Urine Screening Strategy Employing Dipstick Analysis and Selective Culture: An Evaluation

Transcription

1 Urine Screening Strategy Employing Dipstick Analysis and Selective Culture: An Evaluation STANLEY Y. T. LOO, M.D., ALFRED G. SCOTTOLINI, M.D., SOMKENG LUANGPHINITH, MT, AUDREY L. ADAM, MT, LEE D. JACOBS, M.D., AND ALBERT J. MARIANI, M.D. A cost-effective urinalysis test strategy, employing screening dipstick analysis with sediment microscopy performed on urines positive for leukocyte esterase, nitrite, protein, or blood, is evaluated. Screening urine culture is done when 2:5 WBC/HPF, >10 bacteria/hpf, or yeasts are found on sediment microscopy. Predictive value, sensitivity, and specificity of the test strategy in predicting significant bacteriuria is compared with sediment microscopy, Gram staining of uncentrifuged urine and leukocyte chamber counting. Employment of the test protocol for routine urine specimens would decrease sediment microscopy by 49%, while effectively screening for significant bacteriuria with a sensitivity of % and predictive value of a negative result of 98.8%. (Key words: Screening dipstick urinalysis; Selective culture; Urine sediment microscopy) Am J Clin Pathol 1984; 81: THE DELETION of sediment microscopy when macroscopic and dipstick analysis are normal has been proposed previously. In a prior study done in 1979 at this hospital, of 1,100 urinalyses, only 2.4% were found to have abnormal sediment microscopic findings after negative macroscopic and dipstick analysis, 20 a finding similar to the 3% yield reported by Schumann and Greenberg. 19 Dipstick urine screening more recently has been augmented with nitrite and leukocyte esterase test strips available as a nine test dipstick, the Chemstrip-9. The nitrite test is based on the ability of enteric gram-negative bacteria to reduce dietary-derived nitrate to nitrite. Leukocytes can be detected by the action of esterases present in granulocytes and histiocytes, enabling the detection of even lysed leukocytes that may not be recognized on sediment microscopy. 2 Kusumi and associates 12 and Gillenwater 5 have shown the leukocyte esterase strip to be able to detect down to 10 WBC/jiL by chamber counting, reporting sensitivities of 87.9% and 95.3%, respectively. Correlation studies between sediment microscopy leukocyte counts/hpf compared with chamber counting, WBC/^L, have had variable results due to the lack of uniformity of sediment microscopy methods. Alwall 1 found the number of WBC/HPF to be approximately 11% of the number of WBC/jiL, or approximately 9 WBC/ Received July 12, 1983; received revised manuscript and accepted for publication October 20, Address reprint requests to Dr. Loo: Department of Pathology, Kaiser Medical Center, 1967 Ala Moana Boulevard, Honolulu, Hawaii Departments of Pathology, Internal Medicine (Infectious Diseases), and Urology, Kaiser Medical Center, Honolulu, Hawaii and University of Hawaii John A. Burns School of Medicine ul for each WBC/HPF. Thus, the threshold sensitivity of the leukocyte esterase test is on the order of 1 WBC/ HPF. An even greater lack of uniformity of reporting criteria for bacteria seen on sediment microscopy has led many clinicians to disregard this parameter, though one reference" suggests 20 bacteria/hpf as indicative of significant bacteriuria. Gram staining of uncentrifuged urine has been proposed as an alternative for bacteriuria screening with two or more organisms per oil immersion field correlating with the presence of S: 100,000 organisms/ ml with a sensitivity of 94% in the study by Washington and associates. 24 The sensitivity and specificity of Gram stain microscopy was shown to be comparable to that of three automated microbiologic systems 17 (1-General Diagnostics Autobac, 2-AutoMicrobic System AMS, 3-Abbott Laboratories MS-2) with bacteriuria screening capabilities. However, detection time with the systems ranged from 1.8 to 6.1 hours, and the predictive value of a positive screening test was only 27.7% to 40.8%, limiting the usefulness of these methods in immediate outpatient management. Recently introduced are systems employing bioluminescence due to microbial adenosine triphosphate, such as the Lumac /3M system with results available in one hour, 23 and filtration/staining technics (Bac-T-Screen, Marion Laboratories Corp.) for bacteriuria screening. These systems will screen for the presence of significant numbers of bacteria but do not discriminate between significant bacteriuria versus contaminated specimens requiring quantitative culture for confirmation. After a preliminary series, 14 Kaiser Regional Laboratories in Oregon has implemented a cost-effective urinalysis test protocol, employing initial urine testing by Chemstrip-9, with sediment microscopy, unless specifically ordered, performed only on specimens positive for leukocyte esterase, nitrite, protein, or blood. If leukocyte esterase or nitrite is positive, >20 WBC/HPF or many (>10/HPF) bacteria are seen on sediment microscopy, urine culture is automatically performed. With this pro- 634

2 Vol. 81 -No. 5 URINALYSIS TEST STRATEGY 635 tocol, 15 they were able to delete 65.5% of urine sediment microscopy, reduce urine culture volume by 18%, and achieve an annual cost savings of $42,708. At the same time, detection of asymptomatic bacteriuria was improved with the proportion of significant bacteriuria in urine specimens cultured increasing by 25%. This current study undertakes the comparison of a similar optimized urinalysis test protocol with sediment microscopy, quantitative urine culture, uncentrifuged urine Gram staining, and leukocyte chamber counting. The proposed urinalysis dipstick, sediment microscopy, and selective culture algorithm is detailed in Figure 1. Compared with the Kaiser (Oregon) test strategy, the proposed test algorithm initiates urine culture routinely if ^5 WBC/HPF are seen on sediment microscopy rather than with a positive leukocyte esterase test in consideration of the high sensitivity of the esterase test and the potential problem of female patient specimen contamination by genital tract leukocytes. Culture is also initiated if the nitrite test is positive or if yeasts or >10 bacteria/hpf are seen. Urine Collection Materials and Methods Patients were given instructions to obtain a clean-catch midstream urine specimen. Female patients were given a Tomac towelette to wipe the introitus prior to obtaining the specimen. Urine specimens were processed upon receipt in the laboratory, with specimens delayed in transport to the laboratory more than one hour or with visible signs of contamination rejected. Dipstick Urine Testing The Chemstrip-9 (Biodynamics, BMC Division of Boehringer Mannheim, Indianapolis, Indiana) test strips were used in accordance with the manufacturer's instructions. For the first two series of 100 urinalysis each, the earlier Chemstrip leukocyte esterase test was used, reading the color reaction between 15 and 30 minutes. The later series employed the newer leukocyte esterase test strip, reading between one and two minutes. Positive protein dipstick findings were reported only if confirmed on sulfosalicylic acid testing. Urinalysis Sediment Microscopy Sediment microscopy was performed on tenfold concentrated sediment suspension obtained by centrifugation of 10 ml urine at 2,000 rpm (radius of 9 cm) for 5 minutes. The sediment after decantation was resuspended to a volume of 1 ml. Kova (modified Sternheimer-Melbin) stain was used with microscopy performed using. Abnormal color /and appearance / WBC (leukocyte / esterase) Chemstrip-9/ Testing of ( Nitrite Patient's \ Urine \ E Blood - If * or trace If + -If t Protein If + Confirm If - Microscopic -»Microscopic Culture if indicated* Culture if indicated* -» Microscopic and Culture - Microscopic + ^Microscopic * If >5 WBC/HPF, >10 bacteria/hpf or positive yeast Culture if indicated* Culture if indicated* FIG. 1. Summary of proposed screening urinalysis and culture protocol. Kova calibrated disposable slides (ICL Scientific, Fountain Valley, CA). Ten (XlOO) low-power fields (LPF) were scanned for the presence of casts, and then a minimum of ten (X440) high-power fields (HPF) were scanned for other formed elements, the average count of ten fields being reported as an approximate range in the first two series. In the later series, findings were reported according to a standardized reporting scale with an 0-4 WBC/HPF and 0-3 RBC/HPF regarded as the normal range. Bacteria seen on sediment microscopy were recorded as 0-3/HPF, 3-10/HPF, and >10/HPF. Urinalysis Chamber Counting White blood cell chamber counting in WBC/^L of uncentrifuged urine was performed using a Neubauer hemocytometer chamber. Urine Gram Staining Urine Gram stain microscopy as utilized by Washington and associates 24 was performed using one drop of well-mixed uncentrifuged urine applied to a slide. The air-dried and Gram-stained preparation was examined for at least 20 oil immersion (X 1,000) fields, with the positive smear having two or more organisms per oil immersion field. Urine Microbiologic Technics Urine quantitative culture was performed utilizing a mL calibrated loop with inoculation of 5% sheep blood agar plate for enumeration and an eosin methylene blue agar plate. Significant bacteriuria was defined as a pure or nearly pure growth of a potential urinary tract pathogen in quantities of 100,000 organisms or more per milliliter. Mixed growth of three or more types of bacteria was considered indicative of specimen contamination, usually with the presence of gram-positive bacilli (Diphtheroids and Lactobacilli). Bacteria identification was by standard procedures employing the API 20E strip and API Staph Ident System (Analytab Products, Plainview, NY) when appropriate.

3 636 LOO ET AL. A.J.C.P. May 1984 Data Analysis Test performance was tabulated using the standard test parameters of sensitivity, specificity and predictive value of a positive or negative test. Linear regression analysis of chamber count compared with sediment microscopy WBC/HPF was calculated using averaged sediment microscopy range values. Results The ability of dipstick testing to screen for cellular elements was assessed by comparing Chemstrip-9 (1- itiinute leukocyte esterase) analysis with concurrent sediment microscopy in a series of 2,340 urinalyses. Both inpatient and outpatient urine specimens received by the laboratory over a six-week time period were included, the age range of the patients spanning nine decades. Female patients contributed 60.9% of the specimens. During the same time period, 600 of these urinalyses or 25.6% also were cultured by physician request. Detection of Red Blood Cells There were 364 urinalyses with >3 RBC/HPF, of which 356 (sensitivity of 97.8%) were detected by a trace or stronger reaction on the Chemstrip-9 blood strip. Of the eight false-negative cases in 2,340 urinalyses (0.34%), four patients had findings of proteinuria or positive leukocyte esterase dipstick reaction, which would have initiated sediment microscopy according to the dipstick screening protocol. Two of the remaining four patients had 4 RBC/ HPF only. The final two patients with 4-10 RBC/HPF were an adult rheumatoid arthritis inpatient with traumatic compression fracture/disc herniation with 0 RBC/ HPF on a second urinalysis and an eight-year-old male outpatient on his annual physical examination. There were 408 false-positive dipstick analyses (specificity of 79.3%), with 58.5% of these urinalyses having a trace blood dipstick test. Thus, for a series hematuria "prevalence" of 15.6%, a predictive value of a positive test of 46.6% and predictive value of a negative test of 99.5% was found. These findings indicate the ability of the Chemstrip-9 to detect down to apparent physiologic levels of red blood cells in the urine and the possible presence of hemoglobin from lysed red blood cells not detected by sediment microscopy. Detection of Casts Out of the 2340 urinalyses, 123 urinalyses (5.3%) contained casts on sediment microscopy, the vast majority of these being low levels of hyaline casts. Of these urinalyses, 101 would have been detected by the dipstick screening protocol, usually with a positive protein test (sensitivity of 82.1%). Of the 22 instances of casts not detected by dipstick analysis, 17 had only few hyaline casts and 5 had a few coarse granular casts. None of the 22 false-negative urinalyses had RBC casts or WBC casts. Of those cases with more pathologic forms of casts, one urinalysis with 2-5 granular casts/lpf and 0-1 waxy casts/lpf would have been detected by dipstick analysis with a 4+ protein and positive blood tests. Another specimen with 5-10 hyaline casts, 5-10 fine granular casts, and 2-5 fatty casts/lpf was detected by the presence of 4+ proteinuria, positive blood test, and trace positive leukocyte esterase test. The three cases containing 0-1 WBC casts/lpf, hyaline and granular casts all would have been detected with the presence of positive protein, blood, and/or leukocyte esterase tests. These findings were similar to our earlier retrospective review in 1982 of 2,470 Ames Multistix dipstick and sediment microscopy urinalyses. In that series, casts, usually hyaline, were found in 141 urine specimens (5.7%). Of these, 99 had trace or more protein or other dipstick findings that would have initiated sediment microscopy according to a dipstick screening protocol, even without the leukocyte esterase and nitrite test strips (sensitivity of 70.2%). The two urine specimens with WBC casts in the Multistix series were both positive for protein. Detection of Other Formed Elements In the Chemstrip-9 series, 12 urinalyses contained yeast on sediment microscopy, 11 of these having a positive dipstick test result, which would have initiated sediment microscopy. The majority of these had a trace or strong leukocyte esterase reaction correlating with the presence of some WBC on sediment microscopy with some other specimens in this group additionally or alternatively having a positive protein dipstick test or weak blood dipstick reaction. The one urinalysis with a few yeast forms on sediment microscopy not detected by dipstick analysis contained 0-4 WBC/HPF on sediment microscopy and no other findings. Eight urinalyses contained trichomonads on sediment microscopy. Seven of these would have been detected by the dipstick screening protocol, the majority of these having a positive test for blood. The one case not detected by dipstick, a 35-year-old woman on a preoperative admission urinalysis contained a few trichomonads, 0-1 hyaline casts/lpf, and no other findings. Thirty-eight specimens contained crystals on sediment microscopy. Twenty-five would have been detected according to the dipstick screening protocol. The majority of cases had a few calcium oxalate crystals. Six cases had few uric acid crystals, and one case had a few triple phosphate crystals.

4 Vol. 81 No. 5 URINALYSIS TEST STRATEGY 637 Detection of Pyuria by Leukocyte Esterase Test As a preliminary study using the 15-minute leukocyte esterase Chemstrip-9, serial twofold saline dilutions of a peripheral blood buffy coat suspension were prepared with the initial leukocyte concentration determined by chamber counting. The threshold sensitivity of the leukocyte esterase test was approximately 6-12 WBC//iL. As specific methods of sediment microscopy have varied in different series, a correlation study comparing chamber counting with sediment microscopy for leukocytes was undertaken using 200 patients' urine specimens. The chamber count value in WBC/jtL was approximately 10.8 times the value in WBC/HPF by sediment microscopy (Fig. 2). Using >30 WBC/ML by chamber counting as denoting pyuria, sediment microscopy had a sensitivity of 89.5% and 86.8%, respectively, using >3 WBC/HPF and 5 WBC/HPF as the cut-off. Using the same 200 patients' urine specimens, the sensitivity of the 15-minute leukocyte esterase strip was slightly greater than sediment microscopy for leukocytes, using either 3 or 5 WBC/HPF as the cut-off, with a sensitivity of 92.1% and specificity of 84.6%. Using the data by Kusumi and associates, 12 the leukocyte esterase sensitivity (87.9%) was also greater than that of sediment microscopy for 5 or more WBC/HPF (sensitivity of 83.0%). Using the newer 1-minute leukocyte esterase strip in the series of 2,340 urinalyses, (17.4% "prevalence" of patient episodes of pyuria defined as 5 WBC/HPF or more), the leukocyte esterase test had a comparable sensitivity of 91.3%, with a specificity of 90.8% and a predictive value of a negative result of 97.8%. Out of 2,340 urinalyses, there were 42 false-negative cases of pyuria (1.8%). Twenty-eight of these were positive for protein, blood, and/or nitrite, which in the screening dipstick protocol would have initiated sediment microscopy and detection of the microscopic pyuria. All of the remaining 14 false-negative cases had low levels of pyuria in the range of 5-10 WBC/HPF. Following the proposed dipstick screening protocol, the percentage of false-negative cases would be reduced to 0.6% by the above considerations. Detection of Significant Bacteriuria In this part of the study, the utility of Chemstrip-9 dipstick analysis as a screening procedure for significant bacteriuria was assessed with comparison with sediment microscopy and uncentrifuged urine Gram staining. The first two series of 100 urinalyses each employed the 15- minute leukocyte esterase strip, the first series of 100 consecutive urine specimens also being evaluated by leukocyte chamber counting. This first series had a 9% "prevalence" of significant bacteriuria, with 22% of the urines having 100,000 organisms or more of any type or Chlnbcr Count vt. SadtMUlt Mfcroscop) WBt/ul > 10.6 (wec/kp*; * X 240 SWtatnt dlcnxufjr WK/kff CorrtUtian Cotficient <X783 FIG. 2. Comparison of urinary leukocyte detection by chamber counting versus sediment microscopy. mixture per milliliter. The second series of urine specimens with cultures ordered, as expected, had a higher prevalence (21%) of significant bacteriuria, with 28% having 100,000 or more bacteria/ml of any type or mixture. The value of various urinalysis test parameters in predicting significant bacteriuria are summarized in Tables 1 and 2. Gram staining of uncentrifuged urine was evaluated in two ways, one with the positive result defined as two or more organisms per oil immersion field of any type and the second being a positive gram stain finding exclusive of apparently contaminated specimens having the presence of gram-positive bacilli or multiple organisms. The summary data for the combined 200 urine specimens using the 15-minute leukocyte esterase strip is shown in Table 3, with the corresponding values from the literature (also utilizing the 15-minute strip) listed for comparison. In a third series of 600 urinalyses with ordered cultures using the newer 1-minute leukocyte esterase Chemstrip- 9, similar findings were found as summarized in Table 4, with the calculations also listed separately for male and female patient subgroups. Of the urine specimens with significant bacteriuria, 48.4% had >20 WBC/HPF. For the subgroup reporting ranges of <5 WBC/HPF, 5-9 WBC/HPF, WBC/HPF, and ;>20 WBC/HPF, the percentages of urines with significant bacteriuria were

5 LOO ET AL. AJ.C.P. May Table J. First Series of 100 Consecutive Urinalysis (Incidence: Significant Bacteriuria or Yeast 9% [100,000/mL or More Colonies of Any Type or Mixture 22%] [Values (%) Calculated for Significant Bacteriuria]) Sensitivity Specificity Pred.. Val. Pos. Pred. Val. Neg. Sediment microscopy WBC ;> 20/HPF WBC ;> 10/HPF WBC ;> 5/HPF WBC ;> 3/HPF WBC 2/HPF WBC a 5/HPF and/or WBC ;> 5/HPF and Chamber counting > 10 WBC/^L Gram staining > 2 org/oif Gram staining screened Leukocyte esterase Nitrite Esterase and/or nitrite Protocol: (51.0% microscopic exam) (17.0% culture) (47.1% pos. culture) Protocol: nitrite pos. or microscopy with WBC ;> 5/HPF and bacteria > 10/HPF Kaiser Oregon protocol (51% microscopic exam) (33.0% culture) (24.2% pos. culture) %, 31.4%, 44.6%, and 69.4% respectively, lending sup- as seen in Table 3, sensitivity and specificity were optiport for the selection of 2:5 WBC/HPF as one of the mized at the 2-5 WBC/HPF range as a cut-off point, criteria for automatic urine culture in the protocol. Also, Microbiologic findings are summarized in Table 5. Table 2. Second Series of 100 Urinalysis with Ordered Cultures (Incidence: Significant Bacteriuria 21% [100,000/ ml or More Colonies of Any Type or Mixture 28%] [Values (%) Calculated for Significant Bacteriuria]) Sediment microscopy WBC a 20/HPF WBC 10/HPF WBC ;> 5/HPF WBC ;> 3/HPF WBC ;> 2/HPF WBC ;> 5/HPF and/or WBC 2; 5/HPF and bacteria > 10/HPF Gram staining > 2 org/oif Gram staining screened Leukocyte esterase Nitrite Esterase and/or nitrite Protocol (58.0% microscopic exam) (36.0% culture) (52.8% pos. culture) Protocol Nitrite pos. or microscopy with WBC > 5/HPF and Kaiser Oregon protocol (58.0% microscopic exam) (37.0% culture) (51.4% pos. culture) Sensitivity Specificity Pred. Val. Pos. Pred. Val. Neg

6 vol. 81-No. 5 URINALYSIS TEST STRATEGY 639 Table 3. Combined Series of 200 Urinalysis (Incidence of Significant Bacteriuria 15%) (Values in %) Sensitivity Specificity Pred. Val. Pos. Pred. Val. Neg. Sediment microscopy WBC :> 20/HPF WBC a 10/HPF WBC ;> 5/HPF WBC 3/HPF WBC 2; 2/HPF WBC ;> 5/HPF and/or Gram staining 2: 2 org/oif Gram staining screened Leukocyte esterase Nitrite Esterase and/or nitrite Protocol Kaiser Oregon protocol Protocol: Nitrite pos. or microscopy with WBC a 5 and bacteria > 10/HPF (80.0)* (75.0)t 96.7 [94]* 83.3(71.4) 33.3 (69.0) (86.2) 90.0 { } Values in parenthesis from similar study by Wenk and associates. 23 t "Visible" bacteriuria with 15.5% significant bacteriuria on sediment examination. t Values in brackets from study by Washington and associates 24 with 13.4% significant bacteriuria. Cost Analysis In both the initial series of 100 consecutive urinalyses and in the later series of 2,340 urinalyses, employment of the proposed urine screening protocol would decrease sediment microscopy by 49.0% and 52.1%, respectively. In instituting such a cost-effective screening protocol, the (82.9) (52.5) [90] (85.2) 98.8 (90.3) (86.5) 84.7 { } (46.2) (22.5) [90] (46.9) 83.3 (56.6) (53.9) 50.9 { } Table 4. Series of 600 Urinalysis with Ordered Cultures* (95.8) (92.1) [99] (94.3) 89.4(94.1) (97.2) 98.0 { } Values from study by James and associates* for 11.3% incidence bacteriuria. 11 Value ranges for three automated microbiologic urine screening systems with 11.8% significant bacteriuria incidence Pezzlo and associates. 11 option of specifically ordering sediment microscopy or culture regardless of dipstick findings, of course, would be available, so that the actual decrease in sediment microscopy will in part depend on ordering practices. For our laboratory region, approximately 83,000 urinalyses a year are performed. Our actual cost of sediment microscopy is approximately $0.74 for approximately 3 Sensitivity Specificity Pred. Val. Pos. Pred. Val. Neg. Sediment microscopy WBC a 20/HPF 48.4 [50.7] (26.9) 92.6 [90.8] (96.2) 69.4 [71.6] (53.8) 83.0 [80.1] (88.8) WBC a 10/HPF 64.5 [65.7] (46.2) 85.6 [90.8] (91.7) 61.0 [63.3] (48.0) 87.4 [84.0] (91.1) WBC z. 5/HPF 81.9 [81.3] (69.2) 72.4 [67.6] (82.2) 50.8 [53.4] (39.1) 92.0 [88.8] (94.2) 75.9 [74.2] (84.6) 85.6 [80.4] (94.9) 65.2 [63.6] (73.3) 90.9 [87.1] (97.4) WBC a 5 HPF and/or 96.2 [96.2] (96.2) 66.1 [59.1] (79.0) 50.3 [52.0] (43.1) 98.0 [97.1] (99.2) WBC a 5 HPF and 62.0 [59.8] (73.1) 91.9 [88.5] (98.1) 73.7 [71.2] (86.4) 87.2 [82.7] (95.6) Leukocyte esterase 83.4 [84.0] (80.8) 67.8 [59.9] (82.2) 47.8 [48.9] (42.9) 92.0 [89.1] (96.3) Nitrite 35.0 [33.6] (42.3) 98.9 [99.3] (98.7) 93.2 [95.7] (84.6) 81.2 [76.6] (91.2) Leukocyte easterase and/or nitrite pos [91.6] (80.8) 67.1 [59.2] (81.5) 49.1 [50.6] (42.0) 94.9 [93.9] (95.5) Protocol with WBC & 5/HPF and/or bacteria > 10 (66.4% microscopic exam) (45.6% culture) (51.5% pos. culture) 89.2 [90.2] (84.6) 70.4 [63.6] (82.8) 51.5 [52.9] (44.9) 95.4 [94.3] (97.0) Protocol: Nitrite pos. or microscopy with WBC S; 5/HPF and 62.0 [59.8] (73.9) 91.9 [88.5] (97.4) 73.1 [71.2] (82.6) 87.1 [82.7] (95.6) Kaiser Oregon protocol (66.4% microscopic exam) (49.6% culture) (48.7% pos. culture) 90.6 [91.7] (85.2) 65.3 [57.3] (80.0) 48.7 [50.0] (42.6) 95.0 [93.7] (96.9) 69.6% female. 30.4% male patients. 26.6% significant bacteriuria (females 32.0% and males 14.4% significant bacteriuria). Values in %: male and female combined (females only] (males only).

7 640 LOO ET AL. AJ.C.P. May 1984 Table 5. 1st Series of 100 Consecutive Urinalysis Male and Female Significant bacteriuria 100,000 org/ml 9% Contaminants i 100,000 org/ml with coliforms or potential pathogens 3% Contaminants & 100,000 org/ml without coliforms or potential pathogens 10% Single coliform or potential pathogen (alo 4, <10 5 org/ml) 1% Multiple organisms with coliforms (alo 4, <10 5 org/ml) 1% Multiple organisms with coliforms (alo 3, <10 4 org/ml) 3% Nonconform contaminants (a 10", <10 5 org/ml) 14% Nonconform contaminants ( 10 3, <10 4 org/ml) 13% No growth 46% 3rd Series of 600 Urinalysis with Ordered Cultures Significant bacteriuria a 100,000 org/ml Contaminants ^ 100,000 org/ml with coliforms or potential pathogens Single coliform or potential pathogen (:>10 4, <10 5 org/ml) Single coliform or potential pathogen (;>10 3, <10 4 org/ml) Multiple organisms with coliforms (slo 4, <10 5 org/ml) Multiple organisms with coliforms (alo 3, <10 4 org/ml) Noncoliform contaminants (< 100,000 org/ml) Contaminants a 100,000 org/ml without coliforms or potential pathogens No growth Male and Female Female Male 26.6% 32.0% 14.4% 3.4% 3.0% 2.7% 3.0% 3.9% 19.4% 8.4% 29.1% 4.6% 3.4% 2.2% 3.8% 4.3% 20.7% 11.8% 16.6% 0.6% 2.2% 3.9% 1.1% 2.8% 16.6% 0.6% 58.0% minutes of medical technologist time and material costs. Estimating a 52.1% decrease in sediment microscopy will project a cost savings to the laboratory of $31, or $25, if it is assumed that only 80% of urinalyses are ordered under this routine urinalysis protocol. The Chemstrip-9 is slightly more expensive (list price $27.50 per 100 tests) compared with a seven-parameter Chemstrip-7 or Multistix (list price $23.00 per 100 tests) which, using list price, may increase reagent costs by $3, For the series of 2,340 urinalyses using the 1-minute Chemstrip-9, 600 (25.6%) urinalysis requests had concurrent orders for urine culture, of which 26.6% had significant bacteriuria. Employment of the urine screening protocol for the 2,340 urinalyses would have initiated 528 cultures (22.6%) compared with the 600 ordered, a 12% reduction. From this group of 528 cultures, 159 urinalyses had significant bacteriuria identified on culture exclusive of those urines without ordered cultures, which, according to the sediment microscopy findings, probably had significant bacteriuria. Using an estimated 9% significant bacteriuria "prevalence" from the first series of 100 consecutive urinalyses, 210 significant bacteriurias would be expected for the series of 2,340 urinalyses. If one assumes 90% of these would be detected by the screening protocol (189), the percentage of positive cultures initiated would be 35.8%, compared with the 26.6% of ordered cultures. However, changes in number of cultures and percentage positive cultures will be greatly dependent on ordering practices, as there are clinical indications for ordered culture irrespective of urinalyses findings. Discussion Cost-containment in medicine increasingly has been an issue with optimal laboratory utilization requiring knowledge of the clinical situation with regard to the probability and usefulness in patient management of an abnormal test finding. In this study, the feasibility of a cost-effective urinalysis test strategy involving initial urine dipstick screening including leukocyte esterase and nitrite testing is demonstrated, our findings being similar to that found by the Kaiser Oregon Regional Laboratory. Comparatively, for the series of 2,340 urinalyses using the newer 1-minute leukocyte esterase strip, the Kaiser Oregon protocol would have initiated 669 cultures (28.6%), compared with 528 cultures (22.6%) by our protocol. The sensitivity of either protocol is comparable with the Kaiser Oregon protocol, having a "sensitivity" of 90.6% versus 89.2% for our protocol in the series of 600 urinalyses with ordered cultures. As a general screening procedure for routine urine specimens, both protocols are equally effective, the predictive value of a negative result being 98.5% and 98.8%, respectively. As might be expected, the predictive value of a positive test, that is, the percentage of cultures generated that are positive for significant bacteriuria, was less for the Kaiser Oregon protocol compared with ours, this difference being greatest in the first series of 100 consecutive urinalysis specimens with a 9% "prevalence" of significant bacteriuria where the percentage of positive cultures was 24.2% compared with 47.1%. This difference is smaller in the later larger series. The inclusion of a "culture if indicated" provision in both protocols will improve detection of unsuspected urinary tract infection, simplify physician ordering, and potentially decrease the number of urine cultures performed as shown in our study projections and actually observed in the Kaiser Oregon data. We recently have implemented the proposed urinalysis protocol at our institution with achievement of the projected sediment microscopy savings. Physician, nursing staff, and medical technologists' acceptance of the protocol has been good. Not having to

8 Vol. 81 -No. 5 URINALYSIS TEST STRATEGY 641 recall the patient for urine culture specimen when unexpected urinalyses results suggestive of urinary tract infection are reported is a distinct benefit to the protocol. Dipstick screening can be standardized further by employment of automated dipstick readers, one such device, the Urotron, having been updated recently to read the leukocyte esterase strip of the Chemstrip-9. Preliminary field testing of the Urotron reading of the leukocyte esterase test by Gambino 4 indicates the Urotron reading capabilities to equal visual reading of the test strip, and we currently are evaluating use of the Urotron with our protocol. Collection of a noncontaminated "clean catch midstream urine" remains a persistent problem in this study, particularly for female patients, resulting in a lower specificity of the leukocyte esterase test for significant bacteriuria in our study and in that by Perry and associates. 16 In this study, significant bacteriuria was defined along the traditional criterias of Si 100,000 bacteria/ml of a probable pathogen without the presence of three or more organisms indicative of specimen contamination. More recently, Stamm and associates, 22 in studies of acutely dysuric women, compared urine cultures obtained by collection of midstream voided urine with those collected by bladder catheterization or suprapubic aspiration. In their series, the authors found the traditional criteria of : 100,000 bacteria/ml of midstream urine would only identify 41% of women whose bladder urine contained conforms. The study revealed the best diagnostic criteria to be ^ 100 bacteria/ml (sensitivity of 95% and specificity of 85%) for the population of acutely dysuric women. In their series, 32% of clean-catch midstream urine specimens had nonconform organisms (apparent contaminants) without the presence of coliforms, with only 2% of the midstream specimens having no growth, compared with 31-35% with no growth when cultured by urethral catheterization or suprapubic aspiration. These findings underscore the difficulty of obtaining a single isolate with clean-catch midstream urine collection. Special attention to urine specimen collection as in the urine Gram staining study by Washington and associates 24 can minimize specimen contamination and is presumably responsible for the much higher specificity and predictive value of a positive result for Gram staining in their series compared with ours (Table 3). Female urine specimens in their series were collected on a modified gynecologic examining table with the assistance of trained personnel. Logistics, cost, and possible patient resistance might make such a collection system impractical for many institutions. Recently, urine collection systems to facilitate clean-catch midstream urine collection have been introduced that may help reduce specimen contamination. In both Washington's study and in our study, Gram staining of uncentrifuged urine was found to be of value in presumptively confirming urinary tract infection with the identification of gram-negative or gram-positive organisms, providing some discriminative value in excluding contaminated specimens. Its major drawback is its timeconsuming nature in terms of laboratory utilization and inability to detect levels of bacteriuria < 100,000 organisms/ml as pointed out by Stamm. 21 Thus, the uncentrifuged urine Gram stain may be most useful in cases of suspected acute pyelonephritis, where rapid diagnosis may be critical and where, as shown in the benchmark studies of Kass, 7 " 9 95% or more of patients with acute pyelonephritis have heavy bacteriuria ;> 100,000 bacteria/ ml. In this situation, presumptive organism identification of an uncontaminated urine specimen can be attempted by direct innoculation of urine sediment in rapid biochemical testing systems (e.g., Micro ID 18 or if available rapid automated microbiologic systems 10 ), to be confirmed with follow-up organism isolation and identification. While it may be unwise to use a single criterion 2:100,000 organisms/ml to establish the presence of bladder infection, the problems of interpreting lower bacteriuria levels, particularly in mixed culture in asymptomatic patients, favors continued use of the criteria of > 100,000 organisms/ml for screening asymptomatic patients, as still currently recommended by Stamm. 21 Clearly, clinical input as to symptoms and clinical situation is needed for the laboratory to properly apply a decision point of 100,000 organism/ml or a lower decision point for work up as recently discussed by Bartlett and Galen. 3 In the series of Stamm and associates, 22 52% of his acutely dysuric female patients had bladder colonization by coliforms. In their study, pyuria defined by chamber counting of eight or more leukocytes/^l had a sensitivity of 91 % (specificity 50%), with a predictive value of a positive result of 67% forcoliform infection of the bladder, documented by suprapubic or urethral catheterization. Comparatively, in our series of 600 urinalyses with ordered cultures, for female patients the sensitivity of the leukocyte esterase strip that optimally can detect down to 10 WBC/ ml, was 84.0% (specificity 59.9%), with a predictive value of a positive result of 48.9% for significant bacteriuria. The combination of leukocyte esterase and/or nitrite positive in this series had a sensitivity of 91.6% for female patients with a specificity of 59.2% (predictive value of a positive result of 50.6%). Similarly, the presence of ^20 WBC/HPF or a positive nitrite test has a predictive value in our series of 71.6% and 95.7%, respectively for significant bacteriuria. The high predictive value of the nitrite test has also been noted by Wenk and co-workers 25 and James and colleagues. 6 The nitrite test, while having a high specificity, has a low sensitivity for bacteriuria when first-voided morning urine is not used and will not detect

9 642 LOO ET AL. A.J.C.P. May 1984 the presence of Staphylococcus saprophyticus, a non-nitrate-reducing bacterium. The predictive value of these urinalysis findings actually may be higher in symptomatic patients, as our series of 600 urines was not stratified as to symptoms. Preliminary review of Chemstrip-9 urinalysis findings for acutely symptomatic adult female outpatients with dysuria and/or frequency presenting to our acute care clinic with apparently uncomplicated urinary tract infection also reveal similar findings. Of 14 patients seen in clinic with urinalysis and cultures performed, ten patients (71%) had significant bacteriuria S; 100,000 organisms/ml, with 13 patients having > 1,000 organisms/ml of a probable pathogen (lowest value of 5,000/mL of Escherichia coli). Using a > 1,000 organism/ ml criteria for "significant" bacteriuria for this group, the sensitivity of the leukocyte esterase and urinalysis protocol remained similar to the series of 600 urinalyses, with ordered cultures with sensitivities of 84.6% and 92.3%, respectively. Due to the high incidence of "significant" bacteriuria in this group, the predictive value of a positive leukocyte esterase, nitrite, or protocol culture criteria finding was very high in the limited series. Rapid reporting of urinalysis findings therefore can aid considerably in the decision for presumptive antibiotic therapy for symptomatic patients before culture results become available. As an adjunct to the protocol, specific ordering of sediment microscopy or urine culture would be appropriate in more problematic clinical situations, such as with cases of suspected underlying urinary tract disease, states of increased infection susceptibility, e.g. immune compromized patients, cases of possible upper tract disease, patients with prostatitis, catheterized patients, and cases where the urinary tract symptoms or findings may be associated with female genital tract findings. In summary, the proposed urine test protocol will screen effectively for significant urinalysis findings and potentially improve detection of significant bacteriuria while reducing laboratory costs by deletion of unnecessary urine sediment microscopy. As the number of women with symptoms of lower urinary tract infection accounted for more than 5,000,000 office visits per year in the United States, 13 cost savings by the proposals outlined can be substantial. References 1. Alwall N: Pyuria deposit in high power microscopic field WBC/ HPF vs. WBC/mm 3 in counting chamber. Acta Med Scand 1973; 194: Avent J, Schumann GB, Vars L: Comparison of chemstrip leukocyte test with standardized Papanicolaou-stained urine sediment evaluation. Lab Med 1983; 14: Bartlett RC, Galen RS: Editorial: Predictive value of urine culture. Am J Clin Pathol 1983; 79: Gambino SR: Routine and nonroutine urinalysis: A study in costeffective testing. Laboratory Testing, New and Future Procedures International Symposium 1983, Tarpon Springs, Florida 5. Gillenwater JY: Detection of urinary leukocytes by Chemstrip-L. J Urol 1981; 125: James GP, Paul KL, Fuller JB: Urinary nitrite and urinary tract infection. Am J Clin Pathol 1978; 70: Kass EH: Chemotherapeutic and antibiotic drugs in the management of infections of the urinary tract. Am J Med 1955; 18: Kass EH, Finland M: Asymptomatic infections of the urinary tract. Trans Assoc Am Physicians 1956; 69: Kass EH: Bacteriuria and the diagnosis of infections of the urinary tract, with observations on the use of methionine as a urinary antiseptic. Arch Intern Med 1957; 100: Kelly MT, Balfour LC, Latimer JM: Rapid detection and identification of urinary tract pathogens. Rapid methods and automation in microbiology. Edited by RC Tilton. Washington D.C., American Society for Microbiology, 1982, pp Kunin CM: Detection, prevention and management of urinary tract infections. Philadelphia, Lea and Febiger, 1979, p Kusumi RK, Grover PU, Kunin CM: Rapid detection of pyuria by leukocyte esterase activity. JAMA 1981; 245: National Center for Health Statistics: Ambulatory medical care rendered in physicians' offices. United States Adv Data 1977; 12: Overton KM, Jones P: From your laboratory if you processed 2,700 gallons of urine a year what would you do? Viewpoint and Comment. Am Soc of Clin Pathologists Summary Report 1981; 18(7): Overton KM, Jones P, Holahan KP: An effective urinalysis chemical screening program (poster). Am Soc Clin Pathol College of Am Pathol Scientific Assembly, October 1982, Miami Beach, Florida 16. Perry JL, Matthews JS, Weesner DE: Evaluation of leukocyte esterase activity as a rapid screening technique for bacteriuria. J Clin Microbiol 1982; 15: Pezzlo MT, Tan GL, Peterson EM, DeLa Maza LM: Screening of urine cultures by three automated systems. J Clin Microbiol 1982; 15: Randall E: First-day results what is feasible? Rapid methods and automation in microbiology. Edited by RC Tilton. Washington, DC, American Society for Microbiology, 1982, p Schumann GB, Greenberg NF: Usefulness of macroscopic urinalysis as a screening procedure. A preliminary report. Am J Clin Pathol 1979; 71: Scottolini AG, Frankel S, Darm J: Criteria for the reduction of microscopy in urinalysis. Kaiser Medical Center Symposium 1979, Honolulu, Hawaii 21. Stamm WE: Interpretation of urine cultures. Clinical Microbiology Newsletter 1983; 5: Stamm WE, Counts GW, Running K, Fihn S, Turck M, Holmes KK: Diagnosis of coliform infection in acutely dysuria women. N Engl J Med 1982; 307: Szilagyi G, Aning V, Karmen A: Comparative study of two methods of rapid detection of clinically significant bacteriuria. J Clin Lab Auto 1983; 3: Washington JA, White CM, Laganiere M, Smith LH: Detection of significant bacteriuria by microscopic examination of urine. Lab Med 1981; 12: Wenk RE, Dutta D, Rudert J, Kim Y, Steinhagen C: Sediment microscopy, nitrituria and leukocyte esterasuria as predictors of significant bacteriuria. J Clin Lab Auto 1982; 2: