Detection of Pyuria and Bacteriuria in Symptomatic Ambulatory Women RAYMOND N. BLUM, MD, RICHARD A. WRIGHT, MD Objective: To compare the abilities o f two methods f o r rapid detection o f pyuria and three methods o f urinalysis to predict significant bacteriuria in symptomatic ambula. tory women. Design: Prospective simultaneous comparison o f the r e s u i t s o f dipstick urinalysis, standard microscopic urinalysis, and bemocytometric cell counting and Gram staining with the rest*Its o f a standard urine culture. Setting: Two outpatient ambulatory care faciliaes serving predominantly minority and uninsured individual& Selection criteria: Nonpregnant women presenting urith symptoms o f urinary tract infection without symptoms o f vaginal infection. Measurements and main results: 105 women with symptoms w e r e evaluated. The sensitivities o f the dipstick urinalysis and the microscopic urinalysis in predicting pyuria u s defi~ed by hemocytomet~ were 0.76and 0.77, respectively, and their specificities were 0.94 and 0.97, respectively. The sensitivities and specificities o f the three methods o f urinalysis in predicting >--10~ colony-forming u n i t s (CFU)/ntr~ in a urine ctd~re were 0.88 and O.70for the leukocyte esterase- nitrite dipstick urinalysis, 0.98 a n d 0.68for the standard microscopic urinalysis, and 1.00 a n d 0.49 f o r the Gram staining and hemocdaomeO~ c e l l counting. Conclusions: The standard urinalysis was the most accurate single method to predict significant bacteriurta in symptomatic ambulatory women~ Sequencing the dipstick urinalysis with the standard urinalysis may be a cost-effeclive approach to evaluating these patients in clinical practice. Key words: dipstick urinalysis; prediction; urinary tract infection; pyuria; bacteriuria; ambulatory care; women. J GEN INTERN MED 1992;7:140-144.
SYMPTOMS OF URINARYTRACTINFECTIONS(UTIs) remain a c o m m o n clinical p r o b l e m , causing an estimated 5 million physician visits annually. ~ Causes for these s y m p t o m s m a y include bacterial cystourethritis, pyelonephritis, and urethritis due to sexually transmitted pathogens such as C h l a m y d i a t r a c h o m a t i s , N e i s s e r i a g o n o r r h o e a e , T r i c h o m o n a s v a g i n a l i s , and herpes s i m p l e x virus, or there may be no infection at all. 2 The best m e t h o d to evaluate s y m p t o m s of a UTI remains unclear. Traditionally a culture of voided urine showing a quantitative growth of--> 105 colony-forming units (CFU)/mL was interpreted as positive. However, it has
Received from the Division of Infectious Diseases and the Department of Medicine, University of Colorado Health Sciences Center, and the Division of Public and Community Health Services, Denver Department of Health and Hospitals, Denver, Colorado. Address correspondence to Dr. Blum: 1721 East 19th Avenue, Suite 440, Denver, CO 80218. Address reprint requests to Dr. Wright: Public and Community Health Services, 605 Bannock Street, Denver, CO 80204.
140
b e c o m e clear that bacteriuria at a level as l o w as 102 CFU/mL is significant in certain groups such as symptomatic w o m e n . 3, s Preparing cultures is expensive and obtaining results takes time, offering little information to deal with the p r o b l e m at hand. Thus, a p r e s u m p t i v e diagnosis based on rapid diagnostic tests is important in managing this clinical p r o b l e m . It has b e e n suggested that to establish the p r e s e n c e of pyuria is p a r a m o u n t to diagnosing a true infection.4, s Evaluating for pyuria has received little attention in clinical practice. 5 Various methods have b e e n used, including calculating the leukocyte excretion rate, 6"8 p e r f o r m i n g a m i c r o s c o p i c analysis of a centrifuged s p e c i m e n (the so-called " s t a n d a r d " urinalysis), 9 counting cells in uncentrifuged urine using a hemocytometer, 1° and using a leukocyte esterase dipstick, it Calculating the leukocyte excretion rate is very timec o n s u m i n g and impractical for clinical practice. The standard m i c r o s c o p i c evaluation of urine has b e e n shown to be inconsistent and to correlate p o o r l y w i t h cell counting. 6, r, 12 Cell counting w i t h the use o f a hem o c y t o m e t e r has b e e n s h o w n to provide the most accurate measure o f pyuria, with > 1 0 white b l o o d cells ( W B C ) / m m 3 correlating w i t h an excretion rate > 4 0 0 , 0 0 0 leukocytes per hour. 12, 13 It is, however, time-consuming and it may not be routinely available. Recently, the use of leukocyte e s t e r a s e - nitrite ( L - N) chemical strips (dipsticks) has gained favor as a simple, rapid diagnostic p r e d i c t o r of b a c t e r i u r i a ) ~-17 These strips have s h o w n high degrees of sensitivity and specificity at predicting the p r e s e n c e o f - - 105 CFU/mL. However, they have p e r f o r m e d p o o r l y in predicting bacteriuria at levels of 102 to 103 CFU/mL, t8,~9 w i t h a sensitivity of less than 50%. Few prospective studies of the utilities of these various techniques have b e e n d o n e in s y m p t o m a t i c patients. We sought to c o m p a r e the p e r f o r m a n c e of the leukocyte e s t e r a s e - n i t r i t e dipstick with that of standard urinary m i c r o s c o p y in detecting pyuria and to assess the accuracies o f the three methods of urinalysis in predicting bacteriuria in s y m p t o m a t i c a m b u l a t o r y women.
METHODS Patients Two outpatient facilities o f the Denver Departm e n t of Health and Hospitals w e r e used as sources of patients for this study. The clinics serve as p r i m a r y care
JOURNALOFGENERALINTERNALMEDICINE,Volume 7 (March/April), 1992
facilities for primarily indigent and minority patients. Ambulatory, nonpregnant w o m e n more than 18 years old presenting to our clinics with the complaints of dysuria, frequency, urgency, and suprapubic pain were declared eligible. Patients were e x c l u d e d if they had complaints of vaginal discharge or vaginal bleeding. A midstream clean-catch urine sample was collected for each patient and immediately sent to the laboratory, w h e r e the sample was divided into three aliquots for analysis.
w e r e considered contaminated if more than two organisms were cultured in significant amounts. Growth of >- 104 CFU/mL was considered positive. Statistics
Sensitivities, specificities, and positive and negative predictive values were calculated according to Bayes' theorem of conditional probability.
RESULTS
Urinalysis Dipstick.
A dipstick analysis of the urine was performed using the Ames Multi Stix® IOSG (Miles Diagnostics, Elkhart, IN). The results of the leukocyte esterase test and the nitrite test were then recorded.
Cell Counting and Gram Staining. The uncentrifuged urine was examined for leukocytes in the hem o c y t o m e t e r and a WBC count was made. Results expressed as WBC/mL were recorded. In addition, a Gram stain of the uncentrifuged urine was prepared and the slide examined for bacteria and leukocytes. The presence of predominantly gram-negative rods (GNR) on the Gram-stained slide was considered a positive result. Standard Urinalysis. A standard microscopic urinalysis was done by centrifuging 10 mL of urine for five minutes at 2,000 rpm. Approximately 1 mL of urine was used to resuspend the sediment. The urine was then examined for bacteria, p o l y m o r p h o n u c l e a r cells (WBC), and red blood cells (RBC). Semiquantitative results were then recorded. Cutoffs of > 10 WBC/ high-power field (HPF), > 5 RBC/HPF, and either 2 + or 3 + bacteria were considered positive. The results of the various components of these tests were evaluated both individually and in various combinations. Sequencing.
To improve the sensitivity of certain tests, an evaluation of the effect of sequencing of the tests was performed. We defined sequencing in the following manner: If the result of the initial test was positive, then the final result was considered positive and the second test was not performed. If the result of the initial test was negative, then the s e q u e n c e d test was performed. If the result of this second, s e q u e n c e d test was positive, then the final result was considered positive.
Urine Cultures.
The urine sample was cultured by plating the urine on sheep blood agar and MacConkey agar with the use of a platinum (0.001-mL) loop. Cultures were incubated at 3 6 - 37 ° C in 5 - 10% CO2 for 1 8 - 24 hours. Identification of bacterial isolates was done by means of the API® Biochemical Profile (Analytab Products, Plainview, New York). Cultures
141
Specimens from 105 patients were evaluated. Culture results showed >-104 CFU/mL in 43 (41%), -- 10 WBC/mm 3 by hemocytometric counting of uncentrifuged urine. *Sequencingof dipstick urinalysisfollowed by microscopicurinalysis.
14P~
B/urn, Wright, URINALYSIS1N AMBULATORYWOMEN TABLE 3
Characteristicsof RapidDiagnosticTests for Bacteriuria* in Combination and in Sequence Predictive Value Positive Negative
Test
Sensitivity
Specificity
In combination Leukocyte esterasenitrite dipstick (L-N) Hemocytomerry/Gram staining Standard urinalysis
0.88
0.68
0.78
0.82
1.00
0.49
0.69
1.00
0.98
0.68
0.79
0.96
1.00
0.65
0.78
1.00
0.98
0.47
0.70
0.94
0.95
0.62
0.76
0.91
In sequence L- N/standard urinalysis L-N/ hemocytomerry L- N/Gram staining
*As determined by a urine culture with ->104 colony-forming units (CFU)/mL.
TABLE 4
Characteristics of the IndividualComponentsof the RapidDiagnostic Tests for Bacteriuria* Individual Component
Bacterialtests Nitrite dipstick urinalysis Urinalysis-bacteria (cutoff = 2+) Urinalysis-bacteria (cutoff = 3+) Gram staining Leukocytetests Leukocyte esterase dipstick urinalysis Urinalysis- WBCs Urinalysis-RBCs Hemocytometry (cutoff > 10 WBCs)
Predictive Value Positive Negative
Sensitivity
Specificity
0.37
0.97
0.94
0.55
0.70
0.85
0.86
0.69
0.37 0.81
0.97 0.88
0.94 0.90
0.55 0.79
0.88
0.71
0.79
0.83
0.95
0.71
0.80
0.92
0.44
0.88
0.83
0.56
0.99
0.53
0.72
0.95
counting of uncentrifuged urine. When the dipstick urinalysis was s e q u e n c e d with microscopic urinalysis of negative specimens, the sensitivity and specificity were 0.85 and 0.94 in predicting > 10 WBC/mm 3. The performances of the three methods of urinalysis in predicting a culture of ---104 CFU/mL are presented in Table 3. Each o f the tests is a sensitive predictor of a culture of->l 04 CFU/mL, ranging from 0.88 for the dipstick to 1.00 for the h e m o c y t o m e t e r / G r a m stain test. As might be expected, the more sensitive the test, the less specific it is. Sequencing dipstick urinalysis with microscopic urinalysis, hemocytometry, or Gram staining improves the sensitivity to at least 0.95. Each urinalysis m e t h o d is made u p o f individual components. We examined these c o m p o n e n t s for each of the methods (Table 4). They are divided into those components that measure pyuria and those that measure bacteriuria. All leukocyte-based tests p e r f o r m e d better than the bacteria-based tests in predicting bacteriuria. The test for red b l o o d cells is included with the pyuria tests because it is a marker for inflammation. The most sensitive test c o m p o n e n t s were the leukocyte esterase portion of dipstick urinalysis, microscopic urinalysis, and hemocytometry, with sensitivities of 0.88, 0.95, and 0.99, respectively. In relation to bacteria, the nitrite portion of dipstick urinalysis and urinary microscopy with a cutoff of 3 + bacteria p e r f o r m e d similarly, and both are specific predictors o f a positive culture. Gram staining was the most accurate of the tests for bacteria.
DISCUSSION
*As determined by a urine culture with ->104colony-forming units (CFU)/mL.
and the centrifuged urinalysis for leukocytes showed sensitivities of 0.76 and 0.77, respectively, and specificities of 0.94 and 0.97, respectively, in correlating with > 1 0 WBC/mm 3 as measured by h e m o c y t o m e t r i c
Symptoms of UTI are a c o m m o n cause for primary care visits. Traditionally, the " g o l d standard" for diagnosing a UTI has been the culturing of a voided specimen of urine. The presence of an abnormal n u m b e r of bacteria and leukocytes in urinary sediment has constituted a presumptive diagnosis of a UTI. The best m e t h o d for establishing this evidence has remained unclear. A rapid, simple, inexpensive, and accurate test w o u l d be ideal. We found no single test to be ideal, but through sequencing a leukocyte esterase dipstick urinalysis with a microscopic urinalysis for pyuria, an accurate, cost-effective test is available. Quantifying urinary isolates was the primary mechanism for distinguishing a " t r u e infection," as identified by 10 s CFU/mL o f urine, from contamination or no infection. The finding that a sizable n u m b e r of w o m e n with bacterial cystourethritis have c o l o n y counts between 102 and 105 CFU/mL as verified by analysis of urine specimens obtained from catheterized patients greatly increases the yields and sensitivities of these cultures. By not culturing for 1 0 2 CFU/mL is used to determine positive culture results. The performances of the three different methods of urinalysis in predicting bacteriuria in our study w e r e similar to those reported in studies of different populations.16.18.20, 21 As might be expected, as the sensitivity of a test increases, its specificity declines. This occurred with each of the three methods of urinalysis studied. Since this study examined symptomatic ambulatorywomen, it may provide insight into what may be a reasonable approach in the typical clinical practice. We found that the standard microscopic urinalysis performed relatively well. This is probably because these tests were done by the same technicians, improving the likelihood of consistent results. It has been shown that strict adherence to detail can improve the accuracy of this test. 22, 23 A sizable n u m b e r of urines had to be e x c l u d e d because their culture results suggested contamination. The results of the urinalysis of these specimens varied from normal to consistent with infection. Rather than possibly misclassifying them, we chose not to include them in our analysis. In real practice, it is likely that these patients w o u l d be treated based on the results of rapid diagnostic testing and the contaminated culture results w o u l d be known after the fact. Urine cultures add significantly to the cost of diagnosing UTIs. We did not assess their utility in this study. It has b e e n suggested 24, 25 that pretreatment urine cultures are unnecessary in carefully selected nonpregnant, nonelderly women. The purposes of a urine culture include confirming the diagnosis of a UTI and identifying resistant microorganisms. While this information is vital in managing patients w h o have complicated UTIs such as pyelonephritis, for several reasons
143
such information is of less importance in managing patients w h o have bacterial cystourethritis. Such patients rarely are infected with resistant organisms; available oral antimicrobial agents are efficacious in treating infections caused by a wide range of uropathogens; therapy for symptomatic UTIs is routinely initiated prior to the availability of culture results; and urine cultures do not identify the sites of the infections, thus are of little aid in determining the duration of therapy. Therefore, pre- and post-therapy urine cultures are unnecessary in managing episodes of simple bacterial cystourethritis in carefully selected nonpregnant, nonelderly women. With the elimination of urine cultures in this low-risk population, clinicians are left with initiating empiric therapy based either on symptoms alone or on some evidence of urologic inflammation or infection. There are several considerations in determining the best approach for a given clinical setting. These include the importance of not missing any patients with disease, the risk associated with treating patients unnecessarily, and the risk and costs associated with performing the test. The risk of missing a UTI is low. Although there is a finite risk of urosepsis and death, many episodes w o u l d resolve without therapy, and recurrences both with and without therapy are common. Likewise, the risk from therapy for every patient with symptoms is low, associated mainly with the risk of allergic reactions to antimicrobial agents and that of missing other diagnoses (e.g., sexually transmitted diseases). Based on our findings, a practical cost-effective approach to the ambulatory w o m a n presenting with UTI symptoms w o u l d be to begin with a dipstick urinalysis of the urine. If a positive result is obtained, the patient is likely to have pyuria, w h i c h will respond to treatment. Those with negative results w o u l d then undergo microscopic urinalysis and those with positive results could then be considered for treatment. Those with negative results to both tests could then have cultures done to determine w h e t h e r they have infections with < 1 0 4 CFU/mL and treatment withheld until these
TABLE S
Costs of Evaluating5ymptornatic Patientsfor Various Prevalencesof PositiveUrineCultures Prevalence
L- N*
Standard Urinalysist
L- N/Standard Urinalysis*
Per 1,000 patients§
25% 50% 75%
$366 $366 $366
$1,260 $1,260 $1,260
$1,071 $ 882 $ 706
Per correct diagnosis(falsenegatives/falsepositives)
25% 50% 75%
$0.49 (30/225) $0.46 (60/150) $0.44 (90/75)
$1.67 (5/240) $1.51 (10/160) $1.39 (15/80)
$1.45 (0.262) $1.07 (0/175) $0.77 (0/87)
*Leukocyte esterase/nitritedipstick. ~Microscopicurinalysisalone. SLeukocyteesterase/nitritedipstickanalysisfollowedby microscopicanalysisof negatives. §Cost is basedon personneltime of $0.36 for dipstick urinalysisand $1.26 for microscopicurinalysis.
Blum, Wright,
144
URINALYSIS IN AMBULATORYWOMEN
results are obtained. They should also b e carefully reevaluated to rule out other causes for their complaints, because it is unlikely they will have positive cultures for coliforms. 26 They may still have s y m p t o m s that will r e s p o n d to antimicrobial agents, and this g r o u p needs to b e further evaluated to h e l p define the causes of their symptoms. O t h e r considerations center around the v o l u m e of patients and the availability of laboratory resources. In o u r clinics, a p p r o a c h e s that can free u p technician t i m e to p e r f o r m other tasks are important. In an a t t e m p t to do this, an analysis of sequential testing was done. A cost analysis of this a p p r o a c h is presented in Table 5. Cost estimates for various tests are based on laboratory technician time and not the cost to the clinic of the supplies. They do not include the costs of appointments, obtaining specimens, or other laboratory supplies such as m i c r o s c o p e s or cell-counting chambers. The savings of s e q u e n c i n g these tests is evident, especially at a higher prevalence. There is also a minim u m of false negatives. The savings of technician time b y this a p p r o a c h may not be as important in o t h e r clinical settings. The subjects of this study w e r e symptomatic, ambulatory w o m e n . The ability to a p p l y these results to other patient p o p u l a t i o n s such as pregnant w o m e n , men, or a nonindigent p o p u l a t i o n is unclear. The numbers of the study are small. In addition, s o m e of the data analysis was done in a retrospective fashion. There is the potential for inconsistencies in s o m e of the m o r e subjective t e c h n i q u e s such as urinary m i c r o s c o p y and G r a m staining. Evaluation of the degrees of precision o f these tests was not done. Furthermore, the role of rapid diagnostic screening of infections w i t h less than 1 04 CFU/mL or w i t h noncoliforms such as Staphylococcus saprophyticus remains unclear.
CONCLUSION We found that sequencing the dipstick urinalysis w i t h a m i c r o s c o p i c urinalysis may be a cost-effective a p p r o a c h to the initial evaluation o f a m b u l a t o r y w o m e n s u s p e c t e d of having UTIs. Further evaluation needs to be done in a p r o s p e c t i v e fashion b o t h in other populations and with urine cultures of < 1 0 9 CFU/mL to validate this approach. The authors gratefully acknowledge for their technical assistance Mary McAvoy, MT (ASCP), and Elizabeth Ray, MT (ASCP).
REFERENCES 1. Koch HK. The National Ambulatory Medical Care Survey, Department of Health, Education, and Welfare. Hyattsville, MD: National Center for Health Statistics, 1978. 2. Komaroff AL. Acute dysuria in women. N Engl J Med. 1984;310:368-75. 3. Stamm WE, Counts GW, Running KR, Fihn S, Turck M, Holmes KK. Diagnosis of coliform infection in acutely dysuric women. N Engl J Med. 1982;307:463-8. 4. Stamm WE, Wagner KF, Amsel R, et al. Causes of the acute urethral syndrome in women. N Engl J ivied. 1980;303:409-15. 5. Stamm WE. Measurement of pyuria and its relation to bacteriuria. AmJ Med. 1983;75(suppl 1B):53-8. 6. Mabeck CE. Studies in urinary tract infections IV. Urinary leukocyte excretion in bacteriuria. Acta Med Scand. 1969; 186:193-8. 7. StansfieldJM. The measurement and meaning of pyuria. Arch Dis Child. 1982;37:257-62. 8. Little PJ. A comparison of the urinary white cell concentration with the white cell excretion rate. BrJ Urol. 1964;36:360-3. 9. Schumann GB, Schweitzer SC. Examination of urine. In: Henry JB, ed. Clinical diagnosis and management by laboratory methods, 18th ed. Philadelphia: W. B. Saunders, 1991;431. 10. Alwall N. Pyuria: deposit in high-power microscopic fieldWBC/ hpf versus WBC/mL in counting chamber. Acta Med Scand. 1973;194:537-40. 11. Kusumi RK, Grover PJ, Kunin CM. Rapid detection of pyuria by leukocyte esterase activity. JAMA. 1981 ;245:1653-5. 12. Brumtitt W. Urinary cell counts and their value. J Clin Pathol. 1965;18:550-5. 13. Gadeholt H. Quantitative estimation of cells in urine. Acta Med Scand. 1968;183:369-74. 14. Sewell DL, Butt SP, Gabbert NJ, Bumgardner RV. Evaluation of the Chemstrip 9 as a screening test for urinalysis and urine culture in men. AmJ Clin Pathol. 1985;83:740-3. 15. Wenk RE, Dutta D, Rudert J, Kim Y, Steinhagen C. Sediment microscopy, nitrituria, and leukocyte esterasuria as predictors of significant hacteriuria. J Clin Lab Automation. 1982;2:117-21. 16. PerryJL, MatthewsJS, Weesner DE. Evafuation of leukocyte esterase activity as a rapid screening technique for bacteriuria. J Clin Microbiol. 1982;15:852-4. 17. Wenz B, Lampasso JA. Eliminating unnecessary urine microscopy. AmJ Clin Pathol. 1989;92:78-81. 18. Murray PR, Smith TB, McKinney TC. Clinical evaluation of three urine screening tests. J Clin Microbiol. 1987;25:467-70. 19. Jones C. MacPherson DW, Stevens DL. Inability of the chemstrip LN compared with quantitative urine culture to predict significant bacteriuria. J Clin Microbiol. 1986;23:160-2. 20. Crout FV, Tilton RC. Rapid screening of urine for significant bacteriuria by Gram stain, acridine orange stain, and the Autobac Mrs system. Diagn Microbiol Infect Dis. 1984;2:179-86. 21. Jenkins RD, Fenn JP, Matsen JM. Review of urine microscopy for bacteriuria. JAMA. 1986;255:3397-403. 22. Winkel P, Statland BE, Jorgensen K. Urine microscopy, an ill-defined method examined by a muhifactorial technique. Clin Chem. 1974;20:436-9. 23. Becker SM, Ramirez G, Pribor HC, Gillen AL. A quality control product for urinalysis. AmJ Clin Pathol. 1973;59:185-91. 24. Schultz HJ, McCaffrey LA, Keys TF, Nobrega FT. Acute cystitis: a prospective study of laboratory tests and duration of therapy. Mayo Clin Proc. 1984;59:391-7. 25. Johnson JR, Stamm WE. Urinary tract infections in women: diagnosis and treatment. Ann Intern Med. 1989;111:906-17. 26. Komaroff AL. Urinalysis and urine culture in women with dysuria. Ann Intern Med. 1986;104:212-8.
SYMPTOMS OF URINARYTRACTINFECTIONS(UTIs) remain a c o m m o n clinical p r o b l e m , causing an estimated 5 million physician visits annually. ~ Causes for these s y m p t o m s m a y include bacterial cystourethritis, pyelonephritis, and urethritis due to sexually transmitted pathogens such as C h l a m y d i a t r a c h o m a t i s , N e i s s e r i a g o n o r r h o e a e , T r i c h o m o n a s v a g i n a l i s , and herpes s i m p l e x virus, or there may be no infection at all. 2 The best m e t h o d to evaluate s y m p t o m s of a UTI remains unclear. Traditionally a culture of voided urine showing a quantitative growth of--> 105 colony-forming units (CFU)/mL was interpreted as positive. However, it has
Received from the Division of Infectious Diseases and the Department of Medicine, University of Colorado Health Sciences Center, and the Division of Public and Community Health Services, Denver Department of Health and Hospitals, Denver, Colorado. Address correspondence to Dr. Blum: 1721 East 19th Avenue, Suite 440, Denver, CO 80218. Address reprint requests to Dr. Wright: Public and Community Health Services, 605 Bannock Street, Denver, CO 80204.
140
b e c o m e clear that bacteriuria at a level as l o w as 102 CFU/mL is significant in certain groups such as symptomatic w o m e n . 3, s Preparing cultures is expensive and obtaining results takes time, offering little information to deal with the p r o b l e m at hand. Thus, a p r e s u m p t i v e diagnosis based on rapid diagnostic tests is important in managing this clinical p r o b l e m . It has b e e n suggested that to establish the p r e s e n c e of pyuria is p a r a m o u n t to diagnosing a true infection.4, s Evaluating for pyuria has received little attention in clinical practice. 5 Various methods have b e e n used, including calculating the leukocyte excretion rate, 6"8 p e r f o r m i n g a m i c r o s c o p i c analysis of a centrifuged s p e c i m e n (the so-called " s t a n d a r d " urinalysis), 9 counting cells in uncentrifuged urine using a hemocytometer, 1° and using a leukocyte esterase dipstick, it Calculating the leukocyte excretion rate is very timec o n s u m i n g and impractical for clinical practice. The standard m i c r o s c o p i c evaluation of urine has b e e n shown to be inconsistent and to correlate p o o r l y w i t h cell counting. 6, r, 12 Cell counting w i t h the use o f a hem o c y t o m e t e r has b e e n s h o w n to provide the most accurate measure o f pyuria, with > 1 0 white b l o o d cells ( W B C ) / m m 3 correlating w i t h an excretion rate > 4 0 0 , 0 0 0 leukocytes per hour. 12, 13 It is, however, time-consuming and it may not be routinely available. Recently, the use of leukocyte e s t e r a s e - nitrite ( L - N) chemical strips (dipsticks) has gained favor as a simple, rapid diagnostic p r e d i c t o r of b a c t e r i u r i a ) ~-17 These strips have s h o w n high degrees of sensitivity and specificity at predicting the p r e s e n c e o f - - 105 CFU/mL. However, they have p e r f o r m e d p o o r l y in predicting bacteriuria at levels of 102 to 103 CFU/mL, t8,~9 w i t h a sensitivity of less than 50%. Few prospective studies of the utilities of these various techniques have b e e n d o n e in s y m p t o m a t i c patients. We sought to c o m p a r e the p e r f o r m a n c e of the leukocyte e s t e r a s e - n i t r i t e dipstick with that of standard urinary m i c r o s c o p y in detecting pyuria and to assess the accuracies o f the three methods of urinalysis in predicting bacteriuria in s y m p t o m a t i c a m b u l a t o r y women.
METHODS Patients Two outpatient facilities o f the Denver Departm e n t of Health and Hospitals w e r e used as sources of patients for this study. The clinics serve as p r i m a r y care
JOURNALOFGENERALINTERNALMEDICINE,Volume 7 (March/April), 1992
facilities for primarily indigent and minority patients. Ambulatory, nonpregnant w o m e n more than 18 years old presenting to our clinics with the complaints of dysuria, frequency, urgency, and suprapubic pain were declared eligible. Patients were e x c l u d e d if they had complaints of vaginal discharge or vaginal bleeding. A midstream clean-catch urine sample was collected for each patient and immediately sent to the laboratory, w h e r e the sample was divided into three aliquots for analysis.
w e r e considered contaminated if more than two organisms were cultured in significant amounts. Growth of >- 104 CFU/mL was considered positive. Statistics
Sensitivities, specificities, and positive and negative predictive values were calculated according to Bayes' theorem of conditional probability.
RESULTS
Urinalysis Dipstick.
A dipstick analysis of the urine was performed using the Ames Multi Stix® IOSG (Miles Diagnostics, Elkhart, IN). The results of the leukocyte esterase test and the nitrite test were then recorded.
Cell Counting and Gram Staining. The uncentrifuged urine was examined for leukocytes in the hem o c y t o m e t e r and a WBC count was made. Results expressed as WBC/mL were recorded. In addition, a Gram stain of the uncentrifuged urine was prepared and the slide examined for bacteria and leukocytes. The presence of predominantly gram-negative rods (GNR) on the Gram-stained slide was considered a positive result. Standard Urinalysis. A standard microscopic urinalysis was done by centrifuging 10 mL of urine for five minutes at 2,000 rpm. Approximately 1 mL of urine was used to resuspend the sediment. The urine was then examined for bacteria, p o l y m o r p h o n u c l e a r cells (WBC), and red blood cells (RBC). Semiquantitative results were then recorded. Cutoffs of > 10 WBC/ high-power field (HPF), > 5 RBC/HPF, and either 2 + or 3 + bacteria were considered positive. The results of the various components of these tests were evaluated both individually and in various combinations. Sequencing.
To improve the sensitivity of certain tests, an evaluation of the effect of sequencing of the tests was performed. We defined sequencing in the following manner: If the result of the initial test was positive, then the final result was considered positive and the second test was not performed. If the result of the initial test was negative, then the s e q u e n c e d test was performed. If the result of this second, s e q u e n c e d test was positive, then the final result was considered positive.
Urine Cultures.
The urine sample was cultured by plating the urine on sheep blood agar and MacConkey agar with the use of a platinum (0.001-mL) loop. Cultures were incubated at 3 6 - 37 ° C in 5 - 10% CO2 for 1 8 - 24 hours. Identification of bacterial isolates was done by means of the API® Biochemical Profile (Analytab Products, Plainview, New York). Cultures
141
Specimens from 105 patients were evaluated. Culture results showed >-104 CFU/mL in 43 (41%), -- 10 WBC/mm 3 by hemocytometric counting of uncentrifuged urine. *Sequencingof dipstick urinalysisfollowed by microscopicurinalysis.
14P~
B/urn, Wright, URINALYSIS1N AMBULATORYWOMEN TABLE 3
Characteristicsof RapidDiagnosticTests for Bacteriuria* in Combination and in Sequence Predictive Value Positive Negative
Test
Sensitivity
Specificity
In combination Leukocyte esterasenitrite dipstick (L-N) Hemocytomerry/Gram staining Standard urinalysis
0.88
0.68
0.78
0.82
1.00
0.49
0.69
1.00
0.98
0.68
0.79
0.96
1.00
0.65
0.78
1.00
0.98
0.47
0.70
0.94
0.95
0.62
0.76
0.91
In sequence L- N/standard urinalysis L-N/ hemocytomerry L- N/Gram staining
*As determined by a urine culture with ->104 colony-forming units (CFU)/mL.
TABLE 4
Characteristics of the IndividualComponentsof the RapidDiagnostic Tests for Bacteriuria* Individual Component
Bacterialtests Nitrite dipstick urinalysis Urinalysis-bacteria (cutoff = 2+) Urinalysis-bacteria (cutoff = 3+) Gram staining Leukocytetests Leukocyte esterase dipstick urinalysis Urinalysis- WBCs Urinalysis-RBCs Hemocytometry (cutoff > 10 WBCs)
Predictive Value Positive Negative
Sensitivity
Specificity
0.37
0.97
0.94
0.55
0.70
0.85
0.86
0.69
0.37 0.81
0.97 0.88
0.94 0.90
0.55 0.79
0.88
0.71
0.79
0.83
0.95
0.71
0.80
0.92
0.44
0.88
0.83
0.56
0.99
0.53
0.72
0.95
counting of uncentrifuged urine. When the dipstick urinalysis was s e q u e n c e d with microscopic urinalysis of negative specimens, the sensitivity and specificity were 0.85 and 0.94 in predicting > 10 WBC/mm 3. The performances of the three methods of urinalysis in predicting a culture of ---104 CFU/mL are presented in Table 3. Each o f the tests is a sensitive predictor of a culture of->l 04 CFU/mL, ranging from 0.88 for the dipstick to 1.00 for the h e m o c y t o m e t e r / G r a m stain test. As might be expected, the more sensitive the test, the less specific it is. Sequencing dipstick urinalysis with microscopic urinalysis, hemocytometry, or Gram staining improves the sensitivity to at least 0.95. Each urinalysis m e t h o d is made u p o f individual components. We examined these c o m p o n e n t s for each of the methods (Table 4). They are divided into those components that measure pyuria and those that measure bacteriuria. All leukocyte-based tests p e r f o r m e d better than the bacteria-based tests in predicting bacteriuria. The test for red b l o o d cells is included with the pyuria tests because it is a marker for inflammation. The most sensitive test c o m p o n e n t s were the leukocyte esterase portion of dipstick urinalysis, microscopic urinalysis, and hemocytometry, with sensitivities of 0.88, 0.95, and 0.99, respectively. In relation to bacteria, the nitrite portion of dipstick urinalysis and urinary microscopy with a cutoff of 3 + bacteria p e r f o r m e d similarly, and both are specific predictors o f a positive culture. Gram staining was the most accurate of the tests for bacteria.
DISCUSSION
*As determined by a urine culture with ->104colony-forming units (CFU)/mL.
and the centrifuged urinalysis for leukocytes showed sensitivities of 0.76 and 0.77, respectively, and specificities of 0.94 and 0.97, respectively, in correlating with > 1 0 WBC/mm 3 as measured by h e m o c y t o m e t r i c
Symptoms of UTI are a c o m m o n cause for primary care visits. Traditionally, the " g o l d standard" for diagnosing a UTI has been the culturing of a voided specimen of urine. The presence of an abnormal n u m b e r of bacteria and leukocytes in urinary sediment has constituted a presumptive diagnosis of a UTI. The best m e t h o d for establishing this evidence has remained unclear. A rapid, simple, inexpensive, and accurate test w o u l d be ideal. We found no single test to be ideal, but through sequencing a leukocyte esterase dipstick urinalysis with a microscopic urinalysis for pyuria, an accurate, cost-effective test is available. Quantifying urinary isolates was the primary mechanism for distinguishing a " t r u e infection," as identified by 10 s CFU/mL o f urine, from contamination or no infection. The finding that a sizable n u m b e r of w o m e n with bacterial cystourethritis have c o l o n y counts between 102 and 105 CFU/mL as verified by analysis of urine specimens obtained from catheterized patients greatly increases the yields and sensitivities of these cultures. By not culturing for 1 0 2 CFU/mL is used to determine positive culture results. The performances of the three different methods of urinalysis in predicting bacteriuria in our study w e r e similar to those reported in studies of different populations.16.18.20, 21 As might be expected, as the sensitivity of a test increases, its specificity declines. This occurred with each of the three methods of urinalysis studied. Since this study examined symptomatic ambulatorywomen, it may provide insight into what may be a reasonable approach in the typical clinical practice. We found that the standard microscopic urinalysis performed relatively well. This is probably because these tests were done by the same technicians, improving the likelihood of consistent results. It has been shown that strict adherence to detail can improve the accuracy of this test. 22, 23 A sizable n u m b e r of urines had to be e x c l u d e d because their culture results suggested contamination. The results of the urinalysis of these specimens varied from normal to consistent with infection. Rather than possibly misclassifying them, we chose not to include them in our analysis. In real practice, it is likely that these patients w o u l d be treated based on the results of rapid diagnostic testing and the contaminated culture results w o u l d be known after the fact. Urine cultures add significantly to the cost of diagnosing UTIs. We did not assess their utility in this study. It has b e e n suggested 24, 25 that pretreatment urine cultures are unnecessary in carefully selected nonpregnant, nonelderly women. The purposes of a urine culture include confirming the diagnosis of a UTI and identifying resistant microorganisms. While this information is vital in managing patients w h o have complicated UTIs such as pyelonephritis, for several reasons
143
such information is of less importance in managing patients w h o have bacterial cystourethritis. Such patients rarely are infected with resistant organisms; available oral antimicrobial agents are efficacious in treating infections caused by a wide range of uropathogens; therapy for symptomatic UTIs is routinely initiated prior to the availability of culture results; and urine cultures do not identify the sites of the infections, thus are of little aid in determining the duration of therapy. Therefore, pre- and post-therapy urine cultures are unnecessary in managing episodes of simple bacterial cystourethritis in carefully selected nonpregnant, nonelderly women. With the elimination of urine cultures in this low-risk population, clinicians are left with initiating empiric therapy based either on symptoms alone or on some evidence of urologic inflammation or infection. There are several considerations in determining the best approach for a given clinical setting. These include the importance of not missing any patients with disease, the risk associated with treating patients unnecessarily, and the risk and costs associated with performing the test. The risk of missing a UTI is low. Although there is a finite risk of urosepsis and death, many episodes w o u l d resolve without therapy, and recurrences both with and without therapy are common. Likewise, the risk from therapy for every patient with symptoms is low, associated mainly with the risk of allergic reactions to antimicrobial agents and that of missing other diagnoses (e.g., sexually transmitted diseases). Based on our findings, a practical cost-effective approach to the ambulatory w o m a n presenting with UTI symptoms w o u l d be to begin with a dipstick urinalysis of the urine. If a positive result is obtained, the patient is likely to have pyuria, w h i c h will respond to treatment. Those with negative results w o u l d then undergo microscopic urinalysis and those with positive results could then be considered for treatment. Those with negative results to both tests could then have cultures done to determine w h e t h e r they have infections with < 1 0 4 CFU/mL and treatment withheld until these
TABLE S
Costs of Evaluating5ymptornatic Patientsfor Various Prevalencesof PositiveUrineCultures Prevalence
L- N*
Standard Urinalysist
L- N/Standard Urinalysis*
Per 1,000 patients§
25% 50% 75%
$366 $366 $366
$1,260 $1,260 $1,260
$1,071 $ 882 $ 706
Per correct diagnosis(falsenegatives/falsepositives)
25% 50% 75%
$0.49 (30/225) $0.46 (60/150) $0.44 (90/75)
$1.67 (5/240) $1.51 (10/160) $1.39 (15/80)
$1.45 (0.262) $1.07 (0/175) $0.77 (0/87)
*Leukocyte esterase/nitritedipstick. ~Microscopicurinalysisalone. SLeukocyteesterase/nitritedipstickanalysisfollowedby microscopicanalysisof negatives. §Cost is basedon personneltime of $0.36 for dipstick urinalysisand $1.26 for microscopicurinalysis.
Blum, Wright,
144
URINALYSIS IN AMBULATORYWOMEN
results are obtained. They should also b e carefully reevaluated to rule out other causes for their complaints, because it is unlikely they will have positive cultures for coliforms. 26 They may still have s y m p t o m s that will r e s p o n d to antimicrobial agents, and this g r o u p needs to b e further evaluated to h e l p define the causes of their symptoms. O t h e r considerations center around the v o l u m e of patients and the availability of laboratory resources. In o u r clinics, a p p r o a c h e s that can free u p technician t i m e to p e r f o r m other tasks are important. In an a t t e m p t to do this, an analysis of sequential testing was done. A cost analysis of this a p p r o a c h is presented in Table 5. Cost estimates for various tests are based on laboratory technician time and not the cost to the clinic of the supplies. They do not include the costs of appointments, obtaining specimens, or other laboratory supplies such as m i c r o s c o p e s or cell-counting chambers. The savings of s e q u e n c i n g these tests is evident, especially at a higher prevalence. There is also a minim u m of false negatives. The savings of technician time b y this a p p r o a c h may not be as important in o t h e r clinical settings. The subjects of this study w e r e symptomatic, ambulatory w o m e n . The ability to a p p l y these results to other patient p o p u l a t i o n s such as pregnant w o m e n , men, or a nonindigent p o p u l a t i o n is unclear. The numbers of the study are small. In addition, s o m e of the data analysis was done in a retrospective fashion. There is the potential for inconsistencies in s o m e of the m o r e subjective t e c h n i q u e s such as urinary m i c r o s c o p y and G r a m staining. Evaluation of the degrees of precision o f these tests was not done. Furthermore, the role of rapid diagnostic screening of infections w i t h less than 1 04 CFU/mL or w i t h noncoliforms such as Staphylococcus saprophyticus remains unclear.
CONCLUSION We found that sequencing the dipstick urinalysis w i t h a m i c r o s c o p i c urinalysis may be a cost-effective a p p r o a c h to the initial evaluation o f a m b u l a t o r y w o m e n s u s p e c t e d of having UTIs. Further evaluation needs to be done in a p r o s p e c t i v e fashion b o t h in other populations and with urine cultures of < 1 0 9 CFU/mL to validate this approach. The authors gratefully acknowledge for their technical assistance Mary McAvoy, MT (ASCP), and Elizabeth Ray, MT (ASCP).
REFERENCES 1. Koch HK. The National Ambulatory Medical Care Survey, Department of Health, Education, and Welfare. Hyattsville, MD: National Center for Health Statistics, 1978. 2. Komaroff AL. Acute dysuria in women. N Engl J Med. 1984;310:368-75. 3. Stamm WE, Counts GW, Running KR, Fihn S, Turck M, Holmes KK. Diagnosis of coliform infection in acutely dysuric women. N Engl J Med. 1982;307:463-8. 4. Stamm WE, Wagner KF, Amsel R, et al. Causes of the acute urethral syndrome in women. N Engl J ivied. 1980;303:409-15. 5. Stamm WE. Measurement of pyuria and its relation to bacteriuria. AmJ Med. 1983;75(suppl 1B):53-8. 6. Mabeck CE. Studies in urinary tract infections IV. Urinary leukocyte excretion in bacteriuria. Acta Med Scand. 1969; 186:193-8. 7. StansfieldJM. The measurement and meaning of pyuria. Arch Dis Child. 1982;37:257-62. 8. Little PJ. A comparison of the urinary white cell concentration with the white cell excretion rate. BrJ Urol. 1964;36:360-3. 9. Schumann GB, Schweitzer SC. Examination of urine. In: Henry JB, ed. Clinical diagnosis and management by laboratory methods, 18th ed. Philadelphia: W. B. Saunders, 1991;431. 10. Alwall N. Pyuria: deposit in high-power microscopic fieldWBC/ hpf versus WBC/mL in counting chamber. Acta Med Scand. 1973;194:537-40. 11. Kusumi RK, Grover PJ, Kunin CM. Rapid detection of pyuria by leukocyte esterase activity. JAMA. 1981 ;245:1653-5. 12. Brumtitt W. Urinary cell counts and their value. J Clin Pathol. 1965;18:550-5. 13. Gadeholt H. Quantitative estimation of cells in urine. Acta Med Scand. 1968;183:369-74. 14. Sewell DL, Butt SP, Gabbert NJ, Bumgardner RV. Evaluation of the Chemstrip 9 as a screening test for urinalysis and urine culture in men. AmJ Clin Pathol. 1985;83:740-3. 15. Wenk RE, Dutta D, Rudert J, Kim Y, Steinhagen C. Sediment microscopy, nitrituria, and leukocyte esterasuria as predictors of significant hacteriuria. J Clin Lab Automation. 1982;2:117-21. 16. PerryJL, MatthewsJS, Weesner DE. Evafuation of leukocyte esterase activity as a rapid screening technique for bacteriuria. J Clin Microbiol. 1982;15:852-4. 17. Wenz B, Lampasso JA. Eliminating unnecessary urine microscopy. AmJ Clin Pathol. 1989;92:78-81. 18. Murray PR, Smith TB, McKinney TC. Clinical evaluation of three urine screening tests. J Clin Microbiol. 1987;25:467-70. 19. Jones C. MacPherson DW, Stevens DL. Inability of the chemstrip LN compared with quantitative urine culture to predict significant bacteriuria. J Clin Microbiol. 1986;23:160-2. 20. Crout FV, Tilton RC. Rapid screening of urine for significant bacteriuria by Gram stain, acridine orange stain, and the Autobac Mrs system. Diagn Microbiol Infect Dis. 1984;2:179-86. 21. Jenkins RD, Fenn JP, Matsen JM. Review of urine microscopy for bacteriuria. JAMA. 1986;255:3397-403. 22. Winkel P, Statland BE, Jorgensen K. Urine microscopy, an ill-defined method examined by a muhifactorial technique. Clin Chem. 1974;20:436-9. 23. Becker SM, Ramirez G, Pribor HC, Gillen AL. A quality control product for urinalysis. AmJ Clin Pathol. 1973;59:185-91. 24. Schultz HJ, McCaffrey LA, Keys TF, Nobrega FT. Acute cystitis: a prospective study of laboratory tests and duration of therapy. Mayo Clin Proc. 1984;59:391-7. 25. Johnson JR, Stamm WE. Urinary tract infections in women: diagnosis and treatment. Ann Intern Med. 1989;111:906-17. 26. Komaroff AL. Urinalysis and urine culture in women with dysuria. Ann Intern Med. 1986;104:212-8.
