Semi-Automatedvs. Visual Readingof UrinalysisDipsticks

Transcription

1 CLIN. CHEM. 2/2, () Semi-Automatedvs. Readingof UrinalysisDipsticks James D. Peeie, Jr., Richard H. Gadsden, and Rachel Crews Semi-automated reflectance readings of urinalysis dipsticks showed better precision than visual readings in the case of artificially prepared urine samples containing glucose, ketone, and protein. Actual pathological specimens containing glucose, protein, and nitrite also showed that instrumental readings were better than visual. Results of repeated visual readings spread over three different color blocks for certain concentrations of glucose, ketone, and protein, whereas repeated instrumental readings were never spread greater than two color blocks. Subjectivity can be reduced by semi-automated dipstick urinalysis. Additional Keyphrases: variation, source of. screening - analytical systems The most widely used method of routine urinalysis is visual reading of dipsticks that detect the presence of and semi-quantify a variety of substances. The preceding report described the reproducibility of semiautomated dipstick readings by reflectance, tested on multiple samples at the same concentration (). In this report we compare visual vs. reflectance reading of N-Multistix with the Ames Clini-Tek. For the comparison we used contrived samples covering a range of concentrations for glucose, protein, and ketone. In addition, actual pathological urine samples were used to compare visual vs. reflectance readings for glucose, protein, and nitrite. These pathological samples were collected from patients with diabetes, nephrotic Syndrome, and urinary tract infections, respectively. Materials and Methods Aqueous glucose (0 g/liter), acetoacetic acid (20 g/liter), N-Multistix, and the Clini-Tek were provided by the Ames Co., Division of Miles Laboratories, Elkhart, Ind. 5. Quality-control serum ( Ledernorm U ; Lederle Diagnostics, Pearl River, N. Y. 5) was used to prepare urine with protein. Contrived samples were prepared from urine collected from laboratory personnel and qualified as normal as previously described (). Department of Laboratory Medicine, Medical University of South Carolina, Ashley Ave., Charleston, S. C. 20. Clini-tek and N-Multistix are registered trademarks of Ames Co., Division of Miles Laboratories, Inc., Elkhart, md. 5. Received May 0, ; accepted Sept. 2,. Contrived urine containing glucose was prepared as described before () to give the concentrations listed in Table. Contrived samples containing ketone were prepared in the same manner as glucose. Concentrations of ketone prepared are listed in Table 5. We made no correction for endogeneous glucose or ketone. Aliquots of commercial quality-control serum with a total protein value of g/liter was added to Urine- Tek tubes and diluted to 2 ml with normal urine. The resulting concentration of protein is given in Table and is corrected for endogeneous protein. Serum rather than albumin was chosen for supplementing samples with protein because high concentrations of protein in pathological urine include several proteins rather than albumin alone. N-Multistix, however, responds only to albumin (2). Contrived samples were read times each by Clini-Tek, and either or 20 independent visual readings were obtained for each sample. readings were taken by technicians familiar with dipstick techniques, and all readings were taken in the same area of the laboratory bench, with a white background to minimize interpretation based on differences in background color. Proper timing was strictly adhered to, as suggested by the manufacturer. Three or four different readers were used for each concentration and samples were read as blind unknowns. Pathological urine, containing glucose, was collected from diabetic patients. We quantitated glucose by the glucose oxidase method of Trinder (), using the Programachem 0 (American Monitor Corp., Indianapolis, md. 28) and Glucose Autotest reagents (Boehringer Mannheim Corp., Indianapolis, Ind. 28). Pathological urine containing protein was collected from patients with nephrotic syndrome. The concentration of protein was determined by the method of Doetsch and Gadsden (). Pathological samples that read +2 or + for glucose or protein by N-Multistix were diluted with normal urine to achieve lower readings. Urine submitted for bacterial culture was kept refrigerated alter culturing and read for nitrite within 2 h of culturing. Samples that were determined to have a colony count of >0 000, with a single organism present, were diagnosed as having a definite urinary- 222 CLINICAL CHEMISTRY, Vol. 2, No. 2,

2 Glucose concn, b g/iiter Table. vs. Clini-Tek NC Readings for Urine Supplemented with Glucosea Cilni-Tek T 2 N ±SD CV % a20 visual readings were taken. Each concentration was read times with Clini-Tek. 5Glucose concentration was calculated by taking normal (negative testing) trifle as 0 g/iiter and adding a 0 g/iiter solution of glucose to give the concentrations listed above. No correction made for endogeneous glucose. C In these Tabit s, N = negative; T = trace; and, 2, and are increasing degrees of positivity. Table 2. Correlation of and Instrumental Results on Urinary Glucose Glucose concn, /liter Clini-Tek n I ± SD CV, % n I ± SD CV, % Neg ± ±0.5 Tr.. ± ± ±.0.8± ± ± ± ± tract infection. The nitrite readings for these samples were tabulated according to the per cent positive for visual and Clini-Tek. Pathological urine samples were read only once visually and once by the semi-automated method (Clini-Tek), thus simulating routine urinalysis as actually performed in clinical situations. Results Table compares the results of visual vs. semi-automated reflectance reading of N-Multistix for urine containing glucose at different concentrations. Twenty independent visual readings were obtained on each sample and different readings on each sample by Clini-Tek. The 20 visual readings showed overlap of three sections of the stick (N, T, +)2 at a glucose concentration of 2 g/liter and (+, +2, +) at a concentra- 2 Abbreviations are N, negative; T, trace; +, +2, +, increasing concentrations of analyte. tion of.5 g/liter. The instrument did not show overlap of more than two sections. Comparison of means (Table ) showed that results for visual readings varied slightly less than did those for semi-automated readings for all areas (N, T, +, +2, +) of the strip. A comparison of the coefficients of variation showed that the instrument reads more precisely at N, +2, and + whereas visual readings are better at T and +. In Table 2 are compared the semi-quantitative (dipstick) estimates of glucose in diabetic urine for both visual and Clini-Tek readings. These results on pathological urine agree with the results in Table for contrived samples-i.e., visual readings show better precision at the N, T, and + levels whereas reflectance measurements are more precise at the +2 and + levels. Urinary glucose readings on the strip were light blue for negative, light green for trace, darker green for +, and shades of brown for +2 and +. ly, differen- CLINICAL CHEMISTRY, Vol. 2, No. 2, 22

3 Protein concn, a /liter Table. vs. Cilni-Tek Reading for Contrived Urines Containing Protein Ciil-Tek N T 2 NandT i i.28 i ±SD CV, % aurine total protein for pooled twine was 8 mg/liter. This value was added to calculated value of prepared samples to correct for endogenous protein. 5CIini-Tek does not have a trace reading for protein. Samples that would read as trace are read as +. n Table. DetermInation I ± SD of Proteinuria with Multi-Stix, Protein concn, g/llter CV, % n vs. Clini-Tek Cllnl-Tek x ± SD CV, % N 8 0.8± ± T ± ± ± ± ± ± ±. tiation between blue, green, and brown was easier than distinguishing between two different shades of brown. The method for urinary glucose on N-Multistix involves reaction of glucose with glucose oxidase (EC...) to form hydrogen peroxide which reacts with peroxidase (EC...) and KI to form 2,which imparts a brown color to the pad (5). Glucose oxidase methods used on dry reagent sticks have been described (). At trace levelsthe small amount of 2 formed causes the blue background on the pad to appear as green. This could explain why visual readings showed better precision in the T and + blocks. Also, since this reaction involves different colors rather than different shades of the same color in determination of proteinuria by dipsticks, the optimum wavelength would vary according to the color, but the instrument reads each test at one wavelength band. Table compares visual and instrument reading of contrived urine samples containing protein. Ten independent visual readings were taken and different 22 CLINICAL CHEMISTRY, Vol. 2, No. 2,

4 Table 5. vs. Cllni-Tek Readings for Urine Supplemented with Acetoacetlc Acid VIsual Cllnl-Tek Ketonoconcna N 2 N ±SD CV, % a mg acetoacetic acid/liter of urine. readings with Clini-Tek. There were several concentrations Figure shows a plot of the frequency of occurrence where three sections of the stick overlapped of +2 readings for the contrived protein samples versus when read visually but none when the samples were read the concentration of protein for both visual and instrument by reflectance. For protein, the instrument does not readings. Each sample has a maximum fre- have a trace reading. All samples that would be read quency of / readings. This figure ifiustrates that the visually as trace would read + with the instrument. In instrument has a point of unanimous agreement on all cases (N, +, +2, +) the coefficient of variation for / samples at a protein concentration of between the contrived protein samples for Clini-Tek was better.00 and.50 g/liter. In contrast, the visual readings than for visual readings. The color change in the protein showed a maximum value, but no point of unanimous portion of of the stick involves interpreting different agreement between readers. readings also showed shades of green. This accounts for the increased precision more of a tailing effect (wider distribution) than re- seen with the instrument as contrasted with glucose flectance readings. above. Table shows the comparison of visual vs. instrument determination of proteinuria from nephrotic patients. The coefficient of variation for the instrument, for the N and +2 sections,wasbetter than visualreadings.the + Clini-Tek readings include visual readings of both C T and +. The means for the pathological samples.c compare generally with the means for the artificially contrived urine, with the exception of the + section, where the much greater proteinuria in the pathological samples increased the mean. a TableS shows the comparison of visual and Clini-Tek determination of artifically prepared urine samples containing ketone (acetoacetic acid). As observed for glucoseand protein above, there are areas in the visual ,50 readings that showed overlap of three color sections. Instrument PROTEIN CONCENTRATION (G/LITER) readings did not show overlap of more than Fig.. Frequency of occurrence of +2 reading vs. protein two sections. Means and standard deviations for visual concentration for Clini-Tek (solid line) and visual (dashed line) vs. reflectance readings are approximately identical readings (Table 5) and coefficients of variation are also compa- CLINICAL CHEMISTRY, Vol. 2, No. 2, 225

5 Table. Determination of Bacteriuria from PatIents with Urinary Tract Infection8 Technician Clini-Tek Negative 5 5 Positive 2 % positive 8 a All wines had colony count > and Clini-Tekread one time each on the same samples used for culture. rable. Thus, visual and instrument readings on contrived ketone samples compare favorably. To evaluate visual vs. reflectance determinations of positive or negative nitrite, urines from patients with known urinary tract infections were read visually and with Clini-Tek. The results, shown in Table, indicate that the instrument is more sensitive in discriminating positive and negative nitrite. This test is based on the Greiss reaction () and changes from a white background to shades of pink. Low concentrations of nitrite are often difficult to distinguish visually and interpretation of the sticks can vary according to the surrounding color in the room. Discussion Reflectance readingsof urine dipstickson multiple samplesof the sameconcentrationwerereproducible, with >0% of the samples reading at the expected ( optimum ) level (). Readings rarely deviated more than onecolorsectionawayfrom the optimumreading. The question remained of whether the instrument would show the same reproducibility on pathological samples. The data presented in this report demonstrated that the instrument discriminates better between different concentrations than was the case for visual readings. The coefficients of variation for the instrument were better or just as good as visual readings for all sections (N, T, +, +2, +) for protein and ketone and for the N, +2, and + concentrations for glucose. readings were better for the T and + concentrations of glucose. This may be due to the nature of the color changes on the N-Multistix. Actual pathological urine samples collected from diabetic patients were read once visually and once with Clini-Tek. The results (Table 2) agreed with those of contrived samples in Table. A similar comparison for proteinuria showed that pathological urine samples could be read by reflectance just aswell or better than visually. Contrived samples were read consistently better by reflectance. Artificially prepared samples of urine containing acetoacetic acid showed agreement between visual and Clini-Tek readings. For contrived samples of glucose, ketone, and protein, there were several concentrations where visual interpretation was difficult and estimation ranged over three color sections. Reflectance measurements never showed overlap of greater than two sections. Pathological samples from patients with urinary tract infections showed 8% incidence of positive nitrite with Clini-Tek, but only a % incidence of positive nitrite with visual readings. Instrumental reading of urine dipsticks does not improve the chemistry or color reaction associated with the stick. It does, however, offer reproducibility and day-to-day stability that is unattainable in visual readings. Interpersonnel variation and subjective interpretation can be minimized by semi-automated reflectance measurements. In addition to being highly reproducible on multiple samples of optimum concentrations, the instrument also shows better discrimination of samples where the concentration of analyte is between the optimum for two color sections. Reading of pathological samples confirmed results observed on artifically prepared samples and showed a level of precision equal to or better than visual interpretation. It should be pointed out that visual interpretations will vary from person to person and laboratory to laboratory. Therefore, each laboratory should establish its own ranges, means, standard deviations, and coefficients of variation for visual reading of dipsticks. This work was supported by the Ames Co. References. Peele, J. D., Gadsden, R. H., and Crews, R., Evaluation of Ames Clini-Tek. Clin. Chem. 2, 228 (). 2. Gyure, W. L., Comparison of several methods for semiquantitative determination of urinary protein. Clin. C/oem. 2, 8 ().. Trinder, P., Determination of glucose in blood using glucose oxidase with an alternative oxygen acceptor. Ann. Clin. Biochem., 2 ().. Doetsch, K., and Gadsden, R. H., Determination of total urinary protein, combining Lowry sensitivity and biuret specificity. Clin. Chem., 0 (). 5. N-Multistix, product insert (Ames. Co.).. Free, A. H., and Free, H. M., Urinalysis, critical discipline of clinical science. CRC Crit. Rev. Clin. Lab. Sci., 8 (2). 22 CLINICALCHEMISTRY, Vol. 2, No. 2,