Continuous low-dose antibiotic prophylaxis to prevent urinary tract infection in adults who perform clean intermittent self-catheterisation: the AnTIC RCT

Inclusion criteria

• Adult aged ≥ 18 years.

• Established user of CISC who was predicted to continue using for ≥ 12 months.

• Able to give informed consent for participation in the trial.

• Able and willing to adhere to a 12-month follow-up period.

• Had suffered either at least two episodes of symptomatic UTI related to CISC within the previous 12 months or at least one episode of UTI requiring hospitalisation. Or, for those currently prescribed prophylactic antibiotic for UTI, has completed a 3-month washout period without antibiotic prophylaxis. Any symptomatic UTI was treated prior to randomisation.

• Able to take a once-daily oral dose of at least one of 50 mg of nitrofurantoin, 100 mg of trimethoprim or 250 mg of cefalexin.

Exclusion criteria

• Aged < 18 years.

• In a learning phase of CISC.

• Already taking prophylactic antibiotic against UTI and declining or unable to tolerate a 3-month washout period without antibiotic prophylaxis.

• Inability to tolerate all three of the prophylactic antimicrobial agents.

• Women who were pregnant, breastfeeding or who intended to become pregnant during the planned period of trial participation.

• Previous participation in this study.

• Inability to give informed consent or complete study documents in English (the outcome measures had not been validated in other languages).

Potentially eligible patients were given or sent brief trial information. Those interested and eligible were approached by research staff and provided with full trial information. We welcomed participation of otherwise eligible patients who were already taking antibiotic prophylaxis against recurrent UTI, provided that they agreed to have a 3-month washout period without taking prophylaxis.

Participant allocation

Randomisation was administered centrally by the NCTU secure web-based system. Permuted random blocks of variable length were used to allocate participants 1 : 1 to the control and experimental groups. A statistician not otherwise involved with the trial produced the final randomisation schedule. Stratification by three variables [(1) prior frequency of UTI (fewer than four episodes per year and four or more episodes per year), (2) diagnosis of neurological lower urinary tract dysfunction (yes or no) and (3) sex (female or male)] was performed to ensure a balanced allocation within these known UTI risk factors. For those allocated to prophylaxis, an appointment was arranged with the prescribing clinician to commence an individually suitable trial drug: 50 mg of nitrofurantoin (or 100 mg according to clinician preference and local guidance), 100 mg of trimethoprim or 250 mg of cefalexin.

Severity of urinary tract infection

During review of each UTI record form and associated CRF, severity was classified as non-febrile or febrile, and whether or not hospitalisation was required. Febrile UTI was defined as the primary outcome plus presence of a recorded fever of > 38 °C. Hospitalisation due to UTI was defined as an unplanned visit to hospital for treatment of a UTI, which required at least one overnight hospital stay. We also asked participants to self-rate the severity of each UTI as mild, moderate or severe.

Adverse events related to both prophylaxis and treatment antibiotic use

Any AEs possibly related to the trial intervention suffered during the trial participation were recorded as free text by local research staff at 1-, 3-, 6-, 9- and 12-month trial visits on CRFs. Those occurring during treatment courses of antibiotic for UTI (adverse reactions) were recorded on the participant UTI record form [see Questionnaire (10 Jan 2018), Participant UTI Record available at www.journalslibrary.nihr.ac.uk/programmes/hta/117201/#/documentation (accessed 3 April 2018)]. During episodes of antibiotic-treated UTI and by review of health-care records by local research staff [see Interview Material (10 Jan 2018), Participant 3-Month review interview and Health care records 3-month review available at www.journalslibrary.nihr.ac.uk/programmes/hta/117201/#/documentation (accessed 3 April 2018)], AEs were categorised as mild (discomfort is noticed, but there is no disruption of normal daily activities), moderate (discomfort is sufficient to reduce or affect normal daily activities) or severe (discomfort is incapacitating, with inability to work or to perform normal daily activities). The investigator at the relevant site assessed the relationship of the AE to the trial treatment by checking against the Reference Safety Information (RSI) in the Summary of Product Characteristics (SmPCs) specific to the prescribed antibiotic, which was included in the protocol.

Overall satisfaction with prophylactic antibiotic treatment

This was measured by the participant completion of the Treatment Satisfaction Questionnaire for Medication37 (TSQM) at the end of the trial. Separate scores from the four subscales were reported: (1) effectiveness, (2) side effects, (3) convenience and (4) global satisfaction.

The relative effect on health status among trial participants

This was measured by participant completion of the Short Form questionnaire-36 items version 2 (SF-36v2; 1-week acute recall version) at baseline, after 6 and 12 months of participation and at the time of each UTI. The SF-36v2 includes 36 different questions, each of which contributes to the score that can be calculated for eight different dimensions: (1) physical functioning, (2) role limitations due to physical health, (3) bodily pain, (4) general health perceptions, (5) vitality, (6) social functioning, (7) role limitations due to emotional problems and (8) general mental health caused by either physical or emotional problems. From the scores attached to each of these dimensions, two additional summary scores were derived. These are the physical component summary (PCS) score and the mental component summary (MCS) score. The higher the value of the summary scores, the higher the level of functionality of the patient. 38

Change in kidney and liver function at 12 months

Change in kidney function was determined by comparison of the estimated glomerular filtration rate (eGFR) derived from measurement of serum creatinine and accounting for race and sex, at baseline and 12 months for each individual and each group using an online calculator [available at www.kidney.org/professionals/kdoqi/gfr_calculator (accessed 27 October 2017)]. At baseline, research staff were asked to calculate creatinine clearance using an online calculator [available at www.nuh.nhs.uk/staff-area/antibiotics/creatinine-clearance-calculator (accessed 27 October 2017)] to detect participants allocated to prophylaxis with clearance values of < 45 ml/minute who would not be able to use nitrofurantoin. Change in liver function was assessed using the measured value of the liver enzyme alanine transaminase (ALT) at baseline and 12 months, compared within individuals and according to trial group and the agent used for prophylaxis.

Screening

General demographics and eligibility were checked. Trial Information was provided [see Patient Information Sheet (10 Jan 2018) available at www.journalslibrary.nihr.ac.uk/programmes/hta/117201/#/documentation (accessed 3 April 2018)] to the participant and consent taken prior to randomisation. Participants experiencing symptomatic UTI were treated with standard antibiotic therapy and not randomised until symptom free.

Randomisation

Randomisation was performed as close as possible to the date of consent (normally immediately after). The continued eligibility of those consented participants who had completed a 3-month washout period was checked prior to their randomisation.

Follow-up

Participants were contacted by local trial staff 1 month after randomisation by telephone regarding general concerns, understanding of trial documentation and their tolerance of the prophylactic antibiotic agent (if allocated).

At 3, 6, 9 and 12 months after randomisation, telephone or face-to-face contact (according to local circumstance and participant preference) took place. Details of UTI occurrence, UTI symptoms, adverse reactions to antibiotics taken for UTI and other infections requiring treatment with antibiotics during the previous 3 months were recorded in the electronic CRF (e-CRF). Participants were asked to return a urine specimen during an asymptomatic period at 3, 6, 9 and 12 months, and a perianal swab at 6 and 12 months, to the central laboratory in postage pre-paid, safety-compliant sample packaging provided to them.

During episodes of symptomatic antibiotic-treated UTI, participants completed a UTI record form [see Questionnaire (10 Jan 2018) available at www.journalslibrary.nihr.ac.uk/programmes/hta/117201/#/documentation (accessed 3 April 2018)] recording symptoms, antibiotic use, adverse reactions to treatment for UTI, participant rating of severity of UTI and details if hospitalised. They were also asked to post a urine specimen to the central laboratory prior to commencing a treatment course of antibiotics, for analysis and subsequent storage. This was in addition to urine specimens requested by the local treating clinician required for diagnosis and management of the episode locally in line with routine local diagnostic practice.

Other outcome data were collected by participant postal questionnaire sent by the central trial office at 3, 6, 9 and 12 months and completed by participant. This was then returned to the central trial office. Follow-up data were supplemented by regular inspection of health records by local trial site staff for documented visits to clinicians because of UTI, episodes of fever recorded as > 38 °C associated with UTI, antibiotic prescriptions for UTI, hospitalisations and results of local laboratory urine cultures. The schedule of events for the trial are summarised in Table 2.

Primary outcome measure

The primary outcome measure was the difference in incidence of symptomatic antibiotic-treated UTI during the 12-month observation period. Symptomatic UTI was defined as the presence of at least one patient-reported or clinician-recorded symptom from a predefined list encompassing the recommendations of the BIA, the CDC and spinal cord injury UTI consensus statement44 together with taking a discrete treatment course of antibiotics prescribed by a clinician or as part of a patient-initiated self-start policy. 11,35,36

The rate of UTI in each group was defined to be the incidence rate ratio (IRR), that is, the total number of UTIs suffered across all patients, allowing for differing durations of follow-up. Analyses of the primary outcome measure were performed using both the Poisson rate test (as a simple univariate approach with consideration of days in follow-up) and an IRR modelling approach to allow for the different durations of treatment for symptomatic UTIs, which reduce the number of days-at-risk for individuals. Days-at-risk was defined as the total observation period minus days spent taking treatment courses of antibiotics active against urinary tract organisms. When no information on the duration of antibiotic treatment course was available, it was assumed to last 7 days. A Poisson regression approach was used to adjust for the effects of covariates. The model selection process included the stratification factors and other baseline variables. Site was also explored as an interaction term. An analysis of the primary outcome measure was performed both for the full data set and for the separate subgroups defined by high and low baseline UTI rate (as specified during stratification for the randomisation process). The simple univariate analysis was considered to be the primary analysis for reporting purposes.

Secondary outcome measures

For the following secondary outcome measures included in the SAP, rates were defined in a similar way to the primary outcome:

• microbiologically confirmed symptomatic UTI rate

• febrile UTI rate

• hospitalisation rate attributable to UTIs during the 12 months of the trial

• asymptomatic bacteriuria rate

• antibiotic prescription for asymptomatic UTI rate

• AE rate (those related to prophylaxis and treatment antibiotics), including antibiotic resistance.

Statistical analyses of these outcome measures generally used the same approach as for the primary outcome.

The detection rate for resistance of all pathogens isolated from urine and strains of E. coli from perianal swabs to tested antibiotics at any point during 3-month time periods (0–3, 3–6, 6–9 and 9–12 months) was summarised by trial group. When numbers were sufficient, a chi-squared test for the 9- to 12-month samples from the prophylaxis group versus the no-prophylaxis group and tests for trend including baseline for both groups were used for analysis of the resistance patterns over the 12-month trial period. For perianal swab specimens, 6-month time periods and chi-squared tests for the 6–12 months period were used.

For TSQM scores and change in kidney and liver function at 12 months, the two-sample t-test was used as a simple univariate analysis. Furthermore, an analysis of covariance (ANCOVA) approach was employed using the covariates identified during the primary outcome modelling (for the TSQM scores, each subscale was considered separately).

For base-case analysis of the MCS and PCS of the SF-36v2 completed at baseline and 6 and 12 months, a simple Student’s t-test was used. For the adjusted analyses, linear regression was used with the component scores as the dependent variable and with treatment allocation, score at baseline and baseline characteristics as covariates. This analysis allowed assessment of the impact of the interventions after controlling for imbalances at baseline. A two-sample t-test was used as a simple univariate analysis for differences between groups in MCS and PCS. A linear regression was also performed adjusting for the baseline characteristics: age, sex, neurological bladder dysfunction or not, number of UTI episodes in the 12 months before randomisation (fewer than four vs. four or more), renal function (creatinine clearance) and MCS or PCS scores at baseline. Data from completion of the SF-36v2 at the time of UTI were excluded from this analysis but were used in the cost–utility component of the health economic evaluation.

Severity of urinary tract infection

Overall, participants rated symptoms due to UTI as mild for 31% of episodes, moderate for 43% and severe for 23% (no rating was given for 3% of episodes) (see Figure 4c). Of those participants included in the primary analysis, at least one episode of febrile UTI was suffered by 15 participants in the prophylaxis group and 22 in the no-prophylaxis group (IRR 0.71, 95% CI 0.40 to 1.26). This result was unchanged by Poisson modelling analysis with adjustment for frequency of UTI at baseline, days at risk of UTI, sex and neurological cause of bladder dysfunction. Overall, a fever of > 38 °C was recorded for participants in 49 (7%) of the 703 episodes of UTI reported. Six participants in the prophylaxis group were hospitalised as a result of UTI, compared with eight participants (including one participant with two admissions) in the no-prophylaxis group. Instances of hospitalisation were insufficient in both groups to allow for the planned formal statistical comparison.

Alternative definitions of UTI

The reduction seen with the primary trial definition of UTI (symptomatic and antibiotic treated) in the prophylaxis group was also evident using alternative definitions of UTI. For symptomatic microbiologically proven UTI, the IRR was 0.49 (95% CI 0.39 to 0.60) in favour of prophylaxis (Table 9 and see Figure 4b). This result was unchanged by adjustment for days at risk and for prior frequency of UTI, sex and neurological bladder dysfunction. There was also a considerable reduction in the number of participants suffering from four or more microbiologically proven UTIs from 22 participants in the no-prophylaxis group to 10 participants in the prophylaxis group (see Table 9). Of the 703 episodes of symptomatic antibiotic-treated UTI reported for the primary outcome, 405 (58%) were associated with a positive urine culture, with 221 being both central laboratory and local laboratory specimen positive, 114 being local laboratory urine positive with a missing or negative central specimen, and 70 being a positive central laboratory specimen with a negative or missing local laboratory specimen. A total of 298 (42%) episodes of symptomatic antibiotic-treated UTI were not associated with a positive urine sample as tested by either the central or a local laboratory. Of these, 142 had no associated urine specimen submitted for culture, 57 had no growth on culture from either the sample submitted to a central laboratory or the locally submitted sample and 99 had a single specimen submitted to either the central or local laboratory that was negative on culture. A total of 156 (22%) symptomatic, antibiotic-treated UTI episodes were associated with a negative culture.

For asymptomatic bacteriuria detected in samples submitted to the central laboratory by participants at baseline and at 3, 6, 9 and 12 months, there was no difference between the prophylaxis and no-prophylaxis groups in the rate of positive culture in the unadjusted analysis (IRR 0.88, 95% CI 0.74 to 1.04) (see Figure 4d). When adjusted for days at risk, the subgroup of participants with low prior frequency of UTI (fewer than four per year) who were allocated to prophylaxis had a significantly lower rate of asymptomatic bacteriuria, with an IRR of 0.76 (95% CI 0.59 to 0.98). These results were unchanged by adjustment for sex and presence of neurological bladder dysfunction (Table 10). Overall, 110 (61%) participants in the prophylaxis group and 113 (63%) participants in the no-prophylaxis group had at least one positive urine culture during asymptomatic periods during their trial participation (Table 10). Antibiotic treatment for asymptomatic bacteriuria was prescribed for four participants [four in the prophylaxis group (one prescribed twice) and one in the no-prophylaxis group]. Instances of antibiotic treatment for asymptomatic bacteriuria were insufficient in both groups to allow for the planned formal statistical comparison. Isolates cultured from 3-monthly urine specimens (asymptomatic) and at the time of UTI are detailed in Table 11; E. coli dominated in both sets of specimens.

Adverse events

All SAEs reported during the trial are given in Appendix 1 (see Tables 30–33). Overall, there was no deterioration in kidney or liver function (see Appendix 1, Table 34) in either group over the 12 months of participation. One participant suffered an asymptomatic rise in the serum level of a liver enzyme (ALT) resulting from the use of nitrofurantoin as the prophylactic agent. This resolved completely on stopping nitrofurantoin. There were no reported cases of respiratory illness or peripheral neuropathy associated with prophylaxis using nitrofurantoin. One participant who had been on multiple medications, including nitrofurantoin, was admitted to hospital following a fall that was assessed as due to polypharmacy (not listed in the approved RSI as an expected AE for nitrofurantoin). The sponsor took the decision to report this to MHRA as a suspected unexpected serious adverse reaction (SUSAR). The event was submitted as a ‘drug interaction’. None of the other 51 SAEs reported was categorised as being related to study participation. Three participants died during the 12 months of trial observation: one from metastatic oesophageal cancer, one from bladder cancer and one following a fall.

Review of health records showed that 19 participants in the prophylaxis group and four in the no-prophylaxis group suffered an AE related to use of prophylactic antibiotics (Table 12). At each of the 3-monthly reviews, < 10% of the 203 participants allocated to prophylaxis reported adverse effects. Of the participant responses on UTI record forms, 28 participants in the prophylaxis group and 60 in the no-prophylaxis group reported at least one specific AE due to treatment antibiotic. The most frequent side effects were gastrointestinal disturbance (nausea and diarrhoea reported by 14% and 11% of participants respectively) and candidal infection (7% of participants) (see Table 12). The small numbers of AEs precluded formal statistical analysis.

Bacterial resistance to antibiotics

There were no apparent differences between the groups at baseline in resistance rates of urinary isolates to eight oral antibiotics commonly used for UTI: (1) amoxicillin, (2) cefalexin, (3) ciprofloxacin, (4) co-trimoxazole, (5) co-amoxiclav, (6) mecillinam, (7) nitrofurantoin and (8) trimethoprim.

Urine specimens received during asymptomatic periods were split into time categories according to the date the sample was received in relation to the participant’s date of randomisation. Samples were categorised as baseline, 0–3 months, 3–6 months, 6–9 months or 9–12 months. Samples received up to 14 days after the end of these time periods were included in the preceding category to allow for delay in receipt of samples. A similar methodology was applied to perianal swabs received, but with baseline, 0–6 months and 6–12 months categories. To establish the number of participants whose specimens had isolates resistant to each antibiotic in each time category, we considered that the presence of an isolate displaying resistance in any of the surveillance urine samples received from an individual participant during each specified time period indicated that resistance was present for that participant at that time period. Figures 5–7 show resistance rates over time, plotted for eight antibiotics, by group.

FIGURE 5.

Resistance rates (%) plotted over the 12-month trial duration for pathogens isolated from urine specimens taken during symptomatic UTI, by antibiotic and group. (a) Amoxicillin; (b) cefalexin; (c) ciprofloxacin; (d) co-trimoxazole; (e) co-amoxiclav; (f) mecillinam; (g) nitrofurantoin; and (h) trimethoprim. n/N, number of participants with at least one resistant isolate/number of participants with at least one isolate; NA, not applicable.

FIGURE 6.

Resistance rates (%) plotted over time for bacteria isolated from urine specimens taken during asymptomatic state, by antibiotic and group. (a) Amoxicillin; (b) cefalexin; (c) ciprofloxacin; (d) co-trimoxazole; (e) co-amoxiclav; (f) mecillinam; (g) nitrofurantoin; and (h) trimethoprim. n/N, number of participants with at least one resistant isolate/number of participants with at least one isolate.

FIGURE 7.

Resistance rates plotted over time for E. coli strains isolated from perianal swabs, by antibiotic and group. (a) Amoxicillin; (b) cefalexin; (c) ciprofloxacin; (d) co-trimoxazole; (e) co-amoxiclav; (f) mecillinam; (g) nitrofurantoin; and (h) trimethoprim. n/N, number of participants with at least one resistant isolate/number of participants with at least one isolate.

Table 13 shows the results of a test for trend during the asymptomatic state in the prophylaxis group and the no-prophylaxis group for the 3-monthly urine specimens. We observed an increase in the resistance rates with time for the majority of the antibiotics under consideration in the prophylaxis group. We did not find any evidence of increased antibiotic resistance in the no-prophylaxis group. It was considered that insufficient samples or time points were available for a trend analysis to be presented for the data from the urine samples submitted at the time of UTI or the perianal swabs.

Table 14 shows the results of performing chi-squared tests for each antibiotic to compare the resistance rates in the two groups at 9–12 months (using asymptomatic state urine samples) and at 6–12 months (using perianal samples). In the asymptomatic state samples, we observe that at 9–12 months rates of resistance to nitrofurantoin (χ² = 4.31; p = 0.038), trimethoprim (χ² = 13.0; p < 0.001) and co-trimoxazole (χ² = 9.79; p = 0.002) were significantly higher in the prophylaxis group than in the no-prophylaxis group. Using perianal swabs, we found no evidence that at 6–12 months rates of resistance to any antibiotics were significantly higher in the prophylaxis group than in the no-prophylaxis group.

Satisfaction with treatment

Overall, participants allocated to prophylaxis were satisfied with their treatment. Completion of the TSQM by 144 (71%) of the 203 participants allocated to prophylaxis showed a mean (SD) score of 78 points (19 points) for effectiveness and a mean (SD) of 89 points (14 points) for convenience. Overall mean (SD) satisfaction score was 74 points (25 points) (Table 15). The TSQM is not validated for ‘no treatment’ strategies in an open-label RCT setting but, overall, this group rated mean (SD) overall satisfaction as 63 points (24 points).

At 12 months, 77 (78%) of the 99 participants allocated to prophylaxis who expressed a preference stated that they wished to continue with the treatment, whereas 21 (20%) of the 104 participants in the no-prophylaxis group who expressed a preference wanted to commence prophylaxis.

The relative effect on health status among trial participants

Data from participant completion of the SF-36v2 available or missing at each time point for the prophylaxis group and no-prophylaxis group are shown in Table 16. This table shows the number of participants providing data at each time point falling during the 12-month study period. Overall, approximately 70% of participants completed all three assessments.

Subcategory MCS and PCS scores from participant completion of the SF-36v2 are summarised in Table 17. It presents the mean, SD, 95% CI, median, MCS and PCS scores for each group at baseline and at 6 and 12 months, by group.

Details of the unadjusted and adjusted comparison between randomised groups at 6 and 12 months are reported in Table 18. For the PCS, the unadjusted analysis shows a mean increase of 0.38 (95% CI –2.45 to 3.21; p = 0.79) in the prophylaxis group compared with the no-prophylaxis group at 6 months. However, at 12 months, the prophylaxis group had a mean decrease of –0.39 (95% CI –3.39 to 2.60; p = 0.79) compared with the no-prophylaxis group. These differences in PCS scores at 6 and 12 months between groups were not statistically significant. The MCS showed a difference in mean score of 2.96 (95% CI 0.05 to 5.87; p = 0.05) at 6 months in favour of prophylaxis and 1.07 (95% CI –1.96 to 4.08; p = 0.49) at 12 months. Adjusted analysis gave broadly similar results (see Table 18).

Data collection

Inpatient costs

In the health utilisation questionnaire completed at 6 and 12 months, participants detailed how many nights they had spent in hospital over the previous 3 months. To account for variation of NHS costs with length of stay, three different unit costs were applied, for stays of 1 night, 2–7 nights or ≥ 8 nights [see Health Economics Plan (10 Jan 2018), Table 1 available at www.journalslibrary.nihr.ac.uk/programmes/hta/117201/#/documentation (accessed 3 April 2018)]. The cost of the first overnight stay was assumed to be highest as it involves assessment, diagnosis and frequent monitoring. The costs for nights 2–7 are based on the cost per night for a 7-night stay. The cost a stay of ≥ 8 nights is based on the cost of excess bed-days beyond 7 nights. All unit costs were obtained from freely available NHS sources. 50 The count of the number of nights participants spent in hospital over the 3-month recollected period was multiplied by the relevant cost per night. This total cost for the previous 3 months, detailed in the 6-month and 12-month questionnaires, was then doubled to provide an estimate for the 6-month total inpatient-stay cost. The cost for each participant in each study group was then averaged to give a mean (SD) inpatient cost per participant for each trial group.

Outpatient costs

Outpatient costs were categorised as emergency visits such as attendance at urgent care centres and planned visits to a clinic that did not result in an overnight stay. Data collected through the participant-completed health service utilisation questionnaire at 6 and 12 months were combined with relevant unit costs obtained from NHS Reference Costs 2015-1650 [see Health Economics Plan (10 Jan 2018), Health economics supplementary information Table 1 available at www.journalslibrary.nihr.ac.uk/programmes/hta/117201/#/documentation (accessed 3 April 2018)] to calculate a total cost per participant. The average cost per participant in each group of outpatient service use was then estimated.

Primary care costs

Primary care costs were categorised as consultations with GPs or nurse clinicians. These were subcategorised on the health service utilisation questionnaire according to how the consultation was given: health centre consultation, home visit, telephone consultation or out-of-hours consultation. Each type of consultation was associated with a different unit cost. The unit costs for use of primary care services were obtained from PSSRU (Personal Social Services Research Unit) unit costs of community care [see Health Economics Plan (10 Jan 2018), Health economics supplementary information, Table 2 available at www.journalslibrary.nihr.ac.uk/programmes/hta/117201/#/documentation (accessed 3 April 2018). 51 The average cost of primary care utilised by participants in each study group was then estimated.

Medication costs

Details of antibiotics prescribed to participants for the treatment or prevention of UTI were taken from the UTI record forms completed by participants at the time of each UTI episode over the 12-month follow-up. These details were also separately recorded at the 3-monthly review. Information on antibiotic use was collated from the different sources and checked as part of the clinical primary outcome review to avoid double-counting. The unit cost of each antibiotic was obtained from the British National Formulary52 according to route of administration and dosage recorded in trial CRFs. Use of antibiotics for UTI prevention by participants allocated to the prophylaxis group was collected through the 3-monthly review CRFs. The daily dosage of the three antibiotics used for prophylaxis was 50 mg of nitrofurantoin, 100 mg of trimethoprim and 250 mg of cefalexin over 12 months. All data on antibiotic agent, dose and duration of use for treatment or prevention of UTI were combined with the associated unit cost [see Health Economics Plan (10 Jan 2018), Health economics supplementary information, Tables 3a and 3b available at www.journalslibrary.nihr.ac.uk/programmes/hta/117201/#/documentation (accessed 3 April 2018)]. This was added to the total cost per participant and the average cost per trial group of antibiotic medication.

Costs to participants and their main caregiver

Costs to participants were defined as any direct purchase of health care such as prescription charges, over-the-counter medication and the time and travel costs of accessing health care. Responses to the question about payment for private health-care services in the health service utilisation questionnaire were used to estimate the total cost of private health care. The total costs per participant and average cost per participant of private health-care costs were estimated by adding the total private health-care costs over the 12 months for each participant and dividing by the total number of participants in each trial group.

In the time and travel questionnaire, participants indicated out-of-pocket expenses for travelling to health-care appointments and what their usual activities would have been if they had not been travelling to, or attending, medical appointments or receiving medical treatments. Data on mode of transportation were combined with routine data sources, when necessary, to derive the total cost of a return journey for travelling and accessing health-care services. 53 The total cost per participant and average travel cost per participant were then estimated.

Time costs were estimated for participants and, if they also attended appointments with the participant, their main caregiver. The cost of time of paid activities was valued at the national median wage rate per hour. 54 The cost of time of unpaid activities was valued at the national rate of non-working time. 55 Additionally, the time and travel questionnaire asked for the number of days absent from paid employment due to health problems. The national median wage rate per week divided by 37.5 (the average working week in hours) was used to provide an estimate of the participant’s hourly wage rate. The average loss in earnings due to health problems (for participants and carers) was calculated by summing the total number of days absent from paid employment over the 12-month period and dividing this by the total number of participants, by trial group. Full details on the time and travel costs to participants and their main carers are given in Health Economics Plan (10 Jan 2018); Questionnaire (10 Jan 2018), Health Service Utilisation Questionnaire; and Questionnaire (10 Jan 2018), Time and Travel Questionnaire [all available at www.journalslibrary.nihr.ac.uk/programmes/hta/117201/#/documentation (accessed 3 April 2018)].

Estimation of health outcomes for the cost-effectiveness analysis

The clinical effectiveness estimates used in the CEA used the same approach described in Chapter 2 for calculating the primary outcome. An average total frequency of symptomatic antibiotic-treated UTI between the trial groups over 12 months was calculated by taking the total number of symptomatic antibiotic-treated UTIs recorded by each trial participant and dividing by the total number of participants in each trial group.

Estimation of quality-adjusted life-years for the cost–utility analysis

Responses to the SF-36v2 1-week recall questionnaire were used as the basis to calculate QALYs. The SF-36v2 questionnaire was completed by participants at fixed time points (baseline and 6 and 12 months) and within the first 2 days of each episode of symptomatic antibiotic-treated UTI. The completion of the SF-36v2 was prompted by telephone, e-mail or text message every time a UTI episode was reported by either the participant or the treating clinician.

The responses to the SF-36v2 were mapped onto the SF-6D using a standard algorithm to generate utility values. 56 These utility values have a range of 0 (dead) to 1 (perfect health), with the utility value associated with suffering a UTI falling between within these boundaries. QALY values were estimated for each trial participant by using the area under the curve. 57

In order to capture a true reflection of the impact of a UTI on quality of life (QoL), we incorporated the immediate impact on QoL of having a UTI within our secondary CUA. The primary analysis incorporated QALYs calculated at the predefined intervals (baseline and then at 6 and 12 months’ follow-up). The secondary analysis incorporated an adjusted QALY calculation by including the SF-6D score when a UTI was reported. A 5-day duration was assumed for the UTI SF-6D score before symptoms start to resolve.

Estimation of willingness to pay for use in the cost–benefit analysis

Contingent valuation, a stated preference method used to attribute monetary values to health outcomes, was used to inform the CBA. 58 The contingent valuation questionnaire [see Questionnaire (10 Jan 2018) available at www.journalslibrary.nihr.ac.uk/programmes/hta/117201/#/documentation (accessed 3 April 2018)] administered at 13 months’ follow-up collected information on the maximum amount (as a one-off payment) that participants would be willing to pay to avoid a UTI episode over a 12-month time period. For a given level of income, the higher the monetary value that individuals would be willing to pay, the greater the benefit that participants believe they would experience. This method enabled us to place a monetary value on the health outcome, going beyond the QALY framework, and also to conduct a CBA.

Costs

Regression techniques were applied to the data on total costs per participant (costs for the NHS and PSS, costs to the participant and caregiver, and overall costs) to identify whether or not there is a difference between trial groups, with control for potentially modifying factors such as age and sex (and also identify the key predictors of costs). An ordinary least-squares regression technique was applied (Equation 1):

In Equation 1, a dummy variable for the trial group estimates the difference in costs between the groups controlling for all other factors in the model. Estimated beta values describe the direction and magnitude of the relationship between each variable and the dependent variable. For example, if the dummy is specified as intervention = 1 and current practice = 0 and the beta estimate of the cost coefficient is + 500, this indicates that the intervention (coded as 1 in the dummy) is £500 more costly, on average, controlling for all other factors. If the coefficient was negative, then the intervention would, on average, be £500 cheaper than current practice.

Quality-adjusted life-years

As with costs, regression techniques were carried out to derive the drivers of the difference in QALYs between the groups of the trial after controlling for the key predictors of QALYs. The purpose of this regression was to determine the consequences of a UTI episode and its effect on a participant’s overall QoL (Equation 2):

The variables considered were:

• Dependent variable:

  • QALY (total QALY score across the two groups of the trial controlling for the independent factors).

• Independent variables:

  • age (A) in years and months

  • sex (S)

  • SF-6D score (SF) (baseline SF-36v2 score)

  • trial group (T) (dummy for the group of the trial to which the respondent was randomised).

Here, the dummy variable for the trial group estimates the difference in QALYs between the groups controlling for all other factors in the model. For example, if the dummy is specified as intervention = 1 and current practice = 0 and the beta value of the coefficient is + 0.50. this indicates that the intervention (coded as 1 in the dummy) provided 0.50 more QALYs over 12 months than current practice, on average, after controlling for all other factors. If the coefficient was negative then the intervention would, on average, provide a lower QoL than current practice.

Willingness to pay

Willingness-to-pay data are presented as mean WTP to avoid a single UTI episode over a 12-month period. The total benefits per individual are the maximum WTP to avoid one UTI episode measured in GBP, multiplied by the total count of primary outcome UTI episodes experienced by the individual over the 12-month follow-up period. As previously described, a regression model of individual covariates on WTP was specified. Data on costs and benefits (as measured via the contingent valuation survey) were combined in order to calculate the net benefit. A net benefit in terms of NHS costs and benefits, expressed in commensurate units, was calculated for each participant according to Equation 3:

where i = individual and WTP = maximum WTP threshold to avoid a UTI episode.

Therefore, the decision rule for CBA is relatively simple: if the total benefits per individual are more than the costs, this represents a gain in welfare and the strategy is deemed worthwhile. 58 Results are presented as incremental net benefits (net benefits = mean cost of intervention – mean total monetary benefits). Both stochastic and deterministic sensitivity analyses were conducted and the results presented as incremental net benefit curves.

Cost-effectiveness analysis

The CEA is the base-case economic analysis. It takes the form of a standard two-group CEA using symptomatic UTI avoided as the primary outcome measure at 12 months. Costs associated with the prophylaxis and no-prophylaxis strategies including cost of harms comprise treatment costs, health services utilisation over the 12 months’ follow-up and participant costs (as a sensitivity analysis) over the 12 months’ follow-up.

The cost and outcome regressions (detailed above) were run simultaneously using an approach referred to as a seemingly unrelated regression system. At first look, the equations seem unrelated, but the equations are related through the correlation in the errors. With this approach, the two linear regressions were estimated for the costs and count of UTI defined with a contemporaneous cross-equation error correlation (i.e. the error terms in the regression equations are correlated), given that the cost and outcome data are derived from the same individual. This analysis was conducted in Stata^® (StataCorp LP, College Station, TX, USA) using the seemingly unrelated regression command sureg.

Cost–utility analysis

The CUA measured incremental cost per QALY gained through completion of SF-36v2 at baseline and 6 and 12 months. We also measured incremental cost per QALY gained incorporating the repeated completion of SF-36v2 each time a UTI was reported. This analysis also adopted a seemingly unrelated regression system.

Cost–benefit analysis

Data on costs and benefits derived from multiplying the maximum WTP to avoid one UTI over a 12-month period with the total number of UTIs per participant were combined and used to derive a net benefit value for each individual. Given that costs and outcomes were combined into a single net benefit equation, only one regression needed to be estimated. This controlled for other factors in the model, such as income, age and sex.

Deterministic sensitivity analysis

Deterministic sensitivity analysis was carried out to test for the effect of assumptions and variability, such as an exploration of alternative unit costs applied to the different resources used. A number of analyses were performed including one way and/or multiway, depending on the results obtained from the deterministic analysis. When appropriate, these analyses were combined with the stochastic sensitivity analysis.

In order to assess the impact on the overall results, the costs to patient and carers were added to the NHS costs in the primary CEA. In the CUA, in addition to the analyses conducted for the base case, we also assessed the impact of taking into account the disutility arising from episodes of UTI.

Stochastic sensitivity analysis

Non-parametric bootstrapping-simulations that allow a comparison of arithmetic means without making assumptions about the distribution of the costs, UTIs avoided, QALYs and net benefit (using WTP estimates) is a data-based simulation method for assessing statistical precision. Bootstrapping is based on how values of the within-trial cost per UTI avoided, QALY and net benefit would vary if the sampling process could be repeated many times. Random values were selected from the cost, UTI, QALY and net benefit data collected from within the trial, with replacement (i.e. once a random value has been used for the bootstrap resample, it is put back into the original sample). This yields a bootstrap data set derived from the complete case data on costs and outcomes from each trial group.

For the bootstrapping, 1000 reiterations were carried out. This simulation process created a sample of bootstrapped means for costs, UTIs, QALYs and net benefit, with distributions for each. The means and other parametric statistics were then calculated for the bootstrap distribution. 60 Bootstrap estimates of the difference in costs, rate of UTI, QALYs and WTP between the experimental and control groups were used to populate the cost-effectiveness plane [the horizontal axis represents the difference in effectiveness (UTIs, QALYs or WTP) between two interventions and the vertical axis represents the corresponding difference in costs].

For cost-effectiveness and CUA, combining this information with the decision-maker’s maximum WTP over a range of values (e.g. £0 to £100,000 per UTI avoided or QALY) generated a cost-effectiveness acceptability curve (CEAC) quantifying the probability that an intervention is cost-effective based on the decision-maker’s maximum WTP. 60 For the CBA, a similar approach was used. Here, the results were presented as net benefit curves and the probability that the intervention has a net benefit > 0 and is, therefore, more efficient.

Incremental cost-effectiveness analysis

In the unadjusted CEA, the strategy of antibiotic prophylaxis was more effective but also more costly than the strategy of no prophylaxis, with an incremental cost per UTI avoided of £99. A similar pattern was observed for the adjusted analysis although the incremental cost per UTI avoided was reduced to £33 (Table 26). How much a decision-maker believes one avoided UTI is worth in terms of patient benefit is unclear. However, some guidance can be provided by considering the incremental cost per UTI avoided in the light of the WTP data. As indicated in Table 26, the mean amount that patients are willing to pay to avoid one UTI is > £100, and this suggests that the threshold value may be greater than the incremental costs per UTI avoided.

The deterministic results alone are not sufficient to support decision-making. For this, we need to consider the imprecision surrounding estimates of costs, effects and cost-effectiveness and, for this, the probabilistic sensitivity analysis is used. Figure 8 shows the distribution of the cost and effectiveness pairs simulated by bootstrapping the result of the CEA. As this plot illustrates, for 100% of the bootstrap iterations, prophylaxis is associated with fewer UTIs. The picture for cost is more mixed, with a roughly even chance that prophylaxis would be cost saving. The results also show (Figure 9 and Table 27) that there is a 66% chance that prophylaxis would be more cost-effective than no prophylaxis should society be willing to pay £200 to avoid one UTI and that the more we are willing to pay, the more likely it is that prophylaxis would be cost-effective. However, these results are not conclusive, as there was marked variability in costs between participants..

FIGURE 8.

Cost-effectiveness plane for prophylaxis vs. no prophylaxis: adjusted bootstrapped replications for CEA.

FIGURE 9.

Cost-effectiveness acceptability curves for prophylaxis vs. no prophylaxis: adjusted bootstrapped replications for CEA.

Overall, the cost-effectiveness results taking the participant and carer perspective derived from the time and travel questionnaire do not change the main findings (see Table 26). Given that the average total number of contacts with NHS services is higher for prophylaxis, the average costs taking into account the costs to patients and carers are also higher in this group. There is now a reduced chance that prophylaxis would be cost saving, taking the patient and carer perspective, given that the observed outcomes (number of UTIs) do not change across the two groups, but average costs increase across both groups (but to a lesser degree for the no-prophylaxis group).

Incremental cost per quality-adjusted life-year gained

The results of the CUA detailed in Table 27 show the incremental cost per QALY gained. For both the unadjusted and the adjusted analysis, the mean QALYs from prophylaxis are greater than the mean QALYs from no prophylaxis. In the unadjusted analysis, prophylaxis is, on average, more costly, contrasting with the adjusted analysis in which the prophylaxis strategy is less costly (although the CIs around mean costs are very wide). Therefore, the incremental cost per QALY gained is less than conventional thresholds for society’s WTP for a QALY (£20,000–30,000). 61 In the unadjusted analysis and in the adjusted analysis, the prophylaxis group is dominant (less costly and more effective on average); nevertheless, these deterministic results are insufficient to guide decision-making. The results of the probabilistic analysis suggest that there is approximately a 65% chance that prophylaxis treatment would be considered cost-effective at £20,000 per QALY (see Table 27) [see Health Economics Plan (10 Jan 2018), Health economics supplementary information, Figures 1 and 3, available at www.journalslibrary.nihr.ac.uk/programmes/hta/117201/#/documentation (accessed 3 April 2018)].

When the CUA was repeated taking into account the impact on QALYs of UTI together with imputation of missing data, the result in terms of incremental cost per QALY gained was similar (Figure 10 and see Table 27). In both analyses (adjusted and unadjusted), the incremental cost per QALY gained was less than typical values for WTP for a QALY. Similarly, in the probabilistic analysis there was a lower probability that use of prophylaxis would be considered cost-effective at the £20,000 threshold value (58–60%) (see Table 26) [see Health Economics Plan (10 Jan 2018), Health economics supplementary information, Figures 2 and 3, available at www.journalslibrary.nihr.ac.uk/programmes/hta/117201/#/documentation (accessed 3 April 2018)].

FIGURE 10.

Cost-effectiveness plane for prophylaxis vs. no prophylaxis: adjusted bootstrapped replications for primary CUA.

Cost–benefit analysis

In the CBA, the prophylaxis strategy was, again, costlier on average than no prophylaxis (Table 28). However, it was associated with a higher average net WTP to avoid UTIs. As before, the estimates of costs and effect were imprecise. The degree of imprecision was investigated using probabilistic sensitivity analysis (Figure 11 and see Table 28). This showed the comparatively wider spread in incremental benefits. This appears to be a consequence of combining the estimates of UTIs avoided with estimates of WTP as both variables have a degree of statistical imprecision. The results of the probabilistic sensitivity analysis suggest that the strategy of prophylaxis has a 66% chance of being considered more efficient than that of no prophylaxis (in the adjusted analysis).

FIGURE 11.

Cost–benefit plane for prophylaxis vs. no prophylaxis: adjusted bootstrapped replications for CBA.

Normalisation and psychological adjustment: ‘just part of my daily life’

Participants detailed a range of experiences in relation to the impact of using CISC in their daily lives. There was a strong sense of considering the process of CISC as ‘normal’, which led to the acceptance of CISC as a bladder management tool. Many participants believed it to be a no greater burden than the way other people who do not use CISC pass urine and felt that using CISC had not changed their self-image. CISC was labelled as ‘just one of those things’ and assimilated with their own identity:

You just get on with it. To me, it’s just the same as going to the toilet. I don’t think about it really.

Participant 13, female, P

Positive appraisal of CISC intensified the perception of normalisation and eased both the physical and psychological burden of urinary problems. A number of individuals associated CISC with a new ‘lease of life’ that allowed them increased independence and personal autonomy. Two male participants were able to return to employment following CISC initiation and cited benefits such as increased financial security and increased self-esteem:

[CISC] has changed me altogether; it changed my life. In the first week I says ‘this is brilliant, best thing since coloured tellies came out, this is great, I can do what I want to do’. I don’t have a problem with it.

Participant 2, male, NP

The concept of normalisation was also closely linked to participants developing confidence and establishing routines with CISC. Incorporating the practicalities of CISC within their lifestyle enhanced psychological adjustment and often encouraged a favourable attitude towards the process. For many, this was characterised by a pragmatic and organised approach to daily living, helping protect against potential inconveniences, while being mindful of their own abilities:

I just take some [catheters] with me; it [CISC] doesn’t prevent me from getting out and about.

Participant 19, male, P

Although participants generally displayed psychological adjustment to CISC, there were also some who perceived it to pose a negative impact on their QoL. Examples included impeding their ability to enjoy social activities and relax with friends, and avoiding going abroad. One participant expressed his concerns at staying away from home owing to the potential embarrassment at disclosing CISC use and associated practicalities:

The biggest thing is, it’s [CISC] fine at home, it’s just a real pain when you’re not at home. I loath to have to stay overnight ‘cause it’s more of a hassle both for you . . . you don’t want to talk to your hosts about it, but you need reasonably unencumbered access to a bathroom for yourself.

Participant 15, male, P

Time to adapt: ‘it was fine once I got used to it’

A key contributor to the theme of normalisation was the length of time taken to adapt psychologically to using CISC daily. In general, participants acknowledged a change in perspective towards CISC from initial use to how they felt following a period of time (and/or at the time of interview). This prolonged period of adaptation was often preceded by initial resistance:

At first, I didn’t like them. It took me a wee while . . . to get used to what you’re actually doing. I would say it can take as much as a year before you’re comfortable – it takes a while to find your niche I think. Now, they’re just an absolute lifesaver, it’s wonderful!

Participant 3, female, P

A number of participants attributed their immediate aversion to CISC to lack of knowledge about their bodies and the CISC process:

I wasn’t [OK with CISC] because I had no idea what was involved. But now I’m much more comfortable with it, I’m very patient with the process . . . I’m never silly about it and it’s a question of just relaxing you know, and not worrying about it.

Participant 7, male, NP

Alongside the interaction of comorbidities and aversion to CISC in general, participant 16 also expressed the impact of using CISC on her intimate relationships. She described threats to her self-esteem as a direct consequence of using CISC, factors that proved challenging to overcome:

I think the biggest issue that both of us have is that we had a really good sex life before and that has been affected massively because obviously hygiene is paramount and if my back’s not good and with catheterising, oh it’s just awful really. They [catheters] irritate my skin so I get sore skin as well. So, if you’re having intercourse and your skin’s broken you’ve got all that worry as well as all that pain, so that’s difficult. It’s just like not one little thing, it’s a few things together.

Participant 16, female, P

Perceived burden of urinary tract infection, ‘so inconvenient and uncomfortable’

Experiencing UTI was reported as intrinsic to living with bladder problems and using CISC. Symptoms were perceived on a spectrum of severity depending on the individual, the frequency of UTI and perceived disruption on one’s everyday life. One participant described the personal burden of having UTI and the challenges to treatment provision:

. . . well firstly it’s [experiencing a UTI] a bit debilitating, and secondly, it’s a bit depressing, you know, psychologically it’s not good for you. It’s also just a hindrance trying to get a doctor’s appointment, it’s quite a hassle, you can’t guarantee you’re going to get it that quickly and then you’ve got to wait to get the antibiotics, so yeah it’s just a hassle.

Participant 15, male, P

A number of participants implied that, because UTIs were so prevalent in their lives, they were highly attuned to the preliminary stages and symptoms of UTI. For some, this included normalising the feeling and experience of having a UTI to the extent that they did not perceive UTIs to have a notable impact on their lives:

. . . a bit uncomfortable, but not debilitating.

Participant 14, male, P

For these individuals, the experience of having a UTI had been internalised as ‘normal’ and a likely consequence of their bladder problems and/or comorbidities. Having a UTI or, indeed, the potential to have a UTI, did not seem to impede socialising nor going on holidays, or affect their lifestyle.

For others, the ‘hassle’ of having a UTI was more pronounced, such that UTIs were perceived to have a substantial adverse impact on overall QoL. Specifically, in relation to exacerbating comorbidities (e.g. one participant explained that having a UTI can trigger a relapse of his multiple sclerosis), juggling work commitments (e.g. to allow for GP appointments and also the impact of mood on ability to work) and interfering with social activities (e.g. avoiding social gatherings with friends):

. . . you just feel you don’t want to go out.

Participant 22, female, NP

Combined with practical implications of being in the vicinity of public conveniences/toilets:

I need to be close to a toilet all day.

Participant 17, male, NP

Two participants (one male and one female) also voiced the impact of a UTI on their relationships, including the challenges to intimacy and expressing sexuality:

It obviously makes you feel bad about yourself and it makes you feel like you want to keep away from people, go to bed a bit early. It impacts your relationship with your wife, personally, sexually, because you don’t feel good about yourself, you’ve got an infection and it’s not particularly pleasant.

Participant 20, male, P

Nonchalant attitude: ‘I feel fine about taking them’

Participants described a range of attitudes towards antibiotics and taking antibiotics. However, many reported feeling ‘fine’ about taking them with an overarching nonchalant perspective. Antibiotics were perceived as favourable in certain circumstances, particularly if they achieved the desired results or they were deemed necessary by medical professionals:

You just do it [take antibiotics]. If the results achieve what you want it to then you’ll do it.

Participant 3, female, P

Although a relaxed attitude towards taking antibiotics was common, many participants professed that they did not wish to take antibiotics unnecessarily. Instead, they preferred to trust medical opinion and rely on health-care professionals (HCPs) to assess the need for antibiotic treatment. Alternative self-management strategies for UTIs were used, including drinking more water or cranberry juice, or waiting to see if symptoms worsened.

Ambivalence towards antibiotic resistance: ‘I’m obviously aware, but you’ve got the down side to everything’

Participants conveyed mixed awareness and understanding of the concept of antibiotic resistance. A number of participants thought that ‘overusing’ antibiotics was unwise as it could reduce their future effectiveness. The possibility of the body becoming immune or resistant to antibiotics deterred some participants from taking antibiotics, although a minority did express strong concerns specific to the hypothetical development of bacterial pathogen resistance:

If you’re on antibiotics regularly the bacteria just forms resistance. I would only take them if I felt it was absolutely necessary – I think it might be counterproductive. I’m very wary about taking them, I think it’s a big, big worry at the minute.

Participant 7, male, NP

Participant 14 elaborated to say that he would rather take low-dose prophylactic antibiotics than a treatment course of high-dose antibiotics as he prefers the thought of preventing infections with them, as opposed to treating them. A number of individuals reiterated this notion of ‘little but often’ and disclosed a positive perception of prophylaxis.

Habitual tendencies: ‘it’s just another tablet’

For those allocated to taking prophylaxis antibiotics, there was a general sense that this behaviour had become automatic and part of an established routine. Many participants were already taking medications for comorbidities, and, so, taking another tablet became an adjunct behaviour. ‘Adding on’ an additional tablet was perceived as trivial and relatively straightforward and there was a general sense that participants did not perceive this to be a challenge or pose any difficulties:

I just took it in the morning before I went to work; it was only one a day.

Participant 13, female, P

Another mechanism for incorporating prophylaxis consumption into daily life was by exploiting already habitual behaviour. Examples include taking the tablet before going to sleep at night. Some participants relied on partners to ensure that the prophylaxis was taken each day:

I’ve got 18 tablets a day so she’s [wife] got them all morning, afternoon, evening, tea time. And she moans if I miss them.

Participant 2, male, P

A minority of individuals confessed to occasionally forgetting to take the prophylaxis. These lapses were often attributed to mitigating circumstances such as fluctuating work patterns and lifestyle choices, alongside sheer forgetfulness. Not remembering to take the prophylaxis did not seem to cause concern or endure over time, but instead was assumed as a natural event in a year-long commitment such as the AnTIC trial.

Supportive accountability: ‘it was no bother’

Adhering to the prophylactic antibiotic therapy in the AnTIC trial was aided by the perception of supportive accountability. Participants enjoyed having the support of AnTIC trial researchers and research nurses at their disposal, and often cited them as helpful, informative and friendly:

I was on it [AnTIC trial] a year and it was as if someone was listening to me. And it was nice that they kept a check on me every 3 months.

Participant 21, male, NP

Participants also cited curiosity and personal altruism for continuing with the AnTIC trial, such that adhering to the protocol was perceived as ‘giving something back’ and ‘helping others’ that may be in a similar situation. A minority also professed that they were eager to see if prophylaxis antibiotics would help them personally by reducing UTI incidence.