The treatment of asymptomatic bacteriuria in the elderly.

Abstract

Urinary tract infection can occur at any age, but is particularly common in the elderly. Most elderly persons with urinary tract infection do not have symptoms; they have asymptomatic bacteriuria. Overall, of patients between ages of 65 and 70 years, about 15% to 20% of women and 0% to 3% of men are afflicted; in those over 80 years old the numbers rise to 20% to 50% in women and 5% to 20% in men.1–6 Do these common infections have an associated morbidity, or influence mortality, and who, if any, should be treated? The diagnosis of asymptomatic bacteriuria is not difficult. We require two positive urine culture results with more than 105 colony-forming units (CFU)/mL for the same organism in the absence of dysuria, which, if present, would denote symptomatic urinary tract infection. The value in diagnosis of a single positive urine culture result was recently assessed by Nicolle and associates7 who found an error rate (overdiagnosis) of 10%. Pyuria is a marker of urinary tract infection, and our group studied the relationship between pyuria and bacteriuria in the aged.8 That study included 266 ambulatory women with a mean age of 85 years. Clean-caught urine was obtained for culture, and pyuria was defined as showing more than 10 leukocytes/mm3 of uncentrifuged urine in a counting chamber. Bacteriuria was present if more than 105 CFU/mL was found on two occasions one week apart. Low-titer bacteriuria urinary tract infection is a recently described entity in patients with symptoms of dysuria.9 Assuming that asymptomatic low-titer bacteriuria exists in the elderly, we defined asymptomatic low-titer bacteriuria as a finding of 102 to 104 CFU of gram-negative bacilli per mL in subjects without dysuria. Absence of bacteriuria was defined as having less than 105 CFU of gram-positive bacteria per mL and less than 102 CFU of gram-negative bacilli per mL of urine. We found that pyuria was common in the absence of bacteriuria. Of 125 women with pyuria, 47 (38%) had neither low- nor high-titer (≥ 105 CFU/mL) bacteriuria. When pyuria was absent, however, high-titer bacteriuria was unlikely, being found in only 3 of 141 women without pyuria. Cf women with high-titer bacteriuria, 95% (61 of 64) had pyuria; 53% (17 of 32) of those with low-titer bacteria had pyuria; and 28% (47 of 170) with no bacteriuria had pyuria. Thus, pyuria almost always accompanies high-titer bacteriuria, but is a poor predictor of bacteriuria; pyuria per se is not an indication for therapy of asymptomatic bacteriuria. Our findings also suggest that low-titer bacteriuria occurs in the elderly. Several outcomes have been evaluated in attempts to assess the morbidity of bacteriuria.3,6,10–12 First, hypertension has been associated with bacteriuria;11 but this finding has not been confirmed.3,12 Second, decreased renal function was found in a cross-sectional survey in bacteriuric patients compared with those without bacteriuria.12 Another study, however, could not confirm these findings.3 Renal failure, of course, can occur in bacteriuric persons of any age when obstruction is also present. The findings suggest that one can not use concern about renal failure or hypertension as justification for initiating antimicrobial therapy for asymptomatic bacteriuria. Third, incontinence and its relationship to bacteriuria was specifically evaluated in one of our studies10; incontinence in the presence of bacteriuria is occasionally used as an indication to treat presuming that incontinence will improve with eradication of bacteriuria. In our study, subjects without dysuria served as their own controls by answering a questionnaire about symptoms when they were bacteriuric and when they were not bacteriuric. There was no difference in symptom scores between persons with and those without bacteriuria for questions about incontinence patients experienced when awake, asleep, or in association with coughing or sneezing. We also evaluated the frequency of change in symptom scores comparing subjects who were more symptomatic when bacteriuric with those more symptomatic when nonbacteriuric. The percentage of subjects having more incontinence when bacteriuric about equaled the percentage who were more symptomatic when nonbacteriuric. Between 69% and 94% had no change in incontinence symptom scores whatsoever. With an alpha of 0.01, the power of this study to detect clinically relevant differences in symptoms for subjects with and without bacteriuria was high, 0.995. Thus, this study provides strong evidence that incontinence or frequency of change in incontinence in patients with bacteriuria is unrelated to otherwise asymptomatic urinary tract infection. An increase in incontinence should not be used as an indication to treat nondysuric subjects with bacteriuria. Fourth, we also failed to associate bacteriuria with frequency, urgency, or symptoms reflecting a sense of well-being (anorexia, difficulty in falling asleep, difficulty in staying asleep, fatigue, malaise, and weakness). We concluded that bacteriuria in the elderly in the absence of dysuria is asymptomatic. Several studies have shown a relationship between bacteriuria and mortality,12–15 but the mechanism is unclear.6 This association has not been confirmed in some recent careful studies.16–18 Several important methodologic issues must be remembered when reviewing studies about bacteriuria and associated mortality. Age itself must be controlled for, as it may be a confounding variable. Concomitant disease must also be controlled for. It is possible, for example, that bacteriuria per se is not related to mortality, but merely indicates more severe underlying disease that causes mortality.19 It is important to note that subjects in some studies were classified as bacteriuric or nonbacteriuric solely on the basis of culture results obtained on entry to the study12,13,16,17; loss or acquisition of bacteriuria over the period of the study was not considered. Ideally, one should compare mortality among those never bacteriuric with those continuously bacteriuric. The pattern of bacteriuria may be important.6 Persistent bacteriuria occurs in institutionalized males18,20 and females,7 but reinfection also occurs.7,18,20 In our studies of the dynamics of bacteriuria among ambulatory non-hospitalized patients,4 bacteriuria was common, but persistence of bacteriuria was surprisingly uncommon. Rather, in this population acquisition and loss of bacteriuria was frequent, ie, turnover was high. Despite a high turnover, the prevalence overall remained stable. These results suggest that in some persistence is the rule, whereas in others transient bacteriuria is common. The possibility should be considered that mortality differs in subgroups having different patterns of bacteriuria.6 Treatment should be administered to patients of any age with symptomatic urinary tract infection. Many believe, in contrast, that bacteriuria in the elderly, in the absence of symptoms and/or obstruction, is benign and should not be treated.1,2,7,21,22 What leads to this conclusion? First, symptomatic disease does not develop invariably from asymptomatic infection, and the frequency of symptomatic infection in some populations is low. Nicolle et al7 found the incidence of probable genitourinary tract morbidity to be 0.05 episodes per patient year in institutionalized bacteriuric women who were receiving no therapy and were cultured monthly over an extended period. A study of institutionalized males found an incidence of 45 infections per 100 patients per year, but only 1 was symptomatic.20 In ambulatory women enrolled in a trial comparing therapy with no therapy for asymptomatic bacteriuria,23 we found an incidence for subsequent symptomatic infection of 16% and 8% over 6 months in the no-treatment and treatment groups, respectively. Second, in the elderly the recurrence rate for urinary tract infection after therapy appears high, and it remains to be demonstrated that extended infection-free intervals can be maintained. The ability to maintain patients free of infection for long periods is a prerequisite to studies evaluating the effect of therapy on mortality.6 Studies need to show that treatment effectively eradicates bacteriuria and lowers mortality before recommending that treatment be given to lower mortality.6 To illustrate, Nicolle et al20 found that relapse and failure was common in institutionalized males after single-dose therapy. Moreover, 100%, 83%, and 100% of antibiotic treatment courses of 2, 6, and 12 weeks, respectively, failed to cure bacteriuria or were followed by recurrence in those failing single-dose therapy. On the other hand, we compared treatment versus no treatment in ambulatory, nonhospitalized women and found 19 of 55 (35%) of the group receiving no therapy and 35 of 55 (64%) in the treated group (most received single-dose therapy) were free of bacteriuria at 6 months.23 More work is clearly needed to determine the factors promoting success. Third, the adverse reaction rate to antimicrobial agents in this population is high,7 and caution is required before aggressive diagnosis and therapy are recommended. With the current state of knowledge, it is difficult to justify the costs of routine screening and treatment in the elderly. In summary, asymptomatic bacteriuria in the elderly is common; infrequently evolves into symptomatic infection; causes little or no other morbidity including incontinence or change in the sense of well-being; and may or may not be associated with increased mortality. Therapy may be accompanied by adverse drug reactions; and extended bacteriuria-free intervals, in some populations, are hard to achieve. Until highly effective, safe, low-cost therapies are found and the benefits for morbidity and mortality are shown to outweigh the risks and costs, asymptomatic bacteriuria in the elderly should not be treated.