
0.05a
0.24
0.44
0.14
0.49
0.71
0.32
0.74
0.88
0.52
0.87
0.94
0.03
0.10
0.23
0.42
0.11
0.12
0.27
0.30
0.52
0.55
0.72
0.74
Esterase test was neg.
6-15
Esterase test waspos.
6-15
Final post-testprobability.
information with respect to decision making.
Generally, Table 6 shows that, once the esterase test has
been done, little information can be added by microscopy,
especially when the esterase test is positive.
In conclusion: The ability of the esterase and nitrite test
strips to detect or exclude urinary tract infection is comparable with, or better than, that of sediment microscopy.
Once the strip tests have been done, little additional
information can be given by microscopy. Consequently, for
the detection or exclusion of urinary tract infection, microscopy can generally be replaced by the esterase and nitrite
test strips. Moreover, one should always estimate the
prevalence, or pre-test probability, of infection. if both
test-strip results are positive, the urinary tract is likely to
be infected, unless the pre-test probability is very low. If
both are negative, urinary
tract infection is unlikely,
unless the pro-test probability is very high. When esterase
and nitrite test results disagree, infection is likely only if
the pre-test probability is high, as in patients with acute
symptoms; otherwise the examination is inconclusive.In
such cases, urine cultures should be performed.
We thank Mrs. Anne Grete Eriksen and Mr. K#{226}re
S#{248}nstab#{248}
for
skillful
technical assistance.
References
1. Pfaller M, RingenbergB, Rames L, HegemanJ, Koontz F. The
usefulness of screening tests for pyuria in combination
with
culture in the diagnosis of urinary tract infection. Diagn Microbiol
Infect Dis 1987;6:207-15.
2. Sternheimer R, Malbin B. Clinical recognition of pyelonephritis, with a new stain for urinary sediments. Am J Med
1951;11:312-23.
3. Galen RS, Gambino SR. Beyondnormality: the predictivevalue
and efficiency of medical diagnoses. New York/LondoniSydney:
John Wiley & Sons,1975.
4. Griner PF, Mayewski RJ, Mushlin Al, Greenland P. Selection
and interpretation of diagnostic tests and procedures. Principles
and applications. Ann Intern Med 1981;94:553-600.
5. Schwartz JS. Understanding laboratorytest results.Conditions
for appropriatelaboratory tests. Med Clin North Am 1987;71:63952.
6. Bruhl P, Fuchs T, Kattermann R, et al. Diagnosis of leukocytes
in urine by means of a test strip. Dtsch Med Wochenschr
1979;104:1236-40.
7. Gillenwater JY. Detection of urinary leukocytesby ChematripL. J Urol 1981;125:383-4.
8. Kusumi RK, Grover PJ, Kunin CM. Rapid detectionof pyuria
by leukocyte esterase activity. J Am Med Assoc1981;245:1653-5.
9. Bolann B. Levkocyturi. Sanunenligning av screeningmetoder.
(Leukocyturia. A comparison of screening methods.) Tidsskr Nor
Laegeforen 1982;102:1270-1.
10. Colombo JP, PeheimE, Keller H, et al. Rapid demonstration of
urinary leukocytes with a new leukocyte test-strip. A multicenter
cooperativestudy. Dtsch Med Wochenschr 1982;107:853-7.
11. Wilkins EGL, Ratcliff JG, Roberta C. Leukocyte esterasenitrite screeningmethodfor pyuria and bacteriuria. J Clin Pathol
1985;38:1342-5.
12. Avent J, Schumann GB, Vars L. Comparison of Chemstrip
Leukocyte Test” with standardized Papanicolaou-stained
urine
sedimentevaluation.
Lab Med 1983;14:163-6.
13. Smalley DL, Bryan JA. Comparative evaluationof biochemical and microscopic urinalysis. Am J Med Technol 1983;49:237-9.
14. Christenson RH, Tucker JA, Allen E. Results of dipstick tests,
visual inspection,microscopic examination of urine sediment,and
microbiologicalcultures of urine compared for simpli1yingurinalysis. Clin Chem 1985;31:448-.50.
15. Sewell DL, Burt 5?, Gabbert NJ, Bumgardner V. Evaluation
of the Chemstrip 9’ as a screening test for urinalysis and urine
culture in men. Am J Clin Pathol 1985;83:740-3.
16. Shaw ST, Peon SY, Wong ET. “Routine urinalysis.” Is the
CLINICAL CHEMISTRY, Vol.35, No. 8, 1989 1667
dipstick enough?J Am Med Assoc 1985;253:1596-600.
17. Stamm WE, Counts GW, Running KR, Fthn S, Turck M,
Holmes KK. Diagnosis of coliform infection in acutely dysuric
women. N Engl J Med 1982;307:463-8.
18. Lipaky BA, Iret.on RC, Film SD, Hackett R, Berger RE.
Diagnosis of bacteriuria in men: specimen collection and culture
interpretation. J Infect Dis 1987;155:847-54.
19. Leo YT, Scottolim AG, Luangphinith 5, Adam AL, JacobsLD,
Mariani Al. Urine screening strategy employing dipstick analysis
and selective culture: an evaluation.
Am J Clin Pathol
1984;81:634-42.
20. WiseKA, Sagert LA, Grammens GL. Urine leukocyte esterase
and nitrite tests as an aid to predict urine culture results. Lab Med
1984;15:186-7.
21. HughesJG, Snyder RJ, Washington H JA. An evaluation of a
leukocyte esterase/nitrite teststrip and a bioluminescence assay
fordetection ofbacteriuria. Diagn Microbiol Infect Dis 1985;3:13942.
22. Pfaller MA, Koontz F?. Laboratory evaluation of leukocyte
esteraseand nitrite tests for the detection of bacteriuria. J Cliii
Microbiol 1985;21:840-2.
23. Jones C, MacPherson
DW, Stevens DL Inability of the Chemstrip LN compared with quantitative urine culture to predict
significant bacteriuria. J Clin Microbiol 1986;23:160-2.
24. Dobbs FF, Fleming DM. A simple scoringsystemfor evaluating symptoms, history and urine dipstick testing in the diagnosis
of urinary tract infection.J R Coll Gen Pract 1987;37:100-4.
1668
CLINICAL CHEMISTRY, Vol.35, No. 8,1989
25. Warkentin DL, Rawling RA, Koppel WE, Maher GL, Broder
PH, Marchand A. Three techniques comparedfor detecting bacteriuria in symptomaticpatients. Clin Chem 1985;31:1361-3.
26. Murray PR, Smith TB, McKinney TC. Clinical evaluation of
three urine screeningtests.J Clin Microbiol 198725:467-70.
27. Smalley DL, Dittmann AN. Use of leukocyte esterase-nitrate
activity as predictive assays of significant bacteriuria. J Clin
Microbiol 1983;18:1256-7.
28. Bartlett RC, O’Neill D, McLaughlin JC. Detection of bacteriuria by leukocyte esterase,nitrite, and the automicrobicsystem.
Am J Clin Pathol 1984;82:683-7.
29. Pezzlo MT, Wetkowski MA, Peterson EM, de la Maza LM.
Detection of bacteriuria and pyuria within two minutes. J Clin
Microbiol 1985;21:578-81.
30. Epstein SE. Implications of probability analysis on the strategy used for noninvasive detection of coronary artery disease
[Review]. Am J Cardiol 1980;46:491-9.
31. BOhneC. Fruherkennung von Harnweginfekten durchNitritnachweis mit Harn-Testreifen. Erhohung der Trefferquote durch
orale Nitratsubstitution. Med Klin 1969;64:887-90.
32. Sobel JD, Kaye D. Urinary tract infections. In: Mandell GL,
Douglas RG, Bennett JE, eds.Principlesand practice of infectious
diseases, 2nd ed. New York/ChichesterfBrisbane:John Wiley &
Sons, 1985:426-52.
33. Papapetropoulou M, Pappas A. The acute urethral syndrome
in routine practice. J Infect 1987;14:113-8.