Clinical Chemistry 49:4 617 623 (2003) Automation and Analytical Techniques Quantitative Urine Particle Analysis: Integrative Approach for the Optimal Combination of Automation with UF-100 and Microscopic Review with KOVA Cell Chamber Cornelia Ottiger and Andreas R. Huber * Background: Automated systems have enabled the counting of particles in urine to be standardized. Their superiority over traditional sediment analysis has been well documented, but they have not gained wide acceptance. The reasons for this are that sediment analysis has been performed and interpreted for decades. Additionally, pathologic casts and other unknown particles still must be confirmed under the microscope. Furthermore, comparison between the methods has revealed outliers and thus decreased confidence in automation. Methods: We used the standardized KOVA cell chamber system to count particles and compared the results with UF-100 flow cytometry as an alternative to traditional sediment analysis. Results: We compared 252 randomly selected urine samples and obtained a review rate of 33%. Microscopic verification was necessary because of the presence of casts, yeast, sperm, dysmorphic erythrocytes, and some misclassified erythrocytes or leukocytes that were detected by incongruent dipstick results and abnormal scattergrams. We obtained correlation coefficients of 0.966 for erythrocytes and 0.935 for leukocytes. Criteria for an algorithm to identify samples that needed microscopic review were derived from comparisons between the number of particles from UF-100, dipstick results, cell chamber counting, and sediment analysis. Conclusions: Automated cell counting combined with microscopic counting with a standardized cell chamber system is useful. An objective algorithm for review criteria can be developed via systematic comparison of UF-100 flow cytometry and microscopy. Only urine samples that meet these criteria need to be confirmed microscopically. 2003 American Association for Clinical Chemistry Microscopic sediment analysis is still used world-wide to examine cells and particles in the urine, but its limited precision (1) and the large amount of manual labor are drawbacks. Standardization of urinalysis has been recommended by the NCCLS and the European Urinalysis Guidelines (2 4) with use of an automated system or a standardized procedure to count the cells in a chamber within a predefined volume. Flow cytometry allows for more precise counting of particles in the urine than does traditional sediment analysis (5 7). Previous studies that compared the Sysmex UF-100 (TOA Medical Electronics) with microscopy and dipstick results found a good correlation between the methods (8). On the other hand, some outliers were seen in these studies but were never systematically examined (9). Additionally, the identification of pathologic casts and other unknown particles still requires confirmation by microscopy, which is usually performed by traditional sediment analysis, a semiquantitative method. The aim of this study was to determine the microscopic review rate from randomly selected urine samples in a general tertiary care hospital after a simultaneous dipstick measurement with a dipstick reader and automated counting of particles by the UF-100 flow cytometer. Our aim was to establish an algorithm for positive selection of those urine samples that need microscopic confirmation. The benefits of such an algorithm would be a considerable reduction in manual labor and optimization of the workflow without any loss of sensitivity and specificity. The Department of Laboratory Medicine, Kantonsspital Aarau, CH-5001 Aarau, Switzerland. *Author for correspondence. E-mail andreas.huber@ksa.ch. Received September 10, 2002; accepted January 9, 2003. 1 Nonstandard abbreviations: RBC, red blood cell; WBC; white blood cell; EC, squamous epithelial cell; SRC, small round cell; and HPF, high power field. 617
618 Ottiger and Huber: Quantitative Urinalysis with UF-100 plus KOVA Chamber UF-100 flow cytometer was compared with the disposable KOVA cell chamber system (Hycor) (10, 11) and traditional urine sediment analysis. From our results, we propose a standard approach for counting particles in urine that uses the flow cytometer and/or the KOVA cell chamber with the advantage that laboratories of all sizes could integrate one or both methods. Materials and Methods uf-100 flow cytometer We randomly selected 268 urine samples from different departments of our hospital (e.g., medicine, surgery, gynecology, nephrology, and intensive care) and analyzed them microscopically (see below) and with the UF-100 (software version 12; Digitana), as described by the manufacturer. Samples were analyzed within 1 h of arrival. We analyzed 10 samples three to five times with the UF-100 for precision measurements. The number of red blood cells (RBCs), white blood cells (WBCs), squamous epithelial cells (ECs), and other particles were counted in 252 samples. Dipstick results were obtained for all of the urine samples. Blood and leukocyte esterase from the dipsticks and particle counting with the KOVA cell chamber and traditional sediment analysis (see below) were compared with results for RBCs and WBCs from the UF-100 and assessed for concordant/discordant results. Discordant results were defined when the number of RBCs above the cutoff, as measured by the UF-100, was at least two times higher than the number counted in the KOVA cell chamber, when the RBC scattergram was abnormal, or when the dipstick results for blood or leukocyte esterase did not match with the number of RBCs and WBCs, respectively. An incongruent result was defined as follows: (a) the semiquantitative dipstick result was negative/trace and the cell number in the KOVA chamber was 10, but the cell count with UF-100 was 21 and up to 1200 10 6 /L; (b) samples with a positive dipstick result whose cell count was more than twice as high on the KOVA system as on the UF-100, or vice versa, and had, in addition, an abnormal scattergram and/or a flag. Particles such as casts, small round cells (SRCs), yeast, sperm, crystals, and dysmorphic RBC were taken into account because a large number of these particles can influence the classification of cells in the scattergram (5, 12, 13). The reason for each discordant result was determined (e.g., influence of yeast, sperm, or crystals on RBC). The lowest number of each influencing particle that produced discordance was derived for the algorithm accordingly (see Results). microscopy with kova cell chamber and sediment analysis The KOVA system was selected as an alternative microscopic method because it provides a standardized quantitative procedure like the UF-100. The KOVA system includes a conical graduated test tube, in which the urine is centrifuged; the KOVA Petter, a disposable plastic transfer pipette constructed with a bulb-like base, which allows easy and precise decanting of the supernatant; and 10 chambers with grids on one disposable KOVA slide (10). From the urine samples mentioned above, we counted six urine specimen five times in the KOVA cell chamber to assess precision. The same 252 samples assayed by UF-100 were analyzed in parallel with the KOVA system (Axon Lab) and the traditional urine sediment analysis. After the urine was analyzed with the UF-100, the remainder was centrifuged at 400g for 5 min (2). The total urine volume in the conical 8-mL Vacutainer urine tubes (BD Allschwil), which replaced for practical reasons (pneumatic transport system) the original KOVA tubes, was measured against a reference tube. The supernatant was removed with the KOVA Petter after centrifugation, and the volume of the remaining urine in the tube was measured with an Eppendorf pipette, giving a mean (SD) of 0.650 (0.05) ml. The pellet was mixed well with the KOVA Petter, and one drop was placed in the KOVA cell chamber. Irrespective of the UF-100 results, microscopic evaluation took place and was always performed by the same person. The microscopic counting with phase contrast for the detection of dysmorphic RBCs was done after cells were allowed to settle for 5 min. For counting in the KOVA chamber, the middle section in the grid was chosen, and RBCs, WBCs, and ECs were counted at a 400 magnification in 10 small squares. Other particles were evaluated qualitatively in the low power field at 100 magnification. The number of cells was calculated according to the equation below: n Vol Cells 10 6 Centr /L Vol Chamber Vol Tube where n is the number of cells counted in 10 small squares, Vol Centr is the volume of the pellet after centrifugation, Vol Tube is the total volume of urine in the test tube, and Vol Chamber is the volume of 10 small squares (0.1111 10 6 L). Another drop was placed on a glass slide and covered with a 20 20 mm coverslip for traditional urine sediment analysis. For the urine sediment analysis, several high power fields (HPFs) were estimated, and the counts were given as an average per HPF. These values were normalized to a given volume: Cells 10 6 /L n Vol Centr Vol Slide HPF Slide Vol Tube where n is the average number of cells per HPF, and HPF Slide is the ratio of the area of the slide and the area of one HPF, i.e., (20 mm) 2 / (0.175 mm) 2. Vol Slide is the volume under the coverslip (20 10 6 L) (2).
Clinical Chemistry 49, No. 4, 2003 619 dipstick To compare chemical dipstick results (Multistix 10 SG; Bayer) for RBCs and WBCs with those obtained with the UF-100 flow cytometer, all urine samples were analyzed with an automatic dipstick reader (Clinitec Atlas; Bayer). statistics Regression analysis to establish differences between the methods was performed with Deming systematic difference calculations (14). Results that were flagged for review by the UF-100 (Fig. 4), incongruent dipstick results, abnormal scattergrams, urine samples with a volume 5 ml, and samples that had to be diluted because of very high cell numbers were excluded from the regression analysis. Results comparison of uf-100 with kova system We analyzed 252 urine samples and compared the results obtained with the UF-100 and the KOVA system. We included 212 results for RBCs and 241 results for WBCs in the general Deming regression analysis, a weighted leastsquares regression model (14). The remaining samples were excluded from the regression analysis because of flags from the UF-100, incongruent dipstick results, or abnormal scattergrams (see below). Fig. 1 shows the correlation of RBCs and WBCs between the two methods. The Deming regression for RBCs was calculated in the range of 0 to 560 10 6 cells/l between the KOVA and UF-100 (r 0.966). The regression for WBCs was calculated in the range of 0 to 283 10 6 cells/l between the two methods (r 0.935). Additionally, ECs were compared between UF-100 and the KOVA system (r 0.902; Table 1). RBC and WBC results were also compared with the traditional sediment estimation. From the average number of cells/hpf, values with cells 10 6 /L were calculated. The correlation was 0.951 for RBCs and 0.887 for WBCs, respectively (Table 1). precision of uf-100 and kova system Internal quality control of the UF-100 was done with a mixture of latex particles. Interassay imprecision (CV) for 43 measurements was 4.1% at a mean of 195 10 6 RBC/L, 2.8% at 195 10 6 WBC/L, and 7.2% at 213 10 6 BACT (bacteria 2 m)/l. To compare the two methods, precision measurements for RBCs and WBCs were performed with low, intermediate, and high cell numbers with the UF-100 flow cytometer and the KOVA cell chamber system (Fig. 2). SDs were slightly higher for the KOVA system than for the UF-100. As expected, the CV decreased with increasing numbers of cells and asymptotically approached 10% for UF-100 and 15% for the KOVA system. Because the UF-100 (software version 12) enumerates two different bacterial particles, BACT (bacteria 2 m) and H-BACT (bacteria 2 m), the SDs were 1 11 Fig. 1. Deming regression analysis (14) for RBCs (A) and WBCs (B) between the KOVA system and the UF-100. We included 212 results for RBCs and 241 results for WBCs in the regression from a total of 252 urine samples. Regression statistics were: for RBCs, y 1.16x 4.3 10 6 /L (r 0.966); for WBCs, y 1.50x 4.4 10 6 /L (r 0.935). 10 6 /L for BACT in the range of 0.03 3 10 9 /L and 61 641 10 6 /L for H-BACT in the range of 1.2 34 10 9 /L, i.e., the CV was 2 5%. review rate In previous studies, outliers for RBCs and WBCs between the UF-100 and cell chamber counting were noted but not Table 1. Deming regression analysis for RBCs, WBCs, and ECs in urine. KOVA vs UF-100 a Sediment vs UF-100 a RBCs (n 212 samples) b Slope 1.16 (0.02) 0.88 (0.02) Intercept, 10 6 /L 4.3 (1.26) 1.25 (1.58) Correlation coefficient 0.966 0.951 WBCs (n 241 urine samples) b Slope 1.50 (0.04) 1.15 (0.04) Intercept, 10 6 /L 4.4 (1.39) 1.75 (1.70) Correlation coefficient 0.935 0.887 ECs (n 46 urine samples) b Slope 1.26 (0.09) Intercept, 10 6 /L 1.9 (3.22) Correlation coefficient 0.902 a SD of slope and intercept in parentheses. b Of a total of 252 samples.
620 Ottiger and Huber: Quantitative Urinalysis with UF-100 plus KOVA Chamber Fig. 2. Precision of RBC (A) and WBC (B) counts with the UF-100 ( ) and the KOVA system ( ). In the UF-100, each sample was measured three to five times. In the KOVA chamber, for each data point, five aliquots from one centrifuged tube were added to the chamber and counted. specifically examined (8, 9). Our work therefore was initiated to explore the causes of the differences and to obtain an estimate of the fraction of samples requiring microscopic confirmation. Results from 168 (66.7%) of 252 samples were correctly interpreted by UF-100 and would not have needed further evaluation (Fig. 3). Of the remaining 84 samples (33.3%), 23 (9.1%) had to be checked microscopically for casts, SRCs, and high numbers of crystals, yeast, and sperm Fig. 3. Classification of results for 252 urines after analysis with UF-100. Review rate from the interpretation of scattergrams together with results from microscopy and dipstick. cells because classification of RBCs may be influenced by these particles. However, we found that in these cases interference with RBC and WBC counts did not occur and that 6 of 23 samples that contained yeast, sperm, or crystals above the review limits (Fig. 4) were selected unnecessarily. In these six samples, the particles did not overlap with the RBC gate in the scattergram. The presence of pathologic casts or other particles was not confirmed in one-half of the cases because high numbers of hyaline casts, ECs, and cell debris also can cause such flags. A total of 30 samples (11.9%) with RBC concentrations 20 10 6 /L that were flagged for dysmorphic erythrocytes needed microscopic review because we observed that for seven cases in this study, such flags were the only indication of the incorrect counting of RBCs. In 31 samples (12.3%), a review flag from the UF-100, an abnormal distribution of the RBC cloud in the scattergram, or an incongruent dipstick result (see Materials and Methods) for blood or leukocyte esterase made it necessary to recount the cells with the microscope. In all of these cases, the RBC count obtained with the UF-100 was incorrect (Fig. 5), and in five cases, both the RBC and WBC counts were incorrect. RBCs were misclassified when a high amount of bacteria, crystals, sperm, or yeast overlapped with the RBC gate in the scattergram; these particles therefore contributed to erroneously high RBC counts. The presence of naked EC nuclei led to an incorrectly high WBC count. algorithm for microscopic review Ultimately, this study allowed us to determine whether we could find an objective algorithm for screening urine samples that need further confirmation by microscope after combined analysis with the chemical dipstick and counting of particles with the flow cytometer. To achieve this, we adapted the review conditions of the UF-100, originally proposed by the manufacturer. In the proposed algorithm (Fig. 4), we included such items as the cross-check between negative dipstick results and high cell count with the UF-100 (Fig. 5), abnormal scattergrams, undefined or a very high number of particles, RBC information, pathologic casts, SRCs, crystals, yeast, and sperm, whereas conductivity ensured authentic samples in cases of drug abuse. All samples were evaluated for abnormal distribution of the scattergrams and were checked for incongruent results between the dipstick, UF-100, and KOVA system (see Materials and Methods). In those cases in which a particular particle influenced the result, the number of such particles was noted as the limiting factor. Among all such detected samples, the lowest particle number that was found to give no further influence on the scattergram was then used as the review limit in the algorithm. When we examined the data with the proposed algorithm, we found that 6 of 252 samples (2.4%) were selected unnecessarily. However, more importantly, all
Clinical Chemistry 49, No. 4, 2003 621 Fig. 4. Review criteria for UF-100 results to be checked by microscopy. Left, settings that can be adjusted on the instrument. Path. Cast, pathologic casts; X tal, crystals; YLC, yeast. Right, variables that have to be considered. Hb/Prot/Lc/Nitrite, hemoglobin, protein, leukocyte esterase, and nitrite from the dipstick. Total particle count (to dilute if 40 000) and conductivity (check composition of specimen) need review if there are persistent flags. samples that needed confirmation were selected by the algorithm. Two circumstances were critical for determining inappropriate classification of RBCs and WBCs. The first circumstance, incongruent results between the dipstick and flow cytometer, especially the absence or a low concentration of blood or hemoglobin on the dipstick and a high number of RBCs in the scattergram (Fig. 5), gave evidence for false classification of these particles. The second, samples with dysmorphic RBCs 20 10 6 /L, needed to be checked by microscope to avoid missing important renal diseases. In these cases, not only did the number and appearance of the RBCs need to be verified, but RBCs were sometimes wrongly classified and other particles, such as pathologic casts, that had not been flagged by the instrument were detected. Discussion We found an excellent correlation between the KOVA cell chamber and UF-100 for RBCs (r 0.966), WBCs (r 0.935), and ECs (r 0.902). Interestingly, under the condition of a standardized preparation of the traditional sediment analysis within a definite volume, the correlation was still 0.951 for RBCs and 0.887 for WBCs. Sediment analysis has to be regarded as a semiquantitative method. We therefore propose that standard counting of particles in urine with the UF-100 flow cytometer in concordance with the KOVA cell chamber system be introduced to fulfill the recommendations of international guidelines (2, 3). Minor deviations between the two systems have to be considered, e.g., the slopes and intercepts in the regression analysis showed differences from 1 and 0, respectively (Table 1). The intercepts of 4 to5 10 6 cells/l are caused by overestimation by the flow cytometer as a result of an overlapping area in the RBC gate in the scattergram. A conversion factor should be introduced if both methods are used in concert and if the same reference values as proposed (8) are accepted. It must be emphasized that the number of cells found in the sample can be affected not only by the method of counting, but also by the test tube (12, 15), to which cells, casts, and
622 Ottiger and Huber: Quantitative Urinalysis with UF-100 plus KOVA Chamber Fig. 5. Abnormal scattergrams of discordant RBC counts between the KOVA system, UF-100, and dipstick results. Negative ( ), trace (f), (E), 2 (F), and 3 ( ) results were evaluated. In all of these samples, high amounts of bacteria, crystals, yeast, and sperm disturbed the RBC classification. other particles can adsorb, and by transportation and centrifugation, during which mechanical forces and temperature can destroy the particles (1). Therefore, the systematically lower values obtained with the KOVA chamber might explain the differences between our findings and those of Okada et al. (16), who compared the UF-50 and the KOVA chamber filled with uncentrifuged urine. However, because of the small number of cells in the chamber, their precision measurements above the cutoff gave CVs up to 57%. The intraassay CV for patient samples measured by UF-100 was 10% for both RBCs and WBCs above the cutoff of 20 10 6 cells/l. With the KOVA chamber, we obtained slightly higher CVs, up to 15%. This is because we counted a much smaller number of cells microscopically than were counted by the UF-100. Bacteria were counted by UF-100 with high precision (CVs 5%). Flow cytometry for bacterial counting has been recommended for use in the screening of urinary tract infections because its sensitivity is 55% and its specificity is 90% (13). To evaluate a unifying system, we compared the flow cytometer, the dipstick method, and the KOVA system in 252 urine samples. The flow chart in Fig. 4 shows the criteria for a microscopic review. A total review rate of 33% (Fig. 3) was obtained, with 9% of the reviews attributable to flags for casts, SRCs, crystals, yeast, and sperm. Clumped ECs often gave a flag for pathologic casts and SRCs; they therefore need to be confirmed. Large numbers of crystals, yeast, and sperm cells are known to overlap with the RBC gate in the scattergram and may lead to misclassification. This is the main reason to check the number of RBCs by microscope. However, RBCs and WBCs from this group were correctly interpreted by the UF-100 and were not falsely classified. We found dysmorphic RBCs 20 10 6 cells/l in 12% of all urine samples. Considering this flag, 2.8% of all urine samples were misclassified for RBC count. Additionally, the presence of dysmorphic erythrocytes could not be confirmed in all cases by microscopy because there is an overlapping area between normal and dysmorphic RBCs in the RBC Fsc graph of the UF-100. The testing was done with UF-100 software version 12. Although the version has been upgraded to 18 and the differentiation of dysmorphic RBC has been improved, we have not found major changes. In 12% of all samples, RBCs, and in some cases both RBCs and WBCs, were incorrectly classified by the UF-100 (Fig. 5) and thus would have given rise to false results if no attention has been paid to the review criteria (Fig. 4). The reason for the incorrect classification of RBCs was mainly the presence of crystals, yeast, sperm, and a large number of bacteria, as already pointed out by others (12, 13). The selecting gate for RBCs in the scattergram overlaps partly with these particles. One reason for WBC misclassification was the presence of naked EC nuclei. Urine samples from nephrology departments will show higher review rates from flags for casts and dysmorphic RBCs. This could explain the differences in published review rates, 6 35% (7). Expert systems seemed to improve the reliability of results, thereby decreasing microscopic reviews (7, 17). However, it seems that solving the technical details of the urine analysis before using such a system may not have been taken into consideration in these studies. Application of the review criteria (Fig. 4) enabled complete detection of all samples that had to be confirmed by microscopy. Unnecessary confirmation was done in only 2.4% of all cases. It can be concluded that only samples that fall within these parameters have to be checked microscopically. There is thus no objective reason to analyze urine samples systematically by microscopy just because they contain particles above the reference values. Future development of urinalysis systems should aim to attain accurate and precise measurement of the total number of RBCs, WBCs, and bacteria, taking into account parameters for the urine concentration, such as creatinine or conductivity and differential protein analysis (18, 19). With these additional criteria, a more accurate diagnosis and therapy can be achieved. In conclusion, we believe that standardization of cell counting in urine is possible with the UF-100 and KOVA systems and that both methods can be used in lieu of each other under certain conditions. Our systematically performed comparison between UF-100 and microscopy led to an algorithm for the selection of samples for microscopic review. Only those urine specimens specifically identified need microscopic review. We would like to thank Digitana (Horgen, Switzerland) for their granted support of the flow cytometer. We are grateful for the material provided by Axon Lab (Dättwil, Switzerland) and BD (Allschwil, Switzerland).
Clinical Chemistry 49, No. 4, 2003 623 References 1. Winkel P, Statland BE, Jorgensen K. Urine microscopy, an illdefined method, examined by a multifactorial technique. Clin Chem 1974;20:436 9. 2. National Committee for Clinical Laboratory Standards. Urinalysis and collection, transportation and preservation of urine specimens; approved guideline. NCCLS document GP 16-A. Wayne, PA: NCCLS, 1995. 3. European urinalysis guidelines. Summary. Scand J Clin Lab Invest 2000;60(Suppl 231):1 96. 4. Kouri TT, Gant VA, Fogazzi GB, Hofmann W, Hallander HO, Guder WG. Towards European urinalysis guidelines. Introduction of a project under European Confederation of Laboratory Medicine. Clin Chim Acta 2000;297:305 11. 5. Fenili D, Pirovano B. The automation of sediment urinalysis using a new urine flow cytometer (UF-100). Clin Chem Lab Med 1998; 36:909 17. 6. Hannemann-Pohl K, Kampf SC. Automation of urine sediment examination: a comparison of the Sysmex UF-100 automated flow cytometer with routine manual diagnosis (microscopy, test strips, and bacterial culture). Clin Chem Lab Med 1999;37:753 64. 7. Delanghe JR, Kouri TT, Huber AR, Hannemann-Pohl K, Guder WG, Lun A, et al. The role of automated urine particle flow cytometry in clinical practice. Clin Chim Acta 2000;301:1 18. 8. Regeniter A, Haenni V, Risch L, Köchli HP, Colombo JP, Frei R, et al. Urine analysis performed by flow cytometry: reference range determination and comparison to morphological findings, dipstick chemistry and bacterial culture results a multicenter study. Clin Nephrol 2001;55:384 92. 9. Ben-Ezra J, Bork L, McPherson RA. Evaluation of the Sysmex UF-100 automated urinalysis analyzer. Clin Chem 1998;44:92 5. 10. Mahon CR, Smith LA. Standardization of the urine microscopic examination. Clin Lab Sci 1990;3:328 32. 11. McGinley M, Lee Wong L, McBride JH, Rodgerson DO. Comparison of various methods for the enumeration of blood cells in urine. J Clin Lab Anal 1992;6:359 61. 12. Langlois MR, Delanghe JR, Steyaert SR, Everaert KC, De Buyzere ML. Automated flow cytometry compared with an automated dipstick reader for urinalysis. Clin Chem 1999;45:118 22. 13. Kouri TT, Kahkonen U, Malminiemi K, Vuento R, Rowan RM. Evaluation of Sysmex UF-100 urine flow cytometer vs chamber counting of supravitally stained specimens and conventional bacterial cultures. Am J Clin Pathol 1999;112:25 35. 14. Martin RF. General Deming regression for estimating systematic bias and its confidence interval in method-comparison studies. Clin Chem 2000;46:100 4. [Data supplement. GDR: executable program for general Deming regression calculations and graphics. http://www.clinchem.org/cgi/content/full/46/1/100/dc1]. 15. Kim Y, Jin DC, Lee EJ, Lee DH, Chung HH, Kim M, et al. Quantitative analysis of urine sediment using newly designed centrifuge tubes. Ann Clin Lab Sci 2002;32:55 60. 16. Okada H, Sakai Y, Kawabata G, Fujisawa M, Arakawa S, Hamaguchi Y, et al. Automated urinalysis. Evaluation of the Sysmex UF-50. Am J Clin Pathol 2001;115:605 10. 17. Roggeman S, Zaman Z. Safely reducing manual urine microscopy analyses by combining urine flow cytometer and strip results. Am J Clin Pathol 2001;116:872 8. 18. Ivandic M, Hofmann W, Guder WG. The use of knowledge-based systems to improve medical knowledge about urine analysis. Clin Chim Acta 2000;297:251 60. 19. Guder WG. Nephrological basis diagnosis in clinical practice. Ther Umsch 1995;52:328 33.