Alternative to prophylactic antibiotics for the treatment of recurrent urinary tract infections in women: multicentre, open label, randomised, non-inferiority trial - The BMJ

Abstract

Objective To test and compare the efficacy of methenamine hippurate for prevention of recurrent urinary tract infections with the current standard prophylaxis of daily low dose antibiotics.

Design Multicentre, open label, randomised, non-inferiority trial.

Setting Eight centres in the UK, recruiting from June 2016 to June 2018.

Participants Women aged ≥18 years with recurrent urinary tract infections, requiring prophylactic treatment.

Interventions Random assignment (1:1, using permuted blocks of variable length via a web based system) to receive antibiotic prophylaxis or methenamine hippurate for 12 months. Treatment allocation was not masked and crossover between arms was allowed.

Main outcome measure Absolute difference in incidence of symptomatic, antibiotic treated, urinary tract infections during treatment. A patient and public involvement group predefined the non-inferiority margin as one episode of urinary tract infection per person year. Analyses performed in a modified intention-to-treat population comprised all participants observed for at least six months.

Results Participants were randomly assigned to antibiotic prophylaxis (n=120) or methenamine hippurate (n=120). The modified intention-to-treat analysis comprised 205 (85%) participants (antibiotics, n=102 (85%); methenamine hippurate, n=103 (86%)). Incidence of antibiotic treated urinary tract infections during the 12 month treatment period was 0.89 episodes per person year (95% confidence interval 0.65 to 1.12) in the antibiotics group and 1.38 (1.05 to 1.72) in the methenamine hippurate group, with an absolute difference of 0.49 (90% confidence interval 0.15 to 0.84) confirming non-inferiority. Adverse reactions were reported by 34/142 (24%) in the antibiotic group and 35/127 (28%) in the methenamine group and most reactions were mild.

Conclusion Non-antibiotic prophylactic treatment with methenamine hippurate might be appropriate for women with a history of recurrent episodes of urinary tract infections, informed by patient preferences and antibiotic stewardship initiatives, given the demonstration of non-inferiority to daily antibiotic prophylaxis seen in this trial.

Introduction

Recurrent urinary tract infection (UTI) is defined as repeated UTI with a frequency of at least two episodes in the preceding six months or three episodes in the past year.12 Acute UTI is most often uncomplicated cystitis and occurs in 50-80% of women in the general population.3 About one in four women with one UTI episode will go on to develop frequent recurrences,4 representing a substantial global healthcare problem. An economic analysis from the US described UTI as accounting for more than 6.8 million consultations, 1.3 million emergency department visits, and 245 000 hospital admissions with an annual cost of more than US$2.4bn (£1.8bn; €2.1bn).5 National and international guidelines acknowledge the need for preventive strategies, and those from the UK, Europe, and US strongly recommend the use of daily, low dose antibiotics as the standard prophylactic treatment for recurrent UTI.126

The urgent need for demonstration of effective non-antibiotic treatments is underlined by the UK antimicrobial resistance strategy, which comments that "we are heading rapidly towards a world in which our antibiotics no longer work" and recommends a "strong focus on infection prevention."7 One aim of this strategy is to reduce antimicrobial use in people by 15% before 2024; to achieve that, exploration of non-antibiotic preventive treatments in common conditions such as UTI is essential. Methenamine hippurate is one such non-antibiotic treatment, which is hydrolysed to formaldehyde in acidic environments such as the distal tubules of the kidney. Formaldehyde is bacteriocidal and works by denaturing bacterial proteins and nucleic acids.

Methenamine hippurate has been evaluated in previous Cochrane systematic reviews,8 which concluded that "methenamine hippurate may be effective for preventing UTI" but recognised the "need for further large well-conducted RCTs [randomised controlled trials] to clarify." This study aimed to determine whether methenamine hippurate was an effective alternative to the standard treatment of low dose antibiotics for prophylaxis in women with recurrent UTI in a routine clinical setting. We tested the null hypotheses that methenamine hippurate was inferior to daily antibiotics for the prevention of recurrent UTI in women.

Methods

Study design

This pragmatic, multicentre, randomised, open label, non-inferiority trial compared clinical effectiveness of low dose antibiotic prophylaxis, the current standard treatment for recurrent UTI prevention, with the urinary antiseptic methenamine hippurate. The ALTAR trial (alternative to prophylactic antibiotics for the treatment of recurrent urinary tract infections in women) recruited women from secondary care urology and urogynaecology centres in the UK from June 2016 and incorporated a 12 month treatment period followed by a six month follow-up period. Recruitment was completed in June 2018 and the final follow-up visit took place in January 2020. The study protocol has been published elsewhere.910

Participants

Adult women aged 18 years and over with recurrent UTI who had decided, in conjunction with their responsible clinician, that prophylaxis was appropriate, were eligible for inclusion. Recurrent UTI was defined as at least three episodes of symptomatic UTI in the previous 12 months or at least two episodes in the past six months. We excluded women with correctable urinary tract abnormalities contributory to recurrent UTI (eg, urinary tract calculi) and those with neurogenic dysfunction of the lower urinary tract. Women already taking antibiotic prophylaxis or methenamine hippurate were allowed to take part, but a washout period of three months without preventive treatment was required before randomisation. All participants provided written informed consent.

Randomisation and masking

Participants were randomly assigned (1:1) to receive antibiotic prophylaxis or methenamine hippurate. Permuted blocks of variable length (2/4/6/8) were used, stratified by menopausal status (pre-menopausal v peri-menopausal/post-menopausal) and UTI frequency in the preceding year (<4 v ≥4). Randomisation lists were generated by an individual not otherwise involved in the trial and administered centrally via a web based service. This pragmatic trial was designed to reflect contemporary practice, and there was no masking of participants, clinicians, or local research staff.

Procedures

For participants assigned to antibiotic prophylaxis, the drug used was chosen from nitrofurantoin (50 or 100 mg), trimethoprim (100 mg), or cefalexin (250 mg) given orally once daily, depending on previous urine culture results and individuals' history of allergy or intolerance. Methenamine hippurate was prescribed as a twice daily oral dose (1 g). Participants were allowed to switch between antibiotic drugs or between treatment strategies, however, the need to adhere to the allocated intervention was emphasised. Participants experiencing symptomatic UTI episodes were advised to seek discrete treatment courses of antibiotics in their usual way, typically via their general practitioner.

Follow-up assessments took place every three months until month 18. At each visit, participants were asked about the occurrence of any UTIs, treatment adherence, and adverse events. Information on UTI episodes was confirmed where necessary from healthcare records. Blood samples were taken to monitor kidney and liver function in all participants. Urine samples were submitted to the central laboratory at baseline, at scheduled three monthly visits and at the time of UTI episodes. Optional perineal swabs were submitted at baseline and at six monthly routine visits. Participants completed symptom questionnaires every three months and at the time of symptomatic UTI.

Outcomes

The primary clinical outcome measure was the incidence of symptomatic, antibiotic treated, UTI episodes self-reported by participants over the 12 month treatment period. An episode of UTI was defined as the presence of at least one symptom reported by patients or clinicians from a predefined list produced by Public Health England,11 together with the taking of a discrete treatment course of antibiotics for UTI. Consequently, the occurrence of the primary outcome was always defined by the independent prescribing clinician via confirmation of the likely diagnosis and recommendation of antibiotic treatment. The end of one UTI episode was defined as 14 days after the final antibiotic dose. If symptoms restarted or further antibiotics were prescribed within 14 days, this event was counted as the same episode.

Secondary outcomes were the incidence of symptomatic, antibiotic treated UTI in the six months after treatment; microbiologically confirmed UTIs; antibiotic resistance profiles in Escherichia coli isolated from urine and perineal swabs; asymptomatic bacteriuria; total antibiotic use; and hospital admissions due to UTI. Participant satisfaction with treatment was measured using the treatment satisfaction questionnaire for medication, which assessed four domains of treatment satisfaction: effectiveness, side effects, convenience, and global satisfaction.12

Microbiologically confirmed UTIs were episodes defined as per the primary outcome but with an additional criterion of positive urine culture at the time of UTI. A positive urine culture was defined as one isolate of a potential uropathogen at a concentration of ≥10⁴ colony forming units/mL or two species of uropathogens isolated at ≥10⁵ colony forming units/mL.13 Asymptomatic bacteriuria was defined as a positive urine culture from urine samples submitted to the central laboratory in the absence of symptoms. Antibiotic resistance was assessed from urine and perineal swabs with antimicrobial sensitivity tested in triplicate against a panel of antibiotic drugs. Multidrug resistance in E coli was defined as resistance to at least one antibiotic drug in at least three antimicrobial categories (supplementary material, page 1).14

Statistical analysis

The trial was powered to assess non-inferiority of the absolute difference in UTI incidence over the 12 month treatment period. The non-inferiority margin, defined after a series of patient focus group meetings, was a difference of one UTI episode per year. Two meta-analyses of antibiotic prophylaxis and methenamine hippurate815 quoted relative risks of UTI versus placebo of 0.15 and 0.24, respectively. With this information, and local audit suggesting that untreated participants had an average of 6.5 UTI episodes/year, we assumed an average incidence rate of 0.975 and 1.56 episodes/year in the antibiotic prophylaxis and methenamine hippurate groups, respectively, equating to an estimated difference of 0.6 episodes/year (in favour of antibiotics). Using a two sample t test, and assuming a difference of 0.6 episodes/year and a standard deviation of 0.9 (based on placebo groups from meta-analyses), we required 87 participants per group to be 90% sure that the lower limit of a one sided 95% confidence interval (equivalently 90% two sided) was above the non-inferiority limit. To account for an estimated 25% attrition rate over the course of the trial, the target sample size was 240.

For the primary outcome, the incidence rate in each group was the total number of UTI episodes divided by the total follow-up (exposure) time, reported with 95% confidence intervals calculated using a resampling procedure (bootstrap). The absolute difference between groups was calculated and reported with a 90% bootstrap confidence interval. Non-inferiority would be concluded if the upper limit of the confidence interval was below one UTI episode/year. We estimated relative treatment differences using a negative binomial model with follow-up time included as an exposure variable, recruiting centre as a random effect, and baseline stratification factors (menopausal status and prior UTI frequency) as fixed effects. This model yielded an estimate of the incidence rate ratio, presented with a 95% confidence interval.

Sensitivity analyses excluded days when participants took therapeutic antibiotics for UTI from the follow-up time. We analysed a binary indicator of at least one UTI episode using a mixed effects logistic regression model adjusted for centre and baseline stratification factors. Analyses were primarily conducted in a modified intention-to-treat population, which included all participants observed for at least six months, because these participants were assumed to provide a reliable estimate of UTI incidence. Prespecified sensitivity analyses were conducted in intention-to-treat and per protocol populations (participants achieving ≥90% adherence with preventive treatment; switching between treatment strategies was considered adherent). We also conducted a post hoc sensitivity analysis in a strict per protocol population that included only those participants achieving ≥90% adherence with initially randomised treatment.

Symptomatically diagnosed UTI incidence in the post-treatment period and microbiologically confirmed UTIs were analysed as for the primary outcome measure. Secondary outcomes were analysed according to the intention-to-treat principle and included all participants with data available. Rates of asymptomatic bacteriuria and antibiotic resistance were compared by χ² or Fisher's exact tests. These analyses are considered exploratory and should be interpreted with caution. We made no adjustments for multiple testing. Domain scores of the treatment satisfaction questionnaire for drug treatment were compared between groups using a two sample t test and an analysis of covariance model adjusted for baseline stratification factors. To account for switching between treatment strategies, adverse events were reported according to treatment received at the time with data summarised by numbers of participants receiving each intervention. All analyses were conducted using Stata version 16. Independent trial steering and data monitoring committees had oversight throughout the trial. The trial was registered with the ISRCTN registry (ISRCTN70219762).

Patient and public involvement

The host institution has an established UTI patient group and as such patient and public involvement began at the planning stage of this trial. Patient representatives were included on the trial steering committee. The patient and public involvement group helped define the main outcome measure and in particular stressed the importance of a practical UTI definition rather than sole reliance on microbiological tests. The non-inferiority margin was exclusively defined by the patient group who stressed the severity of UTI symptoms and advised a non-inferiority margin of one UTI in a year. This margin was considered a clinically meaningful difference by our patient and public involvement group based on an appreciation of the likely numerical reductions in UTI frequency in both of our trial arms. During recruitment, the study was advertised via patient representatives from Bladder Health UK and results will be disseminated widely among patients including via this groups' regular publications.

Results

Participants

Between 23 June 2016 and 20 June 2018, 240 participants were recruited and randomly assigned to antibiotic prophylaxis (n=120) or methenamine hippurate (n=120). For those allocated to antibiotic prophylaxis, 66 (55%) received nitrofurantoin, 30 (25%) trimethoprim, 24 (20%) cefalexin. A total of 22 (18%) participants allocated to methenamine hippurate switched to receive antibiotic prophylaxis and seven (6%) vice versa. Patient follow-up was completed in January 2020. The modified intention-to-treat analysis included 205 (85%) participants; 102 (85%) in the antibiotic arm and 103 (86%) in the methenamine hippurate arm (fig 1). Supplementary table S1 provides a summary of participants included in the primary and sensitivity analysis populations. Demographics and clinical characteristics at baseline were generally well balanced across treatment groups (table 1).

Fig 1

Trial profile (CONSORT flowchart)

">

Fig 1

Trial profile (CONSORT flowchart)

Table 1

Baseline characteristics. Data are number (%) of participants unless stated otherwise

View this table:

Primary outcome

In the modified intention-to-treat population, 90 symptomatic, antibiotic treated UTI episodes were reported over 101 person years of follow-up in the antibiotic group, and 141 episodes over 102 person years of follow-up in the methenamine hippurate group (supplementary fig S1A). The incidence of symptomatic antibiotic treated UTI over the 12 month treatment period was therefore 0.89 episodes per person year (95% confidence interval 0.65 to 1.12) in the antibiotic group and 1.38 (1.05 to 1.72) in the methenamine hippurate group (absolute difference 0.49 (90% confidence interval 0.15 to 0.84)). With the upper limit of the 90% confidence interval below the non-inferiority limit of one, we can conclude methenamine hippurate to be non-inferior to antibiotic prophylaxis in this setting. This result was confirmed in all sensitivity analysis populations (table 2). To facilitate meta-analyses and comparisons with other studies, a 95% confidence interval for the primary outcome was estimated as 0.49 (0.08 to 0.90). On a relative scale, the adjusted incidence rate ratio was estimated as 1.52 (95% confidence interval 1.16 to 1.98) in favour of antibiotic prophylaxis. Secondary analysis of the primary outcome, excluding time spent taking therapeutic antibiotics for UTI from the follow-up time, showed consistent results (supplementary table S2). Supplementary table S3 shows the number and proportion of participants reporting at least one symptomatic, antibiotic treated UTI episode over the 12 month preventive treatment period.

Table 2

Incidence of episodes of symptomatic, antibiotic treated, urinary tract infection during 12 month preventive treatment period

View this table:

Secondary outcomes

In the six month post treatment follow-up period, the UTI incidence rate was 1.19 (95% confidence interval 0.86 to 1.52) and 1.72 (1.27 to 2.18) episodes per person year in the antibiotic prophylaxis and methenamine hippurate groups, respectively (absolute difference 0.53 (95% confidence interval −0.03 to 1.09); supplementary table S4).

Overall, 183 (79%) of 231 UTI episodes reported in the modified intention-to-treat population were accompanied by a urine sample, and during the 12 month treatment period a positive urine culture was observed in 96 (52%) of these episodes. Incidence of microbiologically confirmed UTIs was 0.41 (95% confidence interval 0.27 to 0.56) in participants allocated to antibiotic prophylaxis and 0.53 (0.34 to 0.72) for those allocated methenamine hippurate (absolute difference 0.11 (−0.12 to 0.35); supplementary table S5). In the six months after treatment, 93 (66%) of 141 UTI episodes were associated with a urine sample, of which 65 (70%) were positive. Incidence of microbiologically confirmed UTI episodes was 0.48 (0.28 to 0.68) and 0.86 (0.59 to 1.14) in the antibiotic prophylaxis and methenamine hippurate groups, respectively (absolute difference 0.38 (0.04 to 0.72)).

During the 18 month trial period, the rate of asymptomatic bacteriuria was similar between treatment groups, however, a post hoc analysis of urine samples submitted during the 12 month treatment period identified a significantly higher rate in the methenamine hippurate group than in the antibiotic prophylaxis group (44 (14%) of 326 samples v 22 (7%) of 323; χ² test, P=0.0048; supplementary table S6).

Therapeutic antibiotics for UTI were received during the 12 month treatment period by 51 (43%) and 67 (56%) participants allocated to the antibiotic prophylaxis and methenamine hippurate groups, respectively. In the six month observation period, the total number of days spent taking therapeutic antibiotics was higher in the methenamine hippurate group than in the antibiotic prophylaxis group (13.5 (interquartile range 6.5-23) v 7.5 (4-15)). Similarly, antibiotics for other infections were received by 28 (29%) of 98 participants receiving methenamine hippurate compared with 15 (15%) of 97 receiving antibiotic prophylaxis (supplementary table S7).

Antimicrobial resistance

Availability of optional six monthly perineal swabs over the 18 month trial period is presented in supplementary table S8. The proportion of participants demonstrating resistance to at least one antibiotic in E coli isolated from perineal swabs was similar between randomised groups at baseline. At six or 12 month follow-up, this proportion was higher in the antibiotic prophylaxis group than in the methenamine hippurate group (46/64 (72%) v 39/70 (56%); χ² test, P=0.05, fig 2, top left panel). However, at month 18, multidrug resistance in E coli isolated from perineal swabs was higher in the methenamine hippurate group than in the antibiotic prophylaxis group (9/45 (20%) v 2/39 (5%), Fisher's exact test, P=0.06, fig 2, top right panel).

">

Fig 2

Antibiotic resistance rates in Escherichia coli isolated from perineal swabs, and urine samples taken during an episode of symptomatic urinary tract infection. Top left panel: proportion of participants (with E coli isolated) showing resistance to at least one antibiotic in E coli isolated from perineal swabs at baseline, six or 12 month follow-up, and 18 month follow-up (P values from χ² test). Top right panel: proportion of participants (with E coli isolated) showing multidrug resistance in E coli isolated from perineal swabs at baseline, six or 12 month follow-up, and 18 month follow-up (P values from Fisher's exact test). Remaining rows: proportion of participants showing resistance in E coli isolated from any symptomatic urine sample submitted during the 12 month preventive treatment period or six month observational period after treatment (out of those participants with E coli isolated from a symptomatic urine sample)

A substantial growth of E coli was isolated from the urine samples of 41 participants during 67 episodes of symptomatic UTI (that is, symptomatic urine samples) over the 12 month treatment period, and from 21 participants during 26 symptomatic episodes over the six months after treatment. The proportion of participants showing resistance to at least one of the antimicrobial drugs tested in E coli isolated from symptomatic urine samples was similar between randomised treatment groups (supplementary fig S2A), as were the rates of multidrug resistance (supplementary fig S2B). However, a higher proportion of participants in the antibiotic prophylaxis group than in the methenamine hippurate showed resistance to co-trimoxazole (6/13 (46%) v 3/14 (21%)) during the 12 month treatment period, and resistance to trimethoprim (6/8 (75%) v 5/13 (38%)) during the six months after treatment (fig 2). Similarly, a higher proportion of E coli isolates from symptomatic urine samples submitted during the 12 month treatment period demonstrated resistance to cephalosporins in those allocated to antibiotic prophylaxis compared to methenamine hippurate (supplementary fig S3). In contrast to the numbers of positive samples collected during symptomatic episodes, E coli was isolated from only a small proportion of urine samples collected routinely at three month intervals from asymptomatic participants (supplementary table S9A). All isolates identified from urine samples are listed in supplementary table S9B.