JOURNAL OF CLINICAL MICROBIOLOGY, June 1988, p. 1115-1119 0095-1137/88/061115-05$02.00/0 Copyright 1988, American Society for Microbiology Vol. 26, No. 6 Urine Specimen Collection with External Devices for Diagnosis of Bacteriuria in Elderly Incontinent Men LINDSAY E. NICOLLE,* GODFREY K. M. HARDING, JAMES KENNEDY, MARGARET MCINTYRE, FREDERICK AOKI, AND DAVID MURRAY Division ofinfectious Diseases, Department of Medicine, and Department of Microbiology and Infectious Diseases, University of Calgary, Sections of Infectious Diseases and Geriatric Medicine, Department of Medicine and Department of Medical Microbiology, University of Manitoba, and Deer Lodge Centre, Winnipeg, Manitoba R3A 1R9, Canada Received 21 September 1987/Accepted 26 February 1988 We determined the validity of using external devices for urine specimen collection from 24 elderly incontinent men residing in a nursing home by collecting three sequential specimens, two with external devices and then one by catheterization. The positive predictive value of organisms isolated in quantitative counts of 2105 CFU/ml in external devices for bladder bacteriuria was 86% for either sterile or clean collecting devices and 93% for the same organism in two consecutive specimens. The negative predictive value for organisms present in quantitative counts of <105 CFU/ml was 90% for both sterile and clean devices and 86% when the organism was present in both specimens. Contamination in external collection devices was not influenced by whether the device was clean or sterile, circumcision of the resident, or duration of time between device application and specimen collection. These data suggest that urine specimens collected by ward nursing staff with external devices are reliable for the diagnosis of bacteriuria in this patient population. Incontinence is a common problem in elderly men residing in nursing homes (5, 10). Collection of valid clean-catch urine specimens for culture from such individuals is frequently difficult or impossible due to lack of voiding control or lack of cooperation. For many of these residents, incontinence is managed through the use of condom drainage and leg bags (10). When urine specimens for culture are required, specimens are sometimes collected from these external drainage devices (1). This method of specimen collection may lead to contamination with flora colonizing the glans penis, compromising the validity of such specimens (2). Thus, it has been suggested that only specimens collected by in-and-out catheter, an invasive method, may be valid for this population. Bacteriuria has a prevalence approaching 50% in elderly institutionalized men (6, 8). Studies to clarify the pathogenesis, natural history, and therapeutic outcome of bacteriuria in this population require a reliable, noninvasive method for the collection of urine specimens. Recently, Ouslander et al. (9) examined the reliability of urine specimens collected with external devices and reported that specimens collected by a single study nurse using a standard protocol were valid for mnicrobiologic analysis. We report here our observations of the validity of such specimen collection where specimens were collected by the ward nursing staff. In addition, factors which might correlate with contamination and the presence of pyuria in specimens were examined. (This study was presented in part at the 26th Interscience Conference on Antimicrobial Agents and Chemotherapy, New Orleans, La., September 1986.) MATERIALS AND METHODS Patient population. The study subjects were 24 elderly men with incontinence who reside at the Deer Lodge Centre, Winnipeg, Manitoba. Fourteen of these had incontinence managed by chronic condom drainage. Characteristics of * Corresponding author. this study population have been previously described (8). The study was approved by the Faculty Committee on The Use of Human Subjects in Research of the University of Manitoba. Consent for participation in the study was obtained from each resident or, when he was incapable of giving consent, from an appropriate family member. Residents who were unable to give consent themselves or who had no suitable family member were excluded from the study. Study design. Three consecutive urine specimens, the initial two specimens obtained with a freshly applied condom and leg bag and the third obtained with a sterile in-and-out catheter, were obtained from each study subject. One condom specimen was obtained from a condom and leg bag sterilized by ethylene oxide, and a second specimen was obtained with the clean condom and leg bag as provided by the manufacturer. The two sets of collecting devices were packaged together under one study number and labeled A (first collection) and B (second collection). The order of clean or sterile specimens was determined randomly by using numbers generated from a random numbers table, and staff members collecting the specimen were blinded to whether the apparatus applied was clean or sterile. The third specimen, obtained by in-and-out catheter with aseptic technique, was collected by a medical house officer. Urine specimens collected with external devices were obtained by the regular ward nursing staff. Before application of the condom, the glans penis was cleaned with soap and water and rinsed with sterile saline. The condom and leg bag were examined every 10 to 15 min until a specimen was obtained, and the time from condom application to specimen collection was recorded. The second condom and leg bag were applied after a repeat cleaning of the glans with soap and water and rinsing with sterile saline. Temperatures of study patients were monitored three times daily for 72 h after specimen collection. All study participants received a single dose of an antimicrobial agent immediately after the catheter collection. 1115
1116 NICOLLE ET AL. Microbiologic methods. Specimens were refrigerated immediately after collection, and the three specimens from each study subject were transported together to the laboratory when collection was complete. Quantitative cultures were performed on all specimens by using 0.001- and 0.01-ml loops to inoculate blood agar and MacConkey agar. Standard microbiologic methods were used for isolation and identification of bacteria (7). Absolute leukocyte counts were performed on uncentrifuged urine with a hemacytometer chamber. The antibody-coated bacteria test was performed as previously described (4). Data analysis. The catheter urine specimen was the reference test, with any organism present in any quantitative count in the catheterized specimen considered an infecting strain. Organisms isolated from specimens collected by external devices but not present in the catheter specimen were contaminants. Sensitivity, specificity, and positive and negative predictive values were calculated for organisms in specimens obtained with clean and sterile condoms. Postcatheterization fever was present when a temperature elevation of at least 1 C above baseline occurred within 24 h after catheterization in a resident with no alternate explanation for fever. Standard statistical methods, including the Chi-square test and Fisher's exact test, were used for proportions. Nonparametric tests were used to compare differences in contaminating organisms, collection time, and leukocyte counts between specimen collection methods. The association between time to specimen collection and number of contaminants was examined by the Spearman Rank Order test. Statistical analyses were performed on an IBM PC computer with a Microstat (Ecosoft, Inc., Indianapolis, Ind.) software package. RESULTS The median age of the 24 study subjects was 78 years (range, 60 to 89 years). Chronic disease diagnoses included dementia (12 men); previous hip fracture and cardiac disease (9 men each); cerebrovascular disease, blindness, chronic pulmonary disease, and other neurologic diseases (6 men each); and peripheral vascular disease (5 men). Three had carcinoma of the bladder or prostate, and two had diabetes mellitus. Twelve (50%) of the study subjects were circumcised. None had maceration of the skin of the penis or any other visible penile skin lesions at the time of their participation in this study. The data for quantitative cultures for all subjects by ail collection methods are shown in Table 1. The first specimen collected was with the sterile device in 13 instances. Twenty of 24 subjects had bacteriuria iri catheter specimens, 7 (35%) with more than one organism. The time from condom application to collection of the urine specimen was not recorded for two residents. The median time to collection of specimens for the remainder was 115 min (range, 3 to 170 min) for the sterile apparatus and 60 min (range, 1 to 165 min) for clean specimens (P = 0.02; Wilcoxon rank sum). Three (15%) of 20 episodes of catheterization in men with bacteriuria were associated with postcatheterization fever. Two of these were transitory and resolved without further intervention, but one was associated with rigors and required extended antimicrobial therapy. Blood cultures were not obtained during this episode. Eleven patients (46%) had specimens collected by either type of external device with organisms isolated which were not present in the catheter specimens. The distribution of the J. CLIN. MICROBIOL. number of contaminating organisms did not differ between specimens collected with clean or sterile devices (P > 0.05; Wilcoxon rank sum). For sterile collecting devices, there was no correlation between time to specimen collection and number of contaminants (r = -0.255). For clean collecting devices, there was an inverse correlation between number of contaminants and time to specimen collection (r = -0.436). Five of 12 circumcised men and 6 of 12 uncircumcised men had contamination for both sterile and clean devices. Infecting and contaminating organisms are summarized in Table 2. Although the small number of organisms precludes analysis by different species, gram-positive organisms tended to occur more frequently as contaminants than gram-negative organisms (chi-square, 3.35; P = 0.067). The leukocyte counts in the urine specimens are shown in Fig. 1. Although there was variation in counts among different specimens from the same individual, there was no statistically significant difference in the range of leukocyte counts between the three collection methods (P = 0.58; Kruskal-Wallis test). Three residents (subjects 11, 15, and 17) had cell counts of less than 10 in the catheter specimen but substantially higher numbers in the externally collected specimens, suggesting that urethritis or prostatitis may contribute to pyuria. The antibody-coated bacteria test was positive for 16 of 19 subjects tested with bacteriuria. In one instance one specimen collected by an external device had a negative antibody-coated bacteria test, with a positive antibody-coated bacteria test in the other two specimens from that resident, but in all other cases antibody-coated bacteria test results were consistent among specimen collection methods. The reliability of specimens collected in external devices compared with those in the catheter specimens was determined by using a count of 2105 CFU of an organism per ml to designate bacteriuria from the external device (Table 3). Four (20%) and 3 (15%) of the 20 patients with bacteriuria had one or more organisms isolated in quantities of 2105 CFU/ml from sterile and clean external devices, respectively, which were not present in the catheter specimen. However, for all organisms isolated at 2105 CFU/ml in external devices, 86% were present in catheter urine with either sterile or clean devices. None of the four patients without bacteriuria would have been identified as bacteriuric by using the quantitative method. The two occasions where the condom collection did not predict bacteriuria (subjects 20 and 23) were instances where counts in the catheter specimen were <105 CFU/ml, with similar counts for condom collection specimens. On one occasion (subject 8), an Escherichia coif was isolated from the catheter specimen which was not identified in either condom specimen. Requiring that an organism be present at 2105 CFU/ml in the two external specimens increased the positive predictive value to 0.93. Organisms isolated with counts of <105 CFU/ml on only one of the two specimens collected by external devices were never identified in the catheter specimen. DISCUSSION These data agree with the observations of Ouslander et al. (9) that urine specimens collected from elderly incontinent men with external devices are reliable for urine culture in this population if a count of 2105 CFU/ml is used as the criterion for infection. In addition, the validity of this criterion is increased to over 90% when the presence of the organism in two consecutive specimens is required. However, although Ouslander et al. reported that collection by
VOL. 26, 1988 URINE SPECIMEN COLLECTION FROM ELDERLY MEN 1117 Patent flo. TABLE 1. Sterile 1 Proteus rettgerib >105 2 Citrobacter sp., >105 3 4 P. mirabilisb >105 Providencia stuartii, >105 Morganella morganii, 105 5 P. rettgeri,b >105 Streptococcus agalactiae, >105 6 P. aeruginosa, >105 P. aeruginosa, >105 7 Staphylococcus aureusb >105 8 Citrobacter sp. >105 Enterobacter agglomerans, >105 S. aureus, 3 x 103 Acinetobacter sp., 4 x 103 9 P. mirabilis, >105 10 Providencia Sp.,b >105 Providencia alcalifaciens, >105 il E. colib >105 S. agalactiae, 105 12 P. stuartii, > 105 13 Citrobacter sp. b >105 14 Pseudomonasfluorescens group, > 105 P. stuartii, 1.4 x 104 P. mirabilis, 104 15 P. mirabilisb >105 16 E. Cojib >105 P. mirabilis, < 104 17 M. morganii, >105 Group D enterococcus, 105 18 S. aureus, >105 19 P. mirabilis,b >10-5 20 P. mirabilisb 2.3 x 104 P. aeruginosa, 104 Group D enterococcus, 1.7 x S. aureus, 2 x 102 21 Diphtheroidsb 102 22 Coagulase-negative-staphylococcus, 102 Diphtheroids, 102 23,-Hemolytic streptococcus, not A, B, D, 1.3 x 104 Klebsiella oxytoca, 2 x 102 Group D enterococcus, 5 x 102 24 M. morganii, 4 x 104 P. mirabilis, 3 x 10' E. coli, 3 x 102 Group D enterococcus, 2 x 102 Coagulase-negative staphylococcus, 102 Quantitative cultures of urine specimens collected from elderly incontinent men' Results with the indicated collection method (species, CFU/ml) Condom and leg bag Clean Citrobacter sp.,b >105 Proteus mirabilis, >105 Coagulase-negative staphylococci, 2 x 103 Acinetobacter sp., > 103 P. mirabilisb 1.4 x103 Pseudomonas aeruginosa, 7 x 104 Coagulase-negative staphylococci, 103 P. mirabilis, >105 P. stuartii, >105 M. morganii, 105 P. rettgeri, >105 P. aeruginosa,b > 105 P. aeruginosa, 105 P. mirabilis, 102 Citrobacter sp.,b > 105 E. agglomerans, >105 P. mirabilis, 104 S. aureus, 9 x 102 Acinetobacter sp., 1.2 X 102,-Hemolytic streptococcus, >105 b S. agalactiae, 105 ' P. fluorescens group >105 P. mirabilis, 2.6 x P. miràbilis,10 E. coli, > 105 P. mirabilis, < 104 E. colib >105 P. mirabilis, 2.5 x 104 Coagulase-negative staphylococcus,b 102 Diphtheroids, 102 P-Hemolytic streptococcus not A, B, D, 3 x 103 K. oxytoca, 3 x 102 Group D enterococcus, 3 x 102 M. morganiib 5.4 x 104 P. mirabilis, 5 x 103 E. coli, 5.8x 104 Group D enterococcus, 2.5 x 104 Catheter Citrobacter sp.," >i105 P. mirabilis,' >10`5 P. stuartii, >105" P. rettgeri, > 105 P. aeruginosa, >105 P. aeruginosa, 105 Citrobacter sp., >105 E. agglomerans, >105 E. coli, 105 S. agalactiae, 6 x 104 ' P. fluorescens group, >105 P. mirabilis, 7.6 x 104 P-Hemolytic streptococcus not A, B, D, 2.7 x 104 a Specimens were collected with external devices (sterile or clear condom drainage) or by catheter drainage. In patients 6 and 14, two different strains of the same species were isolated. b Specimen collected first. C Fever of 38'C within 24 h after catheterization.
1118 NICOLLE ET AL. TABLE 2. Species of organisms' isolated from elderly incontinent men with external collecting devices Species Infecting strains No. of isolates Contaminants Gram negative P. mirabilis 5 8 E. coli 4 1 Citrobacter species 3 P. rettgeri 2 P. stuartii 2 1 P. aeruginosa 2 2 M. morganii 3 Acinetobacter species 1 K. oxytoca 1 E. agglomerans 1 P. alcalifaciens 1 Providencia species 1 P. fluorescens group 1 Gram positive S. aureus 2 2 Coagulase-negative staphylococci 4 S. agalactiae 3 1 Group D enterococci 4 Other streptococci 1 1 Diphtheroids 2 a Strains were designated as infecting if they were present in the catheter specimen and as contaminant if they were isolated from the specimen obtained with an external collecting device but not from the catheter specimen. nursing home staff leads to a high frequency of contamination with multiple organisms in high counts, this study documents that specimens collected by the ward nursing staff can be reliable. The difference between these two studies likely reflects the manner of specimen collection, with the earlier report using specimens collected from drainage bags present for variable duration. Specimens in this investigation were all freshly voided and collected in newly applied condoms and leg bags. Our documentation that collection of specimens by ward nursing staff following a defined procedure is reliable suggests that this method of specimen collection can be widely used in institutions where this population resides. In the study reported by Ouslander et al. (9), only 12 of 26 residents studied were bacteriuric, all with only one organism. Other reports (8, 11) have consistently shown that as many as one-third of elderly men resident in nursing homes have bacteriuria with more than one organism. About onethird of our study population also had bacteriuria with more than one organism, and the specimens collected by external devices reliably identified infecting organisms in residents with polymicrobial bacteriuria when a count of 2105 CFU/ml was used to determine significance. Bacteriuria with counts of <10i CFU/ml may occur in elderly men. In the group studied here 2 (10%) of 20 bacteriuric men had counts of <105 CFU/mI, although one of these two had polymicrobial bacteriuria with the second organism isolated in counts of >105 CFU/ml. These data suggest that in clinical situations where urine cultures collected in external devices have counts of <105 CFU/ml and it is essential to know whether urinary infection is present, an invasive method for specimen collection would be required. Contamination with low counts of bacteria did occur in almost one-half of specimens collected with external de- E 0 0 -.1 100 J. CLIN. MICROBIOL. Sterile Clean Catheter Method of Collection FIG. 1. Pyuria, expressed as leukocytes per milliliter of unspun urine, in specimens collected by external devices and catheter in elderly men resident in an institution. The numbers refer to patient numbers. The lines connect different specimens collected by the three different methods from the same subject. vices. We were, however, unable to identify factors associated with contamination. Use of a sterile rather than clean collecting system and the presence or absence of circumcision of the resident did not correlate with contamination. In addition, isolation of contaminating organisms did not in- TABLE 3. Reliability of urine specimens collected by external device in identifying bladder bacteriuria in a catheter specimen Collection Predictive value for infection method Positive Sterile device 0.86 (25/29) 0.90 (18/20) Clean device 0.86 (25/29) 0.90 (19/21) Two specimens 0.93 (25/27) 0.86 (12/14) a A quantitative count of 2105 CFU/ml in the urine collected by external device was used to designate bacteriuria. The negative predictive value reflects the probability that organisms isolated from the external device in counts of <105 CFU/ml are not present in the bladder. Numbers within parentheses indicate the number positive or negative/total.
VOL. 26, 1988 URINE SPECIMEN COLLECTION FROM ELDERLY MEN 1119 crease with time from application of the collecting device to specimen collection. In fact, for clean specimens an inverse correlation was documented, with a decreased rate of contamination in specimens requiring a longer time to collect. Specimens from study subjects were refrigerated immediately after collection until transport to the laboratory. If these specimens had been maintained at room temperature, the contaminants present would likely have increased in number, leading to false-positive results. Thus, when specimens are collected with external devices optimal storage and transport must be assured. In-and-out catheterization for urine specimen collection was followed by febrile morbidity in 3 (15%) of 20 episodes in bacteriuric residents. Although many episodes of postcatheterization fever are mild, significant sepsis can occur (3). Prophylactic antimicrobial therapy is sometimes given to prevent sepsis due to catheterization in this population, although the efficacy of this intervention has not, to our knowledge, been adequately tested. Prophylactic antimicrobial agents, even if effective, are not an option when catheterization is undertaken for specimen collection for microbiologic diagnosis. Thus, although urine specimen collection by catheter remains the optimal method for ensuring valid microbiologic results, the procedure is associated with significant morbidity. This study documents that a noninvasive method which appears to have little, if any, morbidity associated with it can be an acceptable alternative specimen collection method. Urine specimen collection with external devices would not be appropriate in all instances. In particular, for persons who are acutely ill and require antimicrobial therapy a rapid and reliable means of obtaining a specimen must be used, and in these cases catheterization would be most appropriate. However, when there is no urgency for therapy, such as in obtaining a urine culture before an invasive genitourinary procedure or in investigation of the natural history and therapy of bacteriuria in this population, collection methods with external devices would be appropriate. ACKNOWLEDGMENTS This study was made possible through the enthusiastic support of the nursing staff of Deer Lodge Centre. Expert secretarial assistance was provided by V. Svetina Atkins and B. Kowalczyk. LITERATURE CITED 1. Bentley, D. W. 1983. Management of infection in the elderly, p. 187-206. In R. D. T. Cape (ed.), Fundamentals of geriatric medicine. Raven Press, New York. 2. Fierer, J., and N. Ehstrom. 1981. An outbreak of Providencia stuartii urinary tract infections: patients with condom catheters are a reservoir of the bacteria. J. Am. Med. Assoc. 245:1553-1555. 3. Gleckman, R., N. Blagg, D. Hiert, A. Hall, M. Crowley, A. Pritchard, and W. Warren. 1982. Community-acquired bacteremic urosepsis in the elderly patient: a prospective study of 34 consecutive episodes. J. Urol. 128:79-81. 4. Harding, G. K. M., T. J. Marrie, A. R. Ronald, S. Hoban, and P. Muir. 1978. Urinary tract infection localization in women. J. Am. Med. Assoc. 240:1147-1150. 5. Jewitt, N. A. S., G. R. Fernie, P. J. Holliday, and M. E. Pim. 1981. Urinary dysfunction in a geriatric long-term care population: prevalence and patterns. J. Am. Ger. Soc. 29:211-214. 6. Kaye, D. 1980. Urinary tract infections in the elderly. Bull. N.Y. Acad. Med. 56:209-220. 7. Kelly, M. T., D. J. Brenner, and J. J. Farmer III. 1985. Enterobacteriaceae, p. 263-277. In E. H. Lennette, A. Balows, W. J. Hausler, Jr., and H. J. Shadomy (ed.), Manual of clinical microbiology, 4th ed. American Society for Microbiology, Washington, D.C. 8. Nicolle, L. E., J. Bjornson, G. K. M. Harding, and J. A. MacDonell. 1983. Bacteriuria in elderly institutionalized men. N. Engl. J. Med. 309:1420-1425. 9. Ouslander, J. G., B. A. Greengold, F. V. Silverblatt, and J. P. Garcia. 1987. An accurate method to obtain urine for culture in men with external catheters. Arch. Intern. Med. 147:286-288. 10. Ouslander, J. G., R. L. Kane, and I. B. Abrass. 1982. Urinary incontinence in elderly nursing home patients. J. Am. Med. Assoc. 248:1194-1198. 11. Wolfson, S. A., G. M. Kalmanson, M. E. Rubin, and L. B. Guze. 1965. Epidemiology of bacteriuria in a predominantly geriatric male population. Am. J. Med. Sci. 250:168-173.