British Journalof Urology (1992), 70, 314-317 0 1992 8,ritish Journal of Urology
Screening for Bacteriuria in Urological Patients using Reagent Strips SARAH J. MILLS, M. FORD, F. KATE GOULD, SALLY BURTON and D. E. NEAL Departments of Urology and Microbiology, Freeman Hospital, Newcastle upon Tyne and Department of Surgery, University of Newcastle upon Tyne
Summary-A rapid and accurate ward-based method of diagnosing urinary infection would be of value in determining the prescription of antibiotics in pre-operative urological patients. This study describes the sensitivity and specificity of a screening technique based on commercially available reagent strips in the diagnosis of urinary infection. A total of 222 pre-operative samples and 83 post-operative samples was studied to compare the results of formal urine culture and reagent strips. Using a definition of a positive nitrite or a positive leucocyte esterase on the reagent strips as being suggestive of infection, it was found that the strips had a sensitivity of 91% and a specificity of 85%compared with formal culture in pre-operative samples. The results from post-operative samples were less satisfactory, the strips having a sensitivity of only 71% and specificity of 55%.The strips were insensitive but specific in the identification of pyuria in pre-operative specimens. These results suggest that reagent strips can be used as a ward-based method to identify men at risk of infection before urological procedures, and may allow selectivity in the use of peri-operative antibiotics.
Bacteraemia is found in 60% of men undergoing transurethral prostatectomy who have bacteriuria or indwelling urethral catheters before operation, and septicaemia occurs in 6% of this group: these data argue strongly in favour of the use of perioperative antibiotics in such patients. In men with sterile urine, the incidence of septicaemia and bacteraemia is much lower, being about 1% and 25% respectively (Grabe, 1987), although even in those with sterile urine, bacteria are present in the prostates of 25% (Gorelic et al., 1988). There is also good evidence that the morbidity of instrumentation of the urinary tract is significantly increased in the presence of infected urine (Sullivan et al., 1973; Quintiliani et al., 1978) and that the morbidity of prostatectomy can be decreased in such circumstances by the administration of peri-operative antibiotics (McEntee et al., 1987). On the other hand, evidence that the use of peri-operative parenteral Accepted for publication 20 September 1991
antibiotics can reduce morbidity in men with sterile urine or those without indwelling urethral catheters before operation remains uncertain (Grabe, 1987). Another period of high risk for men undergoing prostatectomy is the time of catheter removal (Gordon et al., 1983), when there is a case to be made for the routine administration of a short course of oral antibiotics to men at high risk-e.g. those with pre-operative infection, those with longterm indwelling catheters and those likely to experience difficulty in achieving complete bladder emptying after prostatectomy, such as men with chronic retention. Many urologists, therefore, operate a selective policy for the administration of peri-operative antibiotics. Crucial to the success of such a policy is the prompt identification of those with preoperative bacteriuria; herein lies a number of potential difficulties. The results of formal urine culture take 24 h to obtain, and therefore several screening methods have been introduced (Stevens, 1989). One commonly used method involves the use
314
315
SCREENING PROCEDURE FOR BACTERIURIA
of a particle counter which permits identification of bacteriuria with a good degree of sensitivity. Unfortunately, urine samples suspected to be positive by this technique will then require formal culture for confirmation. A ward-based method of identifying patients with bacteriuria quickly and reliably would be beneficial. It is important for such tests to have high sensitivity and also desirable for them to be specific, to reduce costs and to diminish the risk of the emergence of resistant organisms which might occur with unnecessary antibiotic administration. The use of reagent strips which detect blood, leucocyte esterase and nitrites has been reported to be useful in the diagnosis of urinary infection in an elderly population (Flanagan, 1989). We have evaluated the use of such strips in a urological population.
Materials and Methods We evaluated a total of 305 urine samples from 222 men (mean age 65 years, range 16-96) admitted to 1 urological ward between September 1990 and February 1991 : these samples were obtained from 222 men studied before operation and 83 of them afterwards. In all, 32 men had indwelling catheters at the time of admission. Of these, 60% were admitted for endoscopic procedures and 15% for open surgery ; the remainder were emergency admissions or undergoing urological investigation. Mid-stream or catheter urine samples were taken on admission, whereas specimens after operation were all taken from the catheters of men undergoing endoscopic procedures. Specimens were collected in the morning by the nursing staff and transported promptly to the laboratory for culture. All specimens were tested with reagent strips immediately after collection and the results read and recorded by the nursing staff. The strips used were the Ames Mull iple Reagent Strips (Ames Division, Miles Laboratory Ltd). These are firm plastic strips with reagent areas for glucose, bilirubin, ketone, specific gravity, blood, protein, nitrite and leucocyte esterase. We limited our observations to results for blood, protein, nitrite and leucocyte esterase, recording any reaction greater than trace positive as representing a positive result. The nitrite test depends upon the conversion of nitrate to nitrite, which is principally a feature of gram-negative bacteria. The test is specific for nitrite and if positive suggests the presence of > 100,000 organisms/ml. False negative results can occur when urinary tract infections are caused by organisms
that do not contain nitrate reductase, when urine has been in the bladder for insufficiently long periods for the reduction of nitrate to occur (e.g. when indwelling catheters are present), or when dietary nitrate is absent. A positive leucocyte esterase test depends on the presence of esterase in granulocytic leucocytes. The sensitivity of this test can be affected by the presence of certain drugs in the urine. Urine samples were cultured by the standard technique of our laboratory. Calibrated loops are used which deliver 1 ~ urine 1 on to blood and MacConkey agar plates; these plates are incubated overnight at 37°C in air. Using the results of formal urine culture results as our gold standard, the sensitivity and specificity were calculated separately and in combination for each of the 4 reagent strip parameters mentioned above. A positive culture result was considered initially as > 100,000 colony forming units per ml (cfu/ml) of a single organism. We also re-calculated the sensitivity and specificity of the strips after broadening the definition of infection to include results of formal cultures when < 100,000 cfu of 2 different organisms were present. Contaminated specimens, where low counts of multiple organisms were present ( < 20,000 cfu/ml), were grouped with the negative cultures which were otherwise defined as containing < 20,000 cfu/ml. Pre-operative and post-operative samples were analysed separately, and amongst the pre-operative samples we reported the results separately for those with or without indwelling catheters.
Results Of the 222 pre-operative urine specimens 3 1 (14%) were infected, using a definition of > 100,000 cfu/ml of a single organism; 7% were contaminated and 79% were sterile. When the definition of a positive culture included cultures of 2 organisms, 16%were infected, 5% were contaminated and 79% were sterile. The types of organisms cultured are reported in Table 1. Thirty-two of the 222 men had indwelling catheters and of these samples, 34% were infected, 25% were contaminated and 41% sterile. The sensitivity and specificity calculated for each of the 4 parameters in the reagent strips in isolation are shown in Table 2 for the pre-operative urine specimens. When a positive reagent result was defined as being either nitrite positive or leucocyte esterase positive, sensitivity was 91% and specificity was 85%. Moreover, when men with indwelling
316
BRITISH JOURNAL OF UROLOGY
Table 1 Results of Formal Culture
esterase positive) gave a sensitivity of 71% and a specificity of 55%. Pyuria was present in 75 patients studied before operation and this was identified by a positive leucocyte esterase test in 47 of them (sensitivity = 63%); 10 of 142 patients had a false positive test (specificity = 93%).
> 100,000 cfulml for single organisms Coliforms* Pseudomonas Klebsiella Enterococcus faecalis Coagulase negative staphylococci Proteus Staph. aureus
Discussion
> 100,000 cfulmlfor 2organisms Pseudomonus Enterococcus Coliform Candidu Coliform Klebsiella
+ +
+
< 100,000 > 1000 cfu/ml Diphtheroidt
1
* Coliform organisms were gram-negative, oxidase-negative, non-lactose fermenting bacilli. t Probable contaminant. Table 2 Pre-operativeUrine Specimens :Reagent Strips Compared with Formal Urine Culture
Nitrite Protein Leucocyte Blood
Sensitivity (%)
Specificity (%)
60.5 52.6 78.9 65.8
94.6 62.2 87.0 71.9
catheters were excluded from these results the sensitivity was 89% and the specificity was 91%. When a positive formal urine culture was defined as > l(30,OOOcfu/ml of 1 or 2 organisms, the sensitivity was 88% and the specificity was 86%. The frequency distribution of bacterial concentrations in urine specimens which grew any organisms at all was as follows: 69% were infected with > 100,000 cfu/ml, 9% grew 2 different organisms at counts of > 100,000 cfu/ml; only 1.5%of specimens grew a single organism at concentrations between > 20,000 and < 100,000 cfu/ml :the remainder were contaminated with mixed growth of > 2 organisms at low counts. After operation, 6% of patients who had sterile urine beforehand and who had undergone endoscopic procedures developed infected urine postoperatively. Of the 83 specimens studied after operation, 18% were infected, 7% were contaminated and 75% were sterile. None of the 4 reagent strip parameters enabled infection to be diagnosed in this group with good sensitivity and specificity. The best combination of parameters in the preoperative specimens (nitrite positive or leucocyte
These results suggest that reagent strips have a good sensitivity and specificity in the diagnosis of urinary infection in pre-operative ward-based samples from men undergoing urological surgery. One problem in the evaluation of the reagent strips arose in deciding the best definition of urinary tract infection in the context of a cohort of urological patients. The conventional criteria are those of Kass (1957), who compared the sensitivity of midstream urine specimens with the culture of suprapubic specimens in the detection of E. coliinfection. It should be noted that he found that more than 100,000 colony forming units (cfuiml) of E. coli in 3 mid-stream specimens was highly significantly associated with the presence of E. coli in suprapubic specimens. In practice, this definition has been broadened to include other organisms and is often taken to include only 1 positive mid-stream specimen. Many studies or urinary infection and prostatectomy (Symes et al., 1972; Jackaman and Chisholm, 1975; Houle et a/. 1989) and the use of reagent strips (Bolann et al., 1989; Flanagan et al., 1989) have accepted this definition. However, some authors have suggested that lower concentrations of bacteria in mid-stream specimens may be significant and that low concentrations of bacteria from catheter specimens in the early post-operative period after prostatectomy may also be significant (Essenhigh and Clayton, 1970; Bruce et al., 1971; Gordon e t a / . , 1983). It has been advocated that the definition should be tailored to the population under analysis (Stevens, 1989) and furthermore, that a frequency distribution of bacterial counts be produced during evaluation of bacteriuria screening systems (Johnson et a/., 1982). In practice, we found that few specimens in our population of men grew organisms at concentrations between 20,000 cfu/ml and > 100,000 cfu/ml. Pyuria was detected in only 63% of patients studied before operation in whom pyuria was detected by microscopy, although the reagent strips were accurate in identifying men without pyuria (specificity: 93%). This study was not designed to determine the use of the strips in detecting sterile
317
SCREENING PROCEDURE FOR BACTERIURIA
pyuria, but the results shown here do not suggest that a negative test could be used with confidence in the out-patient department to exclude sterile pyuria. although a positive test might be useful. In calculating the ability of the reagent strips to detect bacteriuria, we initially looked at the results using it definition of > 100,000 cfu/ml of a single organism and then included specimens with 2 different organisms at > 100,000 cfu/ml, as these are unlikely to represent contamination. We found that il: is possible to identify with accuracy significant bacteriuria in pre-operative specimens by the use of reagent strips. The clinical application of this would be the prescription of prophylactic antibiotic cover for the cohort undergoing endoscopjcsurgery. It is usual practice in this unit to give antibiotics to all patients with indwelling catheters on the assumption that they have an associated bacteriuria. Therefore, we calculated the sensitivity and specificity of the reagent strips after excluding this group. On this basis we detected 89%of patients with pre-operative bacteriuria. This, we feel, is good enough to indictate the prescribing of antibiotic prophylaxis. In addition, with a specificity of 91%, we would not be prescribing unnecessarily in many patients, with the attendant cost and danger of promoting bacterial resistance which that incurs. We feel that these reagent strips have a role to play in the examination of pre-operative specimens from male urological patients.
Flanagan, P. G., Davies, E. A., Rooney, P. G. er al. (1989). Evaluation of four screening tests for bacteriuria in elderly people. Lancet,i, 1117-1118. Grabe,M. (1987). Antimicrobial agents in transurethralprostatic resections. J . Urol., 138,245-252. Gordon, D. L., Bune, A., Grime, B. et al. (1983). Diagnostic criteria and natural history of catheter associated urinary tract infections after prostatectomy. Lancet, ii, 1269-1271. Gorelic, J. I., Senterfit, L. B. and Darracott Vaughan, E. (1988). Quantitative bacterial tissue cultures from 209 prostatectomy specimens. J . Urol., 139, 57-60. Houle, A. M., Mokhless, I., Sarto, N. el d (1989). Perioperative antibiotic prophylaxis for transurethral resection of the prostate: is itjustifiable?J. Urol., 142, 317-319. Jackaman, F. R. and Chisholm, G. D. (1975). Urinary infection and prostatectomy. Br. J . Urol., 47, 545-548. Johnson, H. H., Curtis, G . D. W. and Nichols, W. W. (1982). Current status of bioluminescence as a means of detecting significant bacteriuria. In Rapid Methods and Automation in Microbiology, ed. Tilton, R. C. Pp. 162-165. Washington: American Society for Microbiology. Kass, E. H. (1957). Bacteriuria and the diagnosis of infections of the urinary tract. Arch. Intern. Med., 100, 709-714. McEntee, G . P., McPhail, S., Mulvin, D. et al. (1987). Single dose antibiotic prophylaxis in high risk patients undergoing transurethral prostatectomy. Br. J . Surg., 74, 192-194. Quintiliani, R., K h e k , J., Cunha, B. A. e t d (1978). Bacteraemia after manipulation of the urinary tract. The importance of pre-existing urinary tract disease and compromised host defences. Postgrad. Med. J . , 54,668-671. Stevens, M. (1989). Screening urines for bacteriuria. Med. Lab. Sci., 46, 194-206. Sullivan, N. M., Sutter, V. L., Nms, M. M. etal. (1973). Clinical aspects of bacteraemia after manipulation of the genitourinary tract. J . Infect. Dis., 127,49-55. Symes, J. M., Hardy, D. G., Sutherns, K. et d (1972). Factors reducing the rate of infection after transurethral surgery. Br. J . Urol.,44, 582-586.
Acknowledgements We thank the nursing staff of Ward 3, Freeman Hospital, who helped with the study, and Messrs D. M. Essenhigh and R. R. Hall, who allowed us to study their patients.
References Bolann, J., Sandberg, S. and Digranes, A. (1989). Implications of probability analysis for interpreting results of leucocyte esterase and nitrite test strips. Clin. Chem., 35/8, 1663-1668. Bruce, A. W., Rao, C. R. and Kennedy, W. (1971). Prostatectomy and infection. J . Urol., 106,910-912. Essenhigh, D. M., Clayton, C. B. and Taha, M. A. (1970). The use of Lasix (frusemide) in the prevention of urinary infection following prostatectomy. Br. J . Urol., 42,450-456.
The Authors Sarah J . Mills, MB, BS, Registrar in Urology. M. Ford, FIMLS, Senior Medical Laboratory Scientific Officer, Department of Microbiology. F. Kate Could, MRCPath, Consultant Microbiologist, Department of Pathology. Sally Burton, RGN, Nursing Sister, Department of Urology. D. E. Neal, MS, FRCS, BSc, Senior Lecturer and Honorary Consultant Urological Surgeon, Department of Urology and University Department of Surgery. Requests for reprints to: D. E. Neal, Department of Urology, Freeman Hospital, Freeman Road, High Heaton, Newcastle upon Tyne NE7 7DN.
Screening for Bacteriuria in Urological Patients using Reagent Strips SARAH J. MILLS, M. FORD, F. KATE GOULD, SALLY BURTON and D. E. NEAL Departments of Urology and Microbiology, Freeman Hospital, Newcastle upon Tyne and Department of Surgery, University of Newcastle upon Tyne
Summary-A rapid and accurate ward-based method of diagnosing urinary infection would be of value in determining the prescription of antibiotics in pre-operative urological patients. This study describes the sensitivity and specificity of a screening technique based on commercially available reagent strips in the diagnosis of urinary infection. A total of 222 pre-operative samples and 83 post-operative samples was studied to compare the results of formal urine culture and reagent strips. Using a definition of a positive nitrite or a positive leucocyte esterase on the reagent strips as being suggestive of infection, it was found that the strips had a sensitivity of 91% and a specificity of 85%compared with formal culture in pre-operative samples. The results from post-operative samples were less satisfactory, the strips having a sensitivity of only 71% and specificity of 55%.The strips were insensitive but specific in the identification of pyuria in pre-operative specimens. These results suggest that reagent strips can be used as a ward-based method to identify men at risk of infection before urological procedures, and may allow selectivity in the use of peri-operative antibiotics.
Bacteraemia is found in 60% of men undergoing transurethral prostatectomy who have bacteriuria or indwelling urethral catheters before operation, and septicaemia occurs in 6% of this group: these data argue strongly in favour of the use of perioperative antibiotics in such patients. In men with sterile urine, the incidence of septicaemia and bacteraemia is much lower, being about 1% and 25% respectively (Grabe, 1987), although even in those with sterile urine, bacteria are present in the prostates of 25% (Gorelic et al., 1988). There is also good evidence that the morbidity of instrumentation of the urinary tract is significantly increased in the presence of infected urine (Sullivan et al., 1973; Quintiliani et al., 1978) and that the morbidity of prostatectomy can be decreased in such circumstances by the administration of peri-operative antibiotics (McEntee et al., 1987). On the other hand, evidence that the use of peri-operative parenteral Accepted for publication 20 September 1991
antibiotics can reduce morbidity in men with sterile urine or those without indwelling urethral catheters before operation remains uncertain (Grabe, 1987). Another period of high risk for men undergoing prostatectomy is the time of catheter removal (Gordon et al., 1983), when there is a case to be made for the routine administration of a short course of oral antibiotics to men at high risk-e.g. those with pre-operative infection, those with longterm indwelling catheters and those likely to experience difficulty in achieving complete bladder emptying after prostatectomy, such as men with chronic retention. Many urologists, therefore, operate a selective policy for the administration of peri-operative antibiotics. Crucial to the success of such a policy is the prompt identification of those with preoperative bacteriuria; herein lies a number of potential difficulties. The results of formal urine culture take 24 h to obtain, and therefore several screening methods have been introduced (Stevens, 1989). One commonly used method involves the use
314
315
SCREENING PROCEDURE FOR BACTERIURIA
of a particle counter which permits identification of bacteriuria with a good degree of sensitivity. Unfortunately, urine samples suspected to be positive by this technique will then require formal culture for confirmation. A ward-based method of identifying patients with bacteriuria quickly and reliably would be beneficial. It is important for such tests to have high sensitivity and also desirable for them to be specific, to reduce costs and to diminish the risk of the emergence of resistant organisms which might occur with unnecessary antibiotic administration. The use of reagent strips which detect blood, leucocyte esterase and nitrites has been reported to be useful in the diagnosis of urinary infection in an elderly population (Flanagan, 1989). We have evaluated the use of such strips in a urological population.
Materials and Methods We evaluated a total of 305 urine samples from 222 men (mean age 65 years, range 16-96) admitted to 1 urological ward between September 1990 and February 1991 : these samples were obtained from 222 men studied before operation and 83 of them afterwards. In all, 32 men had indwelling catheters at the time of admission. Of these, 60% were admitted for endoscopic procedures and 15% for open surgery ; the remainder were emergency admissions or undergoing urological investigation. Mid-stream or catheter urine samples were taken on admission, whereas specimens after operation were all taken from the catheters of men undergoing endoscopic procedures. Specimens were collected in the morning by the nursing staff and transported promptly to the laboratory for culture. All specimens were tested with reagent strips immediately after collection and the results read and recorded by the nursing staff. The strips used were the Ames Mull iple Reagent Strips (Ames Division, Miles Laboratory Ltd). These are firm plastic strips with reagent areas for glucose, bilirubin, ketone, specific gravity, blood, protein, nitrite and leucocyte esterase. We limited our observations to results for blood, protein, nitrite and leucocyte esterase, recording any reaction greater than trace positive as representing a positive result. The nitrite test depends upon the conversion of nitrate to nitrite, which is principally a feature of gram-negative bacteria. The test is specific for nitrite and if positive suggests the presence of > 100,000 organisms/ml. False negative results can occur when urinary tract infections are caused by organisms
that do not contain nitrate reductase, when urine has been in the bladder for insufficiently long periods for the reduction of nitrate to occur (e.g. when indwelling catheters are present), or when dietary nitrate is absent. A positive leucocyte esterase test depends on the presence of esterase in granulocytic leucocytes. The sensitivity of this test can be affected by the presence of certain drugs in the urine. Urine samples were cultured by the standard technique of our laboratory. Calibrated loops are used which deliver 1 ~ urine 1 on to blood and MacConkey agar plates; these plates are incubated overnight at 37°C in air. Using the results of formal urine culture results as our gold standard, the sensitivity and specificity were calculated separately and in combination for each of the 4 reagent strip parameters mentioned above. A positive culture result was considered initially as > 100,000 colony forming units per ml (cfu/ml) of a single organism. We also re-calculated the sensitivity and specificity of the strips after broadening the definition of infection to include results of formal cultures when < 100,000 cfu of 2 different organisms were present. Contaminated specimens, where low counts of multiple organisms were present ( < 20,000 cfu/ml), were grouped with the negative cultures which were otherwise defined as containing < 20,000 cfu/ml. Pre-operative and post-operative samples were analysed separately, and amongst the pre-operative samples we reported the results separately for those with or without indwelling catheters.
Results Of the 222 pre-operative urine specimens 3 1 (14%) were infected, using a definition of > 100,000 cfu/ml of a single organism; 7% were contaminated and 79% were sterile. When the definition of a positive culture included cultures of 2 organisms, 16%were infected, 5% were contaminated and 79% were sterile. The types of organisms cultured are reported in Table 1. Thirty-two of the 222 men had indwelling catheters and of these samples, 34% were infected, 25% were contaminated and 41% sterile. The sensitivity and specificity calculated for each of the 4 parameters in the reagent strips in isolation are shown in Table 2 for the pre-operative urine specimens. When a positive reagent result was defined as being either nitrite positive or leucocyte esterase positive, sensitivity was 91% and specificity was 85%. Moreover, when men with indwelling
316
BRITISH JOURNAL OF UROLOGY
Table 1 Results of Formal Culture
esterase positive) gave a sensitivity of 71% and a specificity of 55%. Pyuria was present in 75 patients studied before operation and this was identified by a positive leucocyte esterase test in 47 of them (sensitivity = 63%); 10 of 142 patients had a false positive test (specificity = 93%).
> 100,000 cfulml for single organisms Coliforms* Pseudomonas Klebsiella Enterococcus faecalis Coagulase negative staphylococci Proteus Staph. aureus
Discussion
> 100,000 cfulmlfor 2organisms Pseudomonus Enterococcus Coliform Candidu Coliform Klebsiella
+ +
+
< 100,000 > 1000 cfu/ml Diphtheroidt
1
* Coliform organisms were gram-negative, oxidase-negative, non-lactose fermenting bacilli. t Probable contaminant. Table 2 Pre-operativeUrine Specimens :Reagent Strips Compared with Formal Urine Culture
Nitrite Protein Leucocyte Blood
Sensitivity (%)
Specificity (%)
60.5 52.6 78.9 65.8
94.6 62.2 87.0 71.9
catheters were excluded from these results the sensitivity was 89% and the specificity was 91%. When a positive formal urine culture was defined as > l(30,OOOcfu/ml of 1 or 2 organisms, the sensitivity was 88% and the specificity was 86%. The frequency distribution of bacterial concentrations in urine specimens which grew any organisms at all was as follows: 69% were infected with > 100,000 cfu/ml, 9% grew 2 different organisms at counts of > 100,000 cfu/ml; only 1.5%of specimens grew a single organism at concentrations between > 20,000 and < 100,000 cfu/ml :the remainder were contaminated with mixed growth of > 2 organisms at low counts. After operation, 6% of patients who had sterile urine beforehand and who had undergone endoscopic procedures developed infected urine postoperatively. Of the 83 specimens studied after operation, 18% were infected, 7% were contaminated and 75% were sterile. None of the 4 reagent strip parameters enabled infection to be diagnosed in this group with good sensitivity and specificity. The best combination of parameters in the preoperative specimens (nitrite positive or leucocyte
These results suggest that reagent strips have a good sensitivity and specificity in the diagnosis of urinary infection in pre-operative ward-based samples from men undergoing urological surgery. One problem in the evaluation of the reagent strips arose in deciding the best definition of urinary tract infection in the context of a cohort of urological patients. The conventional criteria are those of Kass (1957), who compared the sensitivity of midstream urine specimens with the culture of suprapubic specimens in the detection of E. coliinfection. It should be noted that he found that more than 100,000 colony forming units (cfuiml) of E. coli in 3 mid-stream specimens was highly significantly associated with the presence of E. coli in suprapubic specimens. In practice, this definition has been broadened to include other organisms and is often taken to include only 1 positive mid-stream specimen. Many studies or urinary infection and prostatectomy (Symes et al., 1972; Jackaman and Chisholm, 1975; Houle et a/. 1989) and the use of reagent strips (Bolann et al., 1989; Flanagan et al., 1989) have accepted this definition. However, some authors have suggested that lower concentrations of bacteria in mid-stream specimens may be significant and that low concentrations of bacteria from catheter specimens in the early post-operative period after prostatectomy may also be significant (Essenhigh and Clayton, 1970; Bruce et al., 1971; Gordon e t a / . , 1983). It has been advocated that the definition should be tailored to the population under analysis (Stevens, 1989) and furthermore, that a frequency distribution of bacterial counts be produced during evaluation of bacteriuria screening systems (Johnson et a/., 1982). In practice, we found that few specimens in our population of men grew organisms at concentrations between 20,000 cfu/ml and > 100,000 cfu/ml. Pyuria was detected in only 63% of patients studied before operation in whom pyuria was detected by microscopy, although the reagent strips were accurate in identifying men without pyuria (specificity: 93%). This study was not designed to determine the use of the strips in detecting sterile
317
SCREENING PROCEDURE FOR BACTERIURIA
pyuria, but the results shown here do not suggest that a negative test could be used with confidence in the out-patient department to exclude sterile pyuria. although a positive test might be useful. In calculating the ability of the reagent strips to detect bacteriuria, we initially looked at the results using it definition of > 100,000 cfu/ml of a single organism and then included specimens with 2 different organisms at > 100,000 cfu/ml, as these are unlikely to represent contamination. We found that il: is possible to identify with accuracy significant bacteriuria in pre-operative specimens by the use of reagent strips. The clinical application of this would be the prescription of prophylactic antibiotic cover for the cohort undergoing endoscopjcsurgery. It is usual practice in this unit to give antibiotics to all patients with indwelling catheters on the assumption that they have an associated bacteriuria. Therefore, we calculated the sensitivity and specificity of the reagent strips after excluding this group. On this basis we detected 89%of patients with pre-operative bacteriuria. This, we feel, is good enough to indictate the prescribing of antibiotic prophylaxis. In addition, with a specificity of 91%, we would not be prescribing unnecessarily in many patients, with the attendant cost and danger of promoting bacterial resistance which that incurs. We feel that these reagent strips have a role to play in the examination of pre-operative specimens from male urological patients.
Flanagan, P. G., Davies, E. A., Rooney, P. G. er al. (1989). Evaluation of four screening tests for bacteriuria in elderly people. Lancet,i, 1117-1118. Grabe,M. (1987). Antimicrobial agents in transurethralprostatic resections. J . Urol., 138,245-252. Gordon, D. L., Bune, A., Grime, B. et al. (1983). Diagnostic criteria and natural history of catheter associated urinary tract infections after prostatectomy. Lancet, ii, 1269-1271. Gorelic, J. I., Senterfit, L. B. and Darracott Vaughan, E. (1988). Quantitative bacterial tissue cultures from 209 prostatectomy specimens. J . Urol., 139, 57-60. Houle, A. M., Mokhless, I., Sarto, N. el d (1989). Perioperative antibiotic prophylaxis for transurethral resection of the prostate: is itjustifiable?J. Urol., 142, 317-319. Jackaman, F. R. and Chisholm, G. D. (1975). Urinary infection and prostatectomy. Br. J . Urol., 47, 545-548. Johnson, H. H., Curtis, G . D. W. and Nichols, W. W. (1982). Current status of bioluminescence as a means of detecting significant bacteriuria. In Rapid Methods and Automation in Microbiology, ed. Tilton, R. C. Pp. 162-165. Washington: American Society for Microbiology. Kass, E. H. (1957). Bacteriuria and the diagnosis of infections of the urinary tract. Arch. Intern. Med., 100, 709-714. McEntee, G . P., McPhail, S., Mulvin, D. et al. (1987). Single dose antibiotic prophylaxis in high risk patients undergoing transurethral prostatectomy. Br. J . Surg., 74, 192-194. Quintiliani, R., K h e k , J., Cunha, B. A. e t d (1978). Bacteraemia after manipulation of the urinary tract. The importance of pre-existing urinary tract disease and compromised host defences. Postgrad. Med. J . , 54,668-671. Stevens, M. (1989). Screening urines for bacteriuria. Med. Lab. Sci., 46, 194-206. Sullivan, N. M., Sutter, V. L., Nms, M. M. etal. (1973). Clinical aspects of bacteraemia after manipulation of the genitourinary tract. J . Infect. Dis., 127,49-55. Symes, J. M., Hardy, D. G., Sutherns, K. et d (1972). Factors reducing the rate of infection after transurethral surgery. Br. J . Urol.,44, 582-586.
Acknowledgements We thank the nursing staff of Ward 3, Freeman Hospital, who helped with the study, and Messrs D. M. Essenhigh and R. R. Hall, who allowed us to study their patients.
References Bolann, J., Sandberg, S. and Digranes, A. (1989). Implications of probability analysis for interpreting results of leucocyte esterase and nitrite test strips. Clin. Chem., 35/8, 1663-1668. Bruce, A. W., Rao, C. R. and Kennedy, W. (1971). Prostatectomy and infection. J . Urol., 106,910-912. Essenhigh, D. M., Clayton, C. B. and Taha, M. A. (1970). The use of Lasix (frusemide) in the prevention of urinary infection following prostatectomy. Br. J . Urol., 42,450-456.
The Authors Sarah J . Mills, MB, BS, Registrar in Urology. M. Ford, FIMLS, Senior Medical Laboratory Scientific Officer, Department of Microbiology. F. Kate Could, MRCPath, Consultant Microbiologist, Department of Pathology. Sally Burton, RGN, Nursing Sister, Department of Urology. D. E. Neal, MS, FRCS, BSc, Senior Lecturer and Honorary Consultant Urological Surgeon, Department of Urology and University Department of Surgery. Requests for reprints to: D. E. Neal, Department of Urology, Freeman Hospital, Freeman Road, High Heaton, Newcastle upon Tyne NE7 7DN.
