JAC Advance Access originally published online on September 23, 2006
Journal of Antimicrobial Chemotherapy 2006 58(5):1000-1008; doi:10.1093/jac/dkl368
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Clinical relevance of laboratory-reported antibiotic resistance in acute uncomplicated urinary tract infection in primary care
1 Health Protection Agency Primary Care Unit, Microbiology Department, Gloucestershire Royal Hospital Great Western Road, Gloucester GL1 3NN, UK 2 Microbiology Department, Norfolk and Norwich University Hospital NHS Trust Norwich, UK 3 Health Protection Agency Centre for Infections London, UK 4 Department of Primary Care, University of Southampton Southampton, UK 5 Gloucestershire Research & Development Support Unit Gloucester, UK 6 School of Medicine, Health Policy & Practice, University of East Anglia Norwich, UK 7 Department of General Practice, University of Aberdeen Aberdeen, UK 8 Minchinhampton Surgery Strand, UK
*Corresponding author. Tel: +44-8454-225061; Fax: +44-1452-526197; E-mail: cliodna.mcnulty{at}hpa.org.uk
Received 8 February 2006; returned 25 March 2006; revised 24 July 2006; accepted 15 August 2006
 |
Abstract |
|---|
 Top Abstract IntroductionPatients and methodsResultsDiscussionTransparency declarationsReferences |
|---|
Objectives: To determine whether patients with an uncomplicated community-acquired urinary tract infection (UTI) and an isolate resistant to trimethoprim had worse clinical outcomes following empirical treatment with trimethoprim 200 mg twice daily for 3 days than did those with a susceptible isolate.
Patients and methods: This was a prospective cohort study of clinical outcome. We enrolled 497 women (
18–70 years) presenting to general practitioner surgeries in Norwich and Gloucester with at least two symptoms of acute (<7 days) uncomplicated UTI. Significant bacteriuria was defined as 104 cfu/mL from a mid-stream urine (MSU).
Results: Of enrolled patients 75% (334/448) had significant bacteriuria, and trimethoprim resistance was present in 13.9% (44/317) of isolates. Patients with resistant isolates had a longer median time to symptom resolution (7 versus 4 days, P = 0.0002), greater reconsultation to the practice (39% versus 6% in first week, P < 0.0001), more subsequent antibiotics (36% versus 4% in first week, P < 0.0001) and higher rates of significant bacteriuria at 1 month (42% versus 20% with susceptible isolate, P = 0.04). Half of patients reconsulting in the first week had a resistant organism.
Conclusions: Patients with uncomplicated UTI caused by trimethoprim-resistant organisms had significantly worse clinical outcomes than those with trimethoprim-susceptible organisms. Nevertheless, trimethoprim resistance was rarer than predicted from routine laboratory submissions and we calculate that 23 women require microbiological investigation to prevent one reconsultation arising from resistance-based treatment failure. We therefore suggest empirical antibiotic treatment in acute, uncomplicated UTIs. If patients reconsult in the first week, we suggest a change of antibiotic treatment with urine culture and susceptibility testing then done. More generally, laboratory resources should concentrate on resistance surveillance to inform empirical antibiotic choice.
Keywords: UTI , clinical outcome , trimethoprim , community/primary care , cohort
|
Introduction |
|---|
|
TopAbstractIntroductionPatients and methodsResultsDiscussionTransparency declarationsReferences |
|---|
Urinary tract infection (UTI) is one of the most frequent conditions encountered by general practitioners (GPs).1–3 Its rational management requires appropriate, cost-effective use of medical resources. Although bacteriological confirmation of the diagnosis may be obtained before treatment is started, the symptoms are characteristic and are often distressing enough to warrant immediate empirical antibiotic therapy.4,5 The use of microbiology laboratories by GPs in England to support this diagnosis and treatment is extremely variable, ranging 10-fold from 29 to 266 urine samples/1000 patients/year.6 Microbiological testing may be unnecessary in acute uncomplicated UTI, except for surveillance purposes, if the clinical response is similar in patients with resistant and susceptible isolates as tested in vitro. This is important because, in most cases, urine culture and susceptibility testing cost more than the antibiotic treatment itself. Not routinely undertaking urine culture could save considerable health service resources, which might be better deployed examining resistance trends in a more structured manner, so as to support empirical treatment practices.
In many infections, laboratory-confirmed resistance correlates with clinical non-response, but there have been few studies examining clinical outcome in UTIs with resistant organisms, as patients with resistant organisms usually are excluded from therapeutic trials. Indeed, only one published study7 has examined the clinical effectiveness of the 3 day trimethoprim regimen now recommended as standard for uncomplicated lower UTI in the UK,8,9 and none has specifically examined clinical outcomes in infections with trimethoprim-resistant strains.5 More generally, we do know that individual-level exposure to trimethoprim increases the probability of trimethoprim-resistant UTIs in the community,10,11 and that trimethoprim retains similar overall clinical efficacy in UTI to other agents with lower resistance prevalence rates.12,13
We therefore aimed to determine whether patients with an uncomplicated, community-acquired, UTI and a urinary isolate resistant to trimethoprim had different clinical outcomes to those with a susceptible isolate, following empirical treatment with trimethoprim 200 mg twice daily for 3 days. Such a study would inform the review of clinical and laboratory use guidelines for the diagnosis and management of urinary symptoms.
|
Patients and methods |
|---|
|
TopAbstractIntroductionPatients and methodsResultsDiscussionTransparency declarationsReferences |
|---|
Design and location
This was a prospective cohort study of clinical outcome in female patients with acute, uncomplicated UTI treated empirically with trimethoprim. In an effort to maximize patient recruitment, general practices in West Gloucestershire with more that three partners were approached to participate. A total of 22 of 54 general practices from rural and urban areas agreed to enrol patients. Nine practices withdrew from the study after enrolling one to four patients each and one practice closed. Those general practices that declined to participate or withdrew cited lack of time. This left 12 out of 54 participating practices (22%). The list sizes of these 12 Gloucestershire practices ranged from 4201 to 16 140 patients (Table 1); they were a mix of rural, semi-rural, urban and city centre practices with a range of deprivation. In Norwich we purposely selected a university practice with a list size of 12 286. The total population served by all 13 practices was 108 392. Only two practices were members of a clinical research network.
|
Participants
Consecutive women (18–70 years) who presented to the GP surgeries in normal surgery hours from October 2000 to October 2003 were invited opportunistically to take part if they had at least two symptoms of acute (<7 days) uncomplicated UTI, for which the clinician intended to prescribe trimethoprim. All the practices approached routinely used trimethoprim as the first-line agent for the treatment of acute uncomplicated UTI. All clinicians stated that they were happy to use this agent empirically, as patients who were pregnant or had recurrent UTI were excluded. Symptoms could include blood in urine, burning or pain when passing urine, urgency, daytime frequency, night-time frequency, and pain in abdomen when not passing urine. At enrolment, patients were also asked about other symptoms, smelly urine, feeling unwell and leaking urine, but these were not inclusion criteria (Table 3). Over the first 18 months of the study, clinicians used positive nitrite urine dipstick tests to increase the likelihood of bacteriologically confirmed UTI (Nephur-test and Leuco, Philip Harris, Ashby de la Zouch, UK).14 However, the many surgeries that did not routinely undertake urine dipstick tests found this step time-consuming; therefore, the protocol was changed so that all patients with two urinary symptoms were invited to participate. Analysis showed that patients recruited under the two protocols were not significantly different in terms of age, outcome or prevalence of clinical symptoms (see Results). We asked practices to keep a record of those eligible patients whom they did not ask to participate.
Patients were excluded if they were pregnant, had had more than four episodes of UTI in the past year, antimicrobials in the past 2 weeks, symptoms suggesting upper UTI (loin pain, fever >39°C, rigors or vomiting), functional or anatomical abnormality of the urinary tract, immunosuppression or hypersensitivity to trimethoprim. Patients were also asked whether they had smelly urine or were feeling unwell, but these were not inclusion criteria. All participants were asked to provide a mid-stream urine (MSU) specimen; they were given a collection pack containing pictorial and written instructions, a soap wipe for cleaning the genital area, a funnel and a sterile universal container without preservative (3S Healthcare, London). MSUs were sent to the local laboratory by normal transport mechanisms on the day of collection or were refrigerated overnight.
Microbiological investigation
Significant UTI was defined according to modified European guidelines15 except that, following UTI UK National Standard Operating Procedures,16 we used a cut-off 104 cfu/mL as indicating a significant count for all organisms, whereas the European guidelines have a lower cut-off of 103 cfu/mL specifically for Escherichia coli and Staphylococcus saprophyticus. Samples were processed in the routine laboratory according to National Standard Operating Procedures,16 and a locally generated report and susceptibility result was sent to the clinician. Trimethoprim susceptibility was determined by the British Society for Antimicrobial Chemotherapy (BSAC) disc diffusion or breakpoint methods, with trimethoprim resistance defined as per BSAC guidelines for urinary infections as an MIC >2 mg/L, or corresponding zone diameter. Susceptibility results were confirmed at the HPA Antibiotic Resistance Monitoring and Reference Laboratory (ARMRL), Colindale, where MICs were determined by the BSAC agar dilution method.17 Local identification was to the ‘coliform’ level, based on growth and morphology on MacConkey agar. At the reference laboratory, Enterobacteriaceae were plated on UTI chromogenic agar (Oxoid, Basingstoke, UK), and those giving pink colonies were accepted as E. coli whereas those not giving pink colonies were further identified using API20E kits (bioMerieux, Basingstoke, UK). As patient recruitment continued over the 3 years from 2000 to 2003, we examined routine laboratory urine isolate antibiotic-susceptibility data for Norwich and Gloucester to determine whether there had been any changes in resistance prevalence over this time period.
Treatment and follow-up
At the surgery patients were given trimethoprim 200 mg to be taken twice daily for 3 days; the study team provided this free of charge. In line with normal clinical practice, patients were advised to contact the surgery if their symptoms worsened or if they were unhappy with their progress. GPs were asked not to routinely contact patients who had a resistant organism reported. If patients did contact the surgery, the local laboratory report was available to the GP, as is usual clinical practice. Patients were given an information sheet explaining the study. The sheet also advised that drinking plenty of fluids usually helps to ease the burning pain when passing urine. Patients were asked to complete a symptom diary for up to 10 days, at least until they had been symptom-free for 2 days; this diary was developed and validated in a Health Technology Assessment-funded UTI study.18,19 Each symptom was scored from 0 (normal), 1 (very little problem), 2 (slight problem), 3 (moderately bad), 4 (bad), 5 (very bad) to 6 (as bad as it could be). A similar approach to recording severity of symptoms has been used for other acute infection studies18 and was sensitive to symptomatic change. Symptom diaries were returned in stamped, addressed envelopes. Symptom resolution was defined as a symptom score of 0 (normal) or 1 (very little problem) for each individual symptom. To determine whether persistent bacteriuria was more common in patients with a UTI due to a resistant organism, all enrolled patients were posted a 4 week follow-up MSU kit and symptom questionnaire. The MSU kit was the same as that used at enrolment and, as before, the sterile universal container did not contain preservative. Patients were asked to complete the request form, specimen pot details and questionnaire and return all three, in the plastic wallet provided, to their GP's surgery. These were then forwarded to the local laboratory by the normal transport system. Data analysis showed that delivery to the laboratories was not significantly slower than for enrolment urine samples. Samples taken outside 21–42 days post-enrolment were excluded from the analysis.
A researcher telephoned each patient 3 days after initiation of treatment, to check on compliance with the diary completion, and at 28 days, to remind them to return the follow-up questionnaire and urine sample. Details of patients who reconsulted were retained within the dataset for analysis.
Outcome measures
Primary outcome measures were the number of patients with symptom resolution at 5 and 7 days, the number of days to complete symptom resolution. Secondary outcome measures were reconsultation with urinary symptoms and the presence of bacteriuria at 4 weeks. GP records for all enrolled patients were retrospectively examined for consultations with urinary symptoms in the 1 month following initial consultation. The researchers were blinded to antibiotic-susceptibility results. Antibiotics prescribed for urinary symptoms were noted.
Sample size calculation
It was estimated at initiation that 
20% of isolates would be resistant to trimethoprim2 and that 90% of patients with a susceptible isolate would be symptom-free at 7 days. GPs were only interested in sending an MSU for microbiological analysis if there was a 15% poorer response rate in patients with resistant isolates. To detect such a 15% difference (at 5% significance level and 80% power) in symptom resolution we predicted that 320 positive samples would be required, equating to 64 resistant and 256 susceptible isolates. In other recent outcome studies, 80% of patients returned the questionnaire if telephoned by the research assistant.20 Taking this figure, and assuming that 80% of patients with symptomatic UTI and a positive nitrite test would yield a positive culture, we calculated that we needed to enrol 500 patients.
Ethics
Ethical approval was obtained via the then PHLS (2000/10), Norwich (2000/104) and Gloucestershire (99/95G) Local Research Ethics Committees. Written informed consent was obtained from patients after a verbal explanation by a clinician.
Analysis of data
This was a prospective observational study, comparing outcomes in terms of symptom resolution in those women who had a trimethoprim-resistant pathogen with those where the pathogen was trimethoprim-susceptible. The 95% confidence interval (CI) around the estimated proportion of trimethoprim-resistant UTIs was calculated using an exact binomial method. Factors associated with culture positivity and trimethoprim resistance were assessed using chi-squared tests and multivariable logistic regression analysis. Those subjects who did not have the particular symptom under analysis during the study period were excluded from that symptom resolution analysis. In those patients where the symptom either relapsed or who developed the symptom after recruitment, the symptom was considered to have resolved only when it did not return. Time to symptom resolution was defined as when the self-reported symptom severity scores fell below 2, as measured on a 0–6 scale. For the ‘All symptoms’ resolution, each individual symptom score needed to fall below 2. Patients who received a second antibiotic were still included in time to resolution. Kaplan–Meier plots of the time to resolution and a log-rank test were used to compare those with resistant and susceptible pathogens. A proportional hazards model21 was used to assess whether the effect of resistance on resolution could be explained by potential confounding factors, which included the testing laboratory (Gloucester or Norwich), age, enrolled by nitrite result or symptoms, duration of symptoms prior to consultation, organism and number of previous UTIs. Patients with resistant isolates who reconsulted were considered separately in this model.
|
Results |
|---|
|
TopAbstractIntroductionPatients and methodsResultsDiscussionTransparency declarationsReferences |
|---|
Study population
The 12 Gloucester practices (list sizes 4201–16 140) enrolled between 9 and 86 patients while the Norwich university practice (list size 12 286) enrolled 88 patients. In total, 497 women agreed to participate. Twenty-six of these were excluded as they did not meet inclusion criteria (e.g. outside the age range, or having recent receipt of antibiotics) and one withdrew after enrolment. The median, lower and upper quartiles of age of the recruited subjects were 39, 24 and 53 years, respectively. A total of 451 initial MSU samples and laboratory forms and 426 enrolment forms were received; 313 subjects returned 10 day symptom diary forms and 232 submitted a second ‘28 day’ urine sample (Figure 1). Reconsultation data were obtained for all enrolled patients. Patients who returned 10 day diaries were significantly older. However, multivariable analysis showed that there was no difference in trimethoprim resistance or reconsultation rates in patients who did or did not return their diaries. None of the participating GPs collected data on consultations of those who may have been eligible who were not asked to participate.
|
The median duration of most symptoms at the time of consultation was 3 days. This applied for burning (present in 83%), urgency (80%), daytime frequency (89%), night-time frequency (67%), abdominal pain (67%), feeling unwell (58%) and leaking urine (21%). In contrast, the median duration of the symptom of blood in the urine (37%) was only 1 day. Patients with blood in urine presented earlier, and the duration of other symptoms was shorter than those without blood in urine (1–2 days versus 3 days).
Microbiological investigation
The local laboratories received 451 MSU samples (Table 2). Three were excluded, as the sample and the patient details did not match, and 114 samples yielded non-significant cultures. The remaining 334 (75%) indicated significant bacteriuria, with 104 cfu/mL; 317 yielded a pure bacterial culture, of which 298 (89%) were identified in the local laboratory as ‘coliforms’, 13 (4%) were S. saprophyticus and 6 yielded other bacteria. A microscopy report was available in 323 patients with significant bacteriuria, 313 (97%) had significant pyuria (104 wbc/mL). Of 130 patients with a positive nitrite test and two urinary symptoms, enrolled in the first 18 months, 108 (83%) had significant bacteriuria compared with 226/318 (71%) of those enrolled with two urinary symptoms and no nitrite test (P = 0.004); however, the bacteriuric patients enrolled in the two phases of the study (with and without the nitrite test) appeared identical in terms of symptom severity and response (not shown).
|
Trimethoprim resistance was present in 44 of 317 isolates from pure cultures (13.9%, 95% CI 10.3–18.2%), and multivariable logistic regression analysis was undertaken to explore those patient characteristics predictive of a resistant isolate. We found a weak association (P = 0.09) between the number of UTIs reported in the previous year and resistance, with an odds ratio of 9.5 (95% CI, 1.5–60.1) for a resistant isolate among those with four UTIs in the past year, compared with those with none. There was no evidence of an increased likelihood of a resistant pathogen from those with one to three previous UTIs in the preceding year compared with those with no previous UTIs in the same period. There was no association between resistance and the severity of any symptom or combination of symptoms recorded at enrolment, nor with age (data not shown). There was no significant difference in trimethoprim resistance rates among practices (Table 1; P = 0.2).
Of the 317 single isolates 256 (81%) were sent to the ARMRL, which confirmed that the local laboratory identification was correct for all isolates, though these included 242 cases where the initial identification was ‘coliform’, without the species indicated. Ten of the 256 (3.9%) local antibiotic-susceptibility results were different to those found by the ARMRL. Based on the ARMRL identifications E. coli accounted for 219/256 (86%) of all the isolates examined and 219/242 (90%) of all ‘coliform’ isolates. Of the 31 trimethoprim-resistant isolates confirmed by ARMRL, 26 (84%) were E. coli and 29/31 (94%) had high-level resistance, with MICs >64 mg/L, whereas two had low-level resistance, with MICs of 4 and 8 mg/L. Local laboratory antibiotic-susceptibility data for all routinely submitted GP urine isolates during the 3 years of study recruitment did not show any changes in trimethoprim resistance (Gloucester 27% and Norwich 24.5% throughout the study period). HPA bacteraemia surveillance data do suggest a recent uptrend in resistance rates for E. coli, which were between 27.3 and 28.4% throughout the years 1996 to 2001, then 31% in 2002, 32% in 2003, 35% in 2004 and 38% in 2005.
Symptom resolution in relation to laboratory-defined resistance
The median time to resolution of all symptoms was shorter for those patients with a susceptible isolate than for those with an isolate found resistant in local laboratory testing (4 versus 7 days, P = 0.0002) (Table 3). Likewise, the time to resolution of each individual symptom was shorter in those with susceptible organisms except for ‘smelly urine’, where there was only weak evidence of a difference.
|
Resolution of all symptoms at 5 days and 7 days occurred in 123/178 (69%) and 147/178 (83%) of assessable patients with a susceptible isolate compared with 7/29 (24%, P
0.0001) and 15/29 (52%, P 0.0001) with a resistant isolate (Table 4). The time to full symptom resolution shown in a Kaplan–Meier plot (Figure 2) was significantly shorter in those patients with a susceptible organism compared with those with a resistant organism (P = 0.0005). This degree of significance remained even when those with resistant organisms were subdivided into those who did reconsult or did not. The hazard rate ratio in patients with resistant isolates who reconsulted at the surgery was 0.40 (95% CI: 0.19–0.82) and 0.54 (95% CI: 0.31–0.95) in those with resistant isolates who did not reconsult compared with those patients with susceptible isolates (Figure 2). Outcome analyses using the ARMRL susceptibility data led to similar conclusions to those with the local laboratory data (data available on request). A proportional hazards model showed that the potential confounding factors of centre, age, enrolment by nitrite test, duration of symptoms prior to consultation, organism and number of previous UTIs did not affect the estimated hazard ratio for symptom resolution in patients with or without a trimethoprim-resistant isolate, which was calculated as 0.49 (95% CI: 0.30–0.76) prior to adjusting for covariates and 0.47 (95% CI: 0.30–0.76) after adjustment (Table 5).
|
|
|
A total of 132 patients with an antibiotic-susceptibility isolate at enrolment provided a follow-up urine sample within 21–42 days. A positive follow-up culture was found in a higher proportion of patients with an initial trimethoprim-resistant organism than those with an initial trimethoprim-susceptible organism (42%, 8/19, compared with 20%, 23/113, P = 0.04). Nevertheless, 58% (11/19) of the patients who originally had a resistant organism were culture-negative on follow-up. Seven (64%) of these had received a further course of antibiotics to which the original isolate was susceptible, compared with only one of eight (12.5%) with a positive culture at follow-up.
Reconsultations with urinary symptoms
We obtained reconsultation data for all 317 patients with antibiotic-susceptibility results at enrolment. Patients with resistant UTIs were significantly more likely to reconsult with urinary symptoms in the month following enrolment (20/44, 45%) than those with UTIs due to trimethoprim-susceptible isolates (41/273 15%, P < 0.0001). Patients with resistant isolates were also more likely to reconsult earlier. Thus, reconsultation in the first week occurred in 17/44 (39%) of patients with resistant isolates versus 17/273 (6%) of patients with susceptible isolates (P < 0.0001). Half (17/34) of all the patients who reconsulted in the first week had resistant isolates. Patients with resistant isolates were significantly more likely to be prescribed a further course of antibiotics for urinary symptoms in the first week and month following enrolment [16/44 (36%) in the first week and 19/44 (43%) in the first month] than those with UTIs due to susceptible organisms [11/273 (4%) in the first week, P < 0.0001; 28/273 (10%) in the first month P < 0.0001].
|
Discussion |
|---|
|
TopAbstractIntroductionPatients and methodsResultsDiscussionTransparency declarationsReferences |
|---|
This is the first study designed specifically to examine the clinical significance of antibiotic resistance in acute uncomplicated UTI. Patients with a UTI due to a trimethoprim-resistant organism had worse clinical outcomes than those with a susceptible isolate. Although 24% of patients with a resistant isolate were asymptomatic at 5 days, and half by 7 days, the general pattern was that symptoms took longer to resolve (7 versus 4 days) and that patients were six times more likely to reconsult in the first week after presentation. Persistent bacteriuria at 1 month was also twice as likely in patients with resistant isolates even though, by then, many of the patients with resistant isolates had received additional antibiotics. In total 64% (7/11) of patients with a resistant isolate and negative follow-up culture had received a second course of antibiotics to which the isolate was susceptible, compared with only one of eight with a positive culture at follow-up.
The study included a broad range of rural, semi-rural, urban and city practices, and a large university practice. These had a range of deprivation scores (Table 1). Despite differences in practice recruitment rates there was no significant difference in trimethoprim resistance rates between practices. Only two practices were members of a GP Research Network. The results therefore should be generalizable to other practices with similar resistance rates. The proportions of different bacterial species that we isolated were very similar to those described in other studies of acute uncomplicated UTI in the USA and Europe,5,7 with 86% being E. coli and 4% S. saprophyticus. Seventy per cent of the women with at least two urinary symptoms had significant bacteriuria. This is higher than studies that include patients with only one symptom.22 However, it is known that the likelihood of infection being present is increased by the detection of specific combinations of symptoms (e.g. dysuria and frequency without vaginal discharge)22 or by using the nitrite test. We used modified European criteria for the definition of UTI, with a lower threshold (104 cfu/mL) than is commonly applied in the UK (usually 105 cfu/mL). This lower cut-off is supported by the recent results of Richards et al.,23 who found that patients with urinary symptoms but with fewer than 105 cfu/mL urine (as indicated by a negative urine dipstick) still benefited from trimethoprim treatment. The European Council of Legal Medicine European Urinalysis Group advises a still lower threshold of 103 cfu/mL specifically for E. coli,15 but this cannot be applied if identification is only to the ‘coliform’ level.
Although we explored the most likely confounders, it is possible that the difference in clinical outcomes between patients with resistant and susceptible isolates was due to some other factor that we failed to identify. Those who returned their 10 day diaries were not more likely to reconsult or have trimethoprim-resistant isolates. Older patients were more likely to return their diaries. However as age is not associated with either outcome, reconsultation or trimethoprim status it cannot confound any association, therefore there is no obvious reason why the findings would not generalize to those who did not return their symptom diary. We did not collect information on co-morbidities such as diabetes and respiratory disease, which may have affected outcome. However, these factors were unlikely to have had any profound effect, and if they were important it is likely that age would be expected to have a greater influence. We have no way of confirming whether patients with resistant isolates were not telephoned by practice staff when the culture result became available, so increasing reconsultations in this group. We do not, however, think this is likely (i) as the reports had an automated comment saying that the patient was in a UTI study and that practice staff should not telephone them with the susceptibility result and (ii) because the notes revealed no documented evidence that any patient had been contacted on this basis.
Largely because recruitment was slower than anticipated, the study extended over the 3 years from October 2000 to October 2003 and another potential confounder is that the agents of UTI may have changed over this period. We know that E. coli shows growing multiresistance.24,25 Nevertheless, local laboratory antibiotic-susceptibility data for all GP urine isolates during the 3 years of study recruitment did not show any changes in trimethoprim resistance, at least up to 2003. Moreover, while there is concern about the accumulation of ESBL-producing and ciprofloxacin-resistant E. coli in the community,25 such isolates are mostly found in complicated cases, with recent hospitalization and/or antibiotics, not (yet) in acute uncomplicated UTI. Patients with such additional risk factors should be managed differently from those with uncomplicated infection, both with respect to microbiological investigation and therapy. In the same context, the present study shows that patients with four UTIs in the past year were more likely to have a trimethoprim-resistant isolate, whereas those with three or fewer were not. We did not collect information on antibiotic use for other infections. Most ESBL-producing E. coli remain susceptible to nitrofurantoin, another narrow-spectrum first-line agent appropriate for lower UTI.26
The lower-than-expected recruitment rate may have been due to a genuinely lower-than-anticipated presentation rate or to failure to recruit potential subjects, either through lack of consent or through practice staff neglecting to offer study participation. Similarly, another study in south-west England exploring strategies for antibiotic resistance surveillance found that the submission of urine specimens from ‘consecutive’ patients with acute urinary symptoms was much slower than expected from routine laboratory urine submissions.6,27 Discussion with GPs and staff in participating practices revealed that practices found the opportunistic recruitment challenging and staff admitted that, at busy times, they failed to recruit due to time constraints. This, unfortunately, is inherent in ‘real world’ primary care studies, particularly those requiring opportunistic recruitment. There was no suggestion of systematic bias in recruitment and the confounders that may have affected recruitment (age, centre, prior duration of symptoms and number of previous UTIs) had no significant effect on outcome (Table 5).
We found that patients with resistant isolates had similar clinical outcomes to those found elsewhere for patients who received placebo in acute uncomplicated UTI studies; Christiaens et al.28 recorded symptomatic cure in 20% and 42% of such patients at 3 and 7 days, respectively. Treatment failure versus resistant isolates is not surprising, because the trimethoprim concentrations achieved in the urine (75 mg/L)5 are far below the MICs (typically >512 mg/L) for most isolates with resistance. Two other clinical studies, by Masterton et al.29 and McCarty et al.,30 showed, in a secondary analysis, that trimethoprim/sulfamethoxazole resistance was still associated with bacterial eradication in 50% of cases and clinical cure in 60%. These rates are greater than those observed at 5 days in our study, but similar to the 7 day resolution. The sample sizes of the studies by Masterton and McCarty were smaller, with fewer than 10 patients, so the CIs would be very wide and would overlap our symptom resolution rate of 24% in infections due to resistant pathogens. Furthermore, it is unclear at what time symptom resolution was measured in these two studies.
Before the study, the participating GPs had indicated that a 15% lower clinical response rate in patients with a resistant UTI would prompt them routinely to send a urine sample to determine antibiotic susceptibility. At first glance it would appear that these conditions were met, as clinical resolution at day 5 was present in almost 70% of patients with a susceptible isolate versus only 24% with a resistant isolate. However, only 4% (20/448) of all the women with an MSU result reconsulted owing to persistent symptoms that related to a resistant isolate. Moreover, trimethoprim resistance was much less frequent than we had expected from local routine laboratory antibiotic-susceptibility data, with resistant isolates from only 13.9% of patients, compared with 24–27% of urinary isolates routinely submitted to the Gloucester and Norwich laboratories. The reasons for this discrepancy cannot be addressed here, but it is likely that GPs send samples to the laboratory more frequently in complicated cases, such as women with repeated UTIs, or those not responding to empirical antibiotic treatment. Since resistance is more likely in patients who have received recent antibiotics,31 it is predictable that the resistance rates based upon routine laboratory susceptibility data will overestimate the true resistance pattern in uncomplicated UTI. It is possible that some GPs only enrolled patients with milder symptoms. However, none of the participating GPs over the 3 years of the study reported doing this and there was no relationship between trimethoprim resistance and severity of symptoms at enrolment. We estimate, based on these figures, that it is necessary to perform urine cultures in 23 symptomatic women to predict one in whom resistance would cause clinical failure leading to reconsultation. Such figures scarcely support routine, and expensive, microbiological investigation of all such patients Rather, we suggest empirical antibiotic treatment with a narrow-spectrum urinary infection agent, e.g. trimethoprim, or as locally recommended,32 in uncomplicated patients with two or more urinary symptoms, as also proposed by Hummers-Pradier and Kocken.4 We suggest that urine culture and susceptibility testing should be reserved for complicated cases, treatment failures, and those with particular risk factors for resistant isolates, e.g. repeated infections and recent hospitalization or antibiotics.31 Patients should be advised to reconsult if their symptoms are not improving by day 4. As 50% of patients who reconsult in the first week have a resistant isolate, an alternative antibiotic treatment should then be prescribed (empirically, unless a urine sample had been sent and the results are available) and a urine specimen submitted for culture and susceptibility testing. Although nitrite tests might also be performed and increases specificity, there is evidence that some patients with a negative nitrite test benefit from antibiotic treatment.33 More generally, laboratory effort in this area should be shifted towards structured surveillance to support empirical therapy, confirming whether or not trimethoprim should remain the preferred first-line agent.
|
Transparency declarations |
|---|
|
TopAbstractIntroductionPatients and methodsResultsDiscussionTransparency declarationsReferences |
|---|
None of the authors has any specific declarations that would influence their opinions in this article. All (except E. F.) work in the field of antibiotic resistance and could be considered to have vested interests in investment in this area, whether by governments, charities or industry.
|
Acknowledgements |
|---|
We are grateful to all the staff in the participating general practices in Gloucestershire and Norwich; without their ongoing support the study would not have been possible (Barnwood Road); Cheltenham Road; Churchdown; Coleford Health Centre; Dursley (Acorn and Walnut surgeries); Frampton on Severn; Forest Health Care; Frithwood; Heathville Road; High Street Medical Centre; Hucclecote; Kingsholm; Minchinhampton; Painswick; Pavillion; Prices Mill; Regent Street; St Michaels; Saintbridge; Severnbank; 102 Stroud Road (now closed); University of East Anglia Health Centre. Thanks are due also to the Research & Development Support Unit in Gloucester, especially Sue Starck, and to the research nurses from the Health Protection Agency Vaccine Evaluation Unit. Finally, many thanks to the BSAC for continuing the funding for the study despite recruitment taking much longer than anticipated.
|
References |
|---|
|
TopAbstractIntroductionPatients and methodsResultsDiscussionTransparency declarationsReferences |
|---|
1 Fry J. (1969) Medicine in Three Societies(Medical and Technical Publishing Co. Ltd, Lancaster).
2 Royal College of General Practitioners Office of Population Consensus & Survey. (1995) Department of Health Morbidity Statistics from General Practice, 1991–1992(HMSO, London).
3
Car J. (2006) Urinary tract infections in women: diagnosis and management in primary care. BMJ 332:94–7.
4 Hummers-Pradier E and Kocken MM. (2002) Urinary tract infections in adult general practice patients. Br J Gen Pract 52:752–61.[ISI][Medline]
5
Gupta K, Hooton TM, Stamm WE. (2001) Increasing antimicrobial resistance and the management of uncomplicated community-acquired urinary tract infections. Ann Intern Med 135:41–50.
6 McNulty CAM, Bowen J, Clark G, et al. (2004) How should general practitioners investigate suspected urinary tract infection? Variations in laboratory-confirmed bacteriuria in South West England. Commun Dis Public Health 7:220–6.[Medline]
7 Gossius G and Vorland L. (1985) The treatment of acute dysuria-frequency syndrome in adult women: double blind randomised comparison of three day versus ten day trimethoprim therapy. Curr Ther Res Clin Exp 37:34–42.
8
Tan TY, McNulty C, Charlett A, et al. (2003) Laboratory antibiotic susceptibility reporting and antibiotic prescribing in general practice. J Antimicrob Chemother 51:379–84.
9 PRODIGY Guidance. Urinary Tract Infection (lower)—Women http://www.prodigy.nhs.uk/guidance.asp?gt=Urinarytractinfection(lower)-women (1 June 2006, date last accessed).
10
Donnan PT, Wei L, Steinke DT, et al. (2004) Presence of bacteriuria caused by trimethoprim resistant bacteria in patients prescribed antibiotics: multilevel model with practice and individual patient data. BMJ 328:1297.
11
Magee JT, Pritchard EL, Fitsgerald KA, et al. (1999) Antibiotic prescribing and antibiotic resistance in community practice: retrospective study 1996–98. BMJ 319:1239–40.
12 Spencer RC, Moseley DJ, Greensmith MJ. (1994) Nitrofurantoin modified release versus trimethoprim or co-trimoxazole in the treatment of uncomplicated urinary tract infection in general practice. J Antimicrob Chemother 33:Suppl A, 121–9.[Medline]
13 Minassian MA, Lewis DA, Chattopadhyay D, et al. (1998) A comparison between single-dose fosfomycin trometamol (Monuril) and a 5-day course of trimethoprim in the treatment of uncomplicated lower urinary tract infection in women. Int J Antimicrob Agents 10:39–47.[CrossRef][ISI][Medline]
14 Hiscoke C, Yoxall H, Greig D, et al. (1990) Validation of a method for the rapid diagnosis of urinary tract infection suitable for use in general practice. Br J Gen Pract 40:403–5.[ISI][Medline]
15 European Confederation of Laboratory Medicine. (2000) European Urinalysis Guidelines. In Kouri T, Fogazzi G, Gant V (Eds.), et al. Scand J Clin Lab Invest 60:Suppl 231, pp. S1–96.
16 PHLS Standard Operating Procedure: Investigation of urine. Ref. No. BSOP41 Issue No. 1 02.04.1998. Technical Services, PHLS HQ, London. HPA Issue No. 5 28.02.05. http://www.hpa-standardmethods.org.uk/documents/bsop/pdf/bsop41.pdf (1 June 2006, date last accessed).
17 Andrews JM. (2001) Determination of minimum inhibitory concentrations. BSAC guide to antimicrobial susceptibility testing. J Antimicrob Chemother 48:Suppl 1, 5–16.[Abstract]
18
Watson L, Little P, Williamson I, et al. (2001) Validation study of a diary for use in acute lower respiratory tract infection. Fam Pract 18:553–4.
19 Development and Randomised Controlled Trial of Dipsticks and Diagnostic Algorithms for the Management of UTI Little PS. NHS Research & Development Health Technology Assessment Programme 2003. International Standard Randomised Controlled Trial Number 03525333. http://www.controlled-trials.com/isrctn/trial/%7C/0/03525333.html (1 June 2006, date last accessed).
20
Little P, Williamson I, Warner G, et al. (1997) Open randomised trial of prescribing strategies for sore throat. Br Med J 314:722–7.
21 Cox DR. (1972) Regression models and life tables (with discussion). J R Stat Soc [Ser B] 34:187–200.
22
Bent S, Nallamouthu BK, Simel DL, et al. (2002) Does this woman have an acute uncomplicated urinary tract infection. JAMA 287:2701–10.
23
Richards D, Toop L, Chambers S, et al. (2005) Response to antibiotics of women with symptoms of urinary tract infection but negative dipstick urine test results: double blind randomised controlled trial. BMJ 331:143–6.
24 Livermore DM. (2005) Minimising antibiotic resistance. Lancet Infect Dis 5:450–9.[CrossRef][ISI][Medline]
25
Calbo E, Romani V, Xercavins M, et al. (2006) Risk factors for community-onset urinary tract infections due to Escherchia coli harbouring extended-spectrum ß-lactamases. J Antimicrob Chemother 57:780–3.
26 Livermore D and Woodford N. (2004) Laboratory detection of bacteria with extended-spectrum beta-lactamases. CDR Weekly 14:27.
27 Lewis D, McNulty CAM, Lovering A, et al. (2003) Antimicrobial resistance surveillance in primary care: practicalities of different methods. Health Protection Agency Inaugural Conference, University of Warwick(London, Health Protection Agency) pp. p208–9 Abstract P109.
28 Christiaens TCM, Meyere M De, Vershcraegen G, et al. (2002) Randomised controlled trial of nitrofurantoin versus placebo in the treatment of uncomplicated urinary tract infection in adult women. Br J Gen Pract 52:729–34.[ISI][Medline]
29
Masterton RG and Bochsler JA. (1995) High-dosage co-amoxiclav in a single dose versus 7 days of co-trimoxazole as treatment of uncomplicated lower urinary tract infection in women. J Antimicrob Chemother 35:129–73.
30 McCarty JA, Richard G, Huck W, et al. (1999) A randomised trial of short-course ciprofloxacin, ofloxacin, or trimethoprim/suflamethoxazole for the treatment of acute urinary tract infection in women. Am J Med 106:292–9.[CrossRef][ISI][Medline]
31
Hay AD, Thomas M, Montgomery A, et al. (2005) The relationship between primary care antibiotic prescribing and bacterial resistance in adults in the community: a controlled observational study using individual patient data. J Antimicrob Chemother 56:146–53.
32 Health Protection Agency. Management of Infection Guidance for Primary Care. http://193.129.245.226/infections/topics_az/primary_care_guidance/Antibiotic_guide_250506.rtf (1 June 2006, date last accessed).
33
Woodford N, Ward ME, Kaufmann ME, et al. (2004) Community and hospital spread of Escherichia coli producing CTX-M extended-spectrum beta-lactamases in the UK. J Antimicrob Chemother 54:735–43.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  A. P. MacGowan and on behalf of the BSAC Working Parties on Resistanc Clinical implications of antimicrobial resistance for therapy J. Antimicrob. Chemother., November 1, 2008; 62(suppl_2): ii105 - ii114. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Hajibagheri, M. Priesemann, I. Morrison, A. Parikh, N. Lessof, and E. Price NICE guidance on urinary tract infection in children abandons routine antibiotic prophylaxis Arch. Dis. Child., April 1, 2008; 93(4): 356 - 356. [Full Text] [PDF]
|
|
|
|
 |
|
 K. Gupta, T. M. Hooton, P. L. Roberts, and W. E. Stamm Short-Course Nitrofurantoin for the Treatment of Acute Uncomplicated Cystitis in Women Arch Intern Med, November 12, 2007; 167(20): 2207 - 2212. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. McNulty, M. Thomas, R. John, A. Lovering, D. Lewis, and A. MacGowan Problems of basing patient recruitment for primary care studies on routine laboratory submissions J. Clin. Pathol., November 1, 2007; 60(11): 1290 - 1293. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. C. Hayward, K. Goldsmith, A. M. Johnson, and on behalf of the Surveillance Subgroup of SACAR Report of the Specialist Advisory Committee on Antimicrobial Resistance (SACAR) Surveillance Subgroup J. Antimicrob. Chemother., August 1, 2007; 60(suppl_1): i33 - i42. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. M. Fleming The state of play in the battle against antimicrobial resistance: a general practitioner perspective J. Antimicrob. Chemother., August 1, 2007; 60(suppl_1): i49 - i52. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Mangin and L. Toop Urinary tract infection in primary care BMJ, March 24, 2007; 334(7594): 597 - 598. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||