Methenamine hippurate compared with antibiotic prophylaxis to prevent recurrent urinary tract infections in women: the ALTAR non-inferiority RCT

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 13/88/21. The contractual start date was in February 2016. The draft report began editorial review in November 2020 and was accepted for publication in May 2021. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Copyright statement

Copyright © 2022 Harding et al. This work was produced by Harding et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This is an Open Access publication distributed under the terms of the Creative Commons Attribution CC BY 4.0 licence, which permits unrestricted use, distribution, reproduction and adaption in any medium and for any purpose provided that it is properly attributed. See: https://creativecommons.org/licenses/by/4.0/. For attribution the title, original author(s), the publication source – NIHR Journals Library, and the DOI of the publication must be cited.

2022 Harding et al.

Scientific background

Urinary tract infections (UTIs) are common in adult women, in both the community and the hospital setting. Inoculation of the periurethral tissues, and subsequently the bladder, by bacteria that originate from the lower gastrointestinal tract or vagina may lead to classic symptoms of infective cystitis such as dysuria, frequency, urgency and suprapubic pain. Around 40% of all adult women will experience a UTI in their lifetime, with peak incidences occurring in the third and ninth decades. Annually, 3% of all adult women will be diagnosed with a UTI and almost half of these women will experience a second episode within 1 year. 2,3 With > 300,000 women per year in the UK requiring treatment for recurrent urinary tract infection (rUTI), this represents a significant health problem. 4 Standard preventative treatment of rUTI usually consists of daily low-dose antibiotic therapy, and a previously published meta-analysis from the Cochrane collaboration5 described a reduction in UTI episodes of > 80%. The relationship between antibiotic prescription and the development of antimicrobial resistance is well described, with documented resistance not only confined to the causative micro-organisms but also observed in commensal flora. 6 The judicious prescribing of antibiotic agents, termed ‘antibiotic stewardship’, is an essential component of national and international strategies to halt the progression of global antimicrobial resistance. 7,8 Current cross-specialty guidelines9 have identified that ‘repeated/prolonged treatment with antibiotics’ is a significant contributing factor to this progression. Consequently, interest has focused on non-antibiotic treatments for the prevention of rUTI, which have the potential to improve public health by minimising the development of antimicrobial resistance in bowel reservoirs.

Summary with implications for trial design

Why this research is needed now

A recent meta-analysis22 reviewed the evidence for non-antibiotic treatments as prophylaxis against rUTI but the results were disappointing, mainly owing to a paucity of evidence. One of the conclusions of this report was that ‘Although sometimes statistically significant, pooled findings for the other (non-antibiotic) interventions should be considered tentative until corroborated by more research’. It would appear that one of the barriers to clinicians recommending non-antibiotic alternatives for the treatment of rUTI is the lack of currently available clinical evidence for these. The campaign for antibiotic stewardship and more prudent prescribing of antibiotic agents can be strengthened only by further work exploring the effectiveness of non-antibiotic alternatives. A further conclusion from this meta-analysis22 was that ‘Large head-to-head trials should be performed to optimally inform clinical decision making’.

One of the most comprehensive guidelines published on the subject of UTI is the 2012 Scottish Intercollegiate Guideline Network (SIGN) Guideline 88. 9 The literature review carried out prior to the formulation of this document identified much evidence of variations both in practice and when antibiotic treatment is started for UTI. 9 In addition, one of the constant themes in this report is the need to avoid prescribing antibiotics unnecessarily, which is associated with adverse events such as CDI, MRSA and antibiotic-resistant UTIs. 9 The UK antimicrobial resistance strategy and action plan states ‘the increasing prevalence of antimicrobial resistant micro-organisms is causing international concern’, and identifies that ‘the emergence of resistance represents adaptive selection by micro-organisms which is an inevitable result of therapeutic use of antimicrobial agents’7 (quotations contain public sector information licensed under the Open Government Licence v3.0). This document reflects an urgent need for prudent antibiotic use as one of three key elements of the strategy to control antibiotic resistance. The ALTAR (ALternatives To prophylactic Antibiotics for the treatment of Recurrent urinary tract infection in women) trial aimed to provide high-level contemporary evidence of the relative effectiveness of a non-antibiotic UTI prevention treatment compared with the standard treatment of prolonged low-dose antibiotics in a UK population of women with rUTI in a routine NHS care setting.

Design/methods

The ALTAR trial tests the null hypothesis that the non-antibiotic treatment methenamine hippurate (Hiprex^®; Mylan NV, Canonsburg, PA, USA) is inferior to the standard treatment of an extended course of prophylactic antibiotic for the prevention of rUTI in women, and is less cost-effective to the NHS. The alternative hypothesis is that methenamine hippurate is as good at preventing rUTI and is as cost-effective as antibiotic prophylaxis. We have chosen to investigate the possible non-inferiority of methenamine hippurate in order to clarify an alternative treatment choice to antibiotics in treating rUTIs, which, if used more routinely, may prevent increases in antibiotic-resistant strains of infection.

Estimates of prevalence, effectiveness and harms from Cochrane reviews5,14 have informed the power calculation conservatively based on what we, guided by a patient panel, considered to be a minimum threshold difference that would drive patient and clinician acceptability together with change of practice prompted by the inclusion of trial results in future meta-analyses and guidance for management of rUTI in the NHS and internationally.

Aims and objectives

Summary of study design

The ALTAR trial is a multicentre, pragmatic, open-label randomised non-inferiority trial evaluating the clinical effectiveness and cost-effectiveness of the urinary antiseptic methenamine hippurate, a non-antibiotic treatment for the prevention of rUTI, and comparing it with the current standard treatment of low-dose antibiotic prophylaxis. Outside the trial medication, both arms received usual care and any breakthrough UTIs were treated by discrete courses of antibiotic treatment, as required.

Progress on trial

The trial duration for each participant was 18 months. The patient flow through the trial is shown in Figure 1; for the schedule of events for trial participants, see Table 3.

FIGURE 1.

Participant flow. LFT, liver function test; TSQM, Treatment Satisfaction Questionnaire for Medication.

Outcome measurement

Outcomes were collected for each participant over the 12-month treatment period following randomisation and during a follow-up period of 6 months after completion of the planned course of preventative treatment (resulting in a total observation period of 18 months for each participant).

Data collection

Trial events

Randomisation

Randomisation was performed as close as possible to the date of consent (normally this was immediately after). For consented participants who completed a 3-month washout period, eligibility was rechecked prior to randomisation.

Follow-up

The schedule of events for the ALTAR trial is shown in Table 3.

Data handling and record-keeping

Case report form data were entered by site staff into the trial-specific clinical data management system, MACRO. Participant-completed questionnaires and UTI records, identifiable only by participant identifier, were returned by post to the central NCTU trial office, where they were entered by NCTU staff and stored securely. As per participant consent processes, identifiable data were stored in a separate, password-protected database within NCTU, with access limited to members of the trial team responsible for the preparation and sending of follow-up questionnaires and logging their return. Two reminders, each with an additional copy of the questionnaires, were sent to participants to prompt their return. Central laboratory results of urine and perineal swab analysis were processed and uploaded into the MACRO database by the database manager. Participants were allocated an individual trial number at randomisation and all trial data and laboratory samples were identified by this unique trial number. Essential data will be retained for a period of at least 10 years following close of the trial, in line with the sponsor’s policy and the latest European Directive on GCP (2005/28/EC). 29 Data were handled, digitalised and stored in accordance with the Data Protection Act 199830 and the General Data Protection Regulation31 (GDPR) from 25 May 2018.

Changes to protocol

Changes made to the protocol during the trial are listed in Table 4.

Sample size calculation

The ALTAR trial had a planned recruitment target of 240 patients, 120 in each of the treatment arms.

A non-inferiority design was chosen, as we believed that an oral urinary antiseptic (methenamine hippurate) would be acceptable to the patient group provided that its effectiveness for UTI prevention was no worse than antibiotic prophylaxis and that the burden of adverse effects was similar or better. There is also the key added potential theoretical benefit of reduced rates of resistant organisms and subsequent collateral harm to the individual and community.

Semistructured interviews with a patient panel of 12 women identified that any reduction in UTI episodes, even one per year, would be deemed worthwhile. Therefore, the minimum clinically important difference between the treatment arms of one UTI per 12 months was set as our strict, non-inferiority margin.

Two existing meta-analyses of studies examining prophylactic antibiotics5 and methenamine hippurate14 both quoted a mean relative risk of UTI compared with placebo of 0.15 and 0.24, respectively. Using these values and data from a local audit (Charlotte Cuff, University of Newcastle, Biomedical Sciences Thesis 2014; n = 200) suggesting that the average number of UTI episodes per year in this patient group is 6.5, we estimated that the average number of UTI episodes per year during preventative treatment would be 0.975 in those randomised to antibiotic prophylaxis and 1.56 in those randomised to methenamine hippurate. This equates to an estimated difference in UTI episodes per year between prophylactic antibiotics and methenamine hippurate of around 0.6 episodes (in favour of antibiotics).

The standard deviation (SD) of UTI episodes per year is taken from the placebo groups in the studies included in the Cochrane meta-analyses and was conservatively estimated at 0.9 episodes per year. 5,14

Using a two-sample t-test and assuming an actual difference of 0.6 UTI episodes per year (in favour of treatment with antibiotics) and a SD of 0.9 UTI episodes per year, then two groups of 87 patients would be required to be 90% sure that the lower limit of a one-sided 95% CI (or equivalently a 90% two-sided CI) was above the non-inferiority limit of one. The attrition rate was conservatively estimated at 25%; therefore, the total sample size required was 232, rounded up to 240.

Statistical analysis

A SAP that includes full details of all statistical analyses was finalised and agreed prior to the final database lock and analysis (see the NIHR Journals Library project web page – URL: www.journalslibrary.nihr.ac.uk/programmes/hta/138821/#/). The statistician responsible for approving the SAP remained blind to treatment allocation until after the final database lock. The main analysis of the primary outcome measure was performed in the modified intention-to-treat (mITT) population and contained all randomised participants with an observational period of at least 6 months. Sensitivity analyses were performed in a strict intention-to-treat (ITT) population and a per-protocol population that included all participants achieving at least 90% compliance with any trial preventative treatment. A post hoc sensitivity analysis in a stricter per-protocol population, including all participants achieving at least 90% compliance with their initial randomised trial treatment, was also performed.

Analyses of symptomatic antibiotic-treated UTIs in the 6-month follow-up period (12–18 months) was conducted in all participants in active follow-up at 15 or 18 months plus those with data available from health-care record review.

For all analyses, with the exception of AE data, participants were analysed according to their randomised treatment assignment.

Adverse events

Adverse events were reported throughout the 18-month trial period. All events were graded as mild, moderate or severe, and relationship to trial medication was reported as unrelated, unlikely, possible, probable or definite. All events were coded systematically at the end of the trial using the Medical Dictionary for Regulatory Activities (MedDRA), version 22.0.

To account for any participants who were switched to the alternative treatment group, AEs are primarily reported by treatment received (antibiotic prophylaxis or methenamine hippurate) at the time of AE onset. Data were also summarised by randomised treatment group.

The numbers of AEs and adverse reactions (ARs) (possibly, probably or definitely related to treatment) reported per participant were summarised descriptively. The worst grade of AE and AR reported per participant was also presented.

The number of participants reporting each type of AE was tabulated. Only those occurring in at least 3% of participants in either treatment group are shown, owing to a large number of different AEs being reported in a small number of participants.

Serious adverse events (SAEs) and serious ARs were listed.

Trial progress and monitoring

The monitoring of trial conduct and data collected was performed by a combination of off-site, on-site and central monitoring to ensure that the trial was conducted in accordance with the trial protocol and GCP trial monitoring. This included the review of consent procedures, source data verification, SAEs, trial conduct and essential documentation, and was undertaken by members of the Trial Management Group (TMG).

Microbiological methods

Definition of end of trial

The definition of the end of the trial was the last recruited participant’s last follow-up visit at 18 months post randomisation, which took place on 27 January 2020. With approval from the TSC, a 9-month extension to the trial recruitment period was granted by the trial funder in October 2017 (variation to contract, January 2018) to accommodate additional time to meet the trial recruitment target.

Compliance and withdrawal

Data monitoring, quality control and assurance

Ethics and governance

The Newcastle upon Tyne Hospitals NHS Foundation Trust Research and Development Directorate sponsored the trial (reference 06867). A favourable ethics opinion for the trial was obtained on 23 December 2015 from the NHS Research Ethics Service Committee North East – Tyne and Wear South (reference 15/NE/0381). Health Research Authority approval was issued for the trial on 20 June 2016 and subsequent local research and development approvals were sought for each participating site. Approval was sought and obtained for all substantive protocol amendments.

Trial registration and protocol availability

The trial was registered as International Standard Randomised Controlled Trial Number (ISRCTN) 70219762 on 31 May 2016 and in the UK NIHR portfolio (reference HTA 13/88/21). The latest version (version 3.0) of the full protocol is available at the NIHR Journals Library project web page (URL: www.journalslibrary.nihr.ac.uk/programmes/hta/138821/#/) and a published version is also available. 1

Serious adverse event reporting

The trial protocol provided guidance on AE and SAE reporting, as well as determining the degree of relatedness and assessment of causality for SAEs that may be related to trial participation. The reference safety information for assessment of expectedness of related events was contained in the summary of product characteristics for trimethoprim, nitrofurantoin, cefalexin and methenamine hippurate. SAEs excluded UTI, since this was the primary outcome collected and documented throughout the trial. All SAEs were reported for the duration of the trial and for 4 weeks after the trial intervention was stopped.

Background

Understanding why patients agree or decline to participate in clinical trials is a crucial factor in the successful delivery of trials on time and to target. Embedded qualitative studies can provide in-depth information on the views and experiences of potential patient participants and recruiting staff that quantitative research cannot provide and highlight issues that may affect recruitment and retention. 34 To identify modifiable factors that could improve the conduct of the ALTAR trial, an embedded qualitative study was proposed. In the first 8 months that sites were open to recruitment, the plan was to conduct up to 15 in-depth interviews with patients approached about the ALTAR trial in each of three groups: (1) those who agreed to participate in the trial and continue on the trial until its end, (2) those who dropped out of the trial and (3) those who declined to participate. In addition, a focus group of eight recruiting staff participants was planned to explore views on recruiting to and delivering the trial.

Aim and objectives

The aim of this embedded qualitative study was to inform the trial team of potential barriers to recruitment and retention experienced by participants and trial staff in the first 8 months of recruitment to the ALTAR trial. The objective was to explore the views of patients approached to participate in, and those recruited to, the ALTAR trial, and collect opinions from staff involved in the study, to identify any factors of the processes and conduct that should be modified.

Methods

Data collection

In-depth telephone interviews were conducted by a trained interviewer. A topic guide was used, which was developed with input from the study team and the Patient and Public Involvement Group. The topics, for both patients/participants and recruitment staff, included experiences of trial recruitment and conduct, and views on the trial interventions. Consent was obtained before the interview commenced. Interviews were digitally recorded with the permission of the interviewee. Two time points were selected for patient interviews: within 2 weeks of being approached to participate in the ALTAR trial and 3–6 months later, or earlier if they decided to withdraw within that period. Recruiting staff were interviewed on one occasion only.

Analysis

Any modifiable factors identified during the interviews were noted and quickly fed back to the Chief Investigator and NCTU. Interviews were transcribed verbatim. Transcripts were checked and anonymised. Data were analysed drawing on the constant comparative method and a thematic coding frame was agreed upon between the members of the qualitative team. NVivo version 12 (QSR International, Warrington, UK) was used as a tool to manage and code transcript data. 35 The overall headline results were made available to inform change in the trial procedures before the end of the first year of the recruitment phase.

Recruitment to the qualitative study

Despite the original plan to recruit up to 15 patients who had declined to take part in the ALTAR trial, only one person from this group returned an EOI form. As a total of 60 patients declined to take part in the trial, this was a cause for concern.

The low recruitment of trial decliners to the qualitative study was explored in some of the interviews with recruiting staff. At one site, one interviewee said that it tended to be characteristic of those who decline trial participation also to do so in response to other aspects of a study. At another site, an interviewee admitted that they had not been asking patients approached to participate in the ALTAR trial about the qualitative study because of the process of identification and recruitment, but promised to do so in future.

In addition to site staff overlooking the need to mention the qualitative study, another explanation could have been the design of the ALTAR trial PIS. The first three pages have information in bullet points about the trial, at the end of which it states, ‘If you are interested in taking part, please continue to read the rest of this leaflet’. Only those who chose to read on would learn about the qualitative study. It is likely that those who were not interested in participating in the trial would not have continued beyond that point. There was a separate information sheet about the qualitative study, but this meant that the burden was on site staff to supply this sheet with the trial information. In retrospect, if space permitted, a bullet point in the first three pages about the qualitative study, and the importance of seeking the views of those who do not participate, would have ensured that patients were aware of it.

Findings: patient interviewees

In the period the embedded qualitative study was conducted (July 2016 to March 2017), 77 patients agreed to participate in the trial and were randomised. For the qualitative study, 34 EOIs were returned via the NCTU and 29 interviews were conducted between July 2016 and March 2017. Despite several attempts, the remaining five patients were not contactable. One of the 29 patients interviewed had declined to participate in the trial; of the remainder, 16 were randomised to an antibiotic and 12 to methenamine hippurate. Ten patients participated in a second interview 3–6 months following the first. The ages of interviewees ranged from 19 to 80+ years.

Recurrent urinary tract infections: past experience

Interviewees talked about their experience of rUTIs. How long interviewees had experienced rUTIs ranged from 2 years to the whole of their lives. A few said that the infections began when they were children:

I was really young and I’ve been having them ever since and I’m 18 now, nearly 19.

Patient 1103

The inclusion criteria for the trial were to have experienced rUTI, defined as at least three episodes of symptomatic UTI in the last 12 months, two episodes in the last 6 months or one episode requiring hospitalisation. Some interviewees described an almost continuous state of having a UTI, although others said that there could be months in between and then they would suffer from a number of infections in a short period:

Since September of last year, I have had them pretty much constantly – one goes, and then I get another one.

Patient 1018

Sometimes I can go 9 months, a year and not have a problem and then perhaps I’ll get two or three in a year.

Patient 1104

The rUTIs were described as ‘terrible’ and ‘horrendous’. In terms of the impact, some said that the rUTIs affected their whole lives. One interviewee, who was experiencing rUTIs every couple of weeks, said that even in between episodes she felt unwell as if the infections had ‘not fully resolved’ (patient 1002). Another interviewee described how they affected her daily life:

A lot of people don’t understand a UTI, they just kind of think ‘waterworks’. But when it’s . . . basic things like walking, sleeping, sitting at work, having a conversation with people, when you’ve got that level of discomfort, it’s difficult to just carry on as normal, to be honest. So yes, it really does affect your quality of life.

Patient 1005

I teach, so (infection) has an impact on lifestyle. Sometimes I was just finishing one course and having to go straight onto another. So it was affecting lifestyles and . . . time off work to go and visit doctors, and put more samples in, and it was just becoming a bit of a burden, to be perfectly honest.

Patient 1011

Interviewees recounted that they had undergone investigation to find the cause of their rUTIs. For some there was a sense of frustration, particularly when no cause could be found and the view from their clinician was that they were ‘just one of these people that had recurring infections’ (patient 1101) or ‘sometimes it just happens’ (patient 1305). Interviewees talked about the numerous courses of antibiotics that they had previously been prescribed for their UTIs, with varied success:

So over the past maybe year or so, I’ve probably had about eight rounds of antibiotics.

Patient 1009

I’d been on sort of, a 3-day course of antibiotics which does clear it but then it keeps coming back.

Patient 1101

Last year when I was pregnant I had them every week. They kept giving me antibiotics and they weren’t responding and even after I had the baby they tried me on different tablets, even the doctors at the hospital and nothing worked.

Patient 1303

Trial processes: views and experiences

Opinions of the ALTAR trial and outcome measures

When asked for their views of the ALTAR trial, one interviewee said that they were pleased someone was exploring a problem ‘that lots of women clearly have’ (patient 1005). Another commented that the trial ‘wasn’t too intrusive’ and wasn’t testing ‘some new drug’, which were positive and reassuring factors (patient 1401). A number of others similarly had no concerns about the trial. Of those who did, the responses were few and disparate. One interviewee was concerned about the length of time they would be taking antibiotics, and about the potential damage to their stomach lining. Another had a question about funding and whether or not money spent on conducting the trial could be detrimental to other key health services. Members of the research team addressed these concerns satisfactorily:

One of my main concerns was that a year was a very long time to do a treatment, so that was it really. The doctor told me that it was a year, just to be safe, I think. I can’t really remember. I think just to give it enough time to really work. I think he said if it was too short, then I would risk having everything come back again.

Patient 1008

Well I just wanted to know if they were taking money away from any urgent cases that people need, because you know they keep saying they are taking money from the elderly and the people in desperate need of cancer treatments and things. I’d hate to think I was taking 12 months of tablets when they could have maybes helped somebody live for another few months.

Patient 1019

Interviewees’ general views of antibiotics were explored. Several were aware of antibiotic resistance and mentioned this as a concern of longer-term use. Others raised the importance of antibiotics more widely without any mention of resistance. A smaller number talked about their long-term use of antibiotics, and the fact that they were no longer effective in treating their rUTIs. A few said that they disliked taking medication per se and commented that antibiotics were ‘bad’ for your body if taken on a long-term basis but did not expand on how:

There’s obviously a definite need for them. I have taken them, and would again take them happily if that’s what I need to cure whatever is wrong with me. [. . .] I realise that there can be drawbacks in terms of long-term use, and that the bugs can become resistant to a certain one.

Patient 1018

I have great faith in them, but they just don’t seem to work the same or if you do you’ve to take a lot of them. It shakes your confidence you know about it.

Patient 1201

While I have nothing against antibiotics, I just think the news that you hear that they’re not effective if you keep taking them all the time and they’re not good for your body.

Patient 1020

The fact that these interviewees may have to take antibiotics if they participated in the ALTAR trial did not impact negatively on their views of the trial.

Views on specific elements of the trial design were explored, specifically the 12-month period on antibiotics or the antiseptic, and the subsequent 6-month period when participants stop the prophylaxis and are monitored. In terms of the 12-month treatment period, apart from the interviewee mentioned above, most did not think that this was too long and understood the rationale:

One year? Mmm I mean, it’s a long time but . . . I’d rather be on some medication that potentially will stop the recurring UTI than not being on anything. I’m just happy to have treatment.

Patient 1013

There was a more varied response to the 6-month period when the prophylaxis stops. Although several interviewees were not worried and appreciated the need for this period to answer the research question, a few others had concerns:

I am just hoping that I don’t start with the recurring UTIs. That’s the only thing . . . the downside of it is that because I will not be on that any more, so, I might . . . hopefully I’ll not, but, I could end up back with the UTIs every 6 weeks.

Patient 1003

Scared. [. . .] I’ll see it through but I do feel a bit concerned about the 6 months where I’ll have no prevention or any kind of treatment whatsoever because I know exactly what’s going to happen.

Patient 1011

Trial participation involved the provision of blood and urine samples at baseline and then every 3 months until the end of the trial. Participants were asked for an additional urine sample if they suffered a UTI during the trial. The completion of a urinary symptom questionnaire, EQ-5D-5L and health resource use questionnaire was required at baseline and at 3, 6, 9, 12, 15 and 18 months. Diaries were supplied to record UTIs as they happened. Finally, at 12 and 18 months, participants were asked to complete the TSQM. The potential burden of the range of measures was explored. Although these interviews were conducted at the beginning of trial participation, most interviewees did not think that the range of measures and what they were expected to do was problematic. One interviewee thought that it was worth it because of the extra monitoring they would receive as part of the trial:

I knew I would be like an active participant in it anyway. I didn’t just expect to get the medication, then just to be left for 12 months. I know I’ve got to keep a diary, and if I do get another UTI, I’ve got to make a lot of notes about that, but I’m totally happy to do that, it’s totally fine.

Patient 1003

Well, I think it is just, you know, touch wood, if I don’t get a water infection I will not have anything to do apart from going back and forward every 3 months, which isn’t a bind really.

Patient 1019

There is stuff to do but I don’t mind. I just think if I’m being monitored and there’s information being provided then it’s surely better for the study. The more information the better, really.

Patient 1013

Although mentioned by a only few interviewees, the only part of the trial that was, at first, perceived as onerous was the submission of urine samples should the interviewee experience an infection:

It seems a bit daunting at first. Once it’s all explained to you, it’s quite simple. Once you realise that the little thing only has to be sent off if you get a urine infection that’s fine.

Patient 1401

I think if you’re fortunate enough not to get any infections the whole way through then you’re going to be sailing through it really. But if you get to the point where you are still getting them and you’re putting a sample in the doctors and sending one in to yourselves, then it gets a little bit more responsibility on you to do that sort of thing. But that’s what you sign up for, isn’t it. It’s part of the trial.

Patient 1016

Only one interviewee commented that the diary would be a burden. They appear to have misunderstood that they complete the diary only when they have symptoms of a UTI rather than daily, and the purpose of this was to assist trial participants to complete the UTI record questionnaire:

It’s this about, just about writing it up every day and all that, I mean you’re not at university, you’re at a hospital and you are a patient. But um I would still help in any way.

Patient 1201

Participating in the ALTAR trial

Feedback to the Trial Management Group

As mentioned at the beginning of this chapter, the main reason for the embedded qualitative research was the early identification of any modifiable issues with trial processes and conduct that may affect recruitment and retention of participants. The factors identified in the 8 months of the qualitative study are listed in Table 5. In summary, these were clarification of the antiseptic, as one interviewee had searched for further information and discovered this was a medication that could eradicate bacteria; the fact that some interviewees expected the trial to identify the cause of and/or cure for their rUTIs; and potential AEs and SAEs reported during the interviews. Some of the other issues identified through the interviews were fed back to the NCTU directly by recruiting staff and were being addressed.

Findings: recruiting staff

Seven RNs and two consultants were interviewed from four of the recruitment sites, and five trial staff members were non-contactable. Recruiting staff were extremely busy and the interviewer had some difficulty securing interviews. This was raised in one of the TMG meetings and the NCTU suggested adding an appeal in the newsletter/bulletin that was sent to sites asking recruiting staff to take part in an interview.

The ALTAR trial site initiation visit

The recruiting staff’s views of the site initiation visits (SIVs) were generally positive, although for most interviewees quite a bit of time had passed since it took place. Seven of the nine interviewees were at the SIV, one was on sick leave and another was unable to attend. The latter interviewee said that their manager conveyed the information from the SIV to them and they had no issues with the trial.

A number of interviewees said that the trial was straightforward and the information provided in the SIV was clear. One interviewee said that they were involved in the development stage and provided feedback on the logistics of running the trial at their own site. There were positive comments about the chief investigator for the ALTAR trial. In addition, a few interviewees mentioned that they had worked on a previous trial with NCTU and this was of benefit:

So I think there is the degree of familiarity with the staff who are involved with this study from the university side . . . helps. Obviously I know the study area, and the chief investigator, Mr Harding [. . .] his knowledge and enthusiasm for the studies was very obvious and helpful.

Recruiting staff 1003

The majority had found the SIV useful. There was one negative comment about the fact that the CRF was not ready at the SIV. Another interviewee commented that assumptions were made about how the trial would run, without acknowledging the differences across trial sites:

I did feel there were a lot of unanswered questions at the SIV that I probably would have liked to have more information about. There were things that weren’t quite set up. For example, the database wasn’t finalised so when they were trying to show the database, there were parts that were . . ., you know, ‘This bit hasn’t been completed yet so we can’t show you that’ or ‘This bit’s going to be getting changed’. It was more to do with actual electronic CRF.

Recruiting staff 1302

There were I think maybe a couple of queries where they presumed that we would do things a certain way which we had looked at different ways of doing things, possibly regarding recruitment and things like that. But generally overall very, very well presented.

Recruiting staff 1601

ALTAR patient identification and recruitment

For the recruiting staff interviewed from four sites, the pool of potential participants were those who attended the outpatient clinic. One site identified eligible patients in a variety of ways: through attendance at clinic, attendance for a cystoscopy and referrals from general practice. This was considered much less labour-intensive than other methods of identifying eligible patients:

So I’m not having to spend hours upon hours searching through lots of different clinics, trying to find people that are eligible for the trial. Because obviously, they’re people that are coming to see them anyway, and obviously they’re getting, they’re kind of first to know. So from primary care and things like that, they contact the hospital, which should be obviously urogyn [urogynaecology], and then refer them on. So it has been, I’ve had a lot of people that I’ve been able to contact.

Recruiting staff 1301

In most sites the clinicians explained the trial to eligible patients and provided them with the PIS and then referred them on to the RNs. The process of the clinician having the first conversation with the patient about the trial (primarily explaining the rationale for it) was considered of benefit in terms of recruitment and ensuring a good understanding. One thought it ‘endorsed’ the trial, particularly as patients have a level of trust in a clinician when they have been under their care for some time (recruiting staff 1302). Another commented that this early discussion ‘primed’ the patient for a more detailed discussion with the RN (recruiting staff 1303). One site also wrote to eligible patients and followed up with a telephone call from the RN to ask if they were interested and if they had any questions.

The majority found the trial information easy to convey, and the fact that there was no placebo, that patients were familiar with antibiotics and that everyone would be having treatment were positive factors. On the surface it could appear that patient participants were required to do a number of things, for example completing questionnaires and submitting samples. The recruiting staff mentioned this early in discussions so that there were no surprises for patients, and pointed out that when viewed over the full period of the trial it was not too burdensome:

So the earlier you introduce that, the better, really. So we do that both when we introduce the study to the patient, but we also reinforce that when they come back to give consent.

Recruiting staff 1003

I just try say, ‘Look, it’s at least every 3 months that we will ask you to provide this but in the meantime if you get an infection, you’ve got this bottle at home, this blue bottle that you can send to Newcastle and fill this in. Hopefully you won’t, but in essence you know it’s four times a year.’ I try and explain that it’s not . . . so time-consuming, if you think about coming in . . . to see [clinician] four times a year.

Recruiting staff 1601

In terms of patient understanding of the trial, the view from those interviewed was that patients were aware of antibiotic resistance and of the use of medication as a prophylactic:

The patients are usually . . . they’re expert patients, to be quite honest. [. . .] I think that the patients are very . . . well informed now, due to the general media, about antibiotic resistance. . . . a lot of them will say, ‘Oh, yes, because the bugs are resistant’, or ‘The bugs are getting stronger’ or something along those lines. [. . .] Many patients will understand what prophylaxis is, but I think we’re just reinforcing it as a preventative medication.

Recruiting staff 1003

Some women have heard about being on prophylactic, some of them have been on prophylactics already. But obviously, they understand it. I mean, you’re not having to teach them what prophylactic means and things like that, they all understand it.

Recruiting staff 1301

Use and views of the patient information sheet

Some recruiting staff sat with patients and explained the ALTAR trial using the PIS as a guide, and other recruiting staff asked patients who had received the PIS, often through the post with a letter, and had had an opportunity to read it, if they had any further questions:

And I think the information sheet lays it out . . . and I have the information sheet there, as we go through, whilst giving a verbal description of the study.

Recruiting staff 1003

When we do meet them face to . . . obviously I would say, ‘You’ve got the information sheet, did you understand it, are there any questions?’ And I think it says in the letter, if there are any queries about any of the information please ring and it has our numbers and things like that.

Recruiting staff 1601

Staff were very complimentary about the content, wording and layout of the PIS. One interviewer said that an ALTAR participant, who is an ex-RN and conscious of the need to make trial information clear and comprehensive, had commented specifically about how good the PIS was. One interviewee said that the PIS was not ‘any bigger or smaller than any of the other ones that we’re used to giving out’ (recruiting staff 1303). A few people agreed that it was long but did not think that this detracted from the quality:

It’s a lengthy PIS, I’ll say that, it’s 14 to 15 pages . . . But to be fair, sometimes I’ve come across PISs that are one page and it doesn’t really tell you a lot. So I think I would rather have overkill . . . I think people need to be completely understanding exactly what their involvement is. So I do like the way that it’s set out, I think it’s got everything, so that every question and every kind of possible query has been answered within that PIS. So no, I quite like the PIS.

Recruiting staff 1301

It’s long but . . . the writing is quite big. The typing is quite a big type. The font is good. Yes, I think it is fine. It’s good because it does give you chance to sit down with the patient and go through it with them properly.

Recruiting staff 1101

There was a general sense that, having read the PIS, patients understood the trial and what was required of them. Interviewees commented that, when they met face to face or talked on the telephone, they had not found patient participants to be unclear about the trial. There was only one criticism; it was felt that a step-by-step guide to submitting a sample for patients who have a urine infection would have been beneficial, and some recruiting staff mentioned another trial they had been involved in where this was provided. Until this issue was addressed, they were improvising with their own written instructions.

And I always try, when I send spare bottles out, to go through a few bullet points, and just handwrite it out and say, ‘Look, if you’ve any issues with all this or if you’ve misplaced these instructions, just please give me a ring, and we can go through it.’ It’s just about making yourself available sometimes myself, [name] and [name] who’s our clinical trials assistant we’ll go through it.

Recruiting staff 1601

Issues with trial processes

Apart from the issues mentioned in Use and views of the patient information sheet about clearer instructions for patients, one site identified another problem once the trial was up and running. RNs can be called on to work on trials in a different specialty, and this was the case for one team that participated in the ALTAR trial. This team’s main area of research was women’s health, and one interviewee did admit that the ALTAR trial had been a steep learning curve. However, the issue was with what they described as an oversight in the trial protocol (and SIV) around the need to calculate a Child–Pugh score before a patient could be randomised:

When you read the protocol . . . it didn’t quite explain what a Pugh score was. We weren’t familiar with it and what we had to do was Google it to find out. [Laugh] . . . and then we realised that as part of the Pugh score we had to have a blood sample, which wasn’t highlighted, in the protocol when we looked at it for clarification [. . .] We have a speciality in women’s health, but not necessarily gynaecology or urology, so I think the protocol has to make sure that it’s very prescriptive. You can’t just assume that that’s what everybody knows and that that’s what everybody does [. . .] when you look at what you have to do per visit, the blood sample for the Pugh score was not in that list.

Recruiting staff 1302

The team was unable to get a response from the NCTU and then discovered that their usual contact had left and the site staff had not been informed:

We would appreciate a bit prompt communication from the site team which we might now get because we appreciate now that [name] has left. That was unfortunate that nobody sent a generic e-mail out to all the sites . . . That never happened and that was a bit frustrating. Because we were sending all these e-mails and they were just not getting anywhere. I think that communication needs to be improved.

Recruiting staff 1302

None of the other interviewees highlighted any issues with the trial processes from the recruiting staff’s perspective.

Supporting ALTAR participants

It was acknowledged that trial participation could be a challenge for elderly patients and those who work and/or who have child care commitments, particularly attending the 3-monthly appointments. A number of interviewees tried to accommodate the needs of trial participants and provide flexibility and support whenever they could:

They work or they’re picking kids up . . . and I always try and say that we can be flexible enough you know to suit you. It’s not all written in tablets of stone that we need to meet on such a date, at such a time. It can be mutually convenient for us both, etc.

Recruiting staff 1601

At one site recruiting staff had identified two issues that could affect the successful delivery of the trial and ensuring that there was as little inconvenience to the participants as possible. First, patients who had to have bloods taken before randomisation could have to wait some time for the results. To reduce any inconvenience to participants, the recruiting staff had arranged for medication to be sent by courier or had delivered it personally:

A lady the other day . . . we had to do all her bloods from scratch before we could randomise her and obviously we send it to the labs as an urgent sample, but the lady could still be sitting there 2 and 3 hours later. So . . . we’ve been letting them go home and then we get the blood samples back, we randomise them, we get the prescription from the doctor, contact the lady and let her know which groups she’s been randomised to and then we drop the medication off to them.

Recruiting staff 1302

The second issue was participant understanding of what to do if they had a UTI during the trial. Recruiting staff also suspected that older participants might struggle with the process of submitting a urine sample if they had had an infection, particularly completing the paperwork. One interviewee described explaining the process to a participant, “I can see their faces just looking at, going like, ‘My goodness, this is an awful lot to remember’ ” (recruiting staff 1301). There were also concerns about older patients ensuring that the box was securely closed, ‘Even I struggle with closing it because it’s quite stiff to get it to shut’ (recruiting staff 1301). Another interviewee expressed potential health and safety risks if the sample is not secure (recruiting staff 1302). This team offered extra help, particularly for the first time a participant had to submit a sample:

We have got round people being unsure about sending back samples by reassuring them and asking them to phone us. In cases we go out to the house and we . . . show them how to pod their urine sample the first time in the hope that they’ll remember how to do it if they have a subsequent UTI [laugh] and I think that does help. And it also provides that reassurance that they’re valued and they’re not left floundering . . . and about being uncertain about how to do this appropriately.

Recruiting staff 1302

Why patients agree to or decline participation in the ALTAR trial

Recruiting staff were asked if, in their view or experience, there were any elements of the ALTAR trial that would encourage or discourage women from participating. One interviewee thought that many of the women at their site were keen to participate and are ‘desperate for something’ and ‘like that additional support that they get as part of this study’ (recruiting staff 1003). The opinion that participants were driven to participate because of the impact the rUTIs were having on their lives was reinforced by an interviewee at another site:

They’re kind of clutching at straws to try and get something to help them. Because they’ve tried everything else and now this is the last opportunity to try and see if there’s something that can help them.

Recruiting staff 1301

Other reasons that the trial was thought to be attractive to patients were that, unlike a placebo-controlled trial, everyone in the ALTAR trial received treatment, and patients are keen to have the chance to try something other than antibiotics that are in common use in the NHS:

I think they are reassured that it is used. It is not a new investigational drug, that it is used around the country. That reassures patients.

Recruiting staff 1102

One interviewee did acknowledge that as a participant progresses through the trial their views that the trial could be the answer to their problems could change, depending on their experience:

And I think, as they progress along their journey, their focus might change in the questions they ask in clinic. I suppose it depends to what degree it has made a difference in reducing the number of infections. You might see a change in their anxiety levels.

Recruiting staff 1003

The actual reasons patients gave for not participating varied. At one site the recruiting staff said that some people are too busy, or their concerns are with other health problems they have that take priority over their rUTIs. In contrast, other patients are well and do not ‘feel the need to do anything extra at the moment’ (recruiting staff 1301). Other patients did not like the idea of being part of a trial. The washout period was another factor that discouraged patients from participating, but one interviewee was surprised that this was not more of an issue:

If they’re already on treatment . . . they don’t want to not take anything for 3 months. [. . .] To be fair actually, they’ve been better than we thought. [Laughter] I thought I was going to have a big problem with that.

Recruiting staff 1303

Based on their experience with patients who have agreed to participate, another interviewee wondered if those who decline are overwhelmed with what is required of them. Others had direct experience of this being a barrier to participation:

I don’t know whether the people declining don’t want any additional hassle, additional visits, additional medication.

Recruiting staff 1302

We have had one lady already come off study, because she thought it would be a burden as well, after having initially being recruited.

Recruiting staff 1002

There is quite a lot for them to do within the trial, so that also puts patients off, but not many.

Recruiting staff 1102

Discussion

This embedded qualitative study achieved its objective of identifying modifiable issues in the ALTAR trial processes. The two main issues were participants’ understanding of the nature of the trial and the risks and benefits of participation. These were easily addressed by ensuring that recruiting staff checked patients’ understanding when they discussed the trial with them.

The majority of patient and recruiting staff interviewees were happy with the PIS, and there was praise for its structure and format. Patient interviewees were complimentary about the recruiting staff, most of whom made exceptional efforts to ensure that participants had a positive experience. The pragmatic nature of the trial, the absence of a placebo arm or experimental intervention facilitated recruitment. The periods on trial medication (12 months) and off trial medication (6 months) were acceptable to patient interviewees (although some were anxious about the latter), and this perhaps is indicative of how keen some were to find an answer to their rUTIs through the trial.

Ensuring trial participants have a positive experience is paramount, but the experiences of the recruiting staff are also important. This qualitative study highlighted some issues for the recruiting staff that were able to be addressed and will also inform the conduct of future trials, particularly where the teams are working in a specialty that is new to them.

The severity of the symptoms experienced by the patients interviewed was salient and was, for many, a motivating factor to enrol in the trial. There was concern about long-term antibiotic use, yet patients demonstrated commitment to the trial by stating that they would continue to participate even if they were randomised to the antibiotic arm. Although not a key focus of the qualitative study, a matrix of factors for trial participation (Figure 2) was developed from the patient and recruiting staff interview data and published before the final report. 33 This analysis of the conduct and processes in the ALTAR trial was beneficial to the current study and may also be utilised by researchers designing UTI trials in the future.

FIGURE 2.

Matrix of factors for trial participation (amended from Lie et al. 33). Reproduced with permission from Lie et al. 33 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

Reproduced with permission from Lie et al. 33 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

Limitations

It was unfortunate that the qualitative study was unable to achieve its target of interviewing up to 15 patients who had declined to participate in the trial and up to 15 who had dropped out. Although it was possible to identify some of the reasons that patients declined from the recruiting staff interviews and the screening log data, the opportunity to explore this with patients could have highlighted issues or misunderstandings about the trial. In hindsight, ensuring that the qualitative study was more prominent in the PIS for the trial may have been beneficial to recruitment. In addition, early contact between the interviewer and the site staff to establish rapport would have helped to raise the profile of the qualitative study. With regard to participants who dropped out of the trial, it could be that this number was small among those who took part in a first interview, but it is possible that the process of informing the interviewer of dropouts was flawed.

Conclusions

The short, embedded qualitative study conducted in the early stages of trial recruitment and the process of rapid feedback to the trial team meant that we were able to identify modifiable factors in trial processes. In future, where such embedded studies are planned, processes to enhance recruitment, particularly of those who decline participation or drop out of the trial, should be carefully thought through, possibly with the input of recruiting staff at sites. Finally, seeking the views of recruiting staff is crucial, and their agreement to participate in interviews could form part of the early scoping of sites to be part of future trials.

Recruitment

The ALTAR trial opened to recruitment on 23 June 2016, with the first participant recruited on 7 July 2016. In total, 240 participants were randomised from eight centres across the UK. The number of sites was increased from six to eight in June 2017 to aid participant recruitment. The last participant was recruited on 20 June 2018, with the final 18-month follow-up visit taking place on 27 January 2020. The recruitment period was originally planned to be 18 months; however, this was extended by 9 months in order to achieve the target sample size (Figure 3).

FIGURE 3.

Observed and predicted accrual.

In total, 480 patients were screened for eligibility. Reasons for non-participation are shown in Table 6. Out of those not recruited into the trial (n = 240), 151 (63%) patients decided not to participate after receiving further information; the main reasons for declining participation were unwillingness to adhere to the 18-month trial protocol (34; 14%) and unwillingness to undertake a 3-month washout of prophylactic treatment (22; 9%). A further 38 (16%) patients did not respond to an invitation letter about the trial or attend an appointment to discuss trial participation. Qualitative data on the willingness of patients to participate in the trial are discussed in Chapter 3. In total, 34 (14%) patients did not meet the eligibility criteria.

Participant flow through the trial

Participant flow through the trial is presented in a Consolidated Standards of Reporting Trials (CONSORT) flow diagram (Figure 4). Participants were randomised (1 : 1) to receive 12 months of antibiotic prophylaxis (n = 120) or the urinary antiseptic methenamine hippurate (n = 120).

FIGURE 4.

Participant flow through the trial (CONSORT flow diagram). Reproduced with permission from Harding et al. 36 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

Reproduced with permission from Harding et al. 36 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

By 12 months, 172 participants remained in active follow-up (83 antibiotic prophylaxis, 89 methenamine hippurate). The primary analysis was conducted in the mITT population, which included all participants with at least 6 months’ follow-up, including data available from review of health-care records. In total, 205 (85%) participants (102 antibiotic prophylaxis, 103 methenamine hippurate) were evaluable for the primary analysis in the mITT population. This surpassed the 87 participants per group required by the sample size calculation.

Over the 18-month trial period, 79 (33%) participants withdrew from the trial or became lost to follow-up: 46 (38%) who were allocated to antibiotic prophylaxis and 33 (28%) who were allocated to methenamine hippurate. Of those, 32 (41%) participants had primary outcome data available from routine health-care records (21 in the antibiotic arm and 11 in the methenamine hippurate arm) (see Appendix 2, Table 27). Reasons for withdrawal, where available, are provided in Appendix 2, Table 28.

Baseline data

Participant demographics and clinical characteristics were generally well balanced across treatment groups (Table 7).

The majority of participants [210 (88%) out of 240] self-reported four or more UTI episodes in the 12 months prior to trial entry, with a median of six episodes [interquartile range (IQR) 4–8 episodes]. The number of positive urine cultures reported in the 12 months prior to trial entry was slightly higher in those allocated to methenamine hippurate (median 3, IQR 1–5 positive urine cultures) than in those allocated to antibiotic prophylaxis (median 2, IQR 1–4 positive urine cultures).

All participants with a previous history of taking antibiotic prophylaxis had to complete a 3-month washout period, without daily antibiotics, prior to randomisation. Overall, 55 (23%) participants had taken antibiotic prophylaxis previously as a preventative treatment for UTI. The completion of the advised washout period was required for only 32 (13%) participants prior to randomisation.

Treatment compliance

Participants were randomised (1 : 1) to receive daily antibiotic (nitrofurantoin, trimethoprim or cefalexin) or twice-daily antiseptic (methenamine hippurate) for 12 months. For those allocated to antibiotic prophylaxis, the majority of participants, 66 (55%), were initially prescribed nitrofurantoin, with 30 (25%) and 24 (20%) receiving trimethoprim and cefalexin, respectively.

Overall, 170 (71%) participants achieved at least 90% compliance with any trial preventative treatment. This was similar in both randomised treatment groups: 84 (70%) antibiotic prophylaxis and 86 (72%) methenamine hippurate.

Switching between the three trial prophylactic antibiotic agents and between treatment strategies (antibiotic or antiseptic) because of AEs or lack of efficacy was permitted as part of the protocol; however, no participant was permitted to use two preventative agents concurrently. In total, 39 (16%) participants received more than one preventative agent during the 12-month trial treatment period, 17 (14%) of whom were allocated to antibiotic prophylaxis and 22 (18%) of whom were allocated to methenamine hippurate. Switching between preventative treatment strategies was more common in those randomised to methenamine hippurate, with 22 (18%) participants switching to antibiotic prophylaxis, compared with seven (6%) participants switching from antibiotic prophylaxis to methenamine hippurate. Switching between preventative treatment strategies occurred slightly earlier in those allocated to methenamine hippurate (median 2.8 months post randomisation) than in those allocated to antibiotic prophylaxis (median 4.1 months), which may be attributed to those allocated to antibiotic prophylaxis first switching between antibiotic agents. A summary of compliance with preventative treatment is given in Table 8.

Numbers analysed

The primary mITT analysis included 205 (85%) randomised participants who had at least 6 months of follow-up data available, including data from routine health-care records. Pre-planned sensitivity analyses were also performed in a strict ITT population, comprising all 240 randomised participants, and a per-protocol population, comprising 170 (71%) participants achieving at least 90% compliance with any trial preventative treatment. A further post hoc, unplanned sensitivity analysis was also performed in a strict per-protocol population, including only those achieving 90% compliance with their allocated preventative treatment strategy (n = 153; 64%).

Analyses in the 6-month follow-up period (i.e. at 12–18 months) were conducted in all participants in active follow-up at 15 or 18 months, plus those with data available from health-care record review.

For all analyses, with the exception of AE data, participants were analysed according to their randomised treatment assignment. A summary of participants included in each analysis population is given in Table 9.

Clinical outcomes

This section reports the results of the clinical outcome measures. In all analyses of the primary outcome, methenamine hippurate was not found to be inferior to the standard daily low-dose prophylactic antibiotics. Results pertaining to cost-effectiveness outcomes are included in Chapter 5.

Secondary outcomes

Symptomatic urinary tract infection in the 6-month follow-up period (12–18 months)

The frequency of symptomatic UTI episodes in the 6-month follow-up period is shown in Appendix 2, Figure 18. The frequency and incidence of symptomatic UTI over the 6-month follow-up period are summarised in Table 12. The incidence of symptomatic UTI episodes in the 6-month follow-up period was higher in both groups than in the 12-month treatment period, but did not return to anywhere close to baseline levels.

TABLE 12 - Incidence of symptomatic UTI episodes in the 6-month follow-up period

Frequency and incidence of UTI	Antibiotic prophylaxis arm (N = 97)	Methenamine hippurate arm (N = 98)
Episodes of symptomatic UTI, n (%)
0	55 (57)	49 (50)
1	31 (32)	27 (28)
2	8 (8)	13 (13)
3	2 (2)	6 (6)
4	1 (1)	2 (2)
5	0 (0)	1 (1)
Mean (SD)	0.59 (0.81)	0.86 (1.10)
Median (IQR)	0 (0–1)	0.5 (0–1)
Incidence rate
Total number of UTI episodes/total follow-up time (years)	57/47.80	84/48.71
Incidence rate (95% CI)	1.19 (0.86 to 1.52)	1.72 (1.27 to 2.18)
Difference (95% CI)	0.53 (–0.03 to 1.09)
Unadjusted IRR (95% CI)	1.45 (0.99 to 2.10)
Adjusted IRR (95% CI)	1.45 (1.16 to 1.81)
Incidence density rate
Total number of UTI episodes/total follow-up time (years)	57/46.25	84/46.59
Incidence rate (95% CI)	1.23 (0.88 to 1.59)	1.80 (1.31 to 2.29)
Difference (95% CI)	0.57 (–0.04 to 1.18)
Unadjusted IRR (95% CI)	1.46 (0.99 to 2.16)
Adjusted IRR (95% CI)	1.47 (1.14 to 1.89)
At least one episode of symptomatic UTI
≥ 1 episode, n (%)	42 (43)	49 (50)
Unadjusted OR (95% CI)	1.31 (0.75 to 2.30)
Adjusted OR (95% CI)	1.32 (0.73 to 2.39)

The incidence rate was 1.19 (0.86–1.52) in those allocated to antibiotic prophylaxis and 1.72 (95% CI 1.27 to 2.18) in those allocated to methenamine hippurate, giving an absolute difference between incidence rates of 0.53 episodes per person-year (95% CI –0.03 to 1.09) in favour of antibiotic prophylaxis. The unadjusted IRR comparing methenamine hippurate with antibiotic prophylaxis was calculated as 1.45 (95% CI 0.99 to 2.10) and the adjusted IRR was estimated as 1.45 (95% CI 1.16 to 1.81).

Over the 6-month follow-up period, 42 (43%) participants in the antibiotic prophylaxis arm and 49 (50%) participants in the methenamine hippurate arm reported at least one episode of symptomatic UTI (unadjusted OR 1.31, 95% CI 0.75 to 2.30). Over half of all trial participants remained UTI free in the 6-month period following completion of preventative treatment.

A post hoc, unplanned analysis comparing the number of participants reporting two or more UTI episodes during the 6-month follow-up period (a definition of rUTI) showed that 11 of 97 (11%) participants in the antibiotic prophylaxis arm, compared with 22 of 98 (22%) participants in the methenamine hippurate arm (chi-squared p-value = 0.039), returned to suffering UTIs that could be classified as recurrent.

Microbiologically proven symptomatic urinary tract infections

In the mITT population, a total of 231 symptomatic UTI episodes were reported during the 12-month treatment period (90 in the antibiotic prophylaxis arm, 141 in the methenamine hippurate arm). Of these, 183 (79%) had a urine sample sent to the central or local laboratory at the time of their UTI episode: 72 (80%) allocated to antibiotic prophylaxis and 111 (79%) allocated to methenamine hippurate. Only 96 (42%) of the 231 symptomatic UTI episodes were confirmed with a positive urine culture: 42 of 90 (47%) in the antibiotic prophylaxis arm and 54 of 141 (38%) in the methenamine hippurate arm.

Consequently, incidence rates for microbiologically proven symptomatic UTI episodes were lower across both arms than at the primary end-point analysis: 0.41 (95% CI 0.27 to 0.56) in the antibiotic prophylaxis arm and 0.53 (95% CI 0.34 to 0.72) in the methenamine hippurate arm (Table 13). The absolute difference between incidence rates showed that those allocated to methenamine hippurate had an excess of 0.11 microbiologically proven UTI episodes per person-year (95% CI –0.12 to 0.35 microbiologically proven UTI episodes per person-year) compared with those allocated to antibiotic prophylaxis. The unadjusted IRR comparing methenamine hippurate with antibiotic prophylaxis was found to be 1.28 (95% CI 0.78 to 2.09). After adjusting for baseline stratification factors, the IRR was estimated as 1.28 (95% CI 1.05 to 1.49). Analyses of the incidence density ratio, excluding time on treatment antibiotics for UTI, gave similar results.

TABLE 13 - Incidence of microbiologically proven symptomatic UTI episodes during the 12-month treatment period (mITT population) and 6-month follow-up period

Frequency and incidence of UTI	12-month treatment period		6-month follow-up period
	Antibiotic prophylaxis arm (N = 102)	Methenamine hippurate arm (N = 103)	Antibiotic prophylaxis arm (N = 97)	Methenamine hippurate arm (N = 98)
Episodes of symptomatic microbiologically confirmed UTI, n (%)
0	73 (72)	70 (68)	76 (78)	65 (66)
1	18 (18)	21 (20)	19 (20)	25 (26)
2	9 (9)	6 (6)	2 (2)	7 (7)
3	2 (2)	3 (3)	0 (0)	1 (1)
4	0 (0)	3 (3)	0 (0)	0 (0)
Mean (SD)	0.41 (0.74)	0.52 (0.95)	0.24 (0.47)	0.43 (0.67)
Incidence rate
Total number of UTI episodes/total follow-up time (years)	42/101.32	54/102.03	23/47.80	42/48.71
Incidence rate (95% CI)	0.41 (0.27 to 0.56)	0.53 (0.34 to 0.72)	0.48 (0.28 to 0.68)	0.86 (0.59 to 1.14)
Difference (95% CI)	0.11 (–0.12 to 0.35)		0.38 (0.04 to 0.72)
Unadjusted IRR (95% CI)	1.28 (0.78 to 2.09)		1.79 (1.08 to 2.96)
Adjusted IRR (95% CI)	1.25 (1.05 to 1.49)		1.86 (1.27 to 2.73)
Incidence density rate
Total number of UTI episodes/total follow-up time (years)	42/99.23	54/98.59	23/46.25	42/46.59
Incidence rate (95% CI)	0.42 (0.27 to 0.58)	0.55 (0.35 to 0.75)	0.50 (0.29 to 0.70)	0.90 (0.61 to 1.20)
Difference (95% CI)	0.12 (–0.13 to 0.37)		0.40 (0.05 to 0.76)
Unadjusted IRR (95% CI)	1.29 (0.79 to 2.13)		1.81 (1.09 to 3.02)
Adjusted IRR (95% CI)	1.29 (1.06 to 1.57)		1.89 (1.28 to 2.78)
At least one episode
≥ 1 episode, n (%)	29 (28)	33 (32)	21 (22)	33 (34)
Unadjusted OR (95% CI)	1.19 (0.65 to 2.16)		1.84 (0.97 to 3.48)
Adjusted OR (95% CI)	1.19 (0.65 to 2.17)		2.00 (1.03 to 3.91)

Analysis in the 6-month follow-up period (12–18 months) showed that 93 (66%) of 141 symptomatic UTI episodes were associated with a urine sample sent to the local or central laboratory: 37 (65%) in the antibiotic prophylaxis arm and 56 (67%) in the methenamine hippurate arm. Only 65 of 141 (46%) symptomatic UTI episodes were confirmed with a positive urine culture: 23 (40%) of 57 in the antibiotic prophylaxis arm and 42 (50%) of 84 in the methenamine hippurate arm. Incidence rates were found to be 0.48 (95% CI 0.28 to 0.68) in the antibiotic prophylaxis arm and 0.86 (95% CI 0.59 to 1.14) in the methenamine hippurate arm, giving a difference between incidence rates of 0.38 (95% CI 0.04 to 0.72) episodes per person-year (see Table 13).

Antibiotic use

During the 12-month treatment period, a total of 22 (18%) participants allocated to methenamine hippurate received prophylactic antibiotics (see Appendix 2, Table 30). During the 12-month treatment period the average proportion of time participants received prophylactic antibiotic treatment was 91% in the antibiotic prophylaxis arm and 12% in the methenamine hippurate arm. Therapeutic antibiotics for UTI were received by 51 (43%) participants allocated to antibiotic prophylaxis and 67 (56%) participants allocated to methenamine hippurate during the 12-month treatment period. The median total days of therapeutic antibiotic treatment was 13 days for those allocated to antibiotic prophylaxis and 16 days for those allocated to methenamine hippurate. Therapeutic antibiotics were prescribed for other reasons to 32 (27%) participants allocated to antibiotic prophylaxis and to 38 (32%) participants allocated to methenamine hippurate.

Only 13 participants were reported to have received prophylactic antibiotic treatment during the 6-month follow-up period (eight randomised to antibiotic prophylaxis and five to methenamine hippurate). During the 6-month follow-up period, therapeutic antibiotics for UTI were received by 48 (49%) and 52 (53%) participants allocated to antibiotic prophylaxis and methenamine hippurate, respectively; the median total days of treatment was 7.5 and 13.5 days, respectively. Therapeutic antibiotics prescribed for other reasons were taken by 15 (15%) participants allocated to antibiotic prophylaxis and 28 (29%) participants allocated to methenamine hippurate during the 6-month follow-up period.

Antimicrobial resistance

Perineal swabs

The availability of perineal swabs at each trial time point is shown in Table 14. In total, 89 participants (37%) had provided a sample at all time points (45 participants in the antibiotic prophylaxis arm and 44 participants in the methenamine hippurate arm) and 37 (15%) participants had E. coli isolated in all four samples (19 participants in the antibiotic prophylaxis arm and 18 participants in the methenamine hippurate arm).

TABLE 14 - Antibiotic resistance in E. coli isolated from routine perineal swabs

	Antibiotic prophylaxis arm (N = 120)				Methenamine hippurate arm (N = 120)
	Baseline	Month 6	Month 12	Month 18	Baseline	Month 6	Month 12	Month 18
Samples available, n (%)	107 (89)	75 (63)	66 (55)	59 (49)	94 (78)	79 (66)	70 (58)	62 (52)
E. coli isolated, n (%)	76 (71)	51 (68)	43 (65)	39 (66)	64 (68)	58 (73)	47 (67)	45 (73)
Antibiotic resistance in E. coli (number of antibiotics)
0, n (%)	32 (42)	19 (37)	12 (29)	24 (62)	29 (45)	27 (47)	23 (49)	26 (58)
1 or 2, n (%)	23 (30)	21 (41)	22 (52)	12 (31)	17 (27)	18 (31)	17 (36)	10 (22)
≥ 3, n (%)	21 (28)	11 (22)	8 (19)	3 (8)	18 (28)	13 (22)	7 (15)	9 (20)
Mean (SD)	1.7 (2.1)	1.5 (1.6)	1.7 (1.8)	0.8 (1.2)	1.7 (2.2)	1.3 (1.6)	1.1 (1.6)	1.2 (1.8)
Median (IQR); range	1 (0–3); 0–8	1 (0–2); 0–6	1 (0–2); 0–7	0 (0–2); 0–4	1 (0–3); 0–9	1 (0–2); 0–6	1 (0–2); 0–6	0 (0–2); 0–8
Antibiotic resistance in E. coli (number of antimicrobial categories)
0, n (%)	32 (42)	19 (37)	12 (29)	24 (62)	29 (45)	27 (47)	23 (49)	26 (58)
1 or 2, n (%)	28 (37)	24 (47)	22 (52)	13 (33)	22 (34)	24 (41)	18 (38)	10 (22)
≥ 3 (MDR), n (%)	16 (21)	8 (16)	8 (19)	2 (5)	13 (20)	7 (12)	6 (13)	9 (20)
Mean (SD)	1.3 (1.5)	1.2 (1.2)	1.4 (1.3)	0.7 (1.0)	1.3 (1.6)	1.1 (1.2)	1.0 (1.2)	1.0 (1.5)
Median (IQR); range	1 (0–2); 0–6	1 (0–2); 0–4	1 (0–2); 0–5	0 (0–2); 0–3	1 (0–2); 0–6	1 (0–2); 0–4	1 (0–2); 0–5	0 (0–2); 0–6

The proportions of participants demonstrating resistance in E. coli to at least one antibiotic at baseline, during the preventative treatment period (at 6 or 12 months) and at 18 months are shown in Figure 5. Comparisons between groups were made using a chi-squared test. Similarly, the rates of MDR are shown in Figure 6 but with comparisons made using Fisher’s exact tests.

FIGURE 5.

Any resistance (per participant) in E. coli isolated from perineal swabs at baseline, 6 or 12 months, and at the 18-month follow-up (chi-squared p-values). Reproduced with permission from Harding et al. 36 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

Reproduced with permission from Harding et al. 36 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

FIGURE 6.

Any MDR (per participant) in E. coli isolated from perineal swabs at baseline, 6 or 12 months, and at the 18-month follow-up (Fisher’s exact p-values). Reproduced with permission from Harding et al. 36 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

Reproduced with permission from Harding et al. 36 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

There were no apparent differences between groups in the antimicrobial susceptibility patterns of the baseline perineal E. coli isolates, as judged by the number of antibiotics to which the isolate was resistant or the percentage of multidrug-resistant isolates. During the preventative treatment period, a higher proportion of participants in the antibiotic prophylaxis arm were found to have at least one E. coli isolate demonstrating resistance to at least one antibiotic (46/64; 72%) than in the methenamine hippurate arm (39/70; 56%) (chi-squared p-value = 0.052). However, at the 18-month time point a higher proportion of participants allocated to methenamine hippurate were colonised with multidrug-resistant E. coli isolates (9/45; 20%) than were participants in the antibiotic prophylaxis arm (2/39; 5%) (Fisher’s exact p-value = 0.06).

Out of 105 participants with E. coli isolated at baseline and in at least one post-baseline sample, 46 (44%) participants developed resistance to at least one different antibiotic in a post-baseline sample; 27 out of 52 (52%) participants in the antibiotic prophylaxis arm and 19 out of 53 (36%) participants in the methenamine hippurate arm (chi-squared p-value = 0.10) (see Appendix 2, Figure 19). The number of participants known to have acquired MDR in E. coli since baseline was similar in both randomised treatment arms, with 7 out of 42 (17%) participants in the antibiotic prophylaxis arm and 6 out of 42 (14%) participants in the methenamine hippurate arm (chi-squared p-value = 0.76) (see Appendix 2, Figure 20).

For eight different antibiotic agents (the three trial prophylactic antibiotic agents, cefalexin, nitrofurantoin and trimethoprim, plus the five agents with the highest resistance rates in our study), the proportions of participants demonstrating ‘known resistance since baseline’ and ‘resistance since baseline’ in perineal isolates are shown in Appendix 2, Figure 21. For each antibiotic, ‘known resistance since baseline’ was defined as participants with at least one E. coli isolate resistant to at least one of the antimicrobial agents tested in a post-baseline perineal sample, out of those who demonstrated sensitivity to the antimicrobial panel in E. coli at baseline. ‘Resistance since baseline’ was defined for each antibiotic as participants with at least one E. coli isolate resistant to at least one of the antimicrobial agents tested in a post-baseline perineal sample out of those who demonstrated sensitivity to the antimicrobial panel in E. coli at baseline or did not have E. coli isolated in their baseline sample. There was a slightly higher proportion of participants allocated to antibiotic prophylaxis showing ‘known resistance since baseline’ to a number of antibiotics (amoxicillin, cefalexin, ciprofloxacin) than there was in those allocated to the methenamine hippurate arm; however, there were no statistically significant differences between treatment arms.

For each of the eight antibiotics, resistance rates in E. coli over time are plotted by antibiotic in Appendix 2, Figure 22. The same data are also provided in Appendix 2, Table 31.

Urine samples

The availability of routine urine samples sent to the central laboratory at baseline and at 3-monthly follow-up visits (3–18 months) is shown in Table 15, along with the number of samples with significant E. coli bacteriuria. E. coli was isolated from relatively small numbers of routine urine samples collected during the 12-month treatment period, particularly from the antibiotic prophylaxis cohort.

TABLE 15 - Antibiotic resistance in E. coli isolated in significant numbers from routine urine samples

Resistant to ≥ 1 antibiotic from ≥ 3 antimicrobial categories.

	Antibiotic prophylaxis arm (N = 120), n (%)							Methenamine hippurate arm (N = 120), n (%)
	Baseline	Month 3	Month 6	Month 9	Month 12	Month 15	Month 18	Baseline	Month 3	Month 6	Month 9	Month 12	Month 15	Month 18
Samples available	111 (93)	95 (79)	81 (68)	79 (66)	74 (62)	64 (53)	62 (52)	111 (93)	96 (80)	89 (74)	81 (68)	78 (65)	71 (59)	71 (59)
E. coli isolated	15 (14)	2 (2)	4 (5)	3 (4)	5 (7)	11 (17)	7 (11)	7 (6)	8 (8)	9 (10)	5 (6)	12 (15)	8 (11)	8 (11)
Antibiotic resistance in E. coli (number of antibiotics)
0	5 (33)	1 (50)	0 (0)	0 (0)	0 (0)	3 (27)	3 (43)	2 (29)	3 (38)	4 (44)	4 (80)	6 (50)	2 (25)	5 (63)
1 or 2	7 (47)	1 (50)	2 (50)	3 (100)	4 (80)	7 (64)	2 (29)	2 (29)	3 (38)	0 (0)	0 (0)	3 (25)	4 (50)	2 (25)
≥ 3	3 (20)	0 (0)	2 (50)	0 (0)	1 (20)	1 (9)	2 (29)	3 (43)	2 (25)	5 (26)	1 (20)	3 (25)	2 (25)	1 (13)
Multidrug resistanta	3 (20)	0 (0)	1 (25)	0 (0)	0 (0)	1 (9)	1 (14)	2 (29)	2 (25)	4 (44)	1 (20)	3 (25)	2 (25)	1 (13)

In total, 124 positive urine cultures were submitted from 62 individual participants during periods of symptomatic UTI, 46 from 29 participants allocated to antibiotic prophylaxis and 78 from 33 participants allocated to methenamine hippurate. Significant E. coli bacteriuria was detected in 93 urine cultures from 51 individual participants: 36 samples from 25 participants allocated to antibiotic prophylaxis and 57 from 26 participants allocated to methenamine hippurate.

The proportions of participants demonstrating resistance in E. coli in any symptomatic urine sample submitted during the 12-month treatment and 6-month follow-up (12–18 months) periods are shown for eight different antibiotics (the three trial prophylactic antibiotic agents plus the five agents with the highest resistance rates in our study) in Figure 7. A higher proportion of participants allocated to antibiotic prophylaxis demonstrated resistance to co-trimoxazole in at least one urinary E. coli isolate during the 12-month preventative treatment period than did those allocated to methenamine hippurate: 6 out of 13 (46%) participants and 3 out of 14 (21%) participants, respectively. Similarly, six out of eight (75%) participants allocated to antibiotic prophylaxis demonstrated resistance to trimethoprim in at least one urinary E. coli isolate during the 12- to 18-month follow-up period, compared with 5 out of 13 (38%) participants allocated to methenamine hippurate. There were no significant differences in the proportion of participants showing resistance to at least one antibiotic in any urinary E. coli isolate or MDR in any E. coli isolate during the preventative treatment or follow-up periods (Figures 8 and 9).

FIGURE 7.

Resistance rates to individual antibiotics (per participant) in E. coli isolated from symptomatic urine samples in the 12-month treatment period and 6-month follow-up period. (a) Amoxicillin; (b) cefalexin; (c) cefuroxime; (d) ciprofloxacin; (e) co-amoxiclav; (f) co-trimoxazole; (g) nitrofurantoin; and (h) trimethoprim. Reproduced with permission from Harding et al. 36 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

Reproduced with permission from Harding et al. 36 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

FIGURE 8.

Any resistance (per participant) in E. coli isolated from symptomatic urine samples in the 12-month treatment period and 6-month follow-up period. Fisher’s exact p-values.

FIGURE 9.

Any MDR (per participant) in E. coli isolated from symptomatic urine samples in the 12-month treatment period and 6-month follow-up period. Fisher’s exact p-values.

Similarly, the proportions of symptomatic urine samples demonstrating resistance in E. coli during the 12-month treatment period and 6-month follow-up period are shown by antibiotic in Figure 10. Data are also presented cumulatively over time in Appendix 2, Figure 23. There was a higher proportion of samples showing resistance to cephalosporins (cefalexin and cefuroxime) and folate pathway inhibitors (trimethoprim and co-trimoxazole) in E. coli isolates in those allocated to antibiotic prophylaxis than in those allocated to methenamine hippurate; however, these did not differ significantly. There was no significant difference in the proportion of samples showing resistance to at least one antibiotic in E. coli (see Appendix 2, Figure 24) or demonstrating MDR in E. coli (see Appendix 2, Figure 25).

FIGURE 10.

Resistance rates to individual antibiotics (per sample) in E. coli isolated from symptomatic urine samples in the 12-month treatment period and 6-month follow-up period. (a) Amoxicillin; (b) cefalexin; (c) cefuroxime; (d) ciprofloxacin; (e) co-amoxiclav; (f) co-trimoxazole; (g) nitrofurantoin; and (h) trimethoprim.

For each antibiotic, the proportions of participants demonstrating ‘resistance since baseline’ (participants with at least one urinary E. coli isolate resistant to at least one of the panel of antibiotics tested in a post-baseline sample who demonstrated sensitivity in E. coli at baseline or did not have E. coli isolated in their baseline sample) and ‘known resistance since baseline’ (participants with at least one urinary E. coli isolate resistant to at least one of the panel of antibiotics tested in a post-baseline sample who demonstrated sensitivity in E. coli at baseline) are shown in Appendix 2, Figure 26. Owing to the small number of positive urine cultures at baseline, the number of participants evaluable for the ‘known resistance since baseline’ analyses was very small. There was no difference between treatment groups in the proportion of participants developing MDR since baseline in E. coli isolates from urine samples.

In addition to E. coli, other isolates were identified in significant growth from routine and symptomatic urine samples; these are tabulated in Appendix 2, Table 32. The analyses in this section have been repeated to explore antimicrobial resistance in any isolate rather than resistance only in E. coli (see Appendix 2, Table 33 and Figures 27–32).

Asymptomatic bacteriuria

The occurrence of asymptomatic bacteriuria over the 18-month trial period was similar in both randomised treatment arms (see Appendix 2, Table 34). Among those allocated to antibiotic prophylaxis, 64 of 557 (11%) routine samples submitted in the absence of symptoms were positive, compared with 79 of 571 (14%) in those allocated to methenamine hippurate (chi-squared p-value = 0.24). In a post hoc comparison of samples submitted during the trial treatment period (i.e. those submitted at the 3-, 6-, 9- and 12-month time points), the occurrence of asymptomatic bacteriuria was 7% in the antibiotic prophylaxis arm, compared with 14% in the methenamine hippurate arm (chi-squared p-value = 0.005).

Hospitalisation due to urinary tract infection and febrile urinary tract infection

Four participants were hospitalised due to UTI; all were allocated to receive methenamine hippurate. Three of the four participants were hospitalised within the 12-month treatment period and one was hospitalised in the 6-month follow-up period.

Six participants reported at least one UTI episode with a recorded fever of ≥ 38 °C. All six participants were allocated to receive methenamine hippurate. In total, 12 febrile UTI episodes were reported from those six participants, eight during the 12-month preventative treatment period and four during the 6-month follow-up period. One participant reported six febrile UTI episodes: three during the 12-month treatment period and three during the 6-month follow-up period.

Participant satisfaction with treatment

Among those completing the TSQM, overall participant satisfaction with preventative treatment was high across both randomised groups, with a mean global satisfaction score at 12 months of 80.6 (SD 22.4) in those randomised to antibiotic prophylaxis and 77.3 (SD 23.9) in those randomised to methenamine hippurate. At 12 months, participants allocated to antibiotic prophylaxis reported a higher score for convenience (mean 91.4; SD 12.7) than those allocated to methenamine hippurate (mean 82.2; SD 18.4); however, there was no evidence of a difference when asked again at 18 months. Scores were similar across treatment groups for all other domain scores (Table 16).

TABLE 16 - Treatment Satisfaction Questionnaire for Medication scores

ANCOVA, analysis of covariance.

Adjusted for baseline stratification factors of menopausal status and prior UTI frequency.

Possible values for each domain range from 0 to 100, with higher scores indicating increased satisfaction. The mean difference is methenamine hippurate minus antibiotic prophylaxis.

Domain/time point	Antibiotic prophylaxis arm (N = 120)	Methenamine hippurate arm (N = 120)	Comparison between arms
	Mean (SD), n	Mean (SD), n	Mean difference (95% CI); two-sample t-test; p-value	ANCOVA,a p-value
Month 12
Effectiveness	80.0 (22.5), 63	77.5 (23.7), 74	–2.5 (–10.4 to 5.4); 0.53	0.43
Side effects	92.9 (19.2), 62	95.8 (13.7), 72	2.9 (–2.7 to 8.5); 0.31	0.29
Convenience	91.4 (12.7), 64	82.2 (18.4), 73	–9.2 (–14.6 to –3.8); 0.001	0.001
Global satisfaction	80.6 (22.4), 64	77.3 (23.9), 73	–3.3 (–11.2 to 4.5); 0.40	0.34
Month 18
Effectiveness	74.4 (28.8), 57	75.0 (24.7), 69	0.7 (–8.8 to 10.1); 0.89	0.95
Side effects	93.2 (18.2), 55	94.9 (15.7), 70	1.7 (–4.3 to 7.7); 0.57	0.56
Convenience	85.7 (17.0), 59	84.6 (15.7), 72	–1.1 (–6.8 to 4.5); 0.69	0.70
Global satisfaction	75.8 (25.5), 60	74.7 (27.1), 72	–1.1 (–10.3 to 8.0); 0.81	0.70

Adverse events

In total, 38 SAEs were reported (23 in the antibiotic prophylaxis arm and 15 in the methenamine hippurate arm) from 28 participants (14 from each arm) (see Appendix 2, Table 35). Two SAEs were classified as being possibly, probably or definitely related to trial treatment. One participant allocated to antibiotic prophylaxis (trimethoprim) was admitted to hospital with severe abdominal pain. One further participant who was allocated to antibiotic prophylaxis reported an increase in ALT to 465 U/l on their month 9 blood test. This resolved to 25 U/l at month 12 following a change in prophylactic antibiotic treatment from nitrofurantoin to trimethoprim.

To account for switching between preventative treatment strategies, AEs are summarised according to the preventative treatment strategy in use at the time the AE started. The mean number of AEs and ARs reported over the 18-month trial period was found to be similar across both treatment groups (Table 17). This is also shown by randomised treatment group in Appendix 2, Table 36. The most commonly occurring AEs (affecting at least 3% of participants in either treatment group) are tabulated by treatment received in Table 18 and by randomised treatment group in Appendix 2, Table 37.

TABLE 17 - Adverse events and ARs by treatment received

Reproduced with permission from Harding et al. 36 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/. The table includes minor additions and formatting changes to the original table.

	Antibiotic prophylaxis arm (N = 142)	Methenamine hippurate arm (N = 127)	Total (N = 269)
All AEs
Number of events per participant
Mean (SD)	1.9 (2.8)	1.8 (2.4)	1.9 (2.7)
Median (IQR)	1 (0–2)	1 (0–3)	1 (0–3)
Minimum, maximum	0, 17	0, 13	0, 17
Worst grade reported per participant, n (%)
None	59 (42)	45 (35)	104 (39)
Mild	41 (29)	47 (37)	88 (33)
Moderate	34 (24)	29 (23)	63 (23)
Severe	8 (6)	6 (5)	14 (5)
ARs (possibly, probably, definitely)
Number of ARs per participant
Mean (SD)	0.4 (0.7)	0.4 (0.7)	0.4 (0.7)
Median (IQR)	0 (0–0)	0 (0–1)	0 (0–1)
Minimum, maximum	0, 3	0, 3	0, 3
Worst grade AR reported per participant, n (%)
None	108 (76)	92 (72)	200 (74)
Mild	24 (17)	26 (20)	50 (19)
Moderate	9 (6)	9 (7)	18 (7)
Severe	1 (1)	0 (0)	1 (< 1)

TABLE 18 - Number of participants affected by each AE (only those occurring in at least 3% of participants in either treatment group are shown), by treatment received

Event term	Antibiotic prophylaxis arm (N = 142)		Methenamine hippurate arm (N = 127)		Total (N = 269)
	n	%	n	%	n	%
Lower respiratory tract infection	10	7	9	7	19	7
Nausea	12	8	5	4	17	6
Abdominal pain	7	5	9	7	16	6
Diarrhoea	8	6	4	3	12	4
Alanine aminotransferase levels increased	5	4	5	4	10	4
Back pain	7	5	3	2	10	4
Headache	3	2	7	6	10	4
Candida infection	4	3	5	4	9	3
Dyspepsia	5	4	4	3	9	3
Rash	3	2	5	4	8	3
Abdominal discomfort	3	2	4	3	7	3
Dyspnoea	5	4	2	2	7	3
Fall	3	2	4	3	7	3
Vomiting	3	2	4	3	7	3
Depressed mood	1	1	4	3	5	2
Herpes zoster	5	4	0	0	5	2

Kidney and liver function was assessed by measuring levels of creatinine eGFR, bilirubin, alkaline phosphatase and ALT at baseline and 3-monthly to month 18. For all measurements, there was little difference in distribution between randomised treatment groups, or over time (Figure 11).

FIGURE 11.

Box plots of creatinine, eGFR and LFTs over time, by randomised treatment group. (a) Creatinine (eGFR); (b) bilirubin; (c) alkaline phosphate; and (d) ALT.

Within-trial analysis

The following outcomes were reported for the within-trial economic evaluation:

• health-care costs to the NHS of managing rUTIs

• participant costs

• mean incidence rate of rUTIs

• QALYs estimated based on responses to the EQ-5D-5L administered at baseline and at 3, 6, 9, 12, 15 and 18 months post randomisation

• regression models which estimated the predictors of costs and effects to inform the calculation of incremental costs, effects and cost-effectiveness

• incremental cost per incidence of rUTI avoided

• incremental cost per QALY gained at the end of the follow-up period (18 months post randomisation).

Results

Data on 205 participants were used in the economic analysis.

Data validity and completeness

Table 19 summarises the response rate to the participant questionnaire and EQ-5D-5L at each time point. There was a progressive decline in the response rate to these questionnaires over the trial follow-up period. The pattern of missing data was similar across both randomised arms. Only 29% of participants (n = 59) responded to the participant questionnaire at all time points and 33% (n = 67) of participants responded to the EQ-5D-5L at all time points.

TABLE 19 - Response rates

N/A, not applicable.

Scheduled time points	Participant questionnaire, n (%)		EQ-5D-5L, n (%)
	Prophylactic antibiotic arm (N = 102)	Methenamine hippurate arm (N = 103)	Prophylactic antibiotic arm (N = 102)	Methenamine hippurate arm (N = 103)
Baseline	97 (95.1)	89 (86.4)	95 (93.1)	98 (95.1)
3 months	75 (73.5)	71 (68.9)	76 (74.5)	76 (73.8)
6 months	59 (57.8)	72 (69.9)	64 (62.7)	76 (73.8)
9 months	66 (64.7)	65 (63.1)	66 (64.7)	76 (73.8)
12 months	59 (57.8)	65 (63.1)	66 (64.7)	73 (70.9)
15 months	N/A	N/A	60 (58.8)	69 (67.0)
18 months	55 (53.9)	60 (58.3)	59 (57.8)	67 (65.0)
Data at all time points	29 (28.4)	30 (29.1)	32 (31.1)	35 (34.0)

Resource use and costs

Over the 18-month follow-up period participants reported using a variety of different primary and secondary health-care services. On average, the health-care resources most frequently reported were outpatient visits, consultations with a GP at a GP practice and telephone consultations with a GP or nurse. The costs and pattern of resource use for primary and secondary health-care resources are presented in Appendix 3, Tables 38 and 39, and were similar across both randomised arms for the 18-month follow-up period. Appendix 3, Table 40, summarises the unit costs used in the estimation of average total costs.

Total NHS costs were estimated for all participants who had at least one completed participant questionnaire, excluding baseline (n = 183). Costs were assumed to be zero for missing items and missing time points. Missing data were assumed to be missing at random. There was no meaningful difference in baseline costs (mean difference £16; p = 0.8806) and QALYs (mean difference 0.04; p = 0.3024) between those with missing data and those with complete data. Table 20 summarises the average total costs for each health-care resource by randomised arm. On average, participants reported higher costs in the methenamine hippurate arm than in the antibiotic prophylaxis arm. The main difference in costs was due to the intervention medication costs. The daily cost associated with methenamine hippurate was slightly higher than the daily cost associated with the antibiotic. In addition, those in the methenamine hippurate arm received more therapeutic antibiotics to treat UTIs than those in the antibiotic prophylaxis arm.

TABLE 20 - Average total costs (£) per randomised arm

Health-care resources	Antibiotic prophylaxis arm (n = 89), mean (SD) (£)	Methenamine hippurate arm (n = 94), mean (SD) (£)
Intervention costs	89 (76)	188 (67)
Concomitant medication costs	2 (8)	1 (7)
Antibiotic costs (due to UTI)	4 (5)	8 (28)
Primary care costs	201 (265)	236 (236)
Secondary care costs	636 (1814)	580 (876)
Average total NHS costs per participant	931 (2015)	1013 (1024)
Private health-care costs	14 (123)	0 (0)
Time and travel costs	331 (657)	359 (430)
Average total NHS costs and participant costs per participant	1276 (2775)	1372 (1438)

Two participants in the antibiotic prophylaxis arm reported incurring private health-care costs related to appointments with a physiotherapist and pain consultant. Both arms incurred similar time and travel costs. Total participant costs were combined with total NHS costs to estimate the average total NHS and participant cost per participant (see Table 20) that was used in a sensitivity analysis (see Appendix 3, Table 41).

Effectiveness outcomes

Incidence of urinary tract infection

On average, the incidence of UTIs was higher in the methenamine hippurate arm than in the antibiotic prophylaxis arm, although the absolute difference did not exceed the predefined non-inferiority margin. Further details on this outcome measure can be found in Chapter 4.

Quality-adjusted life-years

On average, participants reported their health status to be 75% of full health over the 18-month follow-up period. On average, participants in the antibiotic prophylaxis arm reported a higher baseline utility, but this difference was not statistically significant (mean difference 0.063, 95% CI –0.006 to 0.134; p-value = 0.077). Only 33% of participants (n = 67) completed the EQ-5D-5L at all seven time points. Using the assumption that utility data had to be available at five time points (excluding baseline) increased the sample in the base-case analysis to 66% (n = 129).

In total, 38 participants (antibiotic prophylaxis, n = 9; methenamine hippurate, n = 29) completed EQ-5D-5L questionnaires when they experienced a UTI. These women experienced a total of 98 UTIs but there were only utility data available for 86 of these UTIs. QALYs were recalculated incorporating utilities associated with a UTI. The inclusion of these data had no effect on the overall QALY values (Table 21).

TABLE 21 - Average total utilities and QALYs per randomised arm

This is the average utility score associated with a UTI per randomised arm.

This is the average QALY value per randomised arm when the utility score associated with a UTI is incorporated into the QALY equation (see Equation 5).

	Antibiotic prophylaxis arm (N = 102)		Methenamine hippurate arm (N = 103)
	n	Mean (SD)	n	Mean (SD)
Baseline	95	0.813 (0.21)	98	0.750 (0.28)
3 months	76	0.792 (0.28)	76	0.765 (0.27)
6 months	64	0.783 (0.26)	76	0.791 (0.21)
9 months	66	0.780 (0.28)	76	0.768 (0.24)
12 months	66	0.764 (0.27)	73	0.743 (0.26)
15 months	60	0.783 (0.26)	69	0.739 (0.28)
18 months	59	0.795 (0.18)	67	0.746 (0.24)
QALYs (complete case)	32	1.202 (0.33)	35	1.141 (0.34)
QALYs	58	1.182 (0.35)	71	1.133 (0.34)
UTIs
Utility value per reported UTIa	20	0.720 (0.28)	67	0.527 (0.32)
QALYs with UTI utilitiesb	58	1.182 (0.35)	71	1.133 (0.34)

Incremental cost-effectiveness

Incremental cost per urinary tract infection avoided

The results of the cost-effectiveness analysis using UTI avoided as the outcome measure are presented in Table 22. In the unadjusted analysis, on average, antibiotic prophylaxis dominated methenamine hippurate, as it was less costly and more effective. In the adjusted analysis, our conclusions remained the same. However, in both analyses the 95% CI around the difference in costs is wide.

TABLE 22 - Incremental cost per UTI avoided

Results based on adjusted analysis, N = 166 (antibiotic prophylaxis arm, n = 84; methenamine hippurate arm, n = 82).

A negative incremental effect means, on average, the methenamine hippurate arm experienced a higher incidence of UTIs.

Strategy	Cost (SD) (£)	Incremental cost (95% CI),a (£)	Incidence of UTI (SD)	Incremental incidence of UTIs avoided (95% CI)a	ICER	Probability of each treatment strategy being considered cost-effective at different threshold values for society’s WTP for avoiding a UTI
						£0	£100	£150	£200	£250
Antibiotic prophylaxis (costs, n = 89; outcomes, n = 102)	931 (2015)		0.99 (1.13)			0.65	0.73	0.78	0.80	0.81
Methenamine hippurate (costs, n = 94; outcomes, n = 103)	1013 (1024)	83 (–414 to 580)	1.50 (1.62)	–0.64b (–1.0 to 0.2)	Antibiotic prophylaxis dominant	0.35	0.27	0.22	0.20	0.19

These deterministic results alone are not sufficient to inform decision-making; we need to consider the imprecision around costs, effects and cost-effectiveness. If society is not willing to pay to avoid a UTI, then decisions are made on cost alone. In this circumstance, antibiotic prophylaxis had a 65% probability of being considered cost-effective compared with methenamine hippurate. As society’s WTP to avoid an episode of UTI increases, the probability that antibiotic prophylaxis would be considered cost-effective increased (Figures 12 and 13).

FIGURE 12.

Cost-effectiveness plane for methenamine hippurate compared with antibiotic prophylaxis using incidence of UTIs avoided as the measure of outcome.

FIGURE 13.

Cost-effectiveness acceptability curve for methenamine hippurate compared with antibiotic prophylaxis using incidence avoided as the measure of outcome.

Incremental cost per quality-adjusted life-year gained

The results of the cost-effectiveness analysis using QALYs gained as the outcome measure is presented in Table 23. In the unadjusted analysis, antibiotic prophylaxis dominated methenamine hippurate, as it was, on average, less costly and more effective in terms of QALYs gained. However, in the adjusted analysis our conclusions changed and methenamine hippurate dominated antibiotic prophylaxis on average. However, in both circumstances the 95% CIs around the difference in costs and QALYs were wide.

TABLE 23 - Incremental cost per QALY gained

Results based on adjusted analysis, N = 121 (antibiotic prophylaxis arm, n = 57; methenamine hippurate arm, n = 64).

Strategy	Cost (SD) (£)	Incremental cost (95% CI)a (£)	QALYs (SD)	Incremental QALYs (95% CI)a	ICER	Probability of each treatment strategy being considered cost-effective at different threshold values for society’s WTP for a QALY
						£0	£10,000	£20,000	£30,000	£50,000
Antibiotic prophylaxis (costs n = 89; outcomes n = 58)	931 (2015)		1.182 (0.35)			0.49	0.37	0.35	0.34	0.33
Methenamine hippurate (costs n = 94; outcomes n = 71)	1013 (1024)	–40 (–684 to 603)	1.133 (0.35)	0.014 (–0.05 to 0.07)	Methenamine hippurate dominant	0.51	0.63	0.65	0.66	0.67

The bootstrapped results (Figures 14 and 15) illustrate the level of imprecision in our conclusions. If society was not willing to pay for a QALY, methenamine hippurate had a 51% probability of being considered cost-effective; however, this increased to 65% at the UK recommended threshold of £20,000 per QALY gained. 37

FIGURE 14.

Cost-effectiveness plane for methenamine hippurate compared with antibiotic prophylaxis using QALYs as the measure of outcome.

FIGURE 15.

Cost-effectiveness acceptability curve for methenamine hippurate compared with antibiotic prophylaxis using QALYs as the measure of outcome.

Sensitivity analysis

Urinary tract infection utility values

The results of the analysis incorporating utilities associated with UTIs in to the QALY equation are presented in Appendix 3, Table 40 and Figure 33. The result mirrors that of the primary analysis using QALYs as the outcome measure in that, on average in the unadjusted analysis, methenamine hippurate is dominated by antibiotic prophylaxis but this finding is reversed in the adjusted analysis. At current WTP thresholds for a QALY gained, methenamine hippurate has a 65% probability of being considered cost-effective.

Participant costs

The base-case analysis estimating the cost per QALY gained over 18 months was replicated to incorporate participant costs. The results of this analysis are presented in Appendix 3, Table 41 and Figure 34. The result mirrors that of the primary analysis; in the unadjusted analysis, methenamine hippurate was dominated by antibiotic prophylaxis but this reverses in the adjusted analysis. At current WTP thresholds for a QALY gained, methenamine hippurate has a 65% probability of being considered cost-effective.

Changing eligibility criteria

The results of the sensitivity analyses including participants eligible based on the per-protocol criteria are presented in Appendix 3, Table 42 and Figure 35. Again, the conclusions are similar to the base-case analysis using QALYs gained as the outcome measure.

Changing the time horizon

The results of the sensitivity analyses estimating costs and effects (incidence of UTIs and QALYs) over a 12-month time horizon are presented in Appendix 3, Tables 43 and 44 and Appendix 3, Figures 36 and 37. Our conclusions did not change in these analyses.

Antibiotic resistance

On average, when we considered the cost associated with antibiotic resistance, the antibiotic prophylaxis arm was more costly than the methenamine hippurate arm. In the cost-effectiveness analysis, methenamine hippurate was, on average, less costly and less effective at avoiding a UTI and had a 71% probability of being considered cost-effective if we were not willing to pay to avoid a UTI. As the value placed on avoiding a UTI increased so did the probability of methenamine hippurate being considered cost-effective (see Appendix 3, Table 45 and Figure 38). In the cost–utility analysis, methenamine hippurate remained the dominant strategy as it was, on average, less costly and more effective than antibiotic prophylaxis (see Appendix 3, Table 46 and Figure 39). If we were not willing to pay for an additional QALY, methenamine hippurate had a 69% probability of being considered cost-effective. Over the range of threshold values considered for an additional QALY, the probability of methenamine hippurate being considered cost-effective never exceeded 77%.

Model-based analysis

The aim of the economic model was to extrapolate the findings of the trial over the estimated lifetime of a woman suffering from rUTIs. The following outcomes were reported for the economic model:

• health-care costs to the NHS of managing UTIs over the estimated lifetime of women with rUTIs

• QALYs estimated over the estimated lifetime of women with rUTIs

• incremental cost per QALY gained over the estimated lifetime of women with rUTIs.

Methods

An economic model was used to estimate the cost-effectiveness of methenamine hippurate compared with prophylactic antibiotics beyond the 18-month trial period. The model design and parameters estimated used are described in this section.

Economic model

Model structure

A Markov model was developed using R software [heemod (Health Economics Evaluation MODelling) package] (The R Foundation for Statistical Computing, Vienna, Austria) to extrapolate the results of the trial beyond the 18-month period. 51 The model was designed to simulate the pathway of women who experience rUTIs from initial treatment to end of life. Data from the trial alongside UK-relevant data were used to define the transition probabilities (probability of moving between the health states defined in the model) and other model parameter estimates (e.g. costs and effects).

Markov models comprise Markov states (or health states) that individuals stay in for a defined period of time. 51 Each state represents a stage in the pathway of women who experience rUTIs in a 6-month cycle. The model used in the trial, as illustrated in Figure 16, consisted of four health states: asymptomatic, mild symptoms (one UTI episode), moderate symptoms (two or more UTI episodes) and dead. In the model, each woman starts by receiving one of the two trial interventions (antibiotic prophylaxis or methenamine hippurate) and is followed through the model, which describes the process of care and incidence of rUTIs linked by logical and mathematical relationships (Tables 24 and 25).

FIGURE 16.

Model structure. mr = all-cause mortality rate; p_Asy_Mil = probability of transitioning from asymptomatic to mild; p_Asy_Mod = probability of transitioning from asymptomatic to moderate; p_Mil_Asy = probability of transitioning from mild to asymptomatic; p_Mil_Mod = probability of transitioning from mild to moderate; p_Mod_Asy = probability of transitioning from moderate to asymptomatic; and p_Mod_Mil = probability of transitioning from moderate to mild. C, cycle.

TABLE 24 - Utility and cost parameters

Cycle	Health states
	Asymptomatic		Mild		Moderate		Death
	Mean (SD)	n	Mean (SD)	n	Mean (SD)	n	Mean
Costs (£)
1	256 (370)	124	659 (1430)	46	472 (616)	35	0
2	245 (268)	141	278 (249)	39	699 (1222)	25	0
3–100	145 (37)	114	233 (61)	58	298 (108)	33	0
Utilities
1	0.803 (0.24)	124	0.750 (0.27)	46	0.716 (0.237)	35	0
2	0.796 (0.22)	141	0.737 (0.256)	39	0.666 (0.298)	25	0
3–100	0.810 (0.21)	114	0.695 (0.28)	58	0.676 (0.27)	33	0

TABLE 25 - Model-based analysis comparing the difference in costs and QALYs between the two arms

Results based on probabilistic analysis run for 1000 simulations.

Strategy	Lifetime cost (£) (SD)	Incremental cost (£) (95% CI)a	Lifetime QALYs (SD)	Incremental QALYs (95% CI)a	ICER	Probability of each treatment strategy being considered cost-effective at different threshold values for society’s WTP for a QALY
						£0	£10,000	£20,000	£30,000	£50,000
Antibiotic prophylaxis	7231 (10,096)		15.24 (3.03)			0.59	0.59	0.60	0.59	0.59
Methenamine hippurate	7876 (10,468)	645 (359 to 931)	14.96 (2.95)	–0.283 (–0.35 to –0.22)	Antibiotic prophylaxis dominant	0.41	0.41	0.40	0.41	0.41

A model time horizon of 50 years was used, which was broken down into 6-monthly Markov cycles, the length of which have to be relevant to the condition being considered. The main reason for assuming a 6-month cycle was that the post-treatment follow-up period of the trial was 6 months. By using a 6-monthly cycle we could use the data from the trial to estimate the costs, effects and probability of women experiencing UTIs when they were no longer receiving the intervention medication on a regular basis. At the end of each 6-month cycle, women can either remain in the state in which they started the cycle or move to a different state. The rate at which women moved between states was dependent on the transition probabilities. An absorbing state, which is technically impossible to move out of, needs to be included in every model. 51 The death state was included as our absorbing state. The transition probability of a woman entering the death state was informed using UK all-cause mortality rates, given that there was no probability of mortality associated with experiencing a UTI cycle based on the trial data. The model was run for 100 cycles as it was assumed that after this time all of the cohort would be in the death state. A cost and utility weight were assigned to each of the health states. A total cost per participant was estimated depending on how long each individual spent in each health state. Similarly, QALYs were estimated for each woman by summing the utility values associated with the length of time they were in each health state and their length of time in the model.

Within the model, the mean age of the modelled cohort matches that of the trial participants at baseline [mean: 50 (SD 18.6) years].

Model assumptions

The following assumptions were made when developing the model:

• Similarly to the within-trial analysis, the model-based analysis took the perspective of the UK NHS and personal and social services.

• For extrapolation, the cost and utility values (by number of UTI episodes and treatment allocation) incurred beyond the 18-month trial period were assumed to be the same as those incurred in the last 6 months of the trial follow-up period (12–18 months).

• All model parameters were defined as statistical distributions in the model.

• Ranges and distributional assumptions for input parameters were based on the trial data. Gamma distributions were assumed for costs and beta distributions for utility data. Further details on parameter distributions are given in Model parameters.

• Costs and QALYs incurred after 1 year (two cycles) were discounted at the recommended UK rate of 3.5%.

• Cost and utility parameters for each state were homogeneous between the treatment arms, that is, for the same health state, the same values were used irrespective of the treatment received.

• All individuals begin the model in the moderate state, as the inclusion criterion for the trial was individuals who have rUTIs (three episodes of infection in a 12-month period).

• The absorbing state was assumed to be costless and have a utility value of zero.

Model parameters

The model parameters were informed using data from the trial and the distribution of each parameter was defined using the mean, standard error and shape of the distribution. Model parameters included NHS costs, utilities based on responses to the EQ-5D-5L, probabilities of a single or of multiple UTI episodes, and probability of death. Gamma distributions were used for cost parameters, beta distributions were used for utilities and the Dirichlet distribution was used for transition probabilities. Table 24 and Appendix 3, Table 47, describe the model parameters.

Costs

Costs were assigned to each stage in the model, reflecting the NHS costs incurred for each 6-month cycle. The costs considered were those associated with the intervention medications (methenamine hippurate and antibiotic prophylaxis) for the first two cycles only, health-care resource use and concomitant medications reported by those receiving each intervention medication during their time in the trial, and additional antibiotics received to treat UTIs. Regression techniques were applied to the data on total costs per woman obtained from the trial to identify potential differences in costs between each health state. The results of the regression analysis were used to identify the main drivers of costs for women who experience rUTIs over their lifetime. An ordinal least squares regression model was applied to estimate the difference in average total cost between each health state (Equation 6):

⁽⁶⁾ F ( NHS costs ) = β 0 + β 1 S 1 + β 2 S 2 + ê ,

in which β = cost estimate, S₁ = mild health state, S₂ = moderate health state and ê = error term.

In Equation 6, we estimated the difference in health-care costs between the asymptomatic, mild and moderate states. As there are three states (not including the death state), two dummy variables for the mild and moderate states (S₁ and S₂) were included in the model. The cost estimate for the asymptomatic state was derived from the intercept (β₀) in the regression model. Beta coefficients in the equation provide estimates of average total cost in each health state. State health-care costs were assumed to be time variant and, therefore, the regression was run for each cycle in the model. Cost values are presented in Table 24.

Utilities

Utility values were attached to each of the health states in the model, which were used to estimate QALYs. A utility value of zero was assigned to the death health state. The mean QALYs for each arm were estimated by multiplying the length of time in each health state by the health state utility value associated with that health state. The estimated utility data used in the model were based on the utility values estimated in the within-trial analysis, which were based on responses to the EQ-5D-5L.

Similarly to how the cost parameters were estimated, an OLS regression was used to estimate potential differences in utilities between health states (Equation 7):

⁽⁷⁾ F ( utility ) = β 0 + β 1 S 1 + β 2 S 2 + ê ,

in which β = utility estimate, S₁ = mild health state, S₂ = moderate health state and ê = error term.

Equation 7 describes how state utility values were derived. Dummy variables S₁ and S₂ estimated utility values for the mild and moderate states, while the intercept estimated a utility value for the asymptomatic state. State utility values were assumed to be time variant and, therefore, the regression was run for each cycle in the model. Utility values are presented in Table 24.

Transition probabilities

The structure of the model allowed for a cohort of women with rUTIs to enter the model and followed their pathway over time to eventual death. During each 6-month model cycle, a portion of the cohort entered each of the health states based on the probability of experiencing a single episode or multiple episodes of UTIs (trial data) or dying. Transition probabilities are presented in Appendix 3, Table 47.

Model validation

The model was validated by checking the model structure, calculations and data inputs. 52 To further establish model validity, the model was run for three cycles (18 months) to replicate the results of the within-trial analysis (see Appendix 3, Table 48 and Figure 42). Comparing the results for incremental cost per QALY gained between the within-trial and model-based analyses allowed us to verify the accuracy of the model and identify any potential issues in the parameters used to populate the model.

Sensitivity analysis

Probabilistic sensitivity analysis (PSA) was used to quantify any uncertainty in the results due to uncertainty in the parameters of the model. The PSA defined each parameter in the model as a statistical distribution; distributional assumptions were based on trial data. In the PSA, costs and effects were estimated using a set of parameters drawn using a Monte Carlo simulation from each specified distribution. This random sampling was repeated for 1000 iterations, the results of which were plotted onto the cost-effectiveness plane.

Statement and interpretation of results

The results of this randomised trial, which employed a pragmatic methodology in a routine NHS setting, have clearly demonstrated that, for women with recurrent uncomplicated UTIs, a non-antibiotic preventative treatment strategy in the form of methenamine hippurate is not inferior to the current standard of care of daily low-dose prophylactic antibiotics. Uncomplicated UTIs refers to UTIs occurring in patients with no structural or functional urinary tract abnormalities that could be contributory to their infective episodes. The observed numerical difference in clinically diagnosed UTI between the two arms, which favoured antibiotic prophylaxis, did not exceed the strict predefined non-inferiority margin of one UTI episode per year. This finding was consistent across the entire range of conducted analyses:

• Modified ITT analysis – this was the primary analysis and included all patients with at least 6 months of follow-up data, analysed according to their original treatment allocation.

• Strict ITT analysis – this included all patients who were randomised, analysed according to their original treatment allocation.

• Per-protocol analysis – this included all patients with at least 6 months of follow-up data who achieved ≥ 90% compliance with any trial preventative treatment, analysed according to their original treatment allocation.

• Post hoc modified per-protocol analysis – this included only those patients who achieved ≥ 90% compliance with their original allocated treatment, excluding those who changed treatment arm during the trial.

Similar substantial reductions in the primary outcome (rate of clinically diagnosed UTIs) were noted in both trial arms during the 12-month treatment period. These results confirm that both prophylactic daily antibiotics and twice-daily methenamine hippurate are very effective treatments for female patients with recurrent uncomplicated UTI. Overall, during the treatment phase of the trial, the mean number of antibiotic-treated UTIs decreased from a mean of 6.9 to a mean of 1.1 episodes per patient per year, equating to a RR of 0.16 (or an overall risk reduction of around 84%). This level of improvement is very much in line with previously published Cochrane systematic reviews of both trial treatments. 5,14 The review of prophylactic antibiotic treatment describes a RR of UTI of 0.21, whereas a similar review of methenamine hippurate describes a RR of uncomplicated UTI of 0.24. 5,14 Underlining the considerable beneficial effect of both treatments in this trial is the finding that around half of all participants were UTI free during the entire treatment period (UTI-free rates of 43% in the methenamine hippurate arm and 54% in the antibiotic prophylaxis arm).

Using an alternative (urine culture-based) definition of microbiologically confirmed UTI revealed an equivalent proportional change in UTI episodes. However, the use of a microbiological diagnosis of UTI alone would have resulted in failure to capture over half (58%) of all patient-reported, antibiotic-treated UTI episodes. Similarly, during the 6-month follow-up period, only 46% of all patient-reported, antibiotic-treated UTIs were confirmed by positive results from a microbiological culture. This supports previous findings from another recent RCT45 with a primary outcome of clinical (rather than microbiological) UTI, which reports that only 58% of patient-reported symptomatic UTIs were confirmed by a positive urine culture. The finding of such poor correlation between clinical and microbiological culture-based diagnoses of UTI suggests that the use of a clinical definition of UTI allows for greater detection and is better aligned with actual antibiotic-treated UTI episodes. Furthermore, the diagnostic accuracy of MSU samples for acute UTI was recently critically evaluated in a case series of 92 patients, in which the authors concluded that ‘routine MSU culture, adopting the UK interpretation criteria tailored to acute UTI, failed to detect a variety of bacterial species, including recognised uropathogens’. 53 This information will be useful to researchers in this topic area in the future.

Infections, when they occurred in this trial, were almost always not clinically severe, and only four (1.7%) participants (all allocated to the methenamine hippurate arm) required hospitalisation for UTI throughout the entire 18-month trial duration. Overall, the AE rate was low in both arms, as would be expected from a comparative trial involving two licensed treatments for the prevention of rUTI. Only two serious AEs (abdominal pain requiring hospital admission and severe derangement in LFTs) were classified as possibly or probably related to trial medication, and both occurred in the antibiotic arm. All other AEs occurred with a frequency of ≤ 8%. Kidney and liver function were assessed by regular blood tests and, overall, were largely unchanged during the 12 months of treatment in both arms.

In the 6-month follow-up period, the incidence rate of clinical UTI was seen to increase slightly but did not return to anywhere near pre-treatment levels. Over half of the patients (50% in the methenamine hippurate arm and 56% in the antibiotic prophylaxis arm) were UTI free during the entire follow-up period. This is in contrast to the findings from a Cochrane review5 on the subject of antibiotic prophylaxis that demonstrated that the relative risk of having one microbiologically diagnosed UTI after completion of prophylaxis was 0.82. This may be a function of the longer duration of prophylactic treatment in this trial (12 months) compared with those studies included in the Cochrane review, in which the majority of participants used antibiotic prophylaxis for 6 months.

Equivalent rates of overall treatment satisfaction were observed in those taking methenamine hippurate and daily prophylactic antibiotics at both 12 and 18 months. Both arms scored highly for treatment satisfaction, with the only significant difference reported in the convenience domain. Those allocated to once-daily antibiotics found the treatment more convenient than those allocated to methenamine hippurate, which had to be taken twice daily.

Antibiotic use was a secondary outcome in this trial, and the use of methenamine hippurate as a preventative treatment against recurrent uncomplicated UTI was associated with a significant reduction in overall antibiotic use, both prophylactic and therapeutic. First, in the methenamine hippurate arm, only 18% of participants switched to prophylactic antibiotics during the trial, indicating that first-line treatment with the non-antibiotic option of methenamine hippurate would be appropriate for the vast majority of female patients with recurrent uncomplicated UTI. Second, in terms of therapeutic antibiotics, of those patients allocated to methenamine hippurate, 44% did not receive any therapeutic antibiotics during the 12-month treatment period. This is in contrast to their self-reported average incidence of seven antibiotic-treated UTIs per year in the 12 months prior to trial enrolment. This reduction in antibiotic therapy associated with the use of methenamine hippurate as a first-line preventative treatment for women with a history of recurrent uncomplicated UTIs is directly aligned with current antibiotic stewardship strategies designed to reduce antimicrobial resistance. 54

In line with these national and international strategies, the main driver for conducting this research was the exploration of relative efficacy of a non-antibiotic option for the prevention of recurrent uncomplicated UTIs in women (in comparison with the current gold standard of daily antibiotic treatment). The theoretical risk of increased antimicrobial resistance associated with antibiotic treatment was explored through an assessment of bacteria cultured from both perineal swabs and urine samples serially throughout the trial. Significant rates of resistant bacteria were noted in both trial arms, which underlines the importance of simple, conservative measures at presentation prior to the use of pharmacological prophylaxis and emphasises the need for this information to be used when counselling patients regarding treatment options. There were no statistically significant differences between groups in terms of the number of E. coli isolates demonstrating resistance to antibiotics from perineal swabs or urine samples, or in the corresponding rates of MDR from those bacteria at baseline. However, during the treatment period, a higher proportion of patients allocated to daily prophylactic antibiotics exhibited resistance to at least one antibiotic in E. coli isolates from their perineal swabs. This suggests that, despite the observed higher incidence of antibiotic-treated UTIs in the methenamine hippurate arm, the use of continuous low-dose antibiotic prophylaxis was a more significant factor in the induction of antimicrobial resistance in the most common urinary tract pathogen for this trial population. This was similar to findings from a previous RCT1 exploring antibiotic prophylaxis and associated antimicrobial resistance in a population of patients with recurrent complicated UTIs. By the end of the follow-up period, the rate of MDR in E. coli isolates from perineal swabs was higher in the methenamine hippurate arm than in the antibiotic prophylaxis arm, perhaps because of the more prolonged effect of daily antibiotics on the faecal microbiome or greater incidence of antibiotic-treated acute UTIs in the methenamine hippurate arm after cessation of preventative treatment.

Interestingly, these differences in rates of antimicrobial resistance from bacteria isolated from perineal swabs were not always seen when the bacteria isolated in significant concentrations directly from urine samples were analysed. During the treatment period, a higher proportion of participants allocated to antibiotic prophylaxis demonstrated resistance to trimethoprim and co-trimoxazole in at least one E. coli isolate from urine when compared with those allocated to methenamine hippurate. In addition, there was a higher proportion of samples showing resistance to cephalosporins (cefalexin and cefuroxime) in E. coli isolates in those allocated to antibiotic prophylaxis than in those allocated to methenamine hippurate. Antimicrobial resistance occurring in the antibiotic prophylaxis arm since baseline was not confined to the antibiotics prescribed during the trial but covered a host of other commonly used antibiotics for UTI.

The co-primary objective of this trial was to examine the relative cost-effectiveness of methenamine hippurate relative to daily prophylactic antibiotics. The economic evaluation consisted of a within-trial analysis and an economic model that extrapolated the results of the trial beyond the 18-month follow-up period. The within-trial analysis estimated the incremental cost per UTI avoided and incremental cost per QALY gained. In both unadjusted analyses and the adjusted analysis estimating the difference in UTIs avoided, methenamine hippurate was dominated by antibiotic prophylaxis because, on average, it was more costly and less effective. However, in the adjusted analysis estimating the difference in QALYs gained, methenamine hippurate dominated antibiotics, as it was, on average, less costly and more effective in terms of QALYs gained and had a 65% probability of being considered cost-effective at a £20,000 WTP threshold. The change in our unadjusted and adjusted results in this analysis was unsurprising given that those randomised to antibiotic prophylaxis reported higher utility values, on average, at baseline than did those in the methenamine hippurate arm. However, there is a lot of uncertainty in these results, as illustrated on the cost-effectiveness planes for both adjusted analyses. In the longer-term economic model, methenamine hippurate was dominated by antibiotic prophylaxis, as it was, on average, more costly and less effective; however, the probability of antibiotic prophylaxis being considered cost-effective never exceeded 60%. This is because there is considerable uncertainty in these results as the difference in costs and QALYs is being driven by the number of UTIs reported during the 6-month follow-up period, which were higher in the methenamine hippurate arm than in the antibiotic prophylaxis arm.

Strengths and limitations

This trial was conducted in line with current best practice. A central web-based randomisation system using an algorithm written by an independent statistician ensured concealment of allocation. Randomisation was stratified based on important confounders such as previous UTI frequency and menopausal status. The patient characteristics at baseline were well balanced between arms for other possible confounders such as age, weight and previous use of either of the trial treatments. The median number of self-reported infections per year prior to trial enrolment was six in trial arms, reflecting a typical, representative population of female patients attending urology/urogynaecology units with the complaint of rUTI. Although representative, both trial arms contained a mix of pre- and postmenopausal women. It is highly likely that different aetiological factors are responsible for causing rUTIs in pre- compared with postmenopausal women. This trial was therefore unable to define whether the two tested preventative treatment strategies were more beneficial in specific patient groups.

The pragmatic nature of the trial did not allow for the blinding of participants or responsible clinicians to treatment allocation. A computerised adjudication process for attribution of primary outcome was included to avoid observer bias. The primary outcome of clinical UTI rather than reliance on a microbiological diagnosis purposefully aimed to ensure widespread capture of infection episodes. This reflects clinical practice accurately, given that most UTIs are treated prior to knowledge of microbiological culture results. Similar proportional reductions in both clinically and microbiologically confirmed UTIs highlighted the avoidance of detection bias. The trial’s data collection tools allowed for multiple different opportunities to capture the primary outcome. The design of a non-blinded comparative trial of two licensed preventative treatments for rUTI aimed to reduce the risk of differential reporting of outcomes between trial arms, which may occur in a non-blinded placebo-controlled trial. The allocation of primary outcome was also verified using medical records, and a representative sample of cases was given to an independent clinician who was blinded to treatment arm. The independent clinician agreed with the allocation of primary outcome in 100% of cases.

The three antibiotic options included in this trial are the most commonly used prophylactic antibiotic agents in routine clinical practice, and the inclusion of antibiotic choice in this trial reflects the significant prevalence of antimicrobial resistance in patients with rUTI. This degree of choice did, however, hamper the ability to analyse individual antibiotic agents according to benefit. Although the pragmatic nature of this trial allowed for crossover between treatment arms, a range of analyses were performed to ensure that key trial findings were valid. In addition, the compliance of participants with their allocated treatment was regularly assessed by research staff and was recorded in trial documentation. The results showed that the vast majority of participants were compliant with the allocated treatment more than 90% of the time.

An early within-trial qualitative study provided important insight into the recruitment and retention process. The opinions of trial participants, patients declining participation and clinical trial staff were very valuable both in terms of informing the current trial and highlighting important aspects for future UTI trial design.

The trial was conducted according to a pre-published protocol and all outcomes were clearly stated within that publication. 1 A realistic attrition rate was incorporated into the sample size calculation, and the number of withdrawals was in line with predicted rates. Despite the expected significant number of patients dropping out of the trial, which was largely due to an extended 18-month trial period/protocol, every effort was made to capture relevant data from participants after withdrawal. All pre-set thresholds regarding numbers of participants contributing to the primary analysis were met.

The early embedded qualitative study (see Chapter 3) was an important aspect of the trial that highlighted both facilitators of and barriers to trial recruitment and retention, and contributed to the overall success with regards to these fundamental metrics. One of the weaknesses of this qualitative work, however, was the poor recruitment of patients who had declined participation or had withdrawn to the qualitative work package included in the trial.

A major strength of the trial was the robust economic evaluation that was undertaken alongside the trial. The economic analysis incorporated changes in clinical benefits and also changes in quality of life (QoL) based on responses to the EQ-5D-5L, which is arguably more useful for decision-makers. The inclusion of both outcome measures illustrates the importance of looking at clinical and QoL outcomes in economic evaluations, because our conclusions changed depending on the outcome measure chosen; antibiotic prophylaxis, on average, led to a greater reduction in the number of UTIs reported, whereas methenamine hippurate, on average, gained more QALYs (albeit neither difference was statistically significant). Another strength of the economic evaluation was that the trial data were extrapolated over the expected lifetime of participants to estimate the lifetime costs and effects associated with antibiotic prophylaxis and methenamine hippurate.

One of the main challenges of the economic evaluation was the progressive loss of data over the 18-month follow-up period. However, assumptions were made to maximise the data available, which increased the number of participants who could be included in the economic evaluation. Another limitation of the economic analysis was that the perspective chosen in the base-case analysis did not consider the wider economic costs associated with antibiotic resistance. To address this limitation, as part of a sensitivity analysis we estimated the average total cost of antibiotic resistance for each antibiotic prescription and incorporated it into our within-trial results. Incorporating a conservative estimate of antibiotic resistance into the analysis strengthened our conclusions in that methenamine hippurate had a higher probability of being considered cost-effective in the management of rUTIs in both the cost-effectiveness and cost-utility analyses than antibiotic prophylaxis. However, given the crude nature of these cost estimates, it is important that they are interpreted with caution. Although widening our perspective to incorporate the costs associated with antibiotic resistance is important to policy-makers, further research is required to derive robust estimates that can be routinely incorporated into economic evaluations of antibiotic interventions. 55

There were some limitations associated with the economic model. First, the parameters of the model were informed using the trial data, so conservative assumptions had to be made for missing data (e.g. if a participant had partial cost data, missing data were assumed to be zero). Second, few participants were able to provide information to populate the parameters for the moderate model state, which meant that only the randomised variable and model states could be used as covariates in the regression models. 55 Third, the difficulty in estimating the economic impact of antimicrobial resistance meant that these costs were not considered in the model; this is a recognised limitation. Finally, the main difference in costs and effects over the longer term was driven by the difference in rUTIs experienced in the 6-month follow-up period of the trial. The uncertainty caused by the assumptions used to inform the model are illustrated on the cost-effectiveness plane, which shows that the difference in costs and effects over the longer term is uncertain as the spread of iteration is very wide.

To our knowledge, this is the first economic evaluation investigating the effects of urinary antiseptics compared with antibiotics in the management of rUTIs. Although the response rate to the data collection tools used in the economic evaluation was lower than anticipated, this was not surprising, as a similar trend was seen in the AnTIC trial,16 which had 3-monthly data collection points but a shorter trial period (i.e. 12 months). 45 The AnTIC trial16 was a previous UK RCT examining once-daily prophylactic antibiotic treatment for patients with rUTI who perform intermittent bladder catheterisation. Future studies evaluating UTI interventions should be considered while bearing in mind that, although more frequent data collection points are required to capture changes in costs and QoL (owing to the acute nature of rUTIs), the frequency of these data collection points increases participant burden and missing data. The effect of UTIs on QoL was also estimated as part of this trial. On average, participants who experienced a UTI reported a lower utility score, estimated based on responses to the EQ-5D-5L. However, when this was incorporated into the QALY analysis, there was a negligible difference in the results. This is consistent with the results of the AnTIC trial,16 which found no difference in QALYs when utility values associated with a UTI were incorporated into the QALY equation. It is likely that, although participants experience a decrease in QoL associated with their UTI, this has a minimal effect on overall QALYs given its short duration.

Generalisability

The pragmatic design of this trial has ensured widespread applicability and generalisability. The wide inclusion criteria and minimal exclusion criteria encompassed a broad range of eligible participants without introducing excessive heterogeneity into the trial. The included patients accurately represent those women with rUTI seen regularly in routine NHS practice. This is reflected in the baseline characteristics of participants. In addition, the trial recruited patients from a range of geographic and socioeconomic areas. The protocol allowed for shared decision-making between patients and clinicians regarding choice of antibiotic, which is in line with current clinical practice. Patients were allowed to cross over between trial arms, which is again reflective of usual care. The results of this trial will be of interest to guideline writers and policy-makers alike given the global initiative to reduce antimicrobial resistance, which uses antibiotic stewardship as one of its cornerstones. This trial has provided robust evidence of the effectiveness of a non-antibiotic option for the preventative treatment of rUTI that will allow patients and clinicians to avoid extended-course antibiotics in the future. The demonstration that methenamine hippurate is not inferior to daily low-dose prophylactic antibiotics may support a change in practice in terms of guideline-recommended treatments for rUTI prevention.

Recommendations for research

• Evaluation of both methenamine hippurate and other non-antibiotic preventative treatment options for rUTIs in specifically defined populations, such as exclusively pre- or postmenopausal cohorts or those with recurrent complicated UTIs.

• Longer-term studies of UTI prevention treatments given the significant proportion of patients who relapse after cessation of treatment.

• A specific and thorough evaluation of antimicrobial resistance in response to low-dose antibiotic prophylaxis against UTIs, especially in patients on this treatment long term.

• Evaluation of non-antibiotic preventative treatment options in the context of prophylaxis against infections associated with urological or urogynaecological surgery.

• Evaluation of novel diagnostic tests for UTI that may allow for more precise treatment and the avoidance of extended or repeated courses of antibiotics.

• Determination of the longer-term costs and benefits associated with methenamine hippurate and antibiotic prophylaxis. In addition, the potential costs and effects associated with antibiotic resistance need to be considered when comparing the long-term effects of both management strategies.

Contributions of authors

Chris Harding (https://orcid.org/0000-0002-9407-382X) (Chief Investigator) designed the trial, was the principal applicant for funding, co-wrote the study protocol, supervised the overall conduct of the study, interpreted trial data, and co-wrote and edited the final report.

Thomas Chadwick (https://orcid.org/0000-0002-7995-0302) (Clinical Trials Statistician) designed the statistical plan in the study protocol, led the final study statistical analysis and edited the final report.

Tara Homer (https://orcid.org/0000-0002-6664-0671) (Health Economist and Senior Research Associate) designed the health economics analysis plan, led the health economics analysis and co-wrote the final report.

Jan Lecouturier (https://orcid.org/0000-0002-5191-5804) (Senior Research Associate) contributed to the design of the study and protocol (including the embedded qualitative study), was a co-applicant for funding, led the embedded qualitative study and co-wrote the final report.

Helen Mossop (https://orcid.org/0000-0002-6260-3734) (Research Associate in Biostatistics) wrote the SAP, conducted the main study analysis and co-wrote the final report.

Sonya Carnell (https://orcid.org/0000-0002-7648-4297) (Senior Trial Manager) led the NCTU team’s involvement in the trial and co-wrote and edited the final report.

Will King (https://orcid.org/0000-0001-7040-9133) (Health Economist and Training Fellow) wrote the health economics analysis plan, conducted the main study health economics analysis and co-wrote part of the final report.

Alaa Abouhajar (https://orcid.org/0000-0001-7220-6380) (Database Manager) designed and set up trial-specific databases, managed central data processes, reviewed trial results and reviewed the final report.

Luke Vale (https://orcid.org/0000-0001-8574-8429) (Health Foundation Professor of Health Economics) was a co-applicant for funding, contributed to the design of the protocol, designed and supervised the health economics analysis, reviewed the trial results and edited the final report.

Gillian Watson (https://orcid.org/0000-0003-2466-0268) (Trial Manager) was the trial manager for the trial and reviewed the final report.

Rebecca Forbes (https://orcid.org/0000-0002-2391-1371) (Trial Manager) was the trial manager for the trial and reviewed the final report.

Stephanie Currer (https://orcid.org/0000-0002-7074-9370) (Clinical Trial Administrator) was the Clinical Trial Administrator for the trial and reviewed the final report.

Robert Pickard (https://orcid.org/0000-0001-6151-9031) (Professor of Urology) contributed to the design of the protocol and was a co-applicant for funding.

Ian Eardley (https://orcid.org/0000-0002-8411-2987) (Consultant Urologist) contributed to the design of the protocol, was a co-applicant for funding, acted as PI for the Leeds site and reviewed the final report.

Ian Pearce (https://orcid.org/0000-0002-7710-0531) (Consultant Urologist) contributed to the design of the protocol, was a co-applicant for funding, acted as PI for the Manchester site and reviewed the final report.

Nikesh Thiruchelvam (https://orcid.org/0000-0002-2910-2209) (Consultant Urologist) contributed to the design of the protocol, was a co-applicant for funding, acted as PI for the Cambridge site and reviewed the final report.

Karen Guerrero (https://orcid.org/0000-0001-9591-059X) (Consultant Urogynaecologist) contributed to the design of the protocol, was a co-applicant for funding, acted as PI for the Glasgow site and reviewed the final report.

Katherine Walton (https://orcid.org/0000-0002-3434-7753) (Consultant Clinical Microbiologist) contributed to the design of the protocol, was a co-applicant for funding, and led the reporting and interpretation of microbiological results and reviewed the final report.

Zahid Hussain (https://orcid.org/0000-0002-9852-6478) (Consultant Urologist), acted as PI for the Oldham site and reviewed the final report.

Henry Lazarowicz (https://orcid.org/0000-0003-0792-2983) (Consultant Urologist) acted as PI for the Liverpool site and reviewed the final report.

Ased Ali (https://orcid.org/0000-0001-9576-9047) (Consultant Urologist) contributed to the design of the protocol, was a co-applicant for funding, acted as PI for the Wakefield site and reviewed the final report.

Publications

Forbes R, Ali A, Abouhajar A, Brennand C, Brown H, Carnell S, et al. ALternatives To prophylactic Antibiotics for the treatment of Recurrent urinary tract infection in women (ALTAR): study protocol for a multicentre, pragmatic, patient-randomised, non-inferiority trial. Trials 2018;19:616.

Lie MLS, Lecouturier J, Harding C. Should I stay or should I go? A qualitative study exploring participation in a urology clinical trial. Int Urogynecol J 2019;30:9–16.

Harding C, Mossop H, Homer T, Chadwick T, King W, Carnell S, et al. Alternative to prophylactic antibiotics for the treatment of recurrent urinary tract infections in women: multicentre, open label, randomised, non-inferiority trial. BMJ 2022;376:e068229.

Data-sharing statement

All data requests should be submitted to the corresponding author for consideration. Access to anonymised data may be granted following review.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

Disclaimers

This report presents independent research funded by the National Institute for Health and Care Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, the HTA programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, the HTA programme or the Department of Health and Social Care.