224
SAMT
VOL 74
3 SEPT 1988
The authors wish to thank the Medical Superintendent of Tygerberg Hospital for permission to publish.
Discussion A previous survey of blood culmre results reported a low incidence of methicillin-resistant S. aureus at Tygerberg Hospital in comparison with centres in TransvaaJ.3 Our results indicate that when culmres are separated into community- and hospital-acquired groups the incidence is in fact comparable to that reported from other South Mrican centres. Despite the high incidence of methicillin resistance among S. aureus hospital isolates, the use of methicillin or cloxacillin still seems to remain appropriate for community-acquired infections.
REFERENCES 1. Working Party of the Hospital Infection Society and British Society for
Antimicrobial Cbemotherapy. Guidelines for the control of epidemic methicillin-resistant Sraphylococcus aureus. J Hosp Infecc 1986; 7: 193-210. 2. Stokes EJ. Clinical Bacreriology. London: Edward Annold, 1975: 203-261. 3. Van Den Ende J, Rotter ML. An analysis of blood culture isolates from 7 South African teaching hospital centres. S Afr Med J 1986; 69: 89-93.
• Dipstick screening for urInary tract infection J. WIGGELlNKHUIZEN,
D. MAYTHAM,
D. H. HANSLO
Summary In screening for urinary tract infection the leucocyte esterase test will detect almost all samples with significant pyuria and bacteriuria, but is relatively nonspecific. The nitrite test is more specific but less sensitive and about one-third of the urinary tract infections in a large group of children were missed. The combination of screening tests results in greater overall accuracy both in the diagnosis and exclusion of urinary tract infection. Almost all cases of urinary tract infection were detected when either the leucocyte esterase or the nitrite screening test or both were positive. If both tests are negative, urinary tract infection is virtually excluded and unless the child is symptomatic, further urinalysis is unnecessary. Laboratory urinalysis is, however, necessary if anyone screening test for leucocyte esterase or nitrite (or protein or haemoglobin) is positive. Combined biochemical screening for urinary tract infection with dipstick test strips is reliable and allows early diagnosis and management. By avoiding unnecessary urinalysis it is cost-effective for the patient and will significantly reduce the laboratory workload. S AIr Med J 1988; 74: 224-228.
Urinary tract infection is common in infancy and childhood and may indicate underlying strucmral or functional uropathy requiring further management. The diagnosis of urinary tract infection is usually confirmed by microscopy and culmre of a properly collected urine sample. Reliable screening tests for urinary tract infection facilitate early diagnosis and treatment, and if negative may avoid unnecessary laboratory urinalysis.
Renal Unit and Department of Microbiology, University of Cape Town and Red Cross War Memorial Children's Hospital, Cape Town
J. WIGGELINKHUIZEN, M.B. B.CH., M.MED. (PAED.), FCP. (SA) D. MAYTHAM, M.B. CH.B., D.CH. (SA) D. H. HANSLO, M.B. CH.B., M.MED. (PATH.), PATH.
Accepted 11 Apt 1988.
F.F.PATH. (SA), M.R.C
Neutrophil granulocytes contain several esterases which are not normally present in serum, urine or kidney tissue, while most urinary pathogens reduce nitrate present in urine to nitrite. The use of biochemical marker test strips for leucocYte esterase for pyuria, and for nitrite for bacteriuria, proteinuria and haemamria, is evaluated as a rapid screening method in the diagnosis or exclusion of urinary tract infection.
Materials and methods Random urine samples from 1137 children attending medical casualty and outpatient follow-up clinics were screened. The samples were obtained by clean-catch midstream or urine/ ostomy bag collection, bladder catheterisation (at cystourethrography) or suprapubic aspiration. The presence or absence of signs and symptoms of urinary tract infection and antimicrobial therapy was recorded. Screening urinalysis at the bedside was done with both the Combur 9 test (Boehringer Mannheim) and Multistix 9 (Ames Bayer-Miles) dipsticks. Quantitative microscopy of the unstained specimen was performed using the Fuchs-Rosenthal counting chamber. Ten or more pus cells//-Ll (unspun) was considered significant leucocyturia. Routine laboratory urinalysis entailed microscopy of the spun sediment (5 ml at 1500 rpm for 2 min) and semiquantitative culmre using Bacteruritest (Mast Laboratories) filters trip imprints on cystine-lactose-electrolyte-deficient agar (Oxoid CM 423). Isolates were determined by conventional methods, and antimicrobial sensitivity was determined by the Stokes disc diffusion method. The samples were stored at 4°C until they were batch-processed, usually within 1-4 hours of voiding. Significant bacteriuria was defined as > 100000 colonyforming units of one predominant organism/ml urine. Secondary organisms were accepted only if they occurred in concentrations < 10 OOO/tnl; higher counts of three or more organisms were regarded as contaminants and reported as mixed growths. Lower counts were considered significant in samples obtained by bladder catheterisation or aspiration. Significant pyuria was considered present if> 10 white blood cells/high-power field (X 400) were seen. Urinalysis in the laboratory was performed without knowledge of the test strip findings.
SAMJ
The sensltlvlty, specificity and predictive values for each test individually and in combination were calculated.! *
Results In all I 137 urine samples were examined. T~bles I - III detail the age and sex distribution of the patients, the bacteriological findings on urinalysis and the nitrite reaction according to the infecting organism. Significant bacteriuria was found in 153 samples (13,5%). The frequency rate of positive cultures in the randomly collected samples does not necessarily reflect the prevalence of urinary tract infection in this hospital population. Mixed growths (15,4%) were excluded in the calculation of the sensitivity, specificity and predictive values of the tests. The leucocyte esterase reaction was highly sensitive (97,7% and 100%; the first percentage refers to the Combur test and the second to the Mulristix 9 test strip throughout the text) for the presence of white cells on microscopy, but nonspecific * Sensitivity - proportion of true positives labelled as positive. proportion of true negatives labelled as negative. Predictive value - proponion of positive (negative) tests that are true positives (negatives). Accuracy (efficiency) - percentage of all results that are true resulrs whether positive or negative. Specificiry -
VOL 74
3 SEPT 1988
225
(64,5% and 68,1%) with a high number of false-positives (Table IV). The positive predictive value was poor, but pyuria was almost always absent when the esterase reaction was negative (negative predictive value 99,5% and 100%). The low specificity and positive predictive value improved to 86,6% and 57,3% respectively when trace or 1+ positive reactions were read as negative. Fewer samples showed pyuria on laboratory urinalysis than on immediate microscopy of the unspun sample. Positive esterase reactions were found in 30,5% of the boys and 54,5% of the girls although the positive culture rates were similar at 14,3% and 17,7% respectively. Most samples with positive cultures showed positive leucocyte esterase tests (94,1% and 93,4%) with few false-negatives. Specificity was relatively low (68,9% and 69,3%) and a positive test was a poor predictor (36,4% and 36,6%) for the presence of significant bacteriuria. A negative test excluded urinary tract infection in > 98% of instances (Tables V and VI). The nitrite test was highly specific (98,9% and 98,6%) for significant bacteriuria, but relatively insensitive, with only 62,7% of infected urine samples positive for nitrite. The positive and negative predictive values of the nitrite test were ;:, 90% (Tables V and VI). In each age group the proponion of positive nitrite tests correlated with the frequency rate of urinary tract infections (Table I).
TABLE I. AGE AND SEX DISTRIBUTION AND URINE CULTURE FINDINGS Boy_s Age (yrs)
0-2 2-4 4-6 6-8 8-10 10 -12 > 12 Total
Girls
No. of patients
Pos. culture
Neg. culture
Mixed growths
No. of patients
Pos. culture
233 118 76 61 35 46 46
44 9 2 4 3 5 6
113 86 70 57 31 40 39 436
76 23 4 0 1 1 1 106
139 109 63 82 60 35 34 522
40 15 7 7 7 0 4
rn
73
Neg. culture
58 84 49 72
51 33 26 373
80
Mixed growths
Total positive nitrite tests
41 10 7 3 2 2 4
90 27 9 9 11 1 12 159
69
TABLE 11. URINE SAMPLES AND CULTURES
Clean-catch midstream Bag samples Suprapubic aspirates Ostomy samples Catheter samples Total
Positive culture
Negative culture
Mixed growths
Total
60 82 1 3 7
567 189 5 11 37 809
41 (6,1%) 129 (32,3%) 0 5 0 175
668 400 6 19 44 1137
153
TABLE Ill. BACTERIOLOGICAL FINDINGS AND NITRITE REACTIONS Boys (N==615)
E. coli Klebsiella spp Pseudomonas spp Enterococci Proteusspp S. aureus Beta haem. streptococci Total
(N
Girls == 522)
Total
Nitrite test Pos. Neg.
56
72
128
84
44
5
4
1 1
1
9 2 3 9
5 1
4 1 3
9 1
o 73
2
o o
o 5
4
1
1 1
1 1
0 0
80
153
97
56
226
SAMT VOL. 74 3 SEPT 1988
TABLE IV. LEUCOCYTE ESTERASE AND MICROSCOPIC EXAMINATION (%) Leucocyturia on microscopy Corn bur 9 test Multistix 9 Clinic Laboratory Clinic Laboratory Sensitivity Specificity Predictive values Positive test Negative test
100 68
99,2 64,5
98,3 68,1
97,7 64,8
37 100
26,5 99,7
36,7 99,5
26,6 99,5
TABLE V. COMBUR 9 TEST Cultures Pos. Neg. Leucocyte esterase Nitrite Leucocyte esterase or nitrite Leucocyte esterase and nitrite Leucocyte esterase or nitrite· or protein Leucocyte esterase and nitrite and protein
Sensitivity (%)
Specificity (%)
Predictive values (%) Pos. Neg.
Pas. Neg. Pas. Neg. Pas. Neg.
144 9 96 57 146 7
252 557 9 800 253 556
·94,1
68,9
36,4
98,4
62,7
98,9
91,4
93,3
95,4
68,7
36,6
98,8
Pas. Neg.
94 7
8 556
93,1
98,6
92,2
98,8
Pas. Neg.
150 3
372 437
98,0
54,0
28,7
99,3
Pas. Neg.
68 3
4 437
95,8
99,1
94,4
99,3
TABLE VI. MULTISTlX 9 TEST Cultures Pos. Neg. Leucocyte esterase Nitrite Leucocyte esterase or nitrite Leucocyte esterase and nitrite Leucocyte esterase or nitrite or protein Leucocyte esterase and nitrite and protein
Predictive values (%) Pos. Neg.
Sensitivity (%)
Speci/icity (%)
93,4
69,3
36,6
98,2
62,7
98,6
89,7
93,3
94,1
69,2
36,6
98,4
Pas. Neg. Pas. Neg. Pas. Neg.
143 10 96 57 144 9
248 561 11 798 249 560
Pas. Neg.
95 9
10 560
91,3
98,2
90,5
98,4
Pas. Neg.
150 3
418 391
98,0
48,3
26,4
99,2
Pos. Neg.
80 3
5 391
96,4
98,7
94,1
99,2
About 95% of samples with significant bacteriuria were detected when either the leucocyte esterase or the nitrite test or both were positive, but specificity remained low (68,7% and 69,2%). Almost 99% of samples did not have significant bacteriuria when both tests were negative (Tables V and VI).
When the two tests were both positive, specificity (98,6% and 98,2%) and positive predictive value (92,2% and 90,5%) were much improved with a slight loss of sensitivity and no change in negative predictive value. The overall accuracy of the two tests combined Wa\5 > 97% (Tables V and VI).
SAMJ
Sensitivity was greatest (98%) when one or more of the tests for leucocyte esterase, nitrite and protein were positive. Only 3 of the 153 urinary tract infections (1,96%) were missed with this three-test combination (Tables V and VI). Specificity, the predictive values and overall accuracy were greatest when the three tests were either all positive or all negative (Tables V and VI). Inclusion of the test for haematuria in the screen did not further improve detection of urinary tract infection.
Discussion The essential requirements of screening tests for urinary tract infection are high sensitivity so that the diagnosis is not missed and high specificity in order to avoid overdiagnosis. The predictive value of a positive or negative test should be close to 100% so that the question of whether the patient does or does not have a urinary tract infection is reliably answered. I Microscopy of a fresh uncentrifuged sample for leucocytes and bacteria affords a satisfactory method for preselecting urine samples for culture, 2 but requires time, a trained observer and. appropriate facilities. The test strip reaction for leucocyte esterase is specific for detecting the esterase activity of neutrophil granulocytes and results in a colour change if > 10 - 25 white blood cells/,.Ll are present. Both intact and lysed white cells are detected. In this study the new 1- 2-minute leucocyte esterase test was very sensitive in the detection of leucocyturia and almost all samples with pyuria were detected. The test, however, was nonspecific with many false-positives due mainly to trace or 1+ positive reactions. The positive predictive value was low but a negative esterase test excluded pyuria (Table IV). Some false-positive esterase tests may be due to lysis 3 of the leucocytes with release of esterase but negative microscopy.4 Reported studies in adults using the esterase test for the detection of leucocyturia indicate sensitivity levels of 82,495,3% and specificity levels of 71,4 _98%.4-8 Quantitative microscopy on unspun urine was more reliable than the semiquantitative examination of the urine sediment (Table IV).7 The leucocyte esterase test was as efficient as microscopy of the Gram-stained smear for detecting the presence of leucocytes in uncentrifuged urine. 9 Pyuria reflects an inflammatory response within the urinary tract which is most often caused by infection, but sterile pyuria does occur in several other conditions. Over 93% of samples with significant bacteriuria were detected by positive esterase tests, but a large number of false-positives (mistaken diagnosis of urinary tract infection) were found. The negative predictive value, however, was high and a negative leucocyte esterase test correctly indicated a negative culture in > 98% of samples (Tables V and VI). Specificity and the positive predictive value improved to 96,3% and 68,4% respectively when only the maximum reaction on the leucocyte esterase test strip was considered positive. The higher frequency of leucocyturia in girls, due to a greater number of moderate- (false-) positive reactions, is mostly due to contamination with vaginal secretions and makes the leucocyte esterase test on its own an imprecise screening test for urinary tract infection. lo Urinary tract infection may occur with < 100000 organisms/ml and will be detected by the esterase test if pyuria results. 9 When the leucocyte esterase test was evaluated separately in samples from patients on and off antibiotic therapy, overall accuracy in the diagnosis of urinary tract infection improved from 63,1% to 79,2% respectively. This is presumably due to in vitro inhibition of the culture growth by the antimicrobial agent in the' urine. The absence of leucocytes in the
VOL 74
3 SEPT 1988
227
urine does not exclude significant bacteriuria and colonisation, with or without underlying uropathy, is not necessarily followed by invasion of the mucosa and an inflammatory response. -Escherichia coli and most other urinary pathogens (Proteus, Klebsiella, Emerobaccer, Citrobaccer and Salmonella spp. and some enterococci, staphylococci and Pseudomonas spp.) reduce dietary nitrate in the urine to nitrite, which is detected by the reaction on the test strip. The nitrate-to-nitrite conversion takes several hours II and the short time the urine remains in the bladder in young infants, insufficient nitrate in the diet and urine, non-nitrite-producing organisms or antimicrobial substances in the urine preventing bacterial multiplication may cause a false-negative nitrite test in the presence of infection. The use of overnight first-morning samples increases the sensitivity of the nitrite test. The nitrite test was found to be highly specific (98,9% and 98,6%) with few false-positives but was relatively insensitive, with 37,3% of the urinary tract infections missed on the basis of a negative nitrite test. All false-positive nitrite tests were found in samples obtained from patients on antibiotics, which probably lead to the (falsely) negative culture and the positive predictive value of only 90%. ReR0rted sensitivity of the nitrite test varies from 21 % to 93%. 1-13 It is evident that a negative nitrite test does not exclude significant bacteriuria. An adequate dietary intake of nitrate (beans, spinach, cabbage) is necessary to provide sufficient substrate for the enzyme nitrate reductase; however 34/159 positive nitrite tests were obtained from samples of infants < 6 months who presumably were on mostly milk feeds. The number of positive nitrite tests correlated with the prevalence of urinary tract infection in each age group and age of the patient did not influence the result. The presence of frequency of micturition was not specifically recorded but, contrary to the findings of Powell et al.,1l in samples with significant bacteriuria the nitrite test was more often positive in symptomatic (64%) than in asymptomatic children (33%). Contaminants from outside the urinary tract are unlikely to cause a (false-) positive nitrite test due to the brief exposure to urinary nitrate. E. colt~ however, replicates every 20 minutes in urine at room temperature and a delay in screening or plating will result in false-positive findings. Stasis due to obstructive uropathy did not influence sensitivity.11 A rough correlation was observed between the severity of the leucocyturia and bacteriuria, and the time of appearance and depth of the colour change of positive leucocyte esterase and nitrite reactions respectively. Since the leucocyte esterase test has a high sensitivity (but low specificity) and the nitrite test a high specificity (but low sensitivity) the two tests read together should improve overall accuracy. In this study the sensitivity when either the leucocyte esterase or the nitrite test or both were positive was 94,1% and 95,4%, while 2 negative tests correctly identified 98,8% and 98,4% of the negative cultures. Proteinuria is nonspecific as an indicator of urinary tract infection and the improved overall accuracy of the three-test combination (Tables V and VI) was due to the inclusion of patients with proteinuria due to glomerular disease. Nevertheless anyone (or more) positive screening test is a definite indication for microscopy and culture. Patients with urinary tract infection are thus most likely to be detected if one or more of the leucocyte esterase, nitrite or protein tests are positive and in this study only 3/153 « 2%) of urinary tract infections were missed. Significant bacteriuria was the only abnormal finding in these 3 samples; the patients were asymptomatic and the bacteriuria may represent colonisation without invasion or was due to a delay between screening and culturing the sample.
228
SAMT
VOL 74
3 SEPT 1988
The predictive value of three negative tests was > 99%, so that if all three tests are negative, urinary uact infection is virtually excluded. In this hospital most urine samples are sent for routine laboratory urinalysis and only 10- 20% show any abnormality. Our .findings indicate that negative test strip screening predicts with a high degree of confidence that laboratory urinalysis will be negative. If the patient has no clinical features of urinary tract infection and screening tests for leucocyte esterase, nitrite and protein are negative, it is a waste of time and money to request full laboratory urinalysis. The cost-effectiveness of test strip screening is self-evident. In this study, 40 - 45% of the samples (mixed growths excluded) did not require full laboratory urinalysis on the basis of three negative screening tests. Cagle et al.,14 using combined biochemical screening for leucocyres and bacteria (nitrite) before microscopy, reported a reduction of 55% of their workload, and very few microscopic abnormalities (1,3%) were missed when all positive screening tests were followed up. Smalley and Bryan l5 showed that only 0,4% of urines with an abnormal microscopic finding would have been missed, while 25,8% of microscopic examinations of urine could have been eliminated on the basis of a 9-test negative biochemical screen. Screening of urine samples may be taken a step further in that if the test for haemoglobin (blood) is also negative it is unlikely that any form of renal disease is present. Casts are relatively short-lived and disintegrate readily. Samples showing positive reactions on screening for either leucocytes, blood, nitrite or protein should have priority microscopic examination in the busy laboratory. Such samples should immediately be plated for antimicrobial susceptibility testing so that the sensitivity pattern is available within 24 hours. s . The Combur 9 and Multistix 9 test strips are comparable and easy to use, each having minor advantages over the other. The uace-l + reactions are at times difficult to interpret, as are urine samples with a pronounced intrinsic colour. Reflectance photometry (Urotron RL 9, Clinitek 10 or 200) of the test strip may eliminate observer error. The availability of multitest screening tests is a significant step forward in the early diagnosis and ueatment of urinary tract infection, but is still dependent on obtaining the sample which, in the paediatric age group, is too often considered inconvenient. The combination of tests will overdiagnose urinary tract infection but, more important, very few cases will be missed. If all three tests are negative one may be confident that urinary uact infection is absent and unnecessary urinalysis avoided. The high frequency rate of mixed growths in the under-2year age group (31,5%) illustrates the unreliability of bag samples. Suprapubic bladder aspiration is preferable but not always practical. The laboratory will usually report a mixed growth as being due to contamination. Since contaminants do not have sufficient time to reduce nitrate in a fresh sample, >
20% of the mixed growths in this study were probably genuine urinary uact infections as evidenced by the clinical features, positive test strips and significant pyuria on microscopy. All patients with positive leucocyte esterase and/or nitrite tests were treated immediately and the diagnosis could not be confirmed by repeat urinalysis. Positive esterase and nitrite test suip-diagnosed urinary tract infection requires appropriate management irrespective of the laboratory report. Persistently positive nitrite and leucocyte esterase tests while on treatment are indicative of an incorrect choice of antimicrobial agent. In the paediatric age group reliable diagnosis with screening rests also permits early uluasonography for underlying uropathy which predisposes to infection. Our experience is limited to hospital paediatric practice but this screening method should be more widely applied. Biochemical screening tests for urinary tract infection in patients at risk are readily applicable to home testing particularly if an overnight early-morning first specimen is tested. 12 It should be suessed, however, that clinical signs and symptoms should not be ignored because of negative screening tests.
The authors wish to thank Dr R. O. Simpson, Medical Superintendent, Red Cross War Memorial Children's Hospital, for permission to publish and Boehringer Mannheim (SA) for financial assistance.
REFERENCES I. Ransohoff DF, Feinstein AR. Problems of spectrum and hias in evaluating the efficacy of diagnostic tests. N Engl J Med 1978; 299: 926-930.
2. Rohins DG, White RHR, Rogers KB, Osman MS. Urine microscopy as an aid to detection ofhacreriuria. Lancet 1975; i: 476-479. 3. Triger DR, Smith JWG. Survival of urinary leucocytes. J Clin Pathol 1966; 19: 443-447. 4. Kusumi RK, Grover PJ, Kunin CM. Rapid detection of pyuria hy leukocyte esterase activiry.JAMA 1981; 245: 1653-1655. 5. Banauch D. Leukozyten-Nachweis im Urin mit einem Teststreifen. Eine kooperative Studie an elf Zentren. Dcsch Med Wochenschr 1979; 104: 1236-1240. 6. Wilkins EGL, Ratcliffe JG, Roberrs C. Leucocyte esterase-nitrite screening method for pyuria and bacreriuria.J Clin Patho11985; 38: 1342-1345. 7. Gillenwater JY. Detection of urinary leukocytes by Chemstrip-L. J Urol . 1981; 125: 383-384. 8. Schaller G. Test-strip screen for the optimization of microscopic urinalysis. Clin Chem 1983; 29: 1692-1693. 9. LeightOn PM, Linle JA. Leucocyte esterase determination as a secondary procedure for urine screening.J Clin Patho11985; 38: 229-232. 10. Perry JL, Matthews JS, Weesner DE. Evaluation of leukocyte esterase activiry as a rapid screening technique for bacteriuria. J Clin Microbiol1982; 15: 852-854. 11. Powell HR, McCredie DA, Ritchie MA. Urinary nitrite in symptomatic and asymptOmatic urinary infection. Arch Dis Child 1987; 62: 138-140. 12. Scheifele DW, Smith AL. Home-testing for recurrent bacteriuria using nitrite strips. AmJ Dis Child 1978; 132: 46-48. 13. Skelton IJ, Hogan M, Stokes B, Hurst JA. Urinary tracr infections in childhood: the place of the nitrite test. MedJ Aust 1977; I: 882-886. 14. Cagle P, Hurst D, Saleem A. Substitution of biochemical urine screening for routine microscopy. Tex Med 1986; 82: 41-42. 15. Smalley DL, Bryan JA. Comparative evaluation of biochemical and microscopic urinalysis. AmJ Med Techno11983; 49: 237-239.
