adverse consequences of asymptomatic urinary tract infections in adults

Adverse Consequences of Asymptomatic Urinary Tract Infections in Adults

RICHARD PLATT, M.D. Boston, Massachusetts

Asymptomatic urinary tract infection (UTI) in adults can be associated with serious sequelae. During pregnancy, it causes pyelonephritis and low-birth-weight deliveries, both of which can be reduced in frequency by treatment of the bacteriuria. This asymptomatic infection is also associated with an increased risk of psrinatal fetal death; no beneficial effect of therapy has been observed for this condition, however. Asymptomatic UTI has been associated with excess mortality in adult women in a general community setting and in adults of both sexes in a nursing home. To date, there has been no adequate test of treatment of asymptomatlc UTI in these groups. UTls that result from short-term, indwelling bladder catheterization in acute-care hospitals are also associated with a marked Increase (relative odds equal to 2.8) in the risk of dying during the hospitalization. A randomized trial of preventive measures that reduced the risk of catheter-associated UTI demonstrated a decrease in mortality commensurate with the lesser risk of infection. These data suggest that asymptomatic UTI may increase the risk of death under certain circumstances. Additional studies are indicated to confirm the phenomenon, to identify high-risk persons, and to determine whether prevention and/or treatment of asymptomatic UTI reduces these effects.

A large body of evidence documents the adverse manifestations and consequences of symptomatic urinary tract infections (UTls). Local symptoms or manifestations include cystitis, the frequency-dysuria syn- drome, prostatitis, and pyelonephritis. These infections can cause substantial acute discomfort or morbidity and in some cases are a major cause of bacteremia. Conversely, the consequences of asymptomatic bacteriuria are not as readily perceived, but several types of evidence suggest that they also may produce serious adverse effects. This discus- sion therefore touches briefly on the basis for establishing the diagnosis of asymptomatic UTI and then considers some of the potential consequences of such an infection in adults.

DIAGNOSIS OF ASYMPTOMATIC BACTERIURIA

From the Channing Laboratory, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts. Requests for reprints should be addressed to Dr. Richard Platt, Channing Laboratory, 180 Longwood Ave- nue, Boston, Massachusetts 02115.

The introduction of quantitative methods of culturing urine specimens has led to an appreciation of the existence of asymptomatic microbial coloni- zation of the urinary tract. Under most circumstances, the basis for diagnosis is the recovery of greater than or equal to 100,000 colony-forming units (cfu)/ml of a single bacterial species in at least two consecutive, clean-voided urine specimens [l]. It is important to keep in mind that this criterion was selected on empiric grounds because it provides a reason-

June 28,1987 The American Journal of Medlclne Volume 82 (suppl8B) 47

SYMPOSIUM ON NORFLOXACIN-PLAlT

TABLE I Effect of Asymptomatic UTI on Pregnancy more likely to end in perinatal death. Table I shows the Outcome magnitude of these effects to be quite large.

Outcome Relative Risk” Confidence Interval’

Pyelonephritis 26.2 20.2-33.9 Low birth weight 1.7 1.4-2.2 Perinatal death 1.6 1.6-2.4

*Relative risk of a given outcome in women with asymptomatic UTI during pregnancy, compared with women who had no asymptomatic UTI. +95 percent. Modified with permission from (41.

TABLE II Effect of Treatment of Asymptomatic UTI on Outcome of Pregnancy

Outcome Relative Risk* Confidence Interval+

There is relatively strong evidence that asymptomatic UTI is not merely a marker for threatened pregnancies. Several randomized, double-blind trials have shown that antibiotic treatment of asymptomatic UTI significantly reduces the risks of both pyelonephritis and low-birth-weight delivery (Table II). These studies have not found an effect of treatment on perinatal mortality, however, although the absence of such an effect does not exclude the possibility, since the total number of deaths observed in these randomized trials was relatively small and many other factors clearly play major roles in this outcome. The low power of the existing data to address the relation between treatment of asymptomatic UTI and perinatal death is reflected in the broad confidence intervals for its effect (Table II).

Pyelonephritis 10.7 6.9-16.6 Low birth weight 1.4 1.0-l .9 Perinatal death 0.9 0.3-2.5

“Relative risk of a given outcome in women with untreated asymptomatic UTI during pregnancy, compared with women whose asymptomatic UTI was treated during pregnancy. +95 percent. Modified with permission from [4].

In summary, the epidemiology of asymptomatic UTI in pregnancy, its natural history, and the consequences of its treatment are solidly established. Further, this evidence strongly supports the general strategy of screening all pregnant women for bacteriuria and vigorously treating those who are infected.

ASYMPTOMATIC UTI IN THE GENERAL ADULT POPULATION

able separation between the minority of individuals whose urine persistently yields microbial growth and the larger number whose urine does not. The reliability of the method depends on a number of factors, including the cleansing of the perineum and the delay in inoculating the urine sample onto culture medium. It is important to note, however, that the urine sample need not be a midstream specimen.

Several million nonhospitalized persons in the United States have asymptomatic UTls. The majority of these are women, in whom the prevalence increases by roughly 1 percent per decade of age; i.e., approximately 5 percent of 50-year-old women are bacteriuric at any given time. Bacteriuria in men appears principally to be a conse- quence of instrumentation of the urinary tract, which sometimes accompanies treatment of a prostatic obstruc- tion or other condition.

This criterion remains appropriate for the diagnosis of asymptomatic bacteriuria (or, as used interchangeably, asymptomatic UTI) in noncatheterized persons. A criterion of 100,000 cfu/ml has also been used for most studies of catheter-associated UTI, although a report by Stark and Maki [2] suggests that a lower threshold may be appropriate. Recent discussions of different thresholds for symptomatic infections do not apply to the diagnosis of asymptomatic bacteriuria [3].

ASYMPTOMATIC BACTERIURIA IN PREGNANCY

Assessment of the consequences of asymptomatic UTI in this setting is difficult for several reasons. The most important problem is that there is relatively little information on the natural history of persons whose infection status is clearly defined. This paucity is not surprising, since development of such information requires that a sufficiently large group of individuals be identified, have cultures taken repeatedly, and then be followed to determine their subsequent health experience over a period that may involve decades. Such studies have not been conducted.

The epidemiology and sequelae of asymptomatic UTI have been studied most thoroughly in pregnancy. The prevalence of bacteriuria during pregnancy depends on a number of factors, such as parity, and overall findings have ranged from 3.5 to 7.1 percent [4].

Observational studies have consistently demonstrated that in women with this condition, there is a substantial excess risk of pyelonephritis developing before the end of the pregnancy; in addition, these women are more likely to deliver a low-birth-weight infant, and the pregnancy is

Evans and co-workers [5], however, did assess UTls in a defined population of women for periods of up to 13 years. Using door-to-door canvasing techniques, these investigators obtained urine culture specimens from women living in defined census tracts in Wales and Ja- maica. Cultures were obtained at baseline and again after three to six years. The single reliable outcome available to the investigators was mortality from any cause, which was assessed on a third survey conducted after seven to 13 years; the results showed there was a strong association between asymptomatic UTI and mortality in these women.

48 June 28, 1987 The American Journal of Medicine Volume 82 (suppl8B)


This association could not be accounted for by the fact that bacteriuric women were, on average, older than nonbacteriuric women, by differences in weight, or a number of other potential confounders including blood pressure, parity, ponderal index, marital status, or area of resi- dence. After adjustment for age and weight, (the only two factors that significantly affected the association), women who were bacteriuric on both surveys were approximately twice as likely to have died as those who were never bacteriuric; those bacteriuric on one survey were 1.6 times more likely to have died (Table III).

The significant association between asymptomatic UTI and overall mortality demonstrated in the study by Evans et al (51 deserves careful attention, since it has important implications for the care of many individuals. There are several possible explanations for the association, the sim- plest being that it may represent the occasional false- positive result that can occur with any study. Such a possibility is understood for all studies that demonstrate sta- tistically significant findings; the only way to test this possibility is to conduct additional studies that can confirm or deny the association. Another possible explanation of the association is that bacteriuria may have served as a marker for another condition, such as chronic disease, that was the “actual” cause of increased mortality. There are no data at present that allow assessment of this possibility. Evaluation of the causes of death on these patients’ death certificates did not suggest direct complications of UTI as a common event among the bacteriuric women. The absence of such a finding is open to several interpre- tations, however, including the very real difficulty in as- signing an accurate cause of death for individuals who die outside of a hospital where bacteremias, even if they occur, are unlikely to be detected.

If asymptomatic UTI does, in fact, increase mortality in a community setting, it will be necessary to determine whether antimicrobial treatment can affect the outcome. This will be an important test, since current data indicate that permanent eradication of asymptomatic UTI is ex- tremely difficult to achieve and may require multiple- course or long-term suppressive therapy. The decision to treat asymptomatic UTI is, therefore, likely to require a long-term commitment on the part of both physician and patient. Such therapy should be undertaken only if there is reasonable evidence that it will improve health or survival without overall risk. At present, the data do not permit such a conclusion. However, they do argue strongly for a vigorous assessment of the value of therapy for asymptomatic UTI in this setting. (A number of articles dealing with the epidemiology, natural history, and response to therapy of asymptomatic bacteriuria in ambulatory and nursing home settings have been published since this manuscript was prepared in October 1985. A selection of the relevant articles has been added to the references [l l-l 61 and merits the attention of readers interested in this topic.)

TABLE Ill Relation between Asymptomatic UTI and Subsequent Mortality in a Community- Based Population

Bacteriuria Status Risk Ratio’ Confidence Interval’

Positive at both surveys 2.0 1.05-3.92 Positive at one survey 1.6 0.99-2.57

*Relative risk of dying between the second and third surveys, compared with women who were bacteriuric at neither of the first two surveys. +95 percent. Modified with permission from [5].

ASYMPTOMATIC UTI IN NURSING HOME RESIDENTS

Asymptomatic UTI is present in a higher proportion of nursing home residents than in individuals of comparable age who live in the community. Much of this excess is probably attributable to the increased exposure of these individuals to short- or long-term bladder catheterization. The possible role of asymptomatic UTI in these individuals was observed by Dontas and colleagues [6] in a nursing home in Greece. They found that asymptomatic UTI was associated with the doubling of the lo-year risk of dying among 342 ambulatory, long-term nursing home residents who had no chronic diseases requiring frequent medical supervision or long-term antibiotic use, and no medically important renal disease or diabetes. There were 71 deaths among bacteriuric subjects, instead of the 39 that were expected on the basis of the age- and sex-adjusted survival of the total cohort (infected and uninfected). This excess mortality was highly significant and could not be explained by differences in age, sex, blood pressure, is- chemic changes on the electrocardiogram, or the resi- dent’s payor status. The authors noted no difference between the bacteriuric and nonbacteriuric residents with regard to the causes of death. As is true for noninstitut- ionalized individuals, it is difficult to determine how much weight to assign this fact.

The observation that asymptomatic UTI was a strong predictor of shortened survival among nursing home residents is subject to the same concerns that were noted earlier for the community-based study. The major issues to be addressed are whether the association is causal, i.e. does the UTI actually lead to the shortened survival and, if so, does treatment of the UTI modify the course of infection? One study addressed the issue by identifying men in nursing homes who remained bacteriuric after a course of standard therapy, at which point they were randomly as- signed to treatment or observation [7]. The investigators found no difference in the survival rates of the two groups. Several facts temper the interpretation of this study, however. First eradication or suppression of bacteriuria proved to be quite difficult, with 56 percent of the treated subjects being bacteriuric during at least half of the follow-

June 26, 1967 The American Journal of Madlclne Volume 62 (suppl 6B) 49

SYMPOSIUM ON NORFLOXACIN-PLATT

15/108 4/112 321643 261613 No Antibiotic Antibiotic

m Unsealed Collection Junction

n Sealed Collection Junction

291108 111112 481643 461613 No Antibiotic Antibiotic

Figure 1. Relation between catheter-associated UT/ and death during the same hospitalization. The four groups are all patients who had indwelling bladder catheters for short intervals in an acute-care hospital. They are categorized by whether they received an antibiotic or not, plus assignment to sealed-junction cathethers or conventional cathethers. The bottom panel shows the risk of acquiring infection for the four groups; the top panel shows the risk of dying during that hospitalization.

up period and virtually all being bacteriuric at the end of the study. This finding emphasized the difficulty in treating these infections satisfactorily. Second, the duration of observation was relatively brief, with the median duration of observation and therapy being 11 months, and the number of subjects was small (36 patients divided among the two regimens). As a result, this study did not provide a strong test of the value of therapy, since it may have failed to detect important differences in survival, even if they ex- isted.

The high prevalence of asymptomatic UTI among nursing home residents implies that if this condition does cause adverse consequences such as excess mortality, it affects a large number of people. A rigorous assessment of its consequences, and of the implications of therapy, are needed to determine: (1) whether the reported effects are accurate; (2) whether acceptable therapeutic regimens can be devised that will allow eradication or suppression of bacteriuria; and (3) whether successful therapy of the condition reduces these consequences. (A number of articles dealing with the epidemiology, natural history, and response to therapy of asymptomatic bacteri-

uria in ambulatory and nursing home settings have been published since this manuscript was prepared in October 1965. A selection of the relevant articles has been added to the references [l l-l 61 and merits the attention of readers interested in this topic.)

ASYMPTOMATIC UTI IN CATHETERIZED PATIENTS

Convincing evidence that asymptomatic UTI causes excess mortality depends, in part, on the demonstration that treatment or prevention of infection reduces mortality. As noted earlier, there has not yet been a sufficiently large trial in community or nursing home residents to test this hypothesis. Another major limitation of the assessment of asymptomatic UTI in these settings is the difficulty in relat- ing the outcome of infection to its time of onset. Studies by myself and my associates in catheterized patients in an acute-care setting have addressed both of these points; i.e., in studies of infection control techniques aimed at re- ducing the risk of infection during short-term catheterization of the bladder, we were also able to assess the impact of nosocomial UTI on mortality [8,9].

A randomized trial in 1,500 patients of two catheter drainage systems, one with a plastic seal to discourage disconnection of the catheter from the drainage bag and a “conventional” one without a seal, showed that the sealed-junction cathethers significantly reduced the risk of infection in patients who did not receive any antibiotic during the course of catheterization [8]. Those who received no antibiotic had a 27 percent risk of infection if conventional drainage was used, and a 10 percent risk if sealed- junction drainage was used. Patients who received an antibiotic had a 7 percent risk of infection, no matter which drainage system they received. The randomization of patients to treatment regimens, which resulted in different risks of infection, permitted a test of the hypothesis that nosocomial UTI increases mortality.

The percentage of subjects who died in each of the four patient groups was approximately proportional to the risk of acquiring a catheter-associated UTI (Figure 1). Pa- tients who received antibiotic or sealed-junction cathethers had either a 4 percent or 5 percent risk of dying. Patients who received neither had a 14 percent risk of dying. This observation of a dose-response phenomenon provides additional support for the hypothesis. As well, most of the excess mortality in the high-risk group (unsealed catheters and no antibiotic) occurred among patients who were infected. In the high-risk group, six of 15 patients who died had greater than or equal to 100,000 cfu/ml during the catheter course (this was the preset criterion of infection); five more of these patients either had 10,000 cfu/ml during the catheter course, or had greater than or equal to 100,000 cfu/ml later in their hospital course.

In order to assess fhe relation between nosocomial UTI and mortality more closely, we used stepwise multiple lo-

SD June 26, 1667 The American Journal of Medicine Volume 62 (suppl 68)

gistic regression to determine the factors other thah UTI that were significantly associated with death in hospital [9]. We found six such factors: severity of underlying disease, advahced age, treatment by medical (rather than surgical or other specialty) services, increased baseline serum creatinine concentration, longer duration of catheterization, and professional status of the person inserting the catheter. After adjusting for these factors, acquisition of catheter-associated UTI was still associated with a marked excess mortality. Patients who acquired an infection were approximately three times more likely to die during the hospitalization period than those who did not ac- quire infection (adjusted odds ratio equal to 2.8, 95 percent confidence limits equal to 1.5 to 5.1, p <O.OOl).

This finding of substantial excess mortality associated with asymptomatic UTI is striking, in part because it con- tradicts many clinicians’ perception of the experience. Several factors may contribute to this discordance. First, the connection between a death ‘and a catheter-associated UTI may not be recognized because cultures of urine or blood (or both) are not performed. Most catheter- associated UTls are asymptomatic and are not detected unless cultures are obtained on a routine basis; few hospitals have such a policy. Review of the charts of infected patients who died in our series revealed only two who had clearly documented blood stream dissemination of the urinary tract pathogen. However, there were another 10 patients who had unexplained fever, leukocytosis, or hypo- tension as pre-terminal events and who had no blood cultures obtained at the time of clinical deterioration. Thus, it is unclear how many of the deaths could have been linked more directly to the UTI if additional diagnostic studies had been performed. It is of interest, though, that the number of deaths accompanied either by a positive blood culture with a uropathogen, or by unexplained clinical deterioration, corresponded closely to the number of excess deaths predicted by the analysis.

A second factor that may have obscured the relation- ship between UTI and mortality was the fact that much of the excess mortality occurred among patients who were already critically ill. It was, therefore, not surprising to the physician when such a patient died, since he or she was at high risk of dying even in the absence of a UTI. In such an event, there is no compelling reason to attribute the death to a UTI, especially if there is no positive blood culture. At best, the clinician can only say that the cause of death may be multifactorial. Conversely, most infections


occur in patients who are not critically ill. In such patients, even a large excess risk, such as tripling of the mortality rate, would easily go undetected since the absolute increase in the risk of dying is quite small. An appreciation of the contribution of UTI can emerge only from assessment of a large number of cases of infected and uninfected persons. Such an assessment is clearest when it does not depend on a clinical judgment.

A separate finding of interest concerning the relation between catheter-associated UTI and mortality is the observation by Warren et al [lo] that among a population of women with long-term indwelling catheters (all of whom were chronically infected), the incidence of fevers without an obvious nonurinary source was 1 .l per 100 days. Al- though most fevers were clinically mild and brief, and did not result in therapy, six of the 12 deaths that were re- corded while the population was under observation occurred during a fever that had no apparent cause other than a UTI.

COMMENTS

Asymptomatic UTI is a clear cause of morbidity during pregnancy and possibly a cause of prenatal mortality. There is ample evidence that treatment prevents a substantial proportion of pyelonephritis cases during preg nancy as well as low-birth-weight deliveries among women with asymptomatic UTI. Studies in several popula- tions have observed an increased risk of death associated with asymptomatic UTI and premature death. These include community-based studies, studies in nursing homes, and studies in patients undergoing short-term catheterization in acute-care hospitals. One study demonstrated a decrease in mortality rates commensurate with the reduction in the occurrence of nosocomial UTI. The data therefore suggest, but do not prove, that asymptomatic UTI increases the death rate. The presumed mech- anism by which asymptomatic UTI could cause excess mortality is by way of undetected bacteremia or liberation of endotoxin. If UTI does increase the risk of dying, as these data suggest, the most plausible interpretation is that it serves as an extra burden to the patient. Some tol- erate this easily, some with difficulty, and some not at all.

Because of the large number of persons with asymptomatic UTI, and because of the possibility that many of these infections can be prevented or treated, additional investigation of the relation between this infection and death is warranted.

REFERENCES

1. Kass EH: Asymptomatic infections af the urinary tract. Trans 3. Stamm WE, Counts GW, Running KR, Fihn S, Turck M, Holmes Assoc Am Physicians 1956; 69: 56-63. KK: Diagnosis of coliform infection in acutely dysuric women.

2. Stark RP, Maki DG: Bacteriuria in the catheterized patient. What N Engl J Med 1982; 307: 463-468. quantitative level of bacteriuria is relevant? N Engl J Med 4. Kass EH, Platt R: Urinary tract and genital mycoplasmal infec- 1984; 311: 560-564. tion. In: Wald NJ, ed. Antenatal and neonatal screening, 1st

June 26, 1957 The American Journal of WdlClne Volume 62 (suppl 6B) 51


ed. New York: Oxford University Press, 1984; 345-357. 5. Evans DA, Kass EH, Hennekens CH, et al: Bacteriuria and sub-

sequent mortality in women. Lancet 1982; I: 158-158. 8. Dontas AS, Kasviki-Charvati P, Papanayioutou PC, Marketos

SG: Bacteriuria and survival in old age. N Engl J Med, 1981; 304: 939-943.

7. Nicolle LE, Bjornson J, Harding GK, MacDonell JA: Bacteriuria in elderly institutionalized men. N Engl J Med 1983: 309: 1420-1425.

8. Platt R, Polk BF, Murdock B, Rosner B: Reduction of mortality associated with nosocomial urinary tract infection. Lancet 1983; I: 893-897.

9. Platt R, Polk BF, Murdock B, Rosner B: Mortality associated with nosocomial urinary-tract infection. N Engl J Med 1982; 307: 837-842.

10. Warren J, Damron D, DeForge B, Muncie H, Tenney J, Hoopes J: Acute complications of bacteriuria in aged women with long-term urinary catheters (abstr 1099). In: Abstracts of the 1985 Interscience Conference on Antimicrobial Agents and

Chemotherapy, Minneapolis, Minnesota. 11. Kass EH: Bacteriuria and excess mortality: what should the next

steps be? Rev Infect Dis 1985; 7: S782-S766. 12. Heinamaki P, Haavisto M, Hakulinen T, Mattila K, Rajala S: Mor-

tality in relation to urinary characteristics in the very aged. Gerontology 1986; 32: 187-l 71.

13. Nordenstam GR, Brandberg CA, Oden AS, Svanborg-Eden CM, Svanborg A: Bacteriuria and mortality in an elderly population. N Engl J Med 1986; 314: 1152-1158.

14. Boscia JA, Kobasa WD, Knight RA, et al: Therapy versus no therapy for bacteriuria in elderly ambulatory non-hospitalized women. JAMA 1987; 257: 1067-1071.

15. Nicolle LE, Henderson E, Bjornson J, McIntyre M, Harding GKM, MacDonell JA: The association of bacteriuria with resi- dent characteristics and survival in elderly institutionalized men. Ann Intern Med 1987; 106: 682-686.

16. Boscia JA, Abrutyn E, Kaye D: Asymptomatic bacteriuria in elderly persons: treat or do not treat? Ann Intern Med 1987; 106: 764-766.

52 June 26, 1667 The American Journal of Medicine Volume 82 (suppl 68)

adverse consequences of asymptomatic urinary tract infections in adults

Documents