Oral ciprofloxacin or trimethoprim reduces bacteriuria after flexible cystoscopy

Authors


David E. Neal, Oncology, University of Cambridge, Cambridge, UK.e-mail:den22@cam.ac.uk

Abstract

OBJECTIVE

To report a large prospective, pragmatic, double-blind randomized controlled trial to determine whether oral prophylactic antibiotics reduce the risk of bacteriuria after flexible cystoscopy (FC), as up to 10% of patients develop urinary infection afterwards, with significant morbidity and costs for health services.

PATIENTS AND METHODS

In all, 2481 patients were recruited into a three-arm placebo controlled trial and 2083 completed it. Patients were randomly assigned to one of three treatments; (i) placebo; (ii) one oral dose of trimethoprim (200 mg); or (iii) one oral dose of ciprofloxacin (500 mg), each administered 1 h before a FC under local anaesthetic. A mid-stream urine specimen was taken before and 5 days after FC; significant bacteriuria was defined as a pure growth of >105 colony-forming units/mL.

RESULTS

The rate of bacteriuria after FC was reduced from 9% in the placebo group to 5% and 3% in patients receiving trimethoprim and ciprofloxacin prophylaxis, respectively. When rates of bacteriuria before FC were considered the odds of developing bacteriuria after FC relative to baseline were 5, 2 and 0.5 for placebo, trimethoprim and ciprofloxacin, respectively.

CONCLUSION

This large trial shows clearly that one dose of oral ciprofloxacin significantly reduces bacteriuria after FC.

Abbreviation
FC

flexible cystoscopy.

INTRODUCTION

Diagnostic cystoscopy accounted for over 220 000 hospital episodes in 2004/5 in England and Wales [1], although the actual number might be higher. Over half are carried out as an outpatient procedure as a flexible cystoscopy (FC) under a local anaesthetic. About 4% of patients have bacteriuria at the time of FC, and a further 8% will develop new bacteriuria afterwards [2]. It is uncertain if one dose of prophylactic oral antibiotics given at the time of FC reduces the incidence of bacteriuria afterwards, because previous studies were too small to answer the question confidently [3,4], used parenteral antibiotics, or continued oral treatment for up to 3 days after FC [5–7]. Because of the very many FCs carried out each year, a significant reduction in bacteriuria would represent a definite benefit to patients and health services. Our aim was to determine if one dose of an oral antibiotic reduced the incidence of bacteriuria after FC; the principal outcome was the presence of significant bacteriuria 5 days after FC, and account was taken of the presence of asymptomatic bacteriuria before FC.

PATIENTS AND METHODS

Adult patients attending the Freeman Hospital for outpatient diagnostic or surveillance (the follow-up of urothelial carcinoma) FC under local anaesthetic were asked if they would take part in the trial, unless they had any of the following exclusion criteria: (i) Symptoms of UTI on the day of investigation; (ii) hypersensitivity to ciprofloxacin or trimethoprim, potential interaction with other drugs (e.g. methotrexate with trimethoprim; theophylline and warfarin for ciprofloxacin), or contraindications (e.g. epilepsy and ciprofloxacin); (iii) a specific indication for parenteral prophylaxis (e.g. heart valve replacement, cardiac murmur, orthopaedic or vascular prosthesis, etc.), i.e. already taking antibiotics at the time of cystoscopy; (iv) presence of a urethral catheter.

The project was approved by the local research ethics committee. A research nurse provided eligible patients with written and oral information. Following consent, a midstream urine specimen was collected for microscopy and culture. Patients were then given a container containing the trial medication, which they were requested to take immediately, i.e. ≈ 1 h before the planned procedure (median 55–59 min). Each container was numbered and held an oral preparation of either placebo, trimethoprim 200 mg or ciprofloxacin 500 mg. An independent pharmacist randomly assigned the container numbers. The hospital pharmacy alone held the code allowing identification of the contents. The FC was done as per routine after instilling topical lignocaine gel into the urethra. The patient was then discharged, provided with a sterile container and request form, and asked to return a midstream specimen of urine 5 days after the investigation. Significant bacteriuria was defined as a pure growth of >105 colony-forming units/mL.

Patients completed a questionnaire to determine the presence of symptoms that were associated with bacteriuria. These were then classified as follows: 0, asymptomatic significant bacteriuria; 1, mild (presence of dysuria plus significant bacteriuria); 2, moderate (presence of dysuria and loin pain plus significant bacteriuria); 3, severe (any combination of the above plus rigors or admission to hospital for infection).

Published rates of bacteriuria after FC are ≈ 8% and antibiotic prophylaxis is reported to reduce this to ≈ 2%. However, these reports were based on the use of parenteral or multiple-dose antibiotic regimens. A reduction in bacteriuria rates from 8% to 4% can be detected with 90% power at the two-sided 5% significance level with two groups each of 740 patients (uncorrected chi-squared method [8]). As comparisons between each antibiotic group and placebo are of interest, all three groups were planned to be of this size. After allowing for a potential discontinuation rate of ≈ 10% we aimed to recruit a total of 2500 patients. Randomization was done in the pharmacy using random-number tables, and the allocation was concealed from the investigators and patients by providing the tablets in numbered opaque containers. The pre-planned analyses were as follows.

The principal analysis compared the bacteriuria rates after FC in the three treatment groups; differences between groups were assessed using odds ratios and a chi-square test. A second analysis considered the presence of significant bacteriuria before FC; this aspect was considered in an analysis which fitted a logistic model with fixed terms for patient and treatment. The patient terms were eliminated using a conditioning argument (similar to the one which leads to McNemar’s test). This analysis used only the discordant patients whose bacteriuria status changed from before to after FC. If the treatments were equally effective then the ratios of the number of those who were: (a) infected before FC but not afterwards, to (b) those who were infected after FC but not before, would be similar. Descriptive statistics on the time between administration of the tablets and when the FC was performed were computed. The study was not designed to detect decreases in clinically significant infection (presence of rigors or admission to hospital because of infection).

RESULTS

In all, 2481 patients were entered into the study between July 1999 and September 2002; 2083 completed the study, with 687 receiving ciprofloxacin, 712 trimethoprim and 684 placebo, and 398 (16%) were randomized but did not complete the study. A flow diagram outlining the patients who did not complete the study is shown in Fig. 1. The median interval between administration of the treatment (placebo, trimethoprim or ciprofloxacin) and starting FC was 55 min, with no evidence of any significant difference among the treatment groups (P = 0.34, placebo 55, trimethoprim 59 and ciprofloxacin 56 min).

Figure 1.

A flow chart with details of the excluded patients.

The primary outcome was a significant positive urine culture from a sample obtained after FC (Table 1); the proportion of patients with a positive urine culture after FC was highest for those allocated to placebo and lowest for ciprofloxacin. The differences are unlikely to be due to chance (chi-square 27.8, P < 0.001). The odds ratio (95% CI) of bacteriuria for trimethoprim relative to placebo was 0.49 (0.32–0.75), for ciprofloxacin relative to placebo was 0.29 (0.17–0.48) and for ciprofloxacin relative to trimethoprim was 0.59 (0.33–1.04).

Table 1.  The results of urine culture after FC in the three arms of the study, the rate of baseline bacteriuria before FC, the discordant pairs showing the risks of change in bacteriuria status, the odds ratio of bacteriuria after FC relative to baseline, the rate of bacteriuria after FC in patients who had a UTI beforehand, and the symptoms associated with the presence of bacteriuria
VariablePlaceboTrimethoprimCiprofloxacinTotal
  • *

    P values (all chi-square); overall (three treatments) <0.001; trimethoprim vs placebo, 1.0; ciprofloxacin vs placebo <0.001; ciprofloxacin vs trimethoprim, 0.02.

  • †Category for the severity for infection stratified by drug; 0, asymptomatic significant bacteriuria; 1, mild (presence of dysuria); 2, moderate (presence of dysuria and loin pain); 3, severe (any combination of the above plus rigors or admission to hospital for infection).

After FC
Negative622679668 
Positive 62 33 19 
% positive  9.1  4.6  2.8 
Total684712687 
Baseline
Negative658690653 
Positive 26 22 34 
% positive  3.8  3.1  5.0 
Total684712687 
Discordant pairs
Before/after FC    
Present/absent (A)  9 11 28 
Present/absent (B) 45 22 13 
Ratio B:A  5  2  0.46 
Odds ratio (95% CI) of bacteriuria vs baseline
   5.0 (2.44–10.23)  2.0 (0.97–4.12)  0.46 (0.24–0.90) 
Bacteriuria after FC in patients who had a UTI before, n (%)*
No UTI after FC  9 (35) 11 (50) 28 (82) 48 (59)
UTI after FC 17 (65) 11 (50)  6 (18) 34 (41)
Total 26 22 34 82
Categoryof UTI, n (%) within drug
0 44 (73) 23 (70) 12 (67) 79 (71)
1  7 (12)  2 (6)  1 (6) 10 (9)
2  6 (10)  8 (24)  3 (17) 17 (15)
3  3 (5)  –  2 (11)  5 (5)
Total 60 33 18111

Some of the patients started the trial with infected urine; Table 1 shows the baseline rate of asymptomatic bacteriuria in patients immediately before FC. The rates of bacteriuria at baseline and after FC are shown in Table 2. Most patients had negative cultures before and after FC, and there were a few who had bacteriuria before and after FC. The relative merits of the treatments are assessed in the discordant pairs (Table 1), where the discordant patient ratios differed among the treatment arms (P < 0.001).

Table 2.  Comparison of urine culture results before and after FC in the three arms of the study
GroupAfter randomization
NegativePositiveTotal
Placebo
Before randomization
 Negative61345658
 Positive  917 26
Bacteriuria
 Total62262684
Trimethoprim
Before randomization
 Negative66822690
 Positive 1111 22
Bacteriuria
 Total67933712
Ciprofloxacin
Before randomization
 Negative64013653
 Positive 28 6 34
Bacteriuria
 Total66819687

Patients on placebo were 5.0 times more likely to be infected after FC than before; trimethoprim reduced this factor to 2.0, whereas patients given ciprofloxacin were 0.46 times as likely to be infected after FC than before (Table 1). After adjusting for baseline rates of bacteriuria, the odds of bacteriuria on trimethoprim were 4.31 times greater than on ciprofloxacin (95% CI, 1.62–11.46).

The infecting organisms were Escherichia coli (50%), Enterococcus (26%), Klebsiella (9%), with Proteus and Pseudomonas accounting for 5% each. The organism was resistant to the antibiotic in nine of 22 patients (41%) given trimethoprim and in four of 28 (14%) given ciprofloxacin.

The symptoms associated with bacteriuria in the different groups are also shown in Table 1. Asymptomatic bacteriuria was found in 71% of patients overall, ranging from 67% in the ciprofloxacin group to 73% in the placebo group.

DISCUSSION

The present study is by far the largest randomized controlled trial examining the risks of significant bacteriuria after FC and the potential role of oral antibiotic prophylaxis. Routine FC carries a risk of significant bacteriuria of ≈8%[2], which might result in visits to the GP [9]. We confirmed that 3.9% (82 of the present 2083 patients) had significant bacteriuria before FC. Our study is important because we showed that one dose of oral drug is effective, which reduces costs and the risks of resistance. Previous studies used long-term programmes of treatment or parenteral drugs [5–7], or were too small to detect any benefit [10]. Both trimethoprim and ciprofloxacin reduced the risk of significant bacteriuria after FC, but the effect of ciprofloxacin was much more marked. There was evidence that both trimethoprim and ciprofloxacin eliminated bacteriuria when the organism was sensitive to the antibiotic (data not shown). The study was not powered to detect a reduction of severe clinically apparent infections, and there was no significant difference among the groups. Of those with significant bacteriuria, 67–73% were asymptomatic, 6–12% had dysuria alone, and the remainder had loin pain with or without fever. Five patients developed rigors or were admitted to hospital (two in the ciprofloxacin group; three in the placebo group, Table 1).

There are concerns about the possible development of bacterial resistance [11,12] with the increasing use of antibiotic prophylaxis, and the ‘clonal spread’ of organisms resistant to ciprofloxacin has been found in some areas. This is the first study to confirm that one dose of oral ciprofloxacin significantly reduces the risks of bacteriuria, and we think that the risks that the use of one dose of this drug will lead to an increase in the rate of emergence of resistant strains is low. We advocate the use of prophylactic ciprofloxacin taken as one oral tablet in patients undergoing FC under local anaesthetic, provided that this is in keeping with local susceptibility patterns.

ACKNOWLEDGEMENTS

The trial was funded by the NHS R&D programme (Northern and Yorkshire); the Newcastle upon Tyne Trustees.

CONFLICT OF INTEREST

None of the authors have a financial or other conflict of interest.

AUTHORS’ RESPONSIBILITIES

Responsible for medical input during the conduct of the study, MIJ, JM; Daily nursing support for the project, WAR, TL; Later help in the analysis, DM; Statistical advice and data analysis, JNSM; Microbiological advice and support to the trial, KEO; Responsibility for the trial and its conception and performance, DEN.

Ancillary