Interpreting asymptomatic bacteriuriaBMJ 2011; 343 doi: http://dx.doi.org/10.1136/bmj.d4780 (Published 04 August 2011) Cite this as: BMJ 2011;343:d4780
- Martin Cormican, professor of bacteriology1,
- Andrew W Murphy, professor of general practice2,
- Akke Vellinga, epidemiologist2
- 1Centre for Health from Environment, Ryan Institute, and Medical School, NUI Galway, Galway, Ireland
- 2Discipline of general Practice, Medical School, Clinical Science Institute, NUI Galway, Galway, Ireland
- Correspondence to: Professor M Cormican, Antimicrobial Resistance and Microbial Ecology Laboratory, Medical School, Clinical Science Institute, NUI Galway, Newcastle Road, Galway, Ireland
Asymptomatic bacteriuria refers to bacteria in the urine at levels often regarded as clinically significant ( >100 000 colony forming units per millilitre of urine) in patients with no symptoms suggestive of urinary tract infection. It becomes more common with age
Testing for and treating asymptomatic bacteriuria is of established value in pregnant women as it reduces the risk of pyelonephritis later in pregnancy by about 75%
Consider testing for bacteriuria in any patient with clinical features pointing to urinary tract infection (haematuria, dysuria, frequency, urge incontinence, or back pain) or to systemic sepsis without an apparent focus
If children, non-pregnant adults, or people with diabetes or indwelling urinary catheters lack specific symptoms of urinary tract infection or systemic infection, avoid testing them for and treating bacteriuria
Testing for bacteriuria in patients with stable stress incontinence is not appropriate as bacteriuria is not associated with stress incontinence in older people
If you feel obliged to try treating bacteriuria in a patient with non-specific or equivocal symptoms, urine culture can guide selection of the safest and most narrow spectrum agent possible; ensure careful assessment of clinical and microbiological response
Mrs A is a 74 year old woman who saw your locum two days ago with a two to three week history of feeling tired and vaguely unwell. She admitted to “leaking a bit of urine” whenever she coughs or lifts things, but has had this problem for two years. She has not had other urinary tract symptoms and has not had dysuria or frequency in recent years. She was prescribed an antibiotic two or three times in the past year by a doctor in another practice after positive laboratory tests on her urine. Her medical history is otherwise unremarkable and she was not examined. Your locum sent a urine specimen for microscopy and culture as dipstick testing was positive for nitrite (but negative for leucocyte esterase). Mrs A has returned to see you for the results.
What is the next investigation?
Testing urine for evidence of infection
Testing urine for evidence of infection may be appropriate in the following situations.
When clinical features suggest urinary tract infection—A 2002 systematic review concluded that (a) the combination of dysuria and frequency in the absence of vaginal discharge or irritation confirms the diagnosis of urinary tract infection (probability >90%) and (b) dipstick testing cannot lower the post-test probability sufficiently to exclude urinary tract infection if a patient presents with one or more symptoms.1 Urine culture, however, has a value beyond confirming the diagnosis in that it can also direct treatment, on the basis of results of tests for antimicrobial susceptibility. The table⇓ summarises the findings of two systematic reviews.
When features of systemic sepsis are present—Testing of urine in patients presenting with clinical features of systemic sepsis (fever, rigors) is useful as the urinary tract is a common source of blood stream infection. Al-Hasan and colleagues documented an age adjusted incidence rate of 55.3 per 100 000 person years for urinary tract infection complicated by bacteraemia.4
When there are other specific indications—Testing for bacteriuria is recommended in pregnant women without symptoms suggesting urinary tract infection5 6 as 2-7% have clinically significant bacteriuria.5 A systematic review of 14 studies and 2302 women concluded that treatment of asymptomatic bacteriuria in pregnancy reduces the incidence of pyelonephritis later in pregnancy. The overall incidence of pyelonephritis in pregnant women with asymptomatic bacteriuria is 21%, and treatment leads to a reduction in risk of 75%.7 Testing urine for infection is also recommended in the investigation of patients presenting with acute renal failure before major urological procedures.8 9
Avoidance of testing
Other than for the groups outlined above, testing for bacteriuria should generally be avoided in patients who do not have specific features suggesting urinary tract infection. The following groups deserve specific mention as they may cause particular concern and useful evidence is available to guide practice.
Asymptomatic bacteruria is common in, but not confined to, older people. Asymptomatic bacteriuria is present in about 1% of schoolgirls, rising in frequency with age (to >20% of healthy women aged over 80), and affects about 6-15% of men aged over 75.5 A prospective cohort study showed that the condition in older people was associated with about double the risk of symptomatic urinary tract infection over two years of follow-up, with similar findings in studies with longer follow-up.5 10 Treatment of asymptomatic bacteriuria is not recommended in pregnant women of any age because the condition is not associated with adverse long term outcomes and because treatment does not result in durable eradication of bacteriuria or improved clinical outcome.5 9 Although fewer data are available for men, asymptomatic bacteriuria seems to be similarly benign in older men.5 Antimicrobial treatment is associated with increased healthcare costs and avoidable risks, such as the direct adverse effects of antimicrobial agents, disturbance of normal microbial flora leading to mucosal candidiasis and antimicrobial associated diarrhoea, and the promotion of antimicrobial resistance.11 12
Although asymptomatic bacteriuria is associated with urge incontinence in older women, it is not significantly associated with stress incontinence, as described by Mrs A.13 Thus, although bacteriuria is common in older people, testing for bacteriuria is not recommended in older people who do not have specific clinical evidence of urinary infection. If bacteriuria is detected in patients such as Mrs A, be cautious about accepting it as an explanation for stable stress incontinence or for non-specific symptoms such as tiredness. The clinical decision regarding Mrs A might be more difficult if Mrs A had reported a recent deterioration in stress incontinence as specific studies exploring such a presentation are lacking.
In a prospective community based study bacteriuria was detected in 1.9% of 13 464 schoolgirls and 0.2% of 1595 schoolboys aged (5 to 18 years).14 There is no evidence that detection and treatment of bacteriuria is of value in infants and children who do not have a clinical presentation that suggests urinary tract infection or systemic sepsis. Moreover, authoritative guidance exists that neither antimicrobial treatment nor follow-up is appropriate in this setting.15 Therefore it is not appropriate to test children in whom there is no clinical basis for suspecting urinary tract infection. Although most children with urinary tract infection present to primary care with dysuria and frequency it is important also to culture urine in infants and children presenting with fever when no clinical features point to another focus (for example, cough pointing to the respiratory tract).15
Women with diabetes
In women of any age, asymptomatic bacteriuria is more common in those with diabetes than in those without.5 16 Prospective cohort studies show no difference in outcome, including no difference in the incidence of symptomatic urinary tract infection, between diabetic women with and without asymptomatic bacteriuria at 18 months or at 14 years’ follow-up.5 14 Thus, although asymptomatic bacteriuria is more common in women with diabetes, no evidence supports treatment and therefore testing is not appropriate.
People with longstanding urinary catheterisation
Established bacteriuria is almost universal in this group, and defining urinary tract infection in this group is problematic. Consensus guidelines recommend that testing for bacteriuria and antimicrobial treatment of asymptomatic bacteriuria should generally be avoided in patients with indwelling urinary catheters.17 Expert opinion suggests that treatment should be limited to those with at least one of the following: fever, new costovertebral angle tenderness, new onset delirium, or rigors.18 19
Mrs A presents a difficult management problem. Although she has longstanding urinary tract symptoms (stress incontinence), her current presenting symptoms do not include any specific, new urinary tract features or a specific feature of infection such as fever. Her and her doctor’s views may also be shaped by the fact that she had previously had courses of antibiotics for bacteriuria. The doctor may be concerned that intercurrent urinary tract infection may account for or contribute to her malaise.
What tests for bacteriuria are available?
Tests that may be considered include urine dipstick analysis and microscopy and culture of urine.
In general terms urine dipstick analysis provides immediate results to help decision making and tests for other parameters (for example, glucose, protein, and haemoglobin), which may help in evaluating whether the patient has other medical conditions such as diabetes or glomerulonephritis. However, it does not give a specific microbiological diagnosis or guide selection of targeted antimicrobial treatment, as laboratory microscopy and culture do.
For Mrs A there is no clear indication for testing a urine sample by any method and no need for immediate results to guide treatment. In general, if a doctor decides to test a urine sample in a situation where there is no urgency and a relatively complex history including multiple previous antimicrobial exposures, there are advantages in using the more definitive laboratory result, which will include antimicrobial susceptibility testing if appropriate.
For laboratory tests it is usual to ask patients for a midstream urine sample—on the assumption that the initial urine stream is more likely to be contaminated by bacterial flora from the urethra. However, prospective studies comparing results from mid-stream urine samples with routine urine samples (without reference to timing) show that requesting a urine sample midstream is no better than asking the patient to collect a sample at any stage of micturition.20 21
How to interpret test results
Urine dipstick analysis
Reported positive and negative predictive values of dipstick testing vary considerably between studies, partly owing to differences in the pretest probability in the populations studied.22 23 These studies evaluate the performance of dipsticks in patients with symptoms clearly suggestive of urinary tract infection. For example, a recent prospective multicentre study in primary care (including 427 adult female patients from 67 practices with symptoms suggestive of urinary tract infection) found that a positive test result for either nitrite (indicative of bacteriuria) or leucocyte esterase (indicative of pyuria) was associated with an increased probability of urinary tract infection compared with pretest probability. A positive test result both for nitrite and for either leucocyte esterase or red cells had a positive predictive value of 92%, and a negative test for all three parameters had a negative predictive value of 73%.23
However, the relevance of this study to Mrs A is uncertain. An observational cohort study evaluated 200 patients older than 65 presenting to an emergency department.24 In one cohort (referred to as asymptomatic) of 100 patients, the patients were afebrile and presented with complaints such as minor trauma or chest pain, so infection was in no way a consideration. The second cohort (referred to as symptomatic) presented with acute confusion, weakness, or fever but without symptoms specifically pointing to the urinary tract or other specific source of infection. The findings suggested that test results for nitrite and leucocyte esterase correlate poorly with urine culture results in both groups and are therefore of limited value for patients such as Mrs A.
Laboratory microscopy and culture of urine
A laboratory report of pyuria indicates inflammation of the urinary tract. Some laboratories may perform initial microscopy on all samples and not proceed to culture on most urine samples in which pyuria and/or bacteriuria is not observed on microscopy. This practice helps to manage workload. Evidence from a systematic review indicates that absence of pyuria and bacteriuria on microscopy effectively excludes infection, at least in children.25 Semiquantitative urine culture to determine the species of bacteria present, estimate their approximate numbers, and perform susceptibility testing if appropriate remains a standard method of evaluation for urinary tract infection.
Problems of testing and treatment
In clinical practice it can be difficult to decide whether to test for and treat bacteriuria in an individual patient with non-specific symptoms. Difficulties may be compounded, as with Mrs A, by a background of stress incontinence, which may tend to focus attention on the urinary tract even though the available evidence does not suggest an association between this symptom and urinary tract infection. If testing and a trial of treatment of bacteriuria are considered in a non-pregnant patient with symptoms that are not clearly related to urinary tract infection, we suggest that laboratory microscopy and culture are the most likely tests to be useful. If bacteriuria is present it may be appropriate—particularly in women—to determine with a repeat culture whether bacteriuria is persistent. If a decision to treat it is made it is appropriate to use the results of the culture and susceptibility tests wherever possible to guide the selection of a safe and narrow spectrum agent (such as nitrofurantoin or trimethoprim). It is then important to explain the uncertainty of benefit to the patient and critically assess the clinical and microbiological response to treatment. Our experience is that once a patient with bacteriuria has an established conviction that they live with repeated or chronic urinary tract infection it can be difficult to persuade them otherwise. This can distract the patient and doctor from dealing with other important concerns and tends to result in repeated courses of antimicrobial agents or prolonged prophylactic antimicrobial treatment, as well as perhaps resulting in unnecessary specialist referral and investigation.
Culture of Mrs A’s urine as requested by your locum showed ≥100 000 colony forming units per millilitre of Escherichia coli resistant to ampicillin, cefuroxime, co-amoxiclav, trimethoprim, and ciprofloxacin but susceptible to nitrofurantoin. On review, Mrs A’s tiredness and malaise have improved. You explained that the urine sample contained bacteria but that bacteria can be found in the urine of about one in every five women of her age and that this is not a problem for most people. You explained that antibiotics often fail to eradicate the unwanted bacteria in the urine but that they can kill “good” bacteria and so can cause thrush (candida) and diarrhoea. You advised that you would like to hold off the antibiotics while checking out her problem with leaking urine. Mrs A agreed to this plan.
Bacteriuria describes the presence of bacteria in urine. Bacteriuria may result from the presence of bacteria in the urine in the bladder or from contamination of urine during micturition through contact with the normal bacterial flora of the urethra, vagina, or perineum.
Clinically significant bacteriuria
The term clinically significant bacteriuria is used to differentiate laboratory evidence of bacteriuria in the bladder from bacteriuria that probably results from contamination during micturition. Clinically significant bacteriuria is generally accepted as detection of more than 100 000 colony forming units of a single type of bacterium per millilitre of urine. Contamination with normal flora typically results in lower numbers of bacteria per millilitre and/or mixed bacterial species.5 However, the criterion of more than 100 000 colony forming units per millilitre of urine has limitations, in that numbers as low as 1000 colony forming units per millilitre may be clinically significant in patients with symptoms of urinary tract infection.2 9
Asymptomatic bacteriuria refers to clinically significant bacteriuria in those without symptoms that point to a urinary tract infection. The term asymptomatic may be unhelpful as it may be taken literally (the patient is without symptoms of any kind).
Cite this as: BMJ 2011;343:d4780
This series of occasional articles provides an update on the best use of key diagnostic tests in the initial investigation of common or important clinical presentations. The series advisers are Steve Atkin, professor, head of department of academic endocrinology, diabetes, and metabolism, Hull York Medical School; and Eric Kilpatrick, honorary professor, department of clinical biochemistry, Hull Royal Infirmary, Hull York Medical School. To suggest a topic for this series, please email us at.
Competing interests: All authors have completed the ICMJE uniform disclosure form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare: no support from any organisation for the submitted work; AWM received payment for lectures from Mepha LLC, Switzerland, in the previous three years; no other relationships or activities that could appear to have influenced the submitted work.
Contributors: MC prepared the original draft and collated comments. AV and AWM critically reviewed the manuscript and proposed changes through direct and electronic discussion of the content. MC is the guarantor of the article.
Provenance and peer review: Commissioned; externally peer reviewed.
Patient consent not required (patient anonymised, dead, or hypothetical).