A tale of two bladders

Uncomplicated UTI

We define uncomplicated UTI as cases of nonpregnant women (primarily aged 18 to approximately 50 or 60) presenting with irritative bladder symptoms such as frequency, dysuria, urgency, sense of incomplete emptying, and hematuria, in various combinations.

Prevalence and diagnosis

Studies have asked how many women who make appointments with their primary care providers for UTIs have UTIs according to urine culture results, the traditional criterion standard; 50% to 80% of women who think they have a UTI, do.¹ Owing to the limitations of urine culture (discussed below), this prevalence is likely higher. To compare this with the prevalence of strep throat with the highest modified Centor score of 5, the risk of group A streptococcus is around 50%, the bottom of the UTI prevalence range.²

One could argue that 50% to 80% probability of disease is high enough to justify treatment for a condition in which the risk-benefit ratio favours treatment.³ However, let us consider what else is used to make the diagnosis¹: Can specific symptoms help rule in or rule out the diagnosis? Are urine dipstick results helpful? How reliable are urine culture results?

Quick guide to understanding likelihood ratios. Positive likelihood ratios indicate if the symptom, sign, or test result will help make the diagnosis, with scores of 1 meaning they are not helpful at all and larger scores, such as 10, indicating that positive results can be very helpful in making the diagnosis. Negative likelihood ratios indicate if the test will help rule out the diagnosis, with scores of 1 meaning not helpful at all, and lower scores, such as 0.1, indicating that negative results can be very helpful in ruling out the diagnosis.

Symptoms. When considering which symptoms predict the diagnosis of UTI, defined by positive urine culture findings, none stand out as particularly helpful to rule in or rule out the condition (Table 1).¹ Physician-elicited core symptoms such as urgency, dysuria, and frequency do not contribute meaningfully to the diagnosis. Perhaps this is owing to the self-selection process of women presenting with possible UTI and the high prevalence at presentation. Regardless, beyond what has already prompted the woman to present, further physician-elicited symptoms have limited value.

Dipstick urinalysis. In the case of dipstick urinalysis (Table 1),¹ the presence of blood or leukocyte esterase ≥1+ does not appear to meaningfully help rule in or rule out a UTI. In the case of nitrites, a negative test result or result of 0 is not helpful in ruling out a UTI. A nitrite result of ≥1+ indicates a moderately increased probability of UTI. For example, if the prevalence (the probability of UTI) were 60% on presentation, a nitrite result of ≥1+ would yield a posttest probability of approximately 90%. The only caveat is that the 60% probability was likely already high enough to initiate antibiotics.

Urine culture. Perhaps urine culture is not the ideal criterion standard. In one study of 220 symptomatic women, 80% had a positive culture result but 96% were positive for Escherichia coli on polymerase chain reaction testing.¹ This would indicate that the urine culture missed 15% of E coli infections. In another study of 42 untreated symptomatic women with initially negative culture results, 31% had a positive culture finding within 6 weeks.¹ This would suggest that relying on culture results to be positive before treatment would miss women with cystitis.

Given all of that, women presenting for suspected UTI have prevalence higher than 50% to 80%, and eliciting further symptoms or using dipstick urinalysis is likely unnecessary. This would indicate that treating uncomplicated UTIs without requiring an office visit or further testing could save both clinician and patient time and system resources.

Treatment

Can a UTI be left untreated or managed symptomatically only?⁴ In about one-third of nonpregnant women, symptoms will resolve within 4 to 7 days without treatment, and this may be an acceptable risk for some women who prefer to avoid antibiotics. At 3 to 4 days, 46% of nonpregnant women given nonsteroidal anti-inflammatory drugs (NSAIDs) versus 67% given antibiotics were symptom free, for a number needed to treat of 5 for treating with antibiotics over NSAIDs. Treatment with NSAIDs alone increased the risk of fever or pyelonephritis by 1% over 1 month.

Antibiotic treatment choices include β-lactams, fluoroquinolones, fosfomycin, nitrofurantoin, and trimethoprim-sulfamethoxazole. Systematic reviews have compared different antibiotics and found that all antibiotics are similarly effective in symptom resolution with none showing superiority.⁵

Rates of adverse events differed slightly, with rash more common with trimethoprim-sulfamethoxazole (2.6%) and β-lactams (6%) versus nitrofurantoin (0.2%) and fluoroquinolones (0.1%).⁵ In addition, regional differences in resistance are evident, and clinicians can access resistance information through local antibiograms available online (https://www.lifelabs.com/healthcare-providers/reports/antibiograms/) for different provinces and regions.

Duration of treatment has also been examined and differs among antibiotics.⁵ Nitrofurantoin effectiveness seems most certain with 5-day treatment, with very low-quality and indirect evidence for shorter durations. At typical doses β-lactams, fluoroquinolones, cephalosporins, and trimethoprim-sulfamethoxazole are similarly effective with 3-day versus 5- to 10-day treatment. Fluoroquinolone (eg, ciprofloxacin, norfloxacin) studies have compared 1- or 3-day treatment and found that these drugs may be slightly (approximately 5%, relatively) more effective at symptom resolution with 3-day versus 1-day treatment. Fosfomycin has been studied only as a 1-day treatment.

We suggest that treating symptomatic women empirically with antibiotics is a reasonable approach for uncomplicated UTI. This approach has been studied in a primary care–based RCT⁶ among 309 nonpregnant women presenting to primary care with uncomplicated UTI. Women were randomized to 1 of 5 clinical pathways to guide antibiotic treatment; in 2 groups, patients received antibiotics based on empirical evidence either immediately or delayed by 48 hours; in another 2 groups patients received antibiotics immediately if they had positive dipstick results or if they had 2 or more symptoms; and in 1 group patients received antibiotics only after positive results on midstream urine analysis. All approaches provided similar symptom resolution. In addition, an RCT⁷ of 59 nonpregnant women with uncomplicated UTI symptoms and negative findings on dipstick urinalysis randomized women to antibiotics or placebo and found less dysuria after day 3 with antibiotics (24%) versus placebo (74%).

Asymptomatic bacteriuria

Here we move on to the more complicated part of our tale. Asymptomatic bacteriuria (ASB) is defined as the presence of bacteria in urine without symptoms or signs that could be attributed to this presence. The incidence of ASB increases with age. It occurs in 5% to 20% of people in the community older than 80 (in women more commonly than men), and in institutionalized people it occurs in 25% to 50% of women, 15% to 40% of men, and 100% of catheterized patients.⁸

People who are cognitively intact can more reliably report symptoms such as those of an uncomplicated UTI and those commonly ascribed to pyelonephritis (eg, flank pain). But what about people who have cognitive impairment?

Our first question was: Does bacteriuria lead to an altered mental state? We found that much of the evidence is severely flawed by inconsistent or incorrect definitions of UTI. The definition of UTI often includes altered mental state or delirium (without infection symptoms) with bacteriuria. For example, in one study more than 57% of patients with UTI diagnoses had no UTI symptoms. Three SRs of observational studies examined whether bacteriuria or UTI were associated with altered mental state.⁹ Two found an unclear association between “UTI” and delirium, owing in large part to concerns around case definition as described above, while the third pooled the data, despite this limitation, and found an association odds ratio of 2.67. Another observational study found no association between ASB and delirium.⁸

Our next question examined whether treating ASB improved an altered mental state. One small RCT and 2 small observational studies of patients with new-onset delirium and bacteriuria found no difference in mental status or functional recovery when treated with antibiotics.⁸ We also examined whether treating ASB improved any clinical outcomes, reviewing 5 SRs of 3 to 9 RCTs with 328 to 1087 patients, all finding similar results.⁸ The most recent⁸ found no difference in mortality or symptomatic UTI but did find antibiotic use increased the rate of adverse events (eg, diarrhea, rash, candidiasis) from 0.7% to 6.5%.

Finally, we asked whether antibiotic reduction interventions for ASB affected patient outcomes. We found 3 RCTs that looked at antibiotic stewardship interventions in long-term care facilities and all found a reduction in antibiotic use without increases in rates of hospitalizations, adverse events, or mortality.⁹

Overtreatment of ASB is an international problem, and inappropriate prescribing has been found in 35% to 93% of antibiotic prescriptions for UTIs in patients living in nursing homes.¹⁰ Automatic submission of urine for urinalysis or urine culture has a high probability of results being positive for bacteria, as discussed above, and is associated with a concomitant high probability of antibiotic prescribing.¹¹ Antibiotic stewardship initiatives are being instituted in many countries and are included in several Choosing Wisely Canada recommendations.¹² Guidelines recommend avoiding treating elderly (60 or older) patients without clear infectious symptoms and urge assessment for other causes, careful observation, and attention to hydration.^13-15

It is clear that we need better evidence to guide how we manage bacteriuria in confused or noncommunicative elderly patients. Randomized controlled trials for this common problem need to use definitions of UTI that reflect infectious symptoms¹⁶ and use consistent and validated methods of diagnosing delirium, such as the Confusion Assessment Method (CAM).^16,17 The CAM has been found to have a sensitivity of 82% and specificity of 99% for delirium. Box 1 presents criteria for ordering urine cultures for nursing home residents and for using the CAM to enhance consistency in the diagnosis of delirium.^16,18,19

Antibiotic stewardship programs have been initiated in many jurisdictions (Box 2).¹⁴ Another example of a management algorithm for long-term care is available from the Association of Medical Microbiology and Infectious Disease Canada.²⁰