Urine Microscopy, an Ill-Defined Method ... - Clinical Chemistry

2 downloads 0 Views 731KB Size Report
kocytes. Their kind of quantity is number concentra- .... In the first period, each microscope was used by ... alyze each of the four factors independently because.
CLIN. CHEM. 20/4, 436-439

(1974)

Urine Microscopy, an Ill-Defined Method, Examined by a MultifactorialTechnique P. Winkel, B. E. Statland,1 and K. J#{248}rgensen

The precision of routine urine microscopy for counting erythrocytes and leukocytes was evaluated by using a statistical research-planning technique, factorial experimental design. We examined the relative contribution

of

the

technician

preparing

the

urine

specimen, the technician reading the urine slide, the time elapsed since receipt of the urine specimen, and the effect of the microscope used. Results obtained by various technicians differed significantly, because of variation in preparation techniques. This study

demonstrates

that

the

routine

examination

of

for cellular components may be more imprecise than generally believed, and that the technique should be better defined. It is emphasized that the technique of factorial experiments can be used as an urine

important tool control program. Additional periment

for

the

design

of

a rational

statistics e multifactorial control #{149} variation, source of

Keyphrases: #{149} quality

quality-

ex-

Some routine laboratory analyses are performed and reported inadequately as compared to the present state of the art. Microscopic examination of urinary sediment is an illustrative example. In this case a group of quantities are measured. They refer to the same system-i.e., some specified urine-but to separate components-e.g., casts, erythrocytes, and leukocytes. Their kind of quantity is number concentration.2 However, arbitrary kinds of quantity are used instead. Some results are given qualitatively by 0 or +, rather than by stating whether the number concentration is below or above a stated limit. Others are reported “semiquantitatively” by a number of plus signs or by a number or numbers indicating how many cells were seen within a viewfield. Although this practice may serve locally, it is not generally understood, and does not always yield sufficient information. The unnecessary use of arbitrary kinds of quantity may be considered less binding, and suppress the usual application of quality control and related conDepartment of Clinical Chemistry A, Rigshospitalet, Blegdamsvej 9, DK-2100, Copenhagen, Denmark. Present address: Department of Laboratory Medicine, University of Minnesota Hospitals, Minneapolis, Minn. 55405. 2The more logical name “number concentration” is now preferred by the IUPAC/IFCC Commission on Quantities and Units in Clinical Chemistry to “particle concentration” (1) for the kind of quantity: number of component particles divided by volume of system. Received Dec. 13, 1973; accepted Jan. 23, 1974. 436

CLINICAL

CHEMISTRY,

Vol. 20, No. 4, 1974

cepts such as accuracy and precision. Consequently, when changes in method are introduced (new equipment, work simplification, etc.), the meaning of “arbitrary” may change in an unrecognized manner. In our central laboratory, which we consider typical in this respect, “urine microscopy” needed reconsideration for both analytical and clinical reasons. In this paper, we draw attention to the general usefulness of the factorial experimental design in examining laboratory methods. Further, we point to the problem of centralizing methods for which the results rather depend on the analyst.

Methods Analytical

Procedure

Our routine procedure for urine microscopy is: About 10 ml of a urine specimen, in a plastic tube, is centrifuged for 5 mm. at about 1000 rpm in an ordinary laboratory centrifuge. The supernate is discarded, and the sediment is dispersed into the remaining few drops of fluid by shaking. With a glass capillary, a small drop of the mixture is transferred to a slide, and a cover glass (24 x 32 mm) is placed onto the drop. A microscope with a 40x objective lens is used. This technique is taught to the analysts by demonstration.

Experimental

Design

Three experiments were performed. Seven technicians, experienced in urine microscopy, participated in each experiment. Seven pathological urines were examined, with seven microscopes of the same model. Each experiment was divided into seven periods, of about the same time. During each of the periods one technician would examine (i.e., centrifuge, prepare, and read) one urine under one microscope. For each technician-urine-microscope-period combination, two readings were obtained: the average number of erythrocytes and the average number of leukocytes in the 10 viewfields examined. The three experiments differed in respect to the details of the procedure involved in performing the urine microscopies. The total time for one experiment was about 90mm. The purpose of experiment 1 was to study the influence of using different technicians and using different microscopes. As all the necessary readings could not be performed simultaneously, the effect of

the experimental time elapsed also had to be taken into account. Seven different pathological urines were used, to obtain more general conclusions, and therefore it was necessary to take “urine” into account. Thus, four factors were studied: “technician,” “microscope,” “time,” and “urine.” In the first period, each microscope was used by one technician examining one of the urines. In the second period the technicians changed urines and microscopes in such a way that no urine was read at the same microscope or by the same technician as in period one. In the subsequent periods the technicians and urines continued to rotate among the microscopes in such a way that at the end of period seven each technician had examined each urine once and used each microscope once, and so that each urine had been examined once under each of the seven microscopes. The above design enabled the experimenter to analyze each of the four factors independently because the contribution of the other factors is balanced out. Consider for example the technician factor. This is evaluated from the variation among seven mean values, one for each individual technician, i.e., the average of her seven readings. As each mean value includes a reading from each of the seven urines, a reading from each of the seven microscopes, and a reading from each of the seven periods, these three factors are balanced out. Thus any difference among the mean values is ascribed to differences among the technicians in the way they prepare and (or) read the slide. The general experimental plan appears in Table 1. The choice of factors in experiments 2 and 3 was made on the basis of the results of the previous experiment(s). In experiment 2 the procedure was separated into two stages such that the preparation of the slide for urine microscopy and the actual reading of the slide could be studied separately as factors. In the first period a technician prepared and read the same urine, but in the following periods she made a preparation from another urine and read the slide from a third urine prepared by another technician. A given urine was always examined with the same microscope. The experimental design was similar to that in the previous experiment, the four factors being “preparation of slide,” “reading of slide,” “time,” and “microscope.” In experiment 3, the stages of the slide preparation procedure were studied separately: i.e., the centrifuging of the urine and the discarding of the supernate (decanting), the mixing of the precipitate (mixing), and the transferring of a droplet with a cover slip (transfer). In the first period one technician decanted all seven urines, the next period a second technician decanted all the urines, and so forth; in this way the influence of time and decanting were combined. In each period all technicians each performed the mixing procedure for one urine, and the transferring procedure and reading for another urine,

Table 1. Plan for Experiment 1 Based on 7 Greco-Latin Square Design (2)”

x

7

Microscopes

Pe1

2

3

4

5

6

7

1

U1

U2

U3

U4

U5

U6

U7

2

U2

3

U3

riods

T1

T4

4

U4

5

U5

6

U6

7

T2

17

T5 U4 T4

T6

T7

T2

T1

T3 U1

T4 U2

T6

T7

T4

T6

T5 U5

T7 U5

T3

T2 U4

U4

U4 T

T1

T3

15

15 U3

U3

U3

U3

T5

T7

T2

1 U2

U2

U2

U2

T2

T4

T

17 U1

U1

U1

U1

U1

1

13

T5

U7

U7

U7

U7

U7

T

T2

T6

T5

U6

U6

U8

U5

T5

T6

T1

T3

T4

U5

U5

U5

U7

13

U4

U3

T U6

T

T4

“The seven rows symbolize the seven experimental periods; the seven columns refer to the seven microscopes. U, (i = 1,7) symbolize the seven urines used in the experiment and T, (j = 1,7) the seven technicians. The following example illustrates how the diagram Is read. The square of the third row and second column contains the symbols U4 and T1.This means that in period No. 3 technician No. 1 shall read urine No. 4, using microscope No. 2.

except in the first period, where she did both procedures on the same urine. As in experiment 2, a urine was always examined with the same microscope. Thus the four factors studied were “decanting + time,” “mixing,” “transferring + reading of slide,” and “urine + microscope.”

Statistical

Methods

For the analysis of each of the three experiments we used a four-way analysis of variance model with zero interaction terms (2). Four variates were studied: the log of the erythrocyte count, the log of the leukocyte count, the log of the sum of the erythrocyte and leukocyte counts, and the log of the ratio of the erythrocyte count and the leukocyte count. A factor includes seven levels. For instance, in experiment 1 the levels of the “technician factor” are the seven technicians. (A through G in Table 2). For each level of each of the four factors, the mean values for the logarithm of the erythrocyte count and the logarithm of the leukocyte count are computed. For the antilogarithm of these mean values we shall denote the “erythrocyte mean” and the “leukocyte mean,” respectively.

Results The results of experiment 1 are summarized in Table 2. For each technician, each urine, each microscope, and each period the “erythrocyte mean” and “leukocyte mean” are given. For each of the four factors, four F-test values are also presented. It is seen that the erythrocyte counts, the leukocyte counts, and the ratios between these counts vary significantly among the technicians. The intended specCLINICAL CHEMISTRY, Vol. 20. No. 4, 1974

437

Table 2. “Erythrocyte Means” (E) and “Leukocyte Means” (L) of Each of the Four Factors of Experiment 1” Tech-

Technician factor

Microscope Urine factor

nician

L

E

Urine

E

U’ U2 U3

116.8 11.5 6.4 31.2 8.8 2.5 8.7 = 71.57

12.2 15.2 16.8 10.0 9.3 20.5 8.8

A B C D

E F

G FE

9,5 17.6 10.7 9.6 5.5 12.1 10.1 4#{149}g45 FL = 2.50

=

FE+L = 4’ _A

U4

U5 U6 U7 FE

‘EIL

FL,

count

among

and

FE+L,

symbolize

FElL

counts, and the ratio 5P