Despite a lack of definitive data on the relation of bacteriuria to the pathogenesis and severity of incontinence in the nursing home population, recently implemented federal guidelines for the care of incontinence in nursing homes (Resident Assessment Protocol) [13] suggest that bacteriuria should be treated only when symptoms of urinary tract infection other than stable incontinence are present. If eradicating bacteriuria makes the bladder less irritable in this patient population, then the severity of incontinence and the use of expensive adult diapers might be reduced [14]. If, however, eradicating bacteriuria has no effect on the severity of incontinence, no rationale would exist for exposing incontinent nursing home residents with bacteriuria to the potential added morbidity and expense of antimicrobial therapy. We conducted a clinical trial to determine whether eradicating bacteriuria affects the severity of incontinence among nursing home residents. We defined bacteriuria as would a practicing clinician in a nursing home rather than by the strict definition used in most studies of the epidemiology of asymptomatic bacteriuria (that is, two consecutive cultures with growth of more than 10⁵ colony-forming units [CFUs] of the same organism). We hypothesized that sterilizing the urine would have no short-term effect on the severity of chronic incontinence in this population.

Methods

Top Methods Results Discussion Author & Article Info References

Setting

Our study was the first phase of a multifaceted clinical trial addressing the assessment and treatment of incontinence in nursing homes. It was done in one nonprofit and five proprietary nursing homes located close to the offices of the University of California at Los Angeles Borun Center for Gerontological Research at the Jewish Home for the Aging. The nursing homes had a total of 1011 beds (range, 99 to 256 beds) and a total of 832 residents.

Patients

We identified potential patients by asking nursing home staff to identify residents who were incontinent of urine on a regular basis [that is, several times per week to several times per day]. Patients were excluded if 1) their care was being reimbursed by Medicare [indicating either short-term rehabilitation or medical instability]; 2) their prognosis, as determined by the research staff nurse, was so poor that they would probably not live at least 3 to 4 months to complete the protocol; 3) daytime incontinence was not documented by random-hour checks for wetness [described below]; 4) the resident had a permanent indwelling bladder catheter; 5) the resident failed a cognitive status screening test [to pass, the resident had to either say their name or reliably point to one of two objects]; and 6) the resident manifested a severe behavioral disturbance, such as physical aggression or verbal abusiveness, during the wet-checking procedures.

Informed consent was obtained from residents who could provide it (as determined by a facility nurse or social worker). If the resident was not capable of consenting, their assent was obtained (by describing a simple version of the protocol and then observing their cooperation as the protocol was initiated), and consent was then obtained from a responsible party.

Procedures

All patients for whom informed consent was obtained had a focused history, functional status assessment, cognitive assessment with the 30-point Mini-Mental State Examination, targeted physical examination, urinalysis, and urine culture. For the functional assessment, research staff used the Multidimensional Observational Scale for the Elderly Subject [15] to interview nursing home staff who knew the residents well. (When our study was implemented, the Minimum Data Set was not yet available.) Research staff used a standardized scale (Performance on Timed Toileting Instrument [16]) to objectively assess functional status specifically related to toileting skills. The physical examination was done by research staff and a physician coinvestigator and included abdominal, genital, pelvic, rectal, and neurologic examinations. Research staff collected urine from female patients by cleaning the perineal area with Betadine and having them void into a fracture bed pan or measuring hat that had been cleaned with an antiseptic solution. We compared the culture results of urine samples obtained by this procedure to urine samples obtained by catheterization in 101 of our female patients [17]. The prevalence of bacteriuria in the catheterized specimens was 29%. The sensitivity, specificity, and positive and negative predictive values of this procedure in detecting bacteriuria (with a catheter specimen used as the criterion standard) are 90%, 92%, 81%, and 95%, respectively. For men who could not void voluntarily, we used a previously validated technique that involved cleaning the glans penis with Betadine, applying a clean condom catheter, and processing the first voided specimen [18].

Urine specimens were subjected to screening tests done by research staff (including a dipstick method for leukocyte esterase and nitrite and a rapid enzyme-based test for bacteriuria [Uriscreen, Ventrex Laboratories, Portland, Maine]). A bioclinical laboratory used standard techniques to do a microscopic urinalysis and a urine culture and sensitivity test. We defined pyuria as the presence of more than 10 leukocytes per high-power field on microscopic examination of spun urine. Cultured specimens that grew more than 50 000 CFUs were considered to have significant growth. If significant growth of one or more urinary pathogens occurred on a second specimen, the patient was considered to be bacteriuric. Patients whose urine had significant growth of organisms that are not typically pathogens (for example, lactobacillus and -streptococcus) were not considered to be bacteriuric. Our quantitative definition of bacteriuria differs from the standard definition (presence of more than 100 000 CFUs) that has been used in most epidemiologic studies and intervention trials. However, there is controversy about the clinical significance of lower levels of growth [19], and in some studies in the elderly, bacteriuria has been defined as growth of less than 100 000 CFUs [20, 21]. We used the cutoff of 50 000 CFUs because our clinical laboratory reported results at this level, and we believe that most clinicians who practice in nursing homes would consider this to be significant growth when making decisions about treatment. Less than 10% of our cultures that were considered to have significant bacteriuria had growth of more than 50 000 CFUs. In addition, each patient identified as bacteriuric who had growth of more than 50 000 CFUs on one culture did have a second culture with growth of more than 100 000 CFUs; the only exceptions were two patients whose urine samples before treatment were collected by catheterization (see below).

Most specimens obtained after antimicrobial treatment were collected by catheterization, which was done for a determination of postvoid residual volume in the second phase of the longer ongoing clinical trial. Six patients whose urine was initially not bacteriuric by the criteria outlined above were considered to be bacteriuric on the basis of the catheterized specimen and were enrolled in the antimicrobial trial. For these patients, follow-up urine samples were collected by the methods described above rather than by repeat catheterization. Eradication of bacteriuria was documented by culture in all but seven cases; in these cases, the screening tests were used to document that bacteriuria had been eradicated. We have shown that in our population, the presence of at least two negative results on these screening tests has a negative predictive value of more than 90% [22].

Bacteriuric patients were randomly assigned to receive either immediate treatment or delayed treatment (which was administered 2 to 3 weeks after the immediate treatment group had been treated) with a 7-day course of norfloxacin, 400 mg orally twice daily. In three cases, the organism or organisms were not susceptible to norfloxacin, and another antimicrobial agent (trimethoprim-sulfamethoxazole) was used on the basis of results of the sensitivity testing. Wet checks, the outcome measure used in our study and described in detail below, were done before therapy in patients without bacteriuria and in both bacteriuric treatment groups, were repeated in all three groups after the immediate group had been treated, and were repeated again after the delayed group had been treated. We used this design to control for any possible spontaneous fluctuations in incontinence severity in the absence of treatment in the bacteriuric residents. In addition, treatment of the delayed intervention group (which essentially served as a bacteriuric control group for the immediate intervention group), allowed us to maximize the number of treated patients for analyses. Throughout this study protocol, research staff carefully and consistently avoided any behavioral interventions and only provided toileting assistance when the patient specifically asked for assistance. Experience in our previous research has documented that such self-initiated toiletings are uncommon in this population and that wet-check data are reliable in the absence of behavioral intervention [23].

Outcome Measures

Physical checks for wetness were done every hour between 7 a.m. and 7 p.m. for 3 days (11 checks per day; 33 checks total). During these checks, all patients either wore preweighed disposable absorbent briefs or were placed on preweighed disposable absorbent pads. The volume of all incontinent voids was measured except when stool that could not be easily removed was present. In these situations, the volume was treated as missing data. Any volume greater than 5 mL or any increase in pad weight of more than 5 g was counted as a wet episode. Reliability checks were done by senior staff on approximately 2% of all checks to help ensure ongoing accuracy of the wet-check data. Continent voids were measured using a "hat" or urinal. The maximum voided volume was the maximum volume of any one incontinent or continent void during the wet-check period. Wet-check data are reported as percentage of checks that were wet. During each 3-day period, we attempted to obtain all 33 checks. However, this was not possible because patients were commonly in activities, at meals, or occasionally out of the facility. Only 1 or 2 of the 33 checks were missing for most of the patients who had missing checks. In the 17 patients in the first facility, the baseline wet checks were only about two thirds of the expected rate, which was based on data from more than 30 nursing homes; these checks were considered to be unreliable because of the unavailability of an adequate number of trained research staff. The mean baseline rate of wet checks for all other patients was used as the baseline for these 17 patients. Results of repeat wet checks in these patients after additional staff had been hired and trained were consistent with both previous data and data subsequently collected on other patients in this trial. Data analyses were repeated with and without these 17 patients, and the results were essentially identical.

Data Analyses

We did data analyses using SPS software (SPS, Inc., Chicago, Illinois) for personal computers. In a two-way (3 x 3) repeated-measures analysis of variance that was the primary analysis for this study, there were three within-group measures (baseline, immediately after treatment, and after the delayed treatment group was treated) and three between-group levels (sterile, bacteriuric with immediate treatment, and bacteriuric with delayed treatment). The statistic of interest in this analysis was the interaction between groups over time. To control for confounding variables, we did 3 x 3 repeated-measures analyses of covariance using functional status (the self-care subscale of the Multidimensional Observational Scale for the Elderly Subjects and the Performance on Timed Toileting Instrument score) and mental status (Mini-Mental State Examination). We also did a 3-way analysis of variance so that we could use sex as a blocking variable for determining whether sex interacted with study group over time.

Chi-square analysis was used to examine the association of sex, ambulation, and pyuria status with bacteriuria status (sterile patients compared with all bacteriuric patients whose bacteria was eradicated). We used Student t-tests to compare the combined treated bacteriuric group with the sterile group for age, duration of residence, Mini-Mental State Examination score, and score on the self-care subscale of the Multidimensional Observational Scale for the Elderly Subjects.

Results

Top Methods Results Discussion Author & Article Info References

One hundred ninety-one patients were enrolled in the study. Ten patients (5%) were excluded because they were misclassified as having bacteriuria on the basis of erroneous preliminary culture results and were inappropriately treated with an antimicrobial agent. Of the remaining 181 patients, 90 (50%) met the criteria for bacteriuria. Two of these patients dropped out of the study because of illness before completing the protocol, and 17 (19%) still had bacteriuria after antimicrobial treatment. Of the 71 patients in whom bacteriuria was eradicated, 33 were in the immediate treatment group and 38 were in the delayed treatment group. Because we observed no significant differences in any independent or dependent measure between these groups, data are presented for the bacteriuric group as a whole. Of the 91 patients who did not meet the criteria for bacteriuria, 3 dropped out of the study because of illness before completing the protocol.

Selected characteristics of each group of patients are shown in Table 1. Compared with patients without bacteriuria, successfully treated bacteriuric patients were more often female, had been in the nursing home longer, had pyuria more frequently, and had a higher baseline frequency of urinary incontinence and percentage of checks that were wet. Of the 88 patients who were not bacteriuric, 7 (8%) had significant growth of a nonpathogen on initial culture; 3 of these 7 patients had pyuria. The pathogens reported for the bacteriuric group are listed in Table 2. Twelve (14%) of the bacteriuric patients grew more than one pathogen on the specimen that was used to verify bacteriuria.

Figure 1 shows the main outcome results with respect to the frequency of incontinence. Eradicating bacteriuria made no difference in the overall percentage of checks that were wet. In the nonbacteriuric group, the percentage of checks that were wet increased from 29% (95% CI, 26% to 32%) at baseline to 30% (CI, 27% to 34%) on repeated measurement. In the bacteriuric groups, the percentage of checks that were wet increased from 34% (CI, 30% to 38%) before treatment to 35% (CI, 31% to 39%) after the bacteriuria had been eradicated. We also calculated the proportion of patients whose percentage decreased by one third or more. Sixteen percent of the bacteriuric patients had this decrease in the frequency of incontinence after the bacteriuria was eradicated, but a similar proportion of the patients without bacteriuria (19%) had the same degree of change (P = 0.53). Thus, we could find no evidence that eradicating bacteriuria had any significant effect on the frequency of incontinence. Analysis of covariance showed no significant effect of functional or cognitive status. No significant interaction occurred between sex and study group over time, but the small number of men in the study limits our ability to exclude a real sex-related difference in the results. We were especially interested in determining whether patients with pyuria, presumably a sign of inflammation, had a different response to eradication of bacteriuria than did patients without pyuria. As shown in Table 3, changes in all outcome measures were similar in bacteriuric patients with and without pyuria.

View larger version (23K): [in this window] [in a new window]   Figure 1. Incontinence measured by percentage of hourly checks that were wet before and after successful eradication of bacteriuria in 71 incontinent nursing home residents, at baseline, and at a repeated measurement without intervention in 88 incontinent, nonbacteriuric residents. Bars represent 95% CIs.

Discussion

Top Methods Results Discussion Author & Article Info References

Our data clearly show that eradicating bacteriuria has no short-term effects on the severity of chronic incontinence in nursing home residents. Our conclusion supports the strategy, recommended in the Health Care Financing Administration's Resident Assessment Protocol for incontinence, of not treating otherwise asymptomatic bacteriuria in incontinent nursing home residents. However, we must emphasize that our data and conclusion are relevant only for nursing home residents who have chronic, stable symptoms of incontinence and no symptoms or signs of urinary tract infection. The latter may be subtle in this population and may include only increased confusion, anorexia, or decreased functional status [24]. When incontinence suddenly develops or existing incontinence suddenly worsens, urinary tract infection is one of several potentially reversible causes of transient incontinence [21]. Although there are no data on the effects of treating urinary tract infection on new or worsening incontinence, most experts recommend that if a nursing home resident becomes incontinent, or if infrequent incontinence suddenly worsens, bacteriuria should be treated unless other obvious causes are identified [25, 26].

Our study has several limitations. First, we only examined the short-term effects of eradicating bacteriuria. If bacteriuria does, in fact, cause inflammation and make the bladder more irritable in this population, then longer-term treatment may be necessary to diminish the inflammation and bladder irritability and to observe the effects on the severity of urinary incontinence. Because bacteriuria commonly recurs in the nursing home population [27], long-term suppressive treatment would be necessary even if positive effects on incontinence were documented. Second, although we found no effects of pyuria, we used a relatively crude measure for this finding. If we had used a more standardized and sensitive measure of pyuria, we might have identified patients for whom eradicating bacteriuria had beneficial effects. Third, we did not collect catheterized specimens for the cultures that were used to identify bacteriuria, and our definition of bacteriuria differed somewhat from the strict definition used in most epidemiologic studies of asymptomatic bacteriuria (growth of more than 10⁵ CFUs of the same organism on two consecutive specimens). Most of our patients met this strict definition. Although all of our patients had two urine cultures, in about 10% one of the cultures grew more than 50 000 but less than 10⁵CFUs. In addition, the results of the two cultures for about 10% of the bacteriuric patients were not identical. In most of these cases, however, this resulted from the growth of a second pathogen in addition to the same pathogen grown in the matching culture. Thus, our definition of bacteriuria and noncatheter method of specimen collection could have caused a few patients to be erroneously classified as bacteriuric on the basis of the epidemiologic definition [17]. However, we believe that the number of potential false-positive classifications is too small to significantly affect our results or conclusions and that our definition of bacteriuria is consistent with that which a practicing clinician would use to diagnose bacteriuria in the nursing home population. Finally, we did not do urodynamic studies and cannot report the precise types of incontinence seen or the effects of eradicating bacteriuria on any urodynamic variable. We do know from our subsequent studies that almost 80% of these patients had detrusor hyperactivity and urge-type incontinence determined by simple urodynamic testing, but such testing does not precisely define the underlying pathophysiology [28]. Our measurement of maximum voided volume might be considered a partial proxy for bladder capacity, but it does not account for total bladder capacity because we did not measure postvoid residual volume after each void.

Despite these limitations, our data have important implications for the care of nursing home residents. Combined with Nicolle and colleagues' studies documenting the absence of effects of treating bacteriuria on morbidity and mortality in nursing home residents [6, 7], as well as more recent data on very old (mean age, 80 years) women living in retirement communities [29], our data provide further validation for the practice of not treating asymptomatic bacteriuria in the nursing home population. This practice is consistent with the federal guidelines contained in the Resident Assessment Protocol for incontinence and can eliminate the substantial health care expenditure that arises from doing urinalyses and cultures, providing and monitoring antimicrobial therapy, and treating the side effects of such therapy in the nursing home setting.