Asymptomatic Bacteriuria Among Pregnant Women: Overview of Diagnostic Approaches

Felice D. Garingalo-Molina, M.D.

ABSTRACT A review was done to re-evaluate urine culture as a screening test in the diagnosis of asymptomatic bacteriuria among pregnant patients and to determine the diagnostic accuracy of either one or a combination of various rapid urine screening tests compared to quantitative urine culture. Prospective studies comparing the accuracy of any one or a combination of rapid urine screening test/s with quantitative urine culture in pregnant patients with no symptoms of urinary tract infection were included. A search through Medline 1966 to 1999 using the terms asymptomatic bacteriuria (MeSH term) and urine culture (MeSH or textword); sensitivity and specificity (textword) combined and limited to human, female and English language was done. Methodologies of included articles were reviewed. Eight comparative studies were included, resulting in an aggregate sample size of 6,506 patients. The prevalence rate of asymptomatic bacteriuria during pregnancy ranged from 2.3% to 10.3%. All studies compared sensitivity, specificity and positive predictive values (PPV) of rapid urine screening tests against urine culture. Seven indicated negative predictive values (NPV). Seven studies adopted >100,000 colony-forming units/ml (cfu/ml) as cut-off level for significant bacteriuria while one study utilized >10,000 cfu/ml. One study required two consecutive urine cultures with significant bacterial counts for diagnosis. Five studies using reagent strips showed low sensitivity, variable specificity, PPV and NPV. None recommended reagent strips or dipsticks as sufficient screening method. Two studies using reagent strips showed low sensitivity, variable specificity, PPV and NPV. None recommended reagent strips or dipsticks as sufficient screening method. Two studies tested different cut-offs for pyuria. Results showed low sensitivity. Chromogenic Limulus Amoebocyte assay had a high predictive value and low false positive rate but only detected gram-negative organisms and colony counts > 100,000. Two studies using a semi-automated method noted high sensitivity but low specificity and PPV. Technical problems rendered some results uninterpretable. Only one study utilized urine gram stain. Sensitivity and specificity were fairly high. A urine dipslide quantitative kit was evaluated in one study. For detection of gram-negative organisms, sensitivity, specificity and NPV were almost 100%. PPV was 81%. The cysteine-lactose electrolyte deficient dipslide for the detection of all organisms had a sensitivity and NPV of 100% but had very slow specificity and PPV. The test missed 60% of specimens with any group B streptococci. Evidence does not support the use of other urine screening tests for asymptomatic bacteriuria among pregnant women other than urine culture. The closest alternative appears to be urine gram stain. However, future studies are needed to determine the appropriate value for significant bacteriuria in gram stain. A single urine culture seems sufficient basis for the diagnosis of asymptomatic bacteriuria in pregnancy since the probability of true infection with a single determination is already high. (Phil J Microbiol Infect Dis 2000; 29(4):177-186) Keywords: meta-analysis, asymptomatic bacteriuria, UTI

INTRODUCTION Is screening for asymptomatic bacteriuria in pregnant women worth-while? Pregnant women with asymptomatic bacteriuria are at high risk for a number of complications for both mother and the unborn. Maternal complications include overt urinary tract infection for 30 to 40% of patients especially as pregnancy advances. Whether or not symptomatic urinary track infection ensues, the fetus is still at risk for prematurity, low birth weight and even fetal wastage.1 Thus, in the obstetric patients, there is little if at all any doubt, regarding the need for early screening for asymptomatic bacteriuria. A cost-analysis study found that screening is cost-effective when the prevalence of bacteriuria is >2%.2 The condition is detectable and treatable; its consequences are preventable so that screening for asymptomatic bacteriuria is justifiable and ultimately costeffective.3

The debate lies on the manner of screening. Quantitative urine culture of a midstream clean-catch specimen has been widely recognized as the optimal screening test. In the absence of symptoms, the quantitative threshold of >100,000 colony-forming units/ml from a midstream or catheterized urine specimen is useful in differentiating true bacteriuria from contamination.4 Not all studies however, used the same methodology and definition of asymptomatic bacteriuria and often, this was diagnosed on the basis of a single midstream urine specimen. The ideal screening test should be inexpensive, simple and rapid and should have high sensitivity in addition to specificity.5 Inspite of quantitative urine culture being the undisputed gold standard, some clinicians and most training hospitals here and abroad do not utilize urine culture for screening. Among their reasons are: the belief that prevalence of asymptomatic bacteriuria in pregnancy is low and does not justify mandatory urine culture; screening by urine culture is expensive; lack of laboratory facilities and competent personnel to perform culture; urinalysis is a good enough screening method; and the need for a follow-up visit for the results.6 Thus, a number of other bacteriuria screening methods has been proposed. These procedures are predicated on the concept that a prescreen of urine specimens for significant bacteriuria may reduce the need for culture of non-significant samples, increase the cost-effectiveness of culture and assist in immediate patient management.7 This overview intends to highlight the unique features and barriers to improved detection of asymptomatic bacteriuria among pregnant patients and to identify areas of possible revisions in the current recommendations particularly diagnostic laboratory work-up. The following key questions will be addressed: 1. Does the current prevalence rate justify the recommendation to do urine culture during the first prenatal visit? 2. What screening test is the best substitute for urine culture in the event that the former cannot be done? 3. Will a single urine culture of a midstream clean-catch specimen yielding 100,000 or more cfu/ml of a single uropathogen be sufficient basis for diagnosis of asymptomatic bacteriuria in pregnancy? This review aims to re-evaluate the urine culture as a screening test in the diagnosis of asymptomatic bacteriuria among pregnant patients and to determine the diagnostic accuracy of either one or a combination of the various rapid urine screening tests available compared to urine culture. MATERIALS AND METHODS A. Criteria for considering articles for review Types of studies Prospective studies comparing any one or a combination of rapid urine screening test/s with quantitative urine culture. II. Types of participants Pregnant patients with no symptoms of urinary tract infection. III. Types of intervention Experimental: Rapid urine screening tests using one or a combination of the following: urine microscopy, urine gram stain, different dipstick or strip methods, semi-automated techniques, chromogenic assay and quantitative culture kits. Gold standard: Quantitative urine culture IV. Types of outcome measures Sensitivity, specificity, positive predictive value and negative predictive value. I.

B. Search strategy for identification of studies To identify the studies, search was conducted through Medline 1966 to 1999 using the following search terms: asymptomatic bacteriuria (MeSH term), urine culture (MeSH or textword) and sensitivity and specificity (textword) combined and limited to human, female and restricted to English language. Citations of studies, clinical practice guidelines and reviews identified by computer search were also examined to identify other comparative studies. C. Methods of the review I. Inclusion criteria The titles (and abstracts when available) in the search printouts were reviewed. Articles that could meet the criteria or could refer to a related study were all retrieved and examined. II. Quality assessment Methodologies of included articles were reviewed. The Critical Appraisal format was adapted from the validity appraisal of a diagnostic test. (Appendix A) III. Data abstraction The reviewer using the data abstraction form developed for this overview assessed information from each study. Data abstracted included the study period, geographical location of the study, inclusion and exclusion criteria, definition of asymptomatic bacteriuria. The sensitivity, specificity, positive and negative predictive values were likewise abstracted or computed. D. Description of studies The literature search identified 18 studies comparing different rapid urine screening tests either alone or in combination, with urine culture. Of these standard comparative studies, ten were excluded. Details of the included studies are provided in the Table 1. The studies were conducted mainly in obstetrics and family medicine clinics with one study done in an army hospital among pregnant dependents of junior enlisted army personnel. These studies were mostly U.S.-based. One was done in the United Kingdom.9 The studies compared the sensitivity and specificity of urine microscopy, other rapid reagent strip tests and/or semi-automated procedures, chromogenic assay and prepared culture kits with that of urine culture. Two studies used reagent strips and urine microscopy.5,9 One study compared sensitivity and specificity of reagent strip (Chemistrip 9 with leukocyte esterase and nitrite), urinalysis and urine gram stain separately,10 while another used reagent strips only: leukocyte esterase and nitrite separately and in combination.11 Reagent strips and a semi-automated instrument, Bac-T screen was used in a study7 and another evaluated a chromogenic assay.12 Semi-automated instrument and urinalysis were tested in the study by Davis in 1984, while quantitative urine culture kit (Uricult) was used in another.13 Ten studies were excluded for the following reasons: (1) two were retrospective studies (2) five studies had no full texts available (3) one study whose patient population consisted of pregnant women, presently asymptomatic but were previously treated for pyelonephritis in their respective current pregnancies and (4) two were not comparative studies. E. Methodological quality of included studies The quality assessment of individual studies is shown in the Table of Included Studies (Table 1). The methodological qualities of the studies were fairly good. All studies were not clear regarding blind comparison with the gold standard except for the study of Bachman. All included

an appropriate spectrum of patients and applied the reference standard i.e., urine culture on all urine specimens collected.
Table 1. Characteristics of included studies Study ARCHBALD 1984 (USA) Diagnostic Tests Evaluated Urine reagent strip (Microstix) testing for nitrite, semi-quantitation of gram- negative and total bacterial organisms; urinalysis (pyuria=5 or more wbc/hpf; bacteriuria if 20 or more bacteria/hpf) vs.urine culture Significant bacteriuria: 100,000 cfu/ml 1,047 asymptomatic Urine dipstick testing for leukocyte pregnant patients on activity, nitrites individually and in initial and follow-up combination; urinalysis (pyuria:>10 visits at the Obstetrics wbc/hpf);urine gram stain with two and Family Medicine or more organisms/OIF considered departments; excluded as positive vs.urine culture patients taking antibiotics (significant bacteriuria: 100,000 cfu/ml) 1,215 asymptomatic obstetric patients in a hospital 544 consecutive asymptomatic pregnant patients, mostly indigents in an Obstetric outpatient clinic 694 pregnant patients at the obstetric clinic in a regional medical center Semi-automated urine screening (Bac- T screen)vs.urine culture (significant bacteriuria:100,000 cfu/ml) Participants 324 consecutive asymptomatic pregnant women in the Obstetrics Department of a hospital Outcomes Sensitivity Specificity PPV NPV Comments Cannot tell whether there was blind comparison with the gold standard Appropriate spectrum of Patients Urine culture done on all Patients Tests described in detail **Blind comparison with the gold standard Appropriate spectrum of patients Urine culture done on all patients Test described in detail

BACHMAN 1993 (USA)

Sensitivity Specificity PPV Incremental patient costs Clinical outcomes Sensitivity Specificity PPV NPV

DAVIS 1984 (USA) DORSTEN 1986 (USA) MCNEELEY 1987 (USA)

Urine dip-slide quantitative kit Sensitivity (Uricult) vs.urine culture (significant Specificity bacteriuria: 100,000 cfu/ml) PPV NPV Urine dipstick testing for leukocyte esterase and nitrite (significant if either is positive); semi-automated method (Bac-T screen) vs.urine culture; Significant bacteriuria: 10,000 cfu/ml Chromogenic Limulus Amoebocyte assay vs.urine culture (significant bacteriuria:100,000 cfu/ml) Sensitivity Specificity PPV NPV Sensitivity Specificity PPV NPV

Cannot tell whether there was blind comparison with the gold standard Appropriate spectrum of patients Urine culture done on all patients Test described in detail Cannot tell whether there was blind comparison with the gold standard Appropriate spectrum of patients Urine culture done on all patients Test described in detail Cannot tell whether there was blind comparison with the gold standard Appropriate spectrum of patients Urine culture done on all patients Tests described in detail Cannot tell whether there was blind comparison with the gold standard Appropriate spectrum of patients Urine culture done on all patients Test described in detail Can only detect gram-negative organisms >100,000 cfu/ml Cannot tell whether there was blind comparison with the gold standard Appropriate spectrum of patients Urine culture done on all patients Tests described in detail Cannot tell whether there was blind comparison with the gold standard Appropriate spectrum of patients Urine culture done on all patients Tests described in detail.

NACHUM 1986 (USA)

1,039 specimens from 664 obstetric patients in the Obstetrics Department of a hospital

ROBERTSON 750 consecutive 1988 asymptomatic obstetric (USA) patients on first prenatal visit in an army hospital for dependents of enlisted personnel TINCELLO 1998 (UK)

Urine dipstick testing for nitrites and leukocyte esterase activity individually and in combination vs.urine culture (significant bacteriuria: 100,000 cfu/ml, single pathogen in two consecutive cleancatch midstream specimens) Urine reagent strips testing for 893 asymptomatic pregnant patients on their blood, protein, nitrite and leukocyte first pre-natal visit to an esterase tested individually and in inner city maternity combination vs.urine culture hospital; Significant bacteriuria: 100,000 excluded patients taking cfu/ml antibiotics

Sensitivity Specificity PPV NPV

Sensitivity Specificity PPV NPV

RESULTS Eight studies satisfied the inclusion criteria, resulting in an aggregate sample size of 6,506 patients. All included studies compared sensitivity, specificity and positive predictive values of various rapid urine screening tests against urine culture. Except for one study,10 all others indicated

negative predictive values. One study5 included efficiency while only the study of Bachman included patient follow-up and looked into incremental costs and clinical outcomes, i.e., patients who eventually developed over urinary tract infection. Almost all studies were in agreement with 100,000 colony-forming units or more as the level of significant bacteriuria in urine culture, except one study7 where a lower level of 10,000 cfu/ml was already considered significant. Only one study required two consecutive urine cultures with significant bacterial counts for diagnosis of asymptomatic bacteriuria.11 Prevalence rate of asymptomatic bacteriuria among pregnant women fell in the range of 2.3% to 10.3%. (Table 2) Five studies that determined the accuracy of reagent strips (Table 3), consistently showed low sensitivity and variable specificity, positive predictive values as well as negative predictive values.5,7,9,10,11 All these studies did not recommend reagent strips or dipsticks as sufficient screening method. Only one study11 stated that a combination of dipstick methods might improve accuracy and turn out to be cost-effective. In his case, he used a combination of nitrite and leukocyte esterase activity and considered significant when either test showed a positive result. Two studies tested the accuracy of urinalysis using different cut-offs for pyuria, 5 or more leukocytes/hpf5 and more than 10 leukocytes/hpf.10 Neither showed pyuria to be of value in screening for asymptomatic bacteriuria in pregnancy because of low sensitivity (Table 4). One study using Chromogenic Limulus Amoebocyte assay12 showed high positive and negative predictive values (89.9% and 98.6%, respectively), low false positive rate, sensitivity of 87.8% and specificity of 98.9%. These imply that the number of specimens sent for culture can be reduced by sending only those positive in this assay for urine culture. However, this test has two major drawbacks: only gram-negative organisms and colony counts > 100,000 can be detected. Bac-T screen, a semi-automated method was used in two studies.7,14 The McNeeley trial used 10,000/cfu/ml as cut-off for significant bacteriuria. Both studies noted high sensitivity but low specificity and positive predictive values. There were results which were uninterpretable because of technical problems like specimens containing pigments which prevented color interpretation and clogging of the machine filter (Table 5). Only one study utilized urine gram stain.10 Smears with one organism per oil immersion field were considered borderline positives. The presence of two or more organisms per oil immersion field was considered positive. Sensitivity and specificity were fairly high for both borderline positives and definite positives (Table 6).
Table 2. Prevalence rates (%) of included studies Study Archbald 1984 (USA) Bachman 1993 (USA) Davis 1984 (USA) Dorsten 1986 (USA) McNeeley 1987 (USA) Nachum 1986 (USA) Robertson 1988 (USA) Tincello 1998 (USA) Population 324 1,047 1,215 544 694 1,039 750 893 Prevalence rate (%) 9.3 2.3 6.0 10.3 8.1 8.3 8.3 5.4

A urine dip-slide quantitative kit (Uricult) was utilized in one study.13 This kit has a different detector for gram-negative and gram-positive organisms. For detection of gram-negative organisms, sensitivity, specificity as well as negative predictive values were close to 100%. Positive predictive value was 81%. The cysteine-lactose-electrolyte-deficient dipslide for the detection all organisms had a sensitivity and negative predictive value of 100% but a very low specificity (33.2%) and positive predictive value (2.2%). It is noteworthy that the test missed 60% of specimens with any group B streptococci.

Table 3. Performance characteristics of reagent strips (Values are expressed in percentages. LE=leukocyte esterase activity; PPV=positive predictive value;NPV=negative predictive value) Study Archbald 1984 (N=324) Bachman 1993 (N=1,047) Sensitivity 37 33 16.7 45.8 50.0 12.5 69.6 77.4 43.4 92.0 32.2 18.8 33.3 Specificity 100.0 98.0 97.2 99.7 96.9 100.0 83.4 96.1 98.9 95.0 94.2 99.5 91.1 PPV 100 67 12.1 78.6 27.3 100.0 26.9 64.0 79.4 62.6 100.0 69.2 17.6 NPV 93 93

McNeeley 1987 (N=694) Robertson 1986 (N=750)

Tincello 1998 (N=893)

Nitrites Microstix LE Nitrites LE or Nitrites LE and Nitrites LE or Nitrites LE Nitrites LE or Nitrites LE and Nitrites Nitrites Tests combined

96.9 97.9 95.1 99.2 99.2 95.6 96.0

Table 4. Performance characteristics of urine microscopy. Note the different cut-offs used. (Values are expressed in percentages. PPV=positive predictive value) Study Archibald 1984 (N=324) Bachman 1993 (N=1,047) Cut-offs Pyuria:5 or more wbc/hpf Pyuria:>10 wbc/hpf Adjusted cut-off:>50 wbc/hpf Sensitivity 20 25.0 8.3 Specificity 89 99 99.7 PPV 91 37.5 40.0 NPV 91

Table 5. Performance characteristics of the semi-automated method (Bac-T screen) (Values are expressed in percentages. PPV=positive predictive value; NPV=negative predictive value) Study Davis 1984 (N=1,215) McNeeley 1987 (N=694) Significant bacteriuria:100,000 cfu/ml Significant bacteriuria:10,000 cfu/ml Sensitivity Specificity 97.4 70.4 96.4 56.0 PPV 19.0 16.1 NPV 99.7 99.4

Table 6. Performance characteristics of urine gram stain (Values are expressed in percentages. PPV=positive predictive value; NPV=negative predictive value) Study Bachman 1993 Gram stain Sensitivity Definite positive:2 or more organisms/OIF Borderline (1 organism/OIF) or Definite Positive 83.3 91.7 Specificity 94.9 89.2 PPV 27.8 16.5

DISCUSSION Asymptomatic bacteriuria is the presence of actively multiplying bacteria at the time when the patient has no urinary symptoms so that the diagnosis relies upon microbiologic findings. Not all patient populations with asymptomatic bacteriuria warrant treatment. For instance, in school age children, non-pregnant women and elderly, only those with known or presumed abnormal urinary tracts need treatment.15 This is not the case in pregnant women. Thirty to 40% of pregnant asymptomatic bacteriuric women will have acute pyelonephritis usually in their third trimesters if they are left untreated. Treatment lowers this attack rate to 3%.16 Because of the crucial role asymptomatic bacteriuria in pregnancy plays in the causation of acute pyelonephritis, there is not much dispute in the recommendation that all pregnant women should be screened for bacteriuria on their first prenatal visit. There are also attendant morbid consequences for the fetus such as prematurity and low birth weight if the mother is left untreated, whether or not she develops over urinary tract infection later in her pregnancy. This makes detection of the condition all the more

important. Appropriate antimicrobial intervention should be initiated immediately for those found positive. The diagnosis of asymptomatic bacteriuria is based upon quantitative urine culture of a midstream clean-catch specimen with >100,000 colony-forming units/ ml urine and an isolate of a single uropathogen in a patient presenting with no symptoms. Most specimens contaminated during collection contain <10,000 cfu/ml. Most studies in this review utilized a single specimen as basis for the diagnosis. Only one study required two consecutive specimens.11 However, some authors argue that since these are asymptomatic patients, the diagnosis should rest on two consecutive specimens. False positive rate in pregnant asymptomatics is as high as 40%.3 However, there are studies showing 80% probability of true infection with a single positive urine culture and a repeat culture with the same results increases this probability to more than 95%. Conversely, an asymptomatic and untreated pregnant patient with <100,000 gram negative bacilli/ml urine has 98% probability of sterile urine on repeat culture.17 In this review, the prevalence of asymptomatic bacteriuria among pregnant patients was 2.3% to 10.3%. There was no significant variation between the seven studies that based their diagnosis on a single positive urine culture and the sole study using two consecutive positive urine cultures as its basis for diagnosis. Consequences of failure to detect the condition and missing on treatment are more crucial than treating those who actually do not have the infection. With these considerations, the requirement of two consecutive positive cultures should be re-evaluated. It appears that a single culture would suffice because of the already high probability. Requesting the patient to come back for another specimen collection for urine culture and subsequent follow-up of its result would increase the risk of patients being lost to follow-up and their receiving no treatment at all. The urine dipsticks In this review, five studies employed urine reagent strips, all showing low sensitivity, and variable specificity, positive as well as negative predictive values.5,7,9,11 Only one showed an improvement in accuracy when two tests were employed and when positive result for either test was interpreted as a true positive. Nitrite test (Griess) was used in all these studies. Griess test depends on the ability of most enteric bacteria to reduce nitrates to nitrites. The major drawbacks are: 1) not all uropathogens can reduce nitrates and 2) some organisms convert nitrates to ammonia efficiently so that there is not enough nitrites during reduction to permit detection so that nitrite test is useful only when positive. A negative test may signify absence of nitrites in the urine due to absence of infection, absence of nitrate substrate inspite of infection or a reduction of nitrates beyond the nitrite stage.18 There is not much evidence to recommend dipsticks as sufficient screening test nor as a prescreening for those patients who will need urine culture. Urine Microscopy Presence of pus cells is not a reliable marker of bacteriuria. Different cut-offs were used in two studies that evaluated the use of pyuria in screening for asymptomatic bacteriuria.5,10 There is little value for its use because of its high false negative and false positive rates. About 50% of asymptomatic bacteriurics can have <3 leukocytes/hpf in spun urine. Conversely, about one third of pregnant non-bacteriuric women can have 5 or more leukocytes/hpf.18 Chromogenic Limulus Amoebocyte assay High positive predictive value and a false positivity rate were reported.12 Cost reduction is possible when only those specimens positive for this assay are sent for urine culture. However, the major disadvantages of this assay are: 1) only gram-negative organisms are detected so that a

negative test may mean absence of infection or infection caused by a gram-positive organism. While it is true that majority of the urinary pathogen are gram-negative, a significant number are gram-positives. Like the nitrite test, Chromogenic Limulus Amoebocyte assay is only reliable when positive; (2) counts lower than 10,000 cfu/ml cannot be detected. In the case of asymptomatic bacteriuria, this may not be a problem but this limits the use of Limulus assay only to patient populations where the significant level of bacteriuria is >100,000 or more. In symptomatic patients where the significant count considered significant is lower, this assay may be of limited use; 3) availability of this assay. Semi-automated urine screen (Bac-T screen) This method entails a 1 ml aliquot of urine to be automatically filtered through a special test card, entrapping bacteria on the filter. After a washing cycle, safranin stain is used to demonstrate the entrapped bacteria. This test utilizes a semi-automated instrument. Manual interpretations are made by comparing the stained filters with a color chart on the test card. Two studies utilized this screening method.7,14 Both studies reported high sensitivity and negative predictive values (97.4%, 99.7% and 96.4%, 99.4% respectively) but low specificity and positive predictive values (70.4%, 19% and 56%, 16.1% respectively). The test is rapid and simple with little risk of missing true infections. However, technical problems can interfere with the result interpretation. Clogging of the machine filter and certain pigments in the specimens can render the result uninterpretable. Also, McNeeley considered 10,000 cfu/ml as already significant. This lower level of bacteriuria may have enhanced his reported high sensitivity for Bac-T screen. The cost and availability of machine should also be considered. The urine dip-slide quantitative kit (Uricult) The device consists of a plastic paddle coated on both sides with agar media, one side for the detection of gram-negative organisms (coated with MacConkey agar) and the other side which supports the growth of all organisms (coated with cysteine-lactose-electrolyte-deficient agar). The dip-slide is inoculated by dipping the agar-coated slides into the urine specimen, enough to wet both agar surfaces. Incubation for 18 to 24 hours at 350C is required. The colony counts of the Uricult are determined using a color chart. This method cannot identify the specific species of isolates. Once significant growth is detected, traditional urine cultures have to be done for species identification.13 Compared to the results of urine culture, the dip-slide missed 60% of group B streptococci. This is an important detail since group B beta hemolytic streptococci are not merely contaminants but are potential pathogens for the urinary tract during pregnancy, causing 5.1% of infections in this patient population.19 In addition to concerns regarding cost and availability, the Uricult test seems to only duplicate what standard urine culture can do minus the specific identification of isolates. Its use may not be practical and cost-effective. It has no clear advantage over the standard quantitative culture since overnight incubation is also required. Urine Gram stain It is quite surprising that only one study utilized gram stain for urine screening of asymptomatic bacteriuria.10 Gram staining is one basic skill that every proficient medical technologist should master. It offers a rapid and inexpensive method of detecting the presence of microorganisms. Results can be reported in a few minutes. Bachman reported a 83.3% sensitivity and 94.9% specificity. Compared to dipsticks, the cost may be comparable or slightly higher in some countries. It is more reliable and accurate. Compared to the other methods of screening mentioned, it is also simple to perform and does not require special machines for performance.

More studies should be conducted to validate its use in screening for asymptomatic bacteriuria since it holds a potential as an alternative for urine culture if this latter test is deemed expensive and laborious. Gram stain can also offset the other disadvantage of urine culture in that results can be released on the same day so that the danger of patients lost to follow-up is minimized and treatment if needed can be initiated right away. In areas where facilities are inadequate for urine culture, gram stain can be utilized. Urine culture Overall, quantitative urine culture is the only method that has stood the test of time in terms of sensitivity and specificity. The other methods are merely prescreens in an effort to minimize sending unnecessary specimens for culture. For positive cases, urine cultures still have to be done. It is in this same specimen that antimicrobial susceptibility testing of the isolate can be performed. SUMMARY AND CONCLUSIONS This review shows that urine dipstick methods, whether alone or in combination, have low sensitivity variable specificity, positive and negative predictive values. There is a tendency for high false negativity especially for the nitrite test because some uropathogens cannot reduce nitrates and others convert nitrates to ammonia rapidly so that appreciable levels are not detected during nitrite reduction. Urine dipsticks in general, are insufficient as screening tests for asymptomatic bateriuria in pregnancy. Pyuria is a poor predictor of true asymptomatic infections because of low sensitivity and thus, is not recommended as a screening test for asymptomatic bacteriuria. Chromogenic Limulus Amoebocyte assay may not be a good alternative for urine culture because its detecting capability is limited only to gram - negative organisms. The use of semi-automated methods is limited by its low specificity, technical problems in the interpretation of results, cost and availability of special machines needed. Semi-quantitative dipslide urine culture does not offer any advantage over standard urine culture as regards cost, same day result and specific identification of organisms. Gram stain appears a promising alternative to urine culture especially in areas where the latter could not be done. It offers the advantages of high sensitivity and specificity, same day result, simple procedure and less expense. However, urine culture still has to be done whenever possible for confirmation and for subsequent antimicrobial susceptibility testing after empiric treatment has been initiated. Implications for practice The role of urine culture is well established as a screening test for asymptomatic bacteriuria in pregnancy. Inspite of its cost, need for adequate laboratory facilities and delayed results, it is highly sensitive and specific. This review has demonstrated that no urine screening method has come close and accurate enough to be an alternative for urine culture except for urine gram stain. Finally, to answer the questions posted earlier on this review: In an ongoing study on asymptomatic bacteriuria among pregnant Filipino patients at the Philippine General Hospital, the prevalence rate is 2.5% out of over 1,000 patients screened to date. This rate justifies the recommendation of urine culture during the first prenatal visit. Evidence does not support the use of other urine screening tests for asymptomatic bacteriuria among pregnant women other than urine culture. In the event that urine culture cannot be done, the closest alternative appears to be urine gram stain. A single urine culture seems to be sufficient basis for the diagnosis of asymptomatic bacteriuria in pregnant women, since the probability of true infection is already high at 80% with a

single determination, and the consequence of treating a false positive case is not as much as when a true positive infection is missed. Implications for research With urine culture as the gold standard for the diagnosis of asymptomatic bacteriuria among pregnant women, there seems to be no argument about 100,000 cfu/ml of urine as the level for significant bacteriuria. However, there is a need to reconsider the recommendation of two consecutive urine cultures growing the same single isolate at 100,000 or more cfu/ml. Studies should be done comparing the accuracy of a single positive urine culture versus two consecutive cultures. There appears to be an underutilization of urine gram stain as a potential alternative to urine culture, especially in areas where manpower and laboratory facilities for performing urine cultures are lacking. More studies should be conducted to determine the accuracy of urine gram stain and establish a reasonable cut-off value for significant bacteriuria.
ACKNOWEDGEMENTS The author thanks her consultants at the Philippine General Hospital, Infectious Diseases Section: Dr. Mary Ann Lansang, Dr. Myrna T. Mendoza, Dr. Jaime C. Montoya, Dr. Mediadora C. Saniel and Dr. Melecia A. Velmonte for their helpful comments and suggestions on this review, and Ms. Lolita Bugayong for providing computer advise.
REFERENCES 1. Andriole V, Patterson T. Epidemiology, natural history and management or urinary tract infection in pregnancy. Med Clin North Am 1991; 75 (2):359-371. 2. Wadland W, Plante D. Screening for asymptomatic bacteriuria in pregnancy. Fam Phy 1989; 29(4):372-376. 3. Gratacos E, Torres PJ, Villa J, Alonso P, Cararach V. Screening and treatment of asymptomatic bacteriuria in pregnancy to prevent pyelonephritis. J Infect Dis 1994; 169: 1390-1392. 4. Kass E. Bacteriuria and pyelonephritis in pregnancy. Arch Intern Med 1960; 105: 42-46. 5. Archbald F, Verma U, Tejani N. Screening for asymptomatic bacteriuria with Microstix. J Reprod Med 1984; 29(4):272-274. 6. Abyad A. Screening for asymptomatic bacteriuria in pregnancy: urinalysis versus urine culture. J Fam Phys 1991; 33(5):471479. 7. McNeeley SG, Baselski V, Ryan G. An evaluation of two rapid bacteriuria screening procedures. Obstet Gynecol 1987; 69(4):550-553. 8. Task Force on Urinary Track Infection. Philippine Practice Guidelines Group in Infectious Diseases. Urinary Tract Infections: Clinical Practice Guideline. PPGID. Philippine Society for Microbiology and Infectious Diseases 1999; 1(1): 16-19. 9. Tincello D, Richmond D. Evaluation of reagent strips in detecting asymptomatic bacteriuria in early pregnancy: prospective case series. Br Med J 1998; 316(7):435-437. 10. Bachman J, Heise R, Naessens J, Timmerman M. A study of various tests to detect asymptomatic urinary tract infections in an obstetric population. JAMA 1993; 270 (16):1971-1974. 11. Robertson A, Duff P. The nitrite and leukocyte esterase tests for the evaluation of asymptomatic bacteriuria in obstetric patients. Obstet Gynecol 1988; 71(6):878-881. 12. Nachum R, Arce J, Berzofsky R. Gram-negative bacteriuria of pregnancy: rapid detection by a chromogenic limulus amoebocyte lysate assay. Obstet Gynecol 1986; 68 (2):215-219. 13. Dorsten JP, Bannister E. Office diagnosis of asymptomatic bacteriuria in pregnant women. Am J Gynecol 1986; 153(4):777780. 14. Davis J, Stager C. Detection of asymptomatic bacteriuria in obstetric patients with a semi-automated urine screen. Am J Obstet Gynecol 1984:151(8):1069-1073. 15. Zhanel G, Harding G, Guay D. Asymptomatic bacteriuria: which patients should be treated? Arch Intern Med 1990; 150:1389-1395. 16. Hankins G, Whalley P. Acute urinary tract infections in pregnancy. Clin Obstet Gynecol 1985; 28(2):266-278. 17. Patterson T, Andriole V. Detection, significance and therapy of bacteriuria in pregnancy. Infect Dis Clin North Am 1997; 11(3):593-607. 18. Mittendorf R, Williams M, Kass E. Prevention of pre-term delivery and low birth weight associated with asymptomatic bacteriuria. Clin Infect Dis 1992; 14:927-932. 19. Mead P, Harris R. The incidence of group B beta-hemolytic streptococcus in antepartum urinary tract infections. Obstet Gynecol 1978; 51(4):412-414.