The Diagnosis and Significance of Bacteriuria in People with Spinal Cord Injury Marie J. Gribble, MB, ChB, FRCP(C) Division of Infectious Disease, Department of Medicine, University of British Columllia and Vancouver HosPital and Health Sciences Centre, Vancouver, British Columbia, Canada Quantitative urine cultures are used to identify individuals in whom urine in the bladder i5 not sterile, that is, those who have bacteriuria. In people with spinal cord injury the criteria which best distinguish specimen contamination from bacteriuria have been established, and are lower than the traditional ;;:: 105 CFU/ml criterion. Bacteriuria mayor may not be symptomatic, and the benefits of treatment mayor may not outweigh the risks. Abnormal levels of pyuria are present in the great majority of spinal cord injured people who have indwelling catheters or who use intermittent catheterization. In the latter group, catheter urine is not homogeneous, and levels ofpyuria are highest in the terminal fraction ofcatheter urine. Absence ofpyuria is a good indicator of absence of high count gram negative bacteriuria, but pyuria may be present in the absence of bacteriuria in this population. No methods suitable for routine clinical use are currently available to localize the site of urinary infection in this population. Frequently recurrent or persistent bacteriuria is extremely common in people with spinal cord injury. Bacteriuria is most often asymptomatic, yet is clearly significant in that it is responsible for measurable short-term and long-term morbidity. However, at present, quantitative urine cultures and tests for pyuria cannot, by themselves, be used to identify spinal cord injured persons who will reliably benefit from treatment for bacteriuria. Keywords: Bacteriuria; neurogenic bladder; urinary tract infection; diagnosis; spinal cord
injury
MICROBIOLOGIC CRITERIA FOR DIAGNOSIS OF BACTERIURIA Urine in the bladder and the kidneys is normally sterile. Quantitative culture of urine is a wellestablished tool which is capable of discriminating between contamination of a urine specimen with bacteria from the urethra or external surfaces, and true bacteriuria, that is the presence or'bacteria in bladder urine. The best quantitative criterion (or cut-off value) for bacteriuria varies with the prevalence of bacteriuria in the population under study, and with the method of specimen collection (reflecting the likelihood of contamination). Such criteria have been recently established
for persons with spinal cord injury (SCI) managed by intermittent catheterization 1 and for catheterfree SCI males using external condom drainage devices. 2 In addition, criteria for individuals with indwelling catheters 3 have been established. Bacteriuria identified in this fashion mayor may not be symptomatic and mayor may not require treatment. In certain clinical settings, it has been demonstrated that the presence of asymptomatic bacteriuria is associated with excess morbidity or mortality, and that antimicrobial therapy reduces the risk of morbidity or mortality. This has been shown, for example, in pregnant women with asymptomatic bacteriuria and in bacteriuric NeuroRehabill994; 4(4):205-213 Copyright © 1994 by Butterworth-Heinemann
206
NEUROREHABILITATION / OCTOBER 1994
individuals who undergo urologic instrumentation or surgery. In these populations quantitative urine culture is used to identity individuals who will benefit from antimicrobial therapy. In most other clinical settings, however, including persons with spinal cord injury, the presence of "significant bacteriuria" signifies only that bladder urine is not sterile, and does not necessarily predict either invasive infection or that the person will benefit from antimicrobial therapy. Unfortunately, among published studies addressing various aspects of urinary tract infection (UTI) among people with SCI, definitions ofbacteriuria vary widely. Most authors continue to use the traditional criterion, ~ 105 colony forming units per milliliter of urine (cfulml); others require ~ 10 5 cfu/ml on two consecutive occasions; while some have used criteria ~ 10 4 cfulml, or ~ 10 3 cfulml, or any detectable level of bacteriuria. Several have incorporated measures of pyuria into their definition of bacteriuria. This variability in diagnostic criteria may account for at least some of the prevailing lack of agreement regarding the incidence and prevalence of UTI in persons with SCI, the clinical significance of bacteriuria, and the indications for, and efficacy of, antimicrobial prophylaxis and therapy of bacteriuria in this unique population.
cfu/ml, had unacceptably low sensitivity for both gram-positive (0.45) and gram-negative (0.65) bacteriuria. Colony counts in SPA were below 105 cfu/ml for 61 % of gram-positive and 34% of gram-negative isolates. Of organisms which were acquired during the study period, 82% were present in counts :s; 10 3 cfu/ml when first isolated. In 40 of the 50 subjects, the SPA was sterile at commencement of the study. The proportion of specimens yielding low-level bacteriuria may not be as high in subjects who have been on intermittent catheterization for a longer period of time and who have established bacteriuria. Coagulase negative staphylococci were recovered from SPA in 50% of patients. Polymicrobial bacteriuria and/or spontaneously changing flora was documented in 50% of subjects. The subjects in this study were in the early postinjury phase, and external urinary collection devices were not necessary. If such devices promote heavy colonization of the urethra and, thus, contamination of catheter urine specimens, it is likely that the specificity and positive predictive value of the criterion ~ 10 2 cfu/ml will be lower than in our patients. Confirmation of this will require further study.
Specimens Obtained by Intermittent Catheterization
Voided Specimens from Catheter-free Males Who Use External Condom Collection Devices
We have evaluated criteria for diagnosis of bacteriuria in 47 subjects with recent SCI managed by intermittent catheterization. l Two hundred fortynine serial, daily, paired specimens of urine were obtained by suprapubic aspiration and by intermittent catheterization. Presence or absence of bacteriuria was defined by culture of suprapubic aspirate (SPA). The optimal criterion for diagnosis of bacteriuria in specimens obtained by intermittent catheterization was ~ 10 2 cfu/ml. The performance characteristics of this diagnostic criterion for gram-negative bacteriuria were sensitivity .91, specificity .97, positive predictive value .93, negative predictive value .96; and for gram-positive bacteriuria sensitivity .85, specificity .93, positive predictive value .91, negative predictive value .88. The traditional diagnostic criterion, ~ 10 5
Deresinski et al. 2 evaluated diagnostic criteria for bacteriuria in 53 male SCI patients who were catheter-free but used external condom collection devices. Seventy-one paired urine samples were obtained by SPA and by collection of a "clean voided" specimen. Colony counts in SPA and voided specimens correlated closely. Voided urine yielded an organism not present in the paired SPA in 8 (11%) samples; colony counts were ofthe order of 10 1 cfulml in 3 of these, and 104 cfulml in the remaining 5 specimens. The authors do not present data in sufficient detail to permit calculation of the performance characteristics of the various possible quantitative criteria, but they conclude that a criterion of ~ 10 4 cfu/ml has excellent performance characteristics, and that recovery of counts between 10 2
Bacteriuria in People with Spinal Cord Injury
and 10 4 cfulml may require confirmatory SPA. The proportion of SPAs with low-count bacteriuria is not stated in the publication. Polymicrobial bacteriuria was documented by SPA in 3 of 51 samples which yielded bacterial growth. Coagulase-negative staphylococci were recovered from 5 SPAs.
Individuals with Indwelling Catheters For SCI persons with indwelling urethral catheters, diagnostic criteria for bacteriuria can reasonably be extrapolated from those derived for other populations with indwelling catheters. Stark and Maki 3 studied adults in whom an indwelling catheter was inserted in an acute-care hospital setting; serial daily catheter urine specimens were obtained for quantitative culture. No matter what the initial level of bacteriuria, serial cultures showed a very high likelihood of persistance and progression from that level. For organisms initially recovered in low concentrations, 96% increased to > 105 cfulml, usually within 3 days. Thus any measurable level of bacteriuria has potential significance in individuals with short-term « 30 days) indwelling catheters. Twelve percent of such patients have polymicrobial bacteriuria, and coagulase-negative staphylococci are recovered from 34% of patients with low-level bacteriuria and 26% of those with high-level bacteriuria. 3 In persons with long-term (> 30 days) indwelling urethral catheters, a dynamic, polymicrobial bacteriuria with several organisms is usually present. 4 A new organism appears somewhat more often than once every two weeks. Specimens with six or seven organisms at high concentrations are not uncommon. Despite the common practice of assuming that recovery of multiple organisms represents "contamination" of a specimen, a comparison of results of cultures of specimens obtained by SPA and those obtained by catheter puncture showed complete agreement in 76% of 50 sample pairs. 5 Qualitative and quantitative differences between cultures of urine aspirated from long-term catheters and replacement catheters 6 have been observed, but are rarely of clinical significance. 7
207
Low-level bacteriuria « 105 cfu/ml) probably occurs with considerable frequency among subjects with long-term indwelling catheters, but has received little attention. Sapico et al. 8 perfurmed SPA in males with indwelling catheters for ~ 3 weeks, and who had not received recent antimicrobial therapy. Forty percent of isolates were present in low concentrations (s; 104 cfulml) and 27% at 103 cfulml or less. In contrast to the situation in subjects with recently-placed catheters, however, this low-level bacteriuria has not been shown to progress to high-level bacteriuria. Tenney and Warren 9 studied 7 patients with longterm catheters, at the time of catheter replacement and 2 weeks later. Thirty-seven species were recovered, 19 at high concentrations and 18 at low concentrations ( < 105 cfulml). Eighteen of the 19 initially detected in high concentrations were present at high concentrations weeks later, compared to only one of the 18 initially detected in low concentrations. Thus species present at high concentrations apparently represent a barrier to the growth of other organisms.
Specimens Obtained by Suprapubic Aspiration Urine collected by SPA represents the "gold standard" for diagnosis ofbacteriuria. Most students of UTI would agree that recovery of any detectable concentration of bacteria from a specimen obtained by SPA reflect~ the presence of bacteria in bladder urine. Theoretically, specimens obtained by SPA may be contaminated by organisms present. on the skin of the abdomen, or possibly by reflux of urethral flora into the bladder. The frequency of such contamination appears to be low. In our study,l 249 specimens were obtained from 47 patients undergoing serial daily SPA. In 14 instances, organisms were isolated from SPA on one occasion but were not recovered from a subsequent specimen; these isolates may represent either contamination or transient colonization of bladder urine. That the identification and distribution of densities of these organisms was similar to organisms repeatedly isolated, argues against contamination. We believe that the criterion for diagnosis of bacteriuria in specimens obtained by SPA should be any detectable concentration.
208
NEUROREHABILITATION / OCTOBER 1994
Screening Tests for Detection of Bacteriuria Screening tests meant as rapid, inexpensive alternatives to quantitative cultures are highly desirable, and many such screening tests have been developed and evaluated. Virtually all evaluations of urine screening tests have used the presence of ~ 10 5 cfulml as the criterion against which the performance of the screening test is measured, and many have been specifically designed to give negative results with specimens containing lower concentrations of bacteria. As has been stated above, low-level bacteriuria is common among people with SCI. Furthermore, screening tests are most useful in populations in whom the prevalence of disease is low; the prevalence of bacteriuria in the SCI population is very high. For these reasons, the usefulness of screening tests for bacteriuria in persons with SCI is limited.
PYURIA Quantitative or semiquantitative estimates of pyuria are often used by cliiricians in the diagnostic evaluation of patients with symptoms suggestive of UTI, and in the evaluation of the clinical significance of bacteriuria in asymptomatic patients. More recently, tests for pyuria have been recommended as a laboratory-based screening method for identifYing urine specimens in which low colony counts of bacteria may be significant, and which thus require special methods for culture and for interpretation. The methods of measuring pyuria, and the relationship of pyuria to bacteriuria have been reviewed by Stamm. 10 Studies published since the 1960s showed that enumeration of leukocytes in a timed urine collection is a reproducible method that distinguishes the leukocyte excretion rate (LER) in symptomatic patients with bacteriuria from the LER in asymptomatic, abacteriuric individuals. Those studies were performed in ambulatory women, and in patients attending a renal function testing service, and have been extrapolated to other patient populations. Urine leukocyte counts performed in a hemocytometer chamber, using aliquots of uncentrifuged voided urine, correlate well with the
more cumbersome LER. However, the method most commonly used in the clinical setting is enumeration of leukocytes per high power field in centrifuged urine; this method is poorly reproducible and does not correlate with either the LER or with the hemocytometer chamber method. 10 Pyuria is evidence of an inflammatory response in the host urinary tract, and thus its presence in association with bacteriuria intuitively suggests true infection rather than colonization. Nonetheless it must be remembered that noninfective causes of inflammation may cause pyuria, including mechanical and chemical irritation, and several drugs such as acetylsalicylic acid. Furthermore, pyuria may persist for some time after successful treatment of urinary infection. The relationship between pyuria and bacteriuria varies among different patient populations. Populations which have been well studied include asymptomatic nonbacteriuric men and women; symptomatic and asymptomatic bacteriuric women; and symptomatic nonbacteriuric women. 10 Several studies have examined the relationship of pyuria to bacteriuria in persons with SCI, or among subjects with indwelling catheters. 11-17
Persons with Indwelling Catheters Musher et al. II evaluated 226 specimens from 90 hospitalized men who had indwelling catheters present for time periods ranging from hours to more than 6 months. More than 10 leukocytes/mm 3 of urine were present in 62 of69 specimens yielding ~ 105 cfu/ml, in 4 of 9 specimens yielding between 103 and 105 cfulml, and in 40 of 149 specimens with < 103 cfulml. Thus abnormal pyuria was present in almost all specimens with high-level bacteriuria, but also in 27% of specimens with low-level or no bacteriuria. Peterson and Roth 12 studied 32 patients with SCI who had indwelling catheters and whose urine yielded ~ 105 cfulml. 'The subjects were classified, according to results of estimates of pyuria performed by the centrifuged urine cells per high power field method, as having "high pyuria" (> 50 wbclhpf) or "low pyuria," and were followed clinically for 2 to 75 days. Sixty percent of the high pyuria subjects experienced subsequent febrile
Bacteriuria in People with Spinal Cord Injury
episodes versus 13.6% ofthe low pyuria subjects (p < .01). At least one other study in this population suggests that high levels of pyuria are associated with an increased risk of significant morbidity. 13 In view of the known poor reproducibility of the method used to detect pyuria in these reports, the results require confirmatory studies.
Catheter-free Males with SCI, Using Condom Collection Devices Deresinski and Perkash 14 studied catheter-free males and determined levels of pyuria in clean voided urine specimens and in paired SPA specimens, using the hemocytometer chamber method. Levels of pyuria in voided urine accurately reflected those in SPA specimens (r = 0.825). Of samples with abnormal levels of pyuria, 97% had bacteriuria. However, 40% of samples without pyuria had bacteriuria. There was a poor correlation between levels of pyuria and levels of bacteriuria. Similarly, pyuria correlated poorly with symptoms suggestive of UTI in this study.
Persons with SCI Managed by Intermittent Catheterization Hooton et aJ.l5 studied asymptomatic persons with SCI who had ~ 10 5 cfu/ml in at least 2 consecutive urine specimens. Quantitative urinary leukocyte counts were determined, using the hemocytometer chamber method, in 46 infections. Abnormal pyuria was present in 43 of 46 cases. Patients with positive bladder washout tests or positive antibody coated bacteria tests had significantly higher median leucocyte counts than those with negative bladder washout and negative antibody coated bacteria tests. There was, however, substantial overlap between the groups. Anderson and Hsieh-Ma 16 followed 156 hospitalized subjects with SCIon intermittent catheterization. Daily screening dipslide urine cultures were performed. When these were positive, quantitative urine cultures and estimates of pyuria using the hemocytometer chamber method were performed. The mean leukocyte count was < 10 cells/mm 3 for all sterile specimens and for specimens with < 10 5 cfull of gram-positive bacteria. The mean leukocyte count in specimens with
209
> 105 cfu/ml of gram-positive bacteria, was 31 cells/mm, 3 similar to that in specimens with < 10 5 cfu/ml of gram-negative bacteria. Specimens with high counts of gram-negative bacteria had high levels of pyuria (mean 45 cells/mm3). Ranges of counts are not presented in the publication, however, the standard error of the mean was substantial, e.g., 445 ± 147. We have examined the relationship of leukocyte counts (hemocytometer chamber method) in aliquots of mid-catheter urine, terminal catheter urine, and urine obtained by SPA to the leukocyte excretion rate (LER), and the relationship of pyuria to bacteriuria, in persons with SCIon intermittent catheterization. 17 Fifty-two patients were studied serially for an average of 5 days. Leukocyte counts were significantly higher in terminal catheter specimens than in mid-catheter and SPA samples. Twenty-five percent of paired terminal catheter and mid-catheter specimens differed by > 150 cells/mm,3 and 10% differed by > 1300 cells/mm. 3 The statistical correlation between LER and leukocyte counts in catheter specimens was significant, however, terminal catheter urine cell counts frequently overestimated pyuria, and SPA and mid-catheter urine cell counts frequently underestimated pyuria. These data confirm the common observation of SCI people and their attendants that turbidity of catheter urine is most marked in the terminal aliquot. The statistical correlation between bacteriuria and estimates of pyuria was significant but the r values were modest. The strongest correlation was between gram-negative bacteriuria and LER (r = 0.55, P < .001). Examining this relationship more closely however, showed that while abnormal LER was present in all cases with high-level gram-negative bacteriuria, abnormal LER was also present in 60% of cases with low-level gram-negative bacteriuria and in 55% of cases with no gram-negative bacteria. Thus we found that in intermittently catheterized subjects, catheter urine is inhomogeneous, and random aliquots of catheter urine are not suitable specimens for estimation of pyuria. Furthermore, neither leukocyte counts nor the LER provided clear separation of bacteriuric from abacteriuric individuals in this population.
210
NEUROREHABILITATION / OCTOBER 1994
LOCALIZATION OF UTI IN PERSONS WITH SCI Methods of localizing the site of infection, or of differentiating between invasive and noninvasive infections, would be of great value in SCI persons, in view of (1) the high prevalence of asymptomatic bacteriuria; (2) the association between bacteriuria and important morbidity and mortality in some cases 18-24; and (3) the disadvantages of routine antimicrobial therapy for all episodes of asymptomatic bacteriuria. Ureteral catheterization is the "gold-standard" method of localization 25 but is not suitable for routine clinical use or for use in large-scale treatment studies. The bladder washout test 26 ,27 correlates poorly with results of ureteral catheterization in persons with SCI. 28,29 The possibility that it may be useful for distinguishing invasive from noninvasive infections may merit further study. The antibody coated bacteria test 30- 33 correlates poorly with results of both ureteral catheterization and bladder washout tests in this population. 15 ,28-30,34-38 It does not successfully predict outcome of antibiotic treatment of UTI in persons with SCI. 29 No other noninvasive or indirect tests oflocalization or invasiveness are presendy sufficiendy sensitive and specific to be useful in the clinical arena. 39 This may be a fruitful area for future research.
CLINICAL SIGNIFICANCE OF BACTERIURIA IN PERSONS WITH SCI Frequendy recurrent or continuous bacteriuria is extremely common in persons with SCI, regardless of the method of bladder drainage used. Although bacteriuria in this population is most often asymptomatic,2,14,15,17,39,40 several lines of evidence suggest that it is nonetheless responsible for substantial short-term and long-term morbidity and mortality. During initial hospitalization in an acute SCI Unit, bacteriuria with otherwise unexplained fever occurs in up to 40% of subjects. 4o Approximately 10% of persons with SCI are rehospitalized for UTI requiring intravenous antibiotics during
the first year after initial rehabilitation. 18 Within one year of injury, at least one episode ofbacteriuria with fever has occurred in 53% of those undergoing self-intermittent catheterization and in 83% of those undergoing intermittent catheterization by attendants or family members. 19 Autopsy series published in the 1950s and 1960s showed that 65-95% of persons with SCI had evidence of chronic pyelonephritis.20-22 More recendy, Warren et al. 23 have shown that acute inflammation of the renal parenchyma was present in 38% of patients with long-term indwelling catheters. Among more than 5,000 patients who suffered SCI between 1973 and 1980, septicemia was the cause of death with the highest ratio of actual to expected deaths (ratio 140:7),24 and UTI remains the commonest cause of septicemia in this population. There is of course general agreement that antimicrobial therapy is indicated for persons with SCI who have the clinical syndrome of urosepsis, or other manifestations of frank UTI (e.g., epididymoorchitis), and for those with bacteriuria and otherwise unexplained fever. 39 However, there is disagreement regarding the management of asymptomatic bacteriuria and of bacteriuria with nonspecific symptoms (e.g., cloudy or foul urine, increased spasticity, increased voiding between catheters, malaise). On the one hand, the high recurrence rate and frequent appearance of antibiotic-resistant pathogens after antibiotic therapy argue against routine treatment of these infections. On the other hand, asymptomatic infections, particularly those involving the upper-urinary tract, may well be associated with subsequent morbidity. Thus, a practical and inexpensive method capable of distinguishing uppertract infection from bladder bacteriuria, or noninvasive from invasive infection, would be of substantial value in early identification of individuals who may benefit from antibiotic treatment. Thus the weight of evidence compellingly indicates that even asymptomatic bacteriuria is indeed clinically significant, in that it contributes to substantial morbidity and mortality in persons with spinal cord injury. This does not, however, mean that "significant bacteriuria" is necessarily an indication for antimicrobial therapy. Antimicrobials should only be used when it is clear not only that
Bacteriuria in People with Spinal Cord Injury
bacteriuria adversely affects those who acquire it, but also that treatment prevents or improves the adverse outcome, and that the benefits of treatment outweigh the risks. The indications for treatment of bacteriuria in persons with SCI are discussed elsewhere in this issue. 41
SUMMARY Quantitative urine cultures are used to identify persons in whom bladder urine is not sterile, that is, those who have bacteriuria. Criteria for diagnosis of bacteriuria vary according to the prevalence of the condition in the population, and the method of specimen collection. For persons with SCI, quantitative criteria with optimal sensitivity and specificity for diagnosis of bacteriuria are: (1) Catheter specimens from individuals on intermittent catheterization: ;::: 10 2 cfu/ml. (2) Clean-void specimens from catheter-free males who use condom collection devices: ;::: 10 4 cfulml. (In this group, cultures yielding between 10 2 and 10 4 cfu/ml may require confirmatory suprapubic aspiration). (3) Specimens from indwelling catheters: any detectable concentration. (4) Suprapubic aspirates: any detectable concentration. For these four groups the traditional criterion, ;::: 10 5 cfu/ml, has unacceptably low sensitivity for most clinical, epidemiologic, or research purposes. Screening tests for bacteriuria are presently not able to reliably detect low-count bacteriuria, which is common in persons with SCI. In persons with SCI, urine cultures yielding coagulase-negative staphylococci, or multiple organisms, cannot be assumed to be contaminated. Full identification and sensitivity testing of such isolates may be necessary for clinical or infectioncontrol purposes. Abnormal levels of pyuria are present in the great majority of persons with SCI who have indwelling catheters and also those on intermittent catheterization. The most commonly used method of estimating pyuria, enumeration ofleukocytes per high power field in centrifuged urine, is poorly reproducible and does not reliably correlate either with the leukocyte excretion rate or with the recommended method, enumeration of
211
leukocytes in uncentrifuged urine, using a hemocytometer chamber. In persons on intermittent catheterization, catheter urine is inhomogeneous and leukocyte counts are significantly higher in terminal catheter urine than in mid-catheter specimens or suprapubic aspirates. Absence of pyuria is a good predictor of absence of high-level gramnegative bacteriuria, but pyuria may be present in the absence of bacteriuria. 'rests for pyuria have been advocated by some as a method of identifying asymptomatic bacteriuric individuals who require treatment. However, no data currently exists to support the effectiveness of this approach, although it is intuitively appealing to use pyuria as an indication of inflammation and/or invasive infection. Ureteral catheterization is the "gold-standard" method oflocalization of infection, but is not suitable for routine clinical use. The bladder washout test, antibody coated bacteria test, and other current noninvasive or indirect tests oflocalization or invasiveness are presently not sufficiently sensitive or specific to be useful in the clinical arena. Although bacteriuria among persons with SCI is most often asymptomatic, it nonetheless is responsible for substantial morbidity and mortality. Bacteriuria, as defined above, is significant in that the urine in the bladder is not sterile, and in that there may be associated morbidity even when symptoms are absent. However, "significant bacteriuria" is not, by itself, an indication for antimicrobial therapy.
REFERENCES 1. Gribble Mj, McCallum NM, Schechter MT. Eval-
uation of diagnostic criteria for bacteriuria in acutely spinal cord injured patients undergoing intermittent catheterization. Diagn Microbiol InJect Dis 1988; 9:197-206. 2. Deresinski SC, Perkash I. Urinary tract infections in male spinal cord injured patients. Part One: Bacteriologic diagnosis. ] Am Parapl Soc 1985; 8:4-6. 3. Stark RP, Maki DG. Bacteriuria in the catheterized patient: What quantitative level of bacteriuria is relevant? N EnglJ Med 1984; 311 :560564.
212
NEUROREHABILITATION / OCTOBER 1994
4. Warren JW, Tenney JH, Hoopes .1M, Muncie HL. A prospective microbiologic study ofbacteriuria in patients with chronic indwelling urethral catheters.] Infect Dis 1982; 146:719-723. 5. Bergqvist D, Bronnestam R, Hedelin H, et al. The relevance of urinary sampling methods in patients with indwelling foley catheters. BrJ Urol 1980; 52:92-95. 6. Grahn D, Norman DC, White ML, et al. Validity of urinary catheter specimen for diagnosis of urinary tract infection in the elderly. Arch Intern Med 1985; 145:1858-1860. 7. Nicolle LE. Urine cultures and long-term indwelling catheters. Arch Intern Med 1985; 145: 1794-1795. 8. Sapico FL, Wideman PA, Finegold SM. Aerobic and anaerobic flora in bladder urine of patients with indwelling urethral catheters. Urology 1976; 7:382-384. 9. Tenney JH, Warren Jw. Long-term catheterassociated bacteriuria: Species at low concentration. Urology 1987; 30:444-446. 10. Stamm WE. Measurement of pyuria and its relation to bacteriuria. Am] Med 1983; 75(suppl IB):53-58. 11. Musher DM, Thorsteinsson SB, Airola II VM. Quantitative urinalysis: Diagnosing urinary tract infection in men. ]AMA 1976; 236:20692072. 12. Peterson JR, Roth EJ. Fever, bacteriuria, and pyuria in spinal cord injured patients with indwelling urethral catheters. Arch Phys Med Rehabil 1989; 70:839-841. 13. Menon EB, Tan ES. Pyuria: index of infection in patients with spinal cord injuries. Br] Uro11992; 69:144-146. 14. Deresinski SC, Perkash I. Urinary traction infections in male spinal cord injured patients. Part Two: Diagnostic value of symptoms and of quantitative urinalysis.] Am Parapl Soc 1985; 8:7-10. 15. Hooton TM, O'Shaughnessy EJ, Clowers D, et al. Localization of urinary tract infection in patients with spinal cord injury. ] Infect Dis 1984; 150(1):85-9l. 16. Anderson RU, Hsieh-Ma ST. Association ofbacteriuria and pyuria during intermittent catheterization after spinal cord injury.] Urol 1983; 130:299-30l. 17. Gribble MJ, Puterman L, McCallum NM. Pyuria: Its relationship to bacteriuria in spinal cord injured patients on intermittent catheterization. Arch Phys Med Rehabil1989; 70:376-379. 18. Davidoff G, Schultz .IS, Lieb T, et al. Rehospitalization after initial rehabilitation for acute spinal cord injury: Incidence and risk factors. Arch Phys Med Rehabil1990; 71:121-124.
19. Cardenas DD, Mayo ME. Bacteriuria with fever after spinal cord injury. Arch Phys Med Rehabil 1987; 68:291-293. 20. Dietrick RB, Russi S. Tabulation and review of autopsy findings in fifty-five paraplegics. ]AMA 1958; 166:41-44. 21. Tribe CB, Silver JR. Renal failure in paraplegia. London: Pitman Medical Publishing Co., 1969. 22. Talbot HS. Renal disease and hypertension in paraplegics and quadriplegics. Med Seru] Canada 1966; 22:570-575. 23. Warren .JW, Muncie .Jr HL, Hall-Craggs M. Acute pyelonephritis associated with bacteriuria during long-term catheterization: A prospective clinicopathological study. ] Infect Dis 1988; 158:1341-1346. 24. Devivo MJ, Kartus PL, Stover SL, et al. Cause of death for patients with spinal cord injuries. Arch Intern Med 1989; 149:1761-1766. 25. Stamey TA, Govan DE, Palmer JM. The localization and treatment of urinary tract infections: The role of bactericidal urine levels as opposed to serum levels. Medicine (Baltimore) 1965; 44:236. 26. Fairley KF, Bond AG, Brown RB, et al. Simple test to determine the site of urinary-tract infection. Lancet 1967; 2(7513):427-428. 27. Fairley KF, Grounds AD, Carson NE, et al. Site of infection in acute urinary-tract infection in general practice. Lancet 1971; 2(7725):615-618. 28. Kuhlemeier KV, Lloyd LK, Stover SL. Failure of antibody-coated bacteria and bladder washout tests to localize infection in spinal cord injury patients.] Uro11983; 130:729-732. 29. Gribble MJ, Lee L, Fink M, et al. Localization of asymptomatic bacteriuria in spinal cord injured patients. Abstract No. 363. Clinical and Investigative Medicine 1990; 13(4):358. 30. MerrittJL, Keys TE Limitations of the antibodycoated bacteria test in patients with neurogenic bladders.]AMA 82; 247(12):1723-1725. 31. Boutros P, Mourtada H, Ronald AR. Localization of urinary infection. Antimicrob Agents Chemother 1970; 114-116. 32. Thomas V, Shelokov A, Forland M. Antibodycoated bacteria in the urine and the site of urinary-tract infection. N Engl ] Med 1974; 290:588-590. 33. Giamarellou H. Antibody-coated bacteria in urine: When, where and why? ] Antimicrob Chemother 1984; 13(2):95-99. 34. Thorley JD, Barbin GK, Reinarz JA. The prevalence of antibody-coated bacteria in urine. Am] Med Sci 1978; 275(1):75-80. 35. Newman E, Price M, Ederer GM. Urinary tract infection in patients with spinal cord lesions:
Bacteriuria in People with Spinal Cord Injury
Antibody-coated bacteria tests as a diagnostic aid. Arch Phys Med Rehabil1980; 61:406-409. 36. Lindan R. The significance of antibody coated bacteria in neuropathic bladder urines. Paraplegia 1981; 19(4):216-219. 37. Wyndaele jj, Everaert K, Rysselaere M, et al. An evaluation of the bladder wash-out test and the anti-body-coated bacteria test for the localization of a urinary tract infection in patients with and without a neurogenic bladder. Acta Urologica Belgica 1986; 56(4):471-476. 38. Galloway A, Green HT, Menon KK, et al. Antibody coated bacteria in urine of patients with recent spinal injury. J Clin Pathol 1990; 43:953956.
213
39. National Institute or Disability and Rehabilitation Research Consensus Statement: The prevention and management of urinary tract infections among people with spinal cord injuries. J Am Parapl Soc 1992; 15:194-207. 40. Gribble Mj, Puterman ML. Prophylaxis of urinary tract infection in persons with recent spinal cord injury: A prospective, randomized, doubleblind, placebo-controlled study of trimethoprim-sulfamethoxazole. Am J Med 1993; 95:141-152. 41. Montgomorie jZ. Treatment of urinary tract infections in persons with spinal cord injury. Neurorehabill994; 4(4):214-221.
injury
MICROBIOLOGIC CRITERIA FOR DIAGNOSIS OF BACTERIURIA Urine in the bladder and the kidneys is normally sterile. Quantitative culture of urine is a wellestablished tool which is capable of discriminating between contamination of a urine specimen with bacteria from the urethra or external surfaces, and true bacteriuria, that is the presence or'bacteria in bladder urine. The best quantitative criterion (or cut-off value) for bacteriuria varies with the prevalence of bacteriuria in the population under study, and with the method of specimen collection (reflecting the likelihood of contamination). Such criteria have been recently established
for persons with spinal cord injury (SCI) managed by intermittent catheterization 1 and for catheterfree SCI males using external condom drainage devices. 2 In addition, criteria for individuals with indwelling catheters 3 have been established. Bacteriuria identified in this fashion mayor may not be symptomatic and mayor may not require treatment. In certain clinical settings, it has been demonstrated that the presence of asymptomatic bacteriuria is associated with excess morbidity or mortality, and that antimicrobial therapy reduces the risk of morbidity or mortality. This has been shown, for example, in pregnant women with asymptomatic bacteriuria and in bacteriuric NeuroRehabill994; 4(4):205-213 Copyright © 1994 by Butterworth-Heinemann
206
NEUROREHABILITATION / OCTOBER 1994
individuals who undergo urologic instrumentation or surgery. In these populations quantitative urine culture is used to identity individuals who will benefit from antimicrobial therapy. In most other clinical settings, however, including persons with spinal cord injury, the presence of "significant bacteriuria" signifies only that bladder urine is not sterile, and does not necessarily predict either invasive infection or that the person will benefit from antimicrobial therapy. Unfortunately, among published studies addressing various aspects of urinary tract infection (UTI) among people with SCI, definitions ofbacteriuria vary widely. Most authors continue to use the traditional criterion, ~ 105 colony forming units per milliliter of urine (cfulml); others require ~ 10 5 cfu/ml on two consecutive occasions; while some have used criteria ~ 10 4 cfulml, or ~ 10 3 cfulml, or any detectable level of bacteriuria. Several have incorporated measures of pyuria into their definition of bacteriuria. This variability in diagnostic criteria may account for at least some of the prevailing lack of agreement regarding the incidence and prevalence of UTI in persons with SCI, the clinical significance of bacteriuria, and the indications for, and efficacy of, antimicrobial prophylaxis and therapy of bacteriuria in this unique population.
cfu/ml, had unacceptably low sensitivity for both gram-positive (0.45) and gram-negative (0.65) bacteriuria. Colony counts in SPA were below 105 cfu/ml for 61 % of gram-positive and 34% of gram-negative isolates. Of organisms which were acquired during the study period, 82% were present in counts :s; 10 3 cfu/ml when first isolated. In 40 of the 50 subjects, the SPA was sterile at commencement of the study. The proportion of specimens yielding low-level bacteriuria may not be as high in subjects who have been on intermittent catheterization for a longer period of time and who have established bacteriuria. Coagulase negative staphylococci were recovered from SPA in 50% of patients. Polymicrobial bacteriuria and/or spontaneously changing flora was documented in 50% of subjects. The subjects in this study were in the early postinjury phase, and external urinary collection devices were not necessary. If such devices promote heavy colonization of the urethra and, thus, contamination of catheter urine specimens, it is likely that the specificity and positive predictive value of the criterion ~ 10 2 cfu/ml will be lower than in our patients. Confirmation of this will require further study.
Specimens Obtained by Intermittent Catheterization
Voided Specimens from Catheter-free Males Who Use External Condom Collection Devices
We have evaluated criteria for diagnosis of bacteriuria in 47 subjects with recent SCI managed by intermittent catheterization. l Two hundred fortynine serial, daily, paired specimens of urine were obtained by suprapubic aspiration and by intermittent catheterization. Presence or absence of bacteriuria was defined by culture of suprapubic aspirate (SPA). The optimal criterion for diagnosis of bacteriuria in specimens obtained by intermittent catheterization was ~ 10 2 cfu/ml. The performance characteristics of this diagnostic criterion for gram-negative bacteriuria were sensitivity .91, specificity .97, positive predictive value .93, negative predictive value .96; and for gram-positive bacteriuria sensitivity .85, specificity .93, positive predictive value .91, negative predictive value .88. The traditional diagnostic criterion, ~ 10 5
Deresinski et al. 2 evaluated diagnostic criteria for bacteriuria in 53 male SCI patients who were catheter-free but used external condom collection devices. Seventy-one paired urine samples were obtained by SPA and by collection of a "clean voided" specimen. Colony counts in SPA and voided specimens correlated closely. Voided urine yielded an organism not present in the paired SPA in 8 (11%) samples; colony counts were ofthe order of 10 1 cfulml in 3 of these, and 104 cfulml in the remaining 5 specimens. The authors do not present data in sufficient detail to permit calculation of the performance characteristics of the various possible quantitative criteria, but they conclude that a criterion of ~ 10 4 cfu/ml has excellent performance characteristics, and that recovery of counts between 10 2
Bacteriuria in People with Spinal Cord Injury
and 10 4 cfulml may require confirmatory SPA. The proportion of SPAs with low-count bacteriuria is not stated in the publication. Polymicrobial bacteriuria was documented by SPA in 3 of 51 samples which yielded bacterial growth. Coagulase-negative staphylococci were recovered from 5 SPAs.
Individuals with Indwelling Catheters For SCI persons with indwelling urethral catheters, diagnostic criteria for bacteriuria can reasonably be extrapolated from those derived for other populations with indwelling catheters. Stark and Maki 3 studied adults in whom an indwelling catheter was inserted in an acute-care hospital setting; serial daily catheter urine specimens were obtained for quantitative culture. No matter what the initial level of bacteriuria, serial cultures showed a very high likelihood of persistance and progression from that level. For organisms initially recovered in low concentrations, 96% increased to > 105 cfulml, usually within 3 days. Thus any measurable level of bacteriuria has potential significance in individuals with short-term « 30 days) indwelling catheters. Twelve percent of such patients have polymicrobial bacteriuria, and coagulase-negative staphylococci are recovered from 34% of patients with low-level bacteriuria and 26% of those with high-level bacteriuria. 3 In persons with long-term (> 30 days) indwelling urethral catheters, a dynamic, polymicrobial bacteriuria with several organisms is usually present. 4 A new organism appears somewhat more often than once every two weeks. Specimens with six or seven organisms at high concentrations are not uncommon. Despite the common practice of assuming that recovery of multiple organisms represents "contamination" of a specimen, a comparison of results of cultures of specimens obtained by SPA and those obtained by catheter puncture showed complete agreement in 76% of 50 sample pairs. 5 Qualitative and quantitative differences between cultures of urine aspirated from long-term catheters and replacement catheters 6 have been observed, but are rarely of clinical significance. 7
207
Low-level bacteriuria « 105 cfu/ml) probably occurs with considerable frequency among subjects with long-term indwelling catheters, but has received little attention. Sapico et al. 8 perfurmed SPA in males with indwelling catheters for ~ 3 weeks, and who had not received recent antimicrobial therapy. Forty percent of isolates were present in low concentrations (s; 104 cfulml) and 27% at 103 cfulml or less. In contrast to the situation in subjects with recently-placed catheters, however, this low-level bacteriuria has not been shown to progress to high-level bacteriuria. Tenney and Warren 9 studied 7 patients with longterm catheters, at the time of catheter replacement and 2 weeks later. Thirty-seven species were recovered, 19 at high concentrations and 18 at low concentrations ( < 105 cfulml). Eighteen of the 19 initially detected in high concentrations were present at high concentrations weeks later, compared to only one of the 18 initially detected in low concentrations. Thus species present at high concentrations apparently represent a barrier to the growth of other organisms.
Specimens Obtained by Suprapubic Aspiration Urine collected by SPA represents the "gold standard" for diagnosis ofbacteriuria. Most students of UTI would agree that recovery of any detectable concentration of bacteria from a specimen obtained by SPA reflect~ the presence of bacteria in bladder urine. Theoretically, specimens obtained by SPA may be contaminated by organisms present. on the skin of the abdomen, or possibly by reflux of urethral flora into the bladder. The frequency of such contamination appears to be low. In our study,l 249 specimens were obtained from 47 patients undergoing serial daily SPA. In 14 instances, organisms were isolated from SPA on one occasion but were not recovered from a subsequent specimen; these isolates may represent either contamination or transient colonization of bladder urine. That the identification and distribution of densities of these organisms was similar to organisms repeatedly isolated, argues against contamination. We believe that the criterion for diagnosis of bacteriuria in specimens obtained by SPA should be any detectable concentration.
208
NEUROREHABILITATION / OCTOBER 1994
Screening Tests for Detection of Bacteriuria Screening tests meant as rapid, inexpensive alternatives to quantitative cultures are highly desirable, and many such screening tests have been developed and evaluated. Virtually all evaluations of urine screening tests have used the presence of ~ 10 5 cfulml as the criterion against which the performance of the screening test is measured, and many have been specifically designed to give negative results with specimens containing lower concentrations of bacteria. As has been stated above, low-level bacteriuria is common among people with SCI. Furthermore, screening tests are most useful in populations in whom the prevalence of disease is low; the prevalence of bacteriuria in the SCI population is very high. For these reasons, the usefulness of screening tests for bacteriuria in persons with SCI is limited.
PYURIA Quantitative or semiquantitative estimates of pyuria are often used by cliiricians in the diagnostic evaluation of patients with symptoms suggestive of UTI, and in the evaluation of the clinical significance of bacteriuria in asymptomatic patients. More recently, tests for pyuria have been recommended as a laboratory-based screening method for identifYing urine specimens in which low colony counts of bacteria may be significant, and which thus require special methods for culture and for interpretation. The methods of measuring pyuria, and the relationship of pyuria to bacteriuria have been reviewed by Stamm. 10 Studies published since the 1960s showed that enumeration of leukocytes in a timed urine collection is a reproducible method that distinguishes the leukocyte excretion rate (LER) in symptomatic patients with bacteriuria from the LER in asymptomatic, abacteriuric individuals. Those studies were performed in ambulatory women, and in patients attending a renal function testing service, and have been extrapolated to other patient populations. Urine leukocyte counts performed in a hemocytometer chamber, using aliquots of uncentrifuged voided urine, correlate well with the
more cumbersome LER. However, the method most commonly used in the clinical setting is enumeration of leukocytes per high power field in centrifuged urine; this method is poorly reproducible and does not correlate with either the LER or with the hemocytometer chamber method. 10 Pyuria is evidence of an inflammatory response in the host urinary tract, and thus its presence in association with bacteriuria intuitively suggests true infection rather than colonization. Nonetheless it must be remembered that noninfective causes of inflammation may cause pyuria, including mechanical and chemical irritation, and several drugs such as acetylsalicylic acid. Furthermore, pyuria may persist for some time after successful treatment of urinary infection. The relationship between pyuria and bacteriuria varies among different patient populations. Populations which have been well studied include asymptomatic nonbacteriuric men and women; symptomatic and asymptomatic bacteriuric women; and symptomatic nonbacteriuric women. 10 Several studies have examined the relationship of pyuria to bacteriuria in persons with SCI, or among subjects with indwelling catheters. 11-17
Persons with Indwelling Catheters Musher et al. II evaluated 226 specimens from 90 hospitalized men who had indwelling catheters present for time periods ranging from hours to more than 6 months. More than 10 leukocytes/mm 3 of urine were present in 62 of69 specimens yielding ~ 105 cfu/ml, in 4 of 9 specimens yielding between 103 and 105 cfulml, and in 40 of 149 specimens with < 103 cfulml. Thus abnormal pyuria was present in almost all specimens with high-level bacteriuria, but also in 27% of specimens with low-level or no bacteriuria. Peterson and Roth 12 studied 32 patients with SCI who had indwelling catheters and whose urine yielded ~ 105 cfulml. 'The subjects were classified, according to results of estimates of pyuria performed by the centrifuged urine cells per high power field method, as having "high pyuria" (> 50 wbclhpf) or "low pyuria," and were followed clinically for 2 to 75 days. Sixty percent of the high pyuria subjects experienced subsequent febrile
Bacteriuria in People with Spinal Cord Injury
episodes versus 13.6% ofthe low pyuria subjects (p < .01). At least one other study in this population suggests that high levels of pyuria are associated with an increased risk of significant morbidity. 13 In view of the known poor reproducibility of the method used to detect pyuria in these reports, the results require confirmatory studies.
Catheter-free Males with SCI, Using Condom Collection Devices Deresinski and Perkash 14 studied catheter-free males and determined levels of pyuria in clean voided urine specimens and in paired SPA specimens, using the hemocytometer chamber method. Levels of pyuria in voided urine accurately reflected those in SPA specimens (r = 0.825). Of samples with abnormal levels of pyuria, 97% had bacteriuria. However, 40% of samples without pyuria had bacteriuria. There was a poor correlation between levels of pyuria and levels of bacteriuria. Similarly, pyuria correlated poorly with symptoms suggestive of UTI in this study.
Persons with SCI Managed by Intermittent Catheterization Hooton et aJ.l5 studied asymptomatic persons with SCI who had ~ 10 5 cfu/ml in at least 2 consecutive urine specimens. Quantitative urinary leukocyte counts were determined, using the hemocytometer chamber method, in 46 infections. Abnormal pyuria was present in 43 of 46 cases. Patients with positive bladder washout tests or positive antibody coated bacteria tests had significantly higher median leucocyte counts than those with negative bladder washout and negative antibody coated bacteria tests. There was, however, substantial overlap between the groups. Anderson and Hsieh-Ma 16 followed 156 hospitalized subjects with SCIon intermittent catheterization. Daily screening dipslide urine cultures were performed. When these were positive, quantitative urine cultures and estimates of pyuria using the hemocytometer chamber method were performed. The mean leukocyte count was < 10 cells/mm 3 for all sterile specimens and for specimens with < 10 5 cfull of gram-positive bacteria. The mean leukocyte count in specimens with
209
> 105 cfu/ml of gram-positive bacteria, was 31 cells/mm, 3 similar to that in specimens with < 10 5 cfu/ml of gram-negative bacteria. Specimens with high counts of gram-negative bacteria had high levels of pyuria (mean 45 cells/mm3). Ranges of counts are not presented in the publication, however, the standard error of the mean was substantial, e.g., 445 ± 147. We have examined the relationship of leukocyte counts (hemocytometer chamber method) in aliquots of mid-catheter urine, terminal catheter urine, and urine obtained by SPA to the leukocyte excretion rate (LER), and the relationship of pyuria to bacteriuria, in persons with SCIon intermittent catheterization. 17 Fifty-two patients were studied serially for an average of 5 days. Leukocyte counts were significantly higher in terminal catheter specimens than in mid-catheter and SPA samples. Twenty-five percent of paired terminal catheter and mid-catheter specimens differed by > 150 cells/mm,3 and 10% differed by > 1300 cells/mm. 3 The statistical correlation between LER and leukocyte counts in catheter specimens was significant, however, terminal catheter urine cell counts frequently overestimated pyuria, and SPA and mid-catheter urine cell counts frequently underestimated pyuria. These data confirm the common observation of SCI people and their attendants that turbidity of catheter urine is most marked in the terminal aliquot. The statistical correlation between bacteriuria and estimates of pyuria was significant but the r values were modest. The strongest correlation was between gram-negative bacteriuria and LER (r = 0.55, P < .001). Examining this relationship more closely however, showed that while abnormal LER was present in all cases with high-level gram-negative bacteriuria, abnormal LER was also present in 60% of cases with low-level gram-negative bacteriuria and in 55% of cases with no gram-negative bacteria. Thus we found that in intermittently catheterized subjects, catheter urine is inhomogeneous, and random aliquots of catheter urine are not suitable specimens for estimation of pyuria. Furthermore, neither leukocyte counts nor the LER provided clear separation of bacteriuric from abacteriuric individuals in this population.
210
NEUROREHABILITATION / OCTOBER 1994
LOCALIZATION OF UTI IN PERSONS WITH SCI Methods of localizing the site of infection, or of differentiating between invasive and noninvasive infections, would be of great value in SCI persons, in view of (1) the high prevalence of asymptomatic bacteriuria; (2) the association between bacteriuria and important morbidity and mortality in some cases 18-24; and (3) the disadvantages of routine antimicrobial therapy for all episodes of asymptomatic bacteriuria. Ureteral catheterization is the "gold-standard" method of localization 25 but is not suitable for routine clinical use or for use in large-scale treatment studies. The bladder washout test 26 ,27 correlates poorly with results of ureteral catheterization in persons with SCI. 28,29 The possibility that it may be useful for distinguishing invasive from noninvasive infections may merit further study. The antibody coated bacteria test 30- 33 correlates poorly with results of both ureteral catheterization and bladder washout tests in this population. 15 ,28-30,34-38 It does not successfully predict outcome of antibiotic treatment of UTI in persons with SCI. 29 No other noninvasive or indirect tests oflocalization or invasiveness are presendy sufficiendy sensitive and specific to be useful in the clinical arena. 39 This may be a fruitful area for future research.
CLINICAL SIGNIFICANCE OF BACTERIURIA IN PERSONS WITH SCI Frequendy recurrent or continuous bacteriuria is extremely common in persons with SCI, regardless of the method of bladder drainage used. Although bacteriuria in this population is most often asymptomatic,2,14,15,17,39,40 several lines of evidence suggest that it is nonetheless responsible for substantial short-term and long-term morbidity and mortality. During initial hospitalization in an acute SCI Unit, bacteriuria with otherwise unexplained fever occurs in up to 40% of subjects. 4o Approximately 10% of persons with SCI are rehospitalized for UTI requiring intravenous antibiotics during
the first year after initial rehabilitation. 18 Within one year of injury, at least one episode ofbacteriuria with fever has occurred in 53% of those undergoing self-intermittent catheterization and in 83% of those undergoing intermittent catheterization by attendants or family members. 19 Autopsy series published in the 1950s and 1960s showed that 65-95% of persons with SCI had evidence of chronic pyelonephritis.20-22 More recendy, Warren et al. 23 have shown that acute inflammation of the renal parenchyma was present in 38% of patients with long-term indwelling catheters. Among more than 5,000 patients who suffered SCI between 1973 and 1980, septicemia was the cause of death with the highest ratio of actual to expected deaths (ratio 140:7),24 and UTI remains the commonest cause of septicemia in this population. There is of course general agreement that antimicrobial therapy is indicated for persons with SCI who have the clinical syndrome of urosepsis, or other manifestations of frank UTI (e.g., epididymoorchitis), and for those with bacteriuria and otherwise unexplained fever. 39 However, there is disagreement regarding the management of asymptomatic bacteriuria and of bacteriuria with nonspecific symptoms (e.g., cloudy or foul urine, increased spasticity, increased voiding between catheters, malaise). On the one hand, the high recurrence rate and frequent appearance of antibiotic-resistant pathogens after antibiotic therapy argue against routine treatment of these infections. On the other hand, asymptomatic infections, particularly those involving the upper-urinary tract, may well be associated with subsequent morbidity. Thus, a practical and inexpensive method capable of distinguishing uppertract infection from bladder bacteriuria, or noninvasive from invasive infection, would be of substantial value in early identification of individuals who may benefit from antibiotic treatment. Thus the weight of evidence compellingly indicates that even asymptomatic bacteriuria is indeed clinically significant, in that it contributes to substantial morbidity and mortality in persons with spinal cord injury. This does not, however, mean that "significant bacteriuria" is necessarily an indication for antimicrobial therapy. Antimicrobials should only be used when it is clear not only that
Bacteriuria in People with Spinal Cord Injury
bacteriuria adversely affects those who acquire it, but also that treatment prevents or improves the adverse outcome, and that the benefits of treatment outweigh the risks. The indications for treatment of bacteriuria in persons with SCI are discussed elsewhere in this issue. 41
SUMMARY Quantitative urine cultures are used to identify persons in whom bladder urine is not sterile, that is, those who have bacteriuria. Criteria for diagnosis of bacteriuria vary according to the prevalence of the condition in the population, and the method of specimen collection. For persons with SCI, quantitative criteria with optimal sensitivity and specificity for diagnosis of bacteriuria are: (1) Catheter specimens from individuals on intermittent catheterization: ;::: 10 2 cfu/ml. (2) Clean-void specimens from catheter-free males who use condom collection devices: ;::: 10 4 cfulml. (In this group, cultures yielding between 10 2 and 10 4 cfu/ml may require confirmatory suprapubic aspiration). (3) Specimens from indwelling catheters: any detectable concentration. (4) Suprapubic aspirates: any detectable concentration. For these four groups the traditional criterion, ;::: 10 5 cfu/ml, has unacceptably low sensitivity for most clinical, epidemiologic, or research purposes. Screening tests for bacteriuria are presently not able to reliably detect low-count bacteriuria, which is common in persons with SCI. In persons with SCI, urine cultures yielding coagulase-negative staphylococci, or multiple organisms, cannot be assumed to be contaminated. Full identification and sensitivity testing of such isolates may be necessary for clinical or infectioncontrol purposes. Abnormal levels of pyuria are present in the great majority of persons with SCI who have indwelling catheters and also those on intermittent catheterization. The most commonly used method of estimating pyuria, enumeration ofleukocytes per high power field in centrifuged urine, is poorly reproducible and does not reliably correlate either with the leukocyte excretion rate or with the recommended method, enumeration of
211
leukocytes in uncentrifuged urine, using a hemocytometer chamber. In persons on intermittent catheterization, catheter urine is inhomogeneous and leukocyte counts are significantly higher in terminal catheter urine than in mid-catheter specimens or suprapubic aspirates. Absence of pyuria is a good predictor of absence of high-level gramnegative bacteriuria, but pyuria may be present in the absence of bacteriuria. 'rests for pyuria have been advocated by some as a method of identifying asymptomatic bacteriuric individuals who require treatment. However, no data currently exists to support the effectiveness of this approach, although it is intuitively appealing to use pyuria as an indication of inflammation and/or invasive infection. Ureteral catheterization is the "gold-standard" method oflocalization of infection, but is not suitable for routine clinical use. The bladder washout test, antibody coated bacteria test, and other current noninvasive or indirect tests oflocalization or invasiveness are presently not sufficiently sensitive or specific to be useful in the clinical arena. Although bacteriuria among persons with SCI is most often asymptomatic, it nonetheless is responsible for substantial morbidity and mortality. Bacteriuria, as defined above, is significant in that the urine in the bladder is not sterile, and in that there may be associated morbidity even when symptoms are absent. However, "significant bacteriuria" is not, by itself, an indication for antimicrobial therapy.
REFERENCES 1. Gribble Mj, McCallum NM, Schechter MT. Eval-
uation of diagnostic criteria for bacteriuria in acutely spinal cord injured patients undergoing intermittent catheterization. Diagn Microbiol InJect Dis 1988; 9:197-206. 2. Deresinski SC, Perkash I. Urinary tract infections in male spinal cord injured patients. Part One: Bacteriologic diagnosis. ] Am Parapl Soc 1985; 8:4-6. 3. Stark RP, Maki DG. Bacteriuria in the catheterized patient: What quantitative level of bacteriuria is relevant? N EnglJ Med 1984; 311 :560564.
212
NEUROREHABILITATION / OCTOBER 1994
4. Warren JW, Tenney JH, Hoopes .1M, Muncie HL. A prospective microbiologic study ofbacteriuria in patients with chronic indwelling urethral catheters.] Infect Dis 1982; 146:719-723. 5. Bergqvist D, Bronnestam R, Hedelin H, et al. The relevance of urinary sampling methods in patients with indwelling foley catheters. BrJ Urol 1980; 52:92-95. 6. Grahn D, Norman DC, White ML, et al. Validity of urinary catheter specimen for diagnosis of urinary tract infection in the elderly. Arch Intern Med 1985; 145:1858-1860. 7. Nicolle LE. Urine cultures and long-term indwelling catheters. Arch Intern Med 1985; 145: 1794-1795. 8. Sapico FL, Wideman PA, Finegold SM. Aerobic and anaerobic flora in bladder urine of patients with indwelling urethral catheters. Urology 1976; 7:382-384. 9. Tenney JH, Warren Jw. Long-term catheterassociated bacteriuria: Species at low concentration. Urology 1987; 30:444-446. 10. Stamm WE. Measurement of pyuria and its relation to bacteriuria. Am] Med 1983; 75(suppl IB):53-58. 11. Musher DM, Thorsteinsson SB, Airola II VM. Quantitative urinalysis: Diagnosing urinary tract infection in men. ]AMA 1976; 236:20692072. 12. Peterson JR, Roth EJ. Fever, bacteriuria, and pyuria in spinal cord injured patients with indwelling urethral catheters. Arch Phys Med Rehabil 1989; 70:839-841. 13. Menon EB, Tan ES. Pyuria: index of infection in patients with spinal cord injuries. Br] Uro11992; 69:144-146. 14. Deresinski SC, Perkash I. Urinary traction infections in male spinal cord injured patients. Part Two: Diagnostic value of symptoms and of quantitative urinalysis.] Am Parapl Soc 1985; 8:7-10. 15. Hooton TM, O'Shaughnessy EJ, Clowers D, et al. Localization of urinary tract infection in patients with spinal cord injury. ] Infect Dis 1984; 150(1):85-9l. 16. Anderson RU, Hsieh-Ma ST. Association ofbacteriuria and pyuria during intermittent catheterization after spinal cord injury.] Urol 1983; 130:299-30l. 17. Gribble MJ, Puterman L, McCallum NM. Pyuria: Its relationship to bacteriuria in spinal cord injured patients on intermittent catheterization. Arch Phys Med Rehabil1989; 70:376-379. 18. Davidoff G, Schultz .IS, Lieb T, et al. Rehospitalization after initial rehabilitation for acute spinal cord injury: Incidence and risk factors. Arch Phys Med Rehabil1990; 71:121-124.
19. Cardenas DD, Mayo ME. Bacteriuria with fever after spinal cord injury. Arch Phys Med Rehabil 1987; 68:291-293. 20. Dietrick RB, Russi S. Tabulation and review of autopsy findings in fifty-five paraplegics. ]AMA 1958; 166:41-44. 21. Tribe CB, Silver JR. Renal failure in paraplegia. London: Pitman Medical Publishing Co., 1969. 22. Talbot HS. Renal disease and hypertension in paraplegics and quadriplegics. Med Seru] Canada 1966; 22:570-575. 23. Warren .JW, Muncie .Jr HL, Hall-Craggs M. Acute pyelonephritis associated with bacteriuria during long-term catheterization: A prospective clinicopathological study. ] Infect Dis 1988; 158:1341-1346. 24. Devivo MJ, Kartus PL, Stover SL, et al. Cause of death for patients with spinal cord injuries. Arch Intern Med 1989; 149:1761-1766. 25. Stamey TA, Govan DE, Palmer JM. The localization and treatment of urinary tract infections: The role of bactericidal urine levels as opposed to serum levels. Medicine (Baltimore) 1965; 44:236. 26. Fairley KF, Bond AG, Brown RB, et al. Simple test to determine the site of urinary-tract infection. Lancet 1967; 2(7513):427-428. 27. Fairley KF, Grounds AD, Carson NE, et al. Site of infection in acute urinary-tract infection in general practice. Lancet 1971; 2(7725):615-618. 28. Kuhlemeier KV, Lloyd LK, Stover SL. Failure of antibody-coated bacteria and bladder washout tests to localize infection in spinal cord injury patients.] Uro11983; 130:729-732. 29. Gribble MJ, Lee L, Fink M, et al. Localization of asymptomatic bacteriuria in spinal cord injured patients. Abstract No. 363. Clinical and Investigative Medicine 1990; 13(4):358. 30. MerrittJL, Keys TE Limitations of the antibodycoated bacteria test in patients with neurogenic bladders.]AMA 82; 247(12):1723-1725. 31. Boutros P, Mourtada H, Ronald AR. Localization of urinary infection. Antimicrob Agents Chemother 1970; 114-116. 32. Thomas V, Shelokov A, Forland M. Antibodycoated bacteria in the urine and the site of urinary-tract infection. N Engl ] Med 1974; 290:588-590. 33. Giamarellou H. Antibody-coated bacteria in urine: When, where and why? ] Antimicrob Chemother 1984; 13(2):95-99. 34. Thorley JD, Barbin GK, Reinarz JA. The prevalence of antibody-coated bacteria in urine. Am] Med Sci 1978; 275(1):75-80. 35. Newman E, Price M, Ederer GM. Urinary tract infection in patients with spinal cord lesions:
Bacteriuria in People with Spinal Cord Injury
Antibody-coated bacteria tests as a diagnostic aid. Arch Phys Med Rehabil1980; 61:406-409. 36. Lindan R. The significance of antibody coated bacteria in neuropathic bladder urines. Paraplegia 1981; 19(4):216-219. 37. Wyndaele jj, Everaert K, Rysselaere M, et al. An evaluation of the bladder wash-out test and the anti-body-coated bacteria test for the localization of a urinary tract infection in patients with and without a neurogenic bladder. Acta Urologica Belgica 1986; 56(4):471-476. 38. Galloway A, Green HT, Menon KK, et al. Antibody coated bacteria in urine of patients with recent spinal injury. J Clin Pathol 1990; 43:953956.
213
39. National Institute or Disability and Rehabilitation Research Consensus Statement: The prevention and management of urinary tract infections among people with spinal cord injuries. J Am Parapl Soc 1992; 15:194-207. 40. Gribble Mj, Puterman ML. Prophylaxis of urinary tract infection in persons with recent spinal cord injury: A prospective, randomized, doubleblind, placebo-controlled study of trimethoprim-sulfamethoxazole. Am J Med 1993; 95:141-152. 41. Montgomorie jZ. Treatment of urinary tract infections in persons with spinal cord injury. Neurorehabill994; 4(4):214-221.
