Systematic voiding programme in adults with urinary incontinence following acute stroke: the ICONS-II RCT

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 16/111/31. The contractual start date was in May 2018. The draft report began editorial review in September 2021 and was accepted for publication in February 2022. 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 Watkins et al. This work was produced by Watkins 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 Watkins et al.

Background

This study addressed the management of urinary incontinence (UI) in patients admitted to hospital with acute stroke. UI affects around half of stroke survivors in the acute phase. 3–5 As many as 44% and 38% of stroke survivors remain incontinent at 3 months and 1 year, respectively. 6 UI often presents as a new problem after stroke or, if pre-existing, worsens significantly, adding to the disability and helplessness caused by neurological deficits. 7

The more severe a stroke is, the greater the likelihood of UI. 8,9 Other factors that increase the likelihood of UI include older age and cognitive impairment. 10 Urge incontinence (involuntary leakage immediately following or concurrent with an urgent sensation of needing to void10) is the most common type of UI after stroke11 and is generally the result of detrusor overactivity. 11

It is important to study UI in this population, as the symptoms are more severe and have more of an effect on stroke survivors than on other groups. 7 Furthermore, associated stroke impairments compound bladder control difficulties, with motor, visual or speech problems making the task of accessing toilet facilities a challenge. 12

Urinary incontinence is distressing for individuals and families, and depression is twice as common in stroke survivors who are incontinent. 13 Negative social consequences for survivors and carers cannot be ignored; both may become isolated and marginalised. 14 Continuing incontinence is associated with poor outcomes in both stroke survivors and carers. 4,5,15

Although clinical guidelines state that indwelling urinary catheters (IUCs) should be used to relieve retention only,16 there is an over-reliance on catheterisation as a management strategy for UI in stroke units, especially in the acute phase. 17,18 This puts patients at risk of IUC-associated urinary tract infection (UTI) and its consequences,19–22 including increased morbidity, mortality rate and resource use. 21,23,24 In the Identifying Continence OptioNs after Stroke (ICONS)-I feasibility trial, 48% of patients had been catheterised prior to recruitment in the acute phase. 25 The number of UTIs and extent of antibiotic use are considerably greater in patients with IUCs than in those without, with an increasing risk of infection associated with IUC removal. 25 IUCs should be used not to manage incontinence, but rather to relieve urinary retention. Only by including all patients with IUCs and encouraging a trial without catheter (TWOC) were we able to include all patients with potential incontinence. The ICONS-II trial promoted catheter avoidance and aimed to reduce the number of UTIs and the need for antibiotic treatment.

We expected our intervention to reduce the number of patients with UI by ≥ 5% and improve continence in at least a further 5–10% of patients. If the intervention had been shown to be effective and were to have been adopted across the UK, assuming that 40,000 stroke survivors per annum have UI at 3 months,6 around 2000 patients would have become continent and around 2000–4000 would have had improved continence. Currently, patients with UI after stroke typically receive care focused on containment, using strategies that do not promote continence (e.g. pads) and that are likely to be harmful (e.g. IUCs17,18,26). The ICONS-II trial was designed to provide high-quality evidence regarding the clinical and economic effects of a new approach to assessing, managing and treating UI after stroke.

Rationale for the trial

Stroke is the third-largest cause of death and the largest single cause of severe adult disability,27 with up to 95,000 people per annum surviving a stroke in the UK. The incidence of stroke is unlikely to decline given the ageing population,28 and its prevalence continues to rise. 29,30 Stroke patients with UI have considerably worse outcomes than those without UI: there are clear associations between UI after stroke and death, disability and an increased likelihood of being discharged to residential care. 4,8,31 Addressing UI early and effectively could have a major impact on these outcomes and significantly improve the quality of life (QoL) of patients. 32

Improving the management of UI after stroke has been identified as an urgent priority in successive Sentinel Stroke National Audit Programme (SSNAP) reports,33,34 but there is a lack of evidence that this aspect of care has improved. Consequently, the National Institute for Health and Care Excellence (NICE) has recommended further research into improved continence care after neurological events. 35

To the best of our knowledge, this trial was the first to test the effect of a programme to assess and treat UI after stroke in hospital, building on our feasibility trial. 25,35,36 Although the feasibility trial was not powered to demonstrate effectiveness, there were indications that the intervention may work, particularly for urge incontinence. Staff believed that the programme improved patient outcomes and was sustainable: five out of eight sites reported continuing core aspects of continence promotion after the research was completed.

The update of our Cochrane review37 revealed several new studies,38–43 although the conclusion – that data from the available trials are insufficient to guide continence care – was unlikely to change without a definitive trial. The ICONS-II trial was intended to address this gap in the evidence base.

Original and modified trial design

The ICONS-II trial was preceded by the ICONS-I trial, a cluster-randomised feasibility trial of a similar intervention compared with usual care, funded by the National Institute for Health and Care Research (NIHR) Programme Grants for Applied Research (PGfAR) programme. 25 This was implemented as a change in practice in intervention sites, but analysis of recruitment patterns showed differences in participant characteristics between control and intervention sites. Feedback from a rejected NIHR Health Technology Assessment (HTA) application for a cluster-randomised effectiveness trial led to a change in design to individual patient randomisation (IPR) for the ICONS-II trial. It was considered that IPR would reduce the possibility of selection bias, indicated by the difference in participant characteristics between the cluster trial groups. It would also avoid the large inflation of sample size required by clustering (despite an upwards adjustment to the sample size to allow for potential contamination).

As a result, the trial was designed as a pragmatic, multicentre, individual-patient-randomised (1 : 1), parallel-group trial with an internal pilot, with 512 participants in the intervention group undergoing the systematic voiding programme (SVP) and 512 participants in the usual-care group. The original plan was that the results from an internal pilot, with an initial target of recruiting 355 participants and a total of 72 site-months of recruitment, would be used to determine progression to full trial. In May 2019, just before recruitment started, a decision was taken to shorten the duration of the pilot owing to the late opening of the trial. The expected number of site-months in the internal pilot was reduced from 72 to 45, and the number of participants was reduced from 355 to 225–315. A few other minor changes to the trial conduct were introduced, particularly after the first meeting of the ICONS-II Trial Steering Committee (TSC). The list of the minor changes to the ICONS-II trial protocol is provided in Appendix 6. The trial included a process evaluation that investigated fidelity to the intervention and usual care, and an economic evaluation.

Trial summary

Outcomes

The severity of UI was measured using the ICIQ-UI-SF51 total score. Higher scores indicate greater severity of UI. Participants who were catheterised did not complete the ICIQ-UI-SF.

The primary outcome was severity of UI at 3 months post randomisation, measured using the ICIQ-UI-SF for those not catheterised. Participants who were catheterised were given a maximum ICIQ-UI-SF score of 21.

The secondary outcomes were measured at 3 and 6 months post randomisation. Outcomes 1, 3, 4, 7 and 8 were also measured at discharge from the stroke unit:

1. Severity of UI, measured using the ICIQ-UI-SF (at discharge and 6 months).

2. Urinary symptoms, measured using the LUSQ47 (questions ‘Do you ever leak when you do any of the following?’ and ‘When you get the urge to pass urine, does any leak before you get to the toilet?’ only). The LUSQ was not administered to participants with an IUC.

3. Number of UTIs. A UTI was defined as (1a) symptoms (fever, indicated by temperature of > 37.5 °C, on two occasions, suprapubic tenderness, costovertebral angle pain or dysuria) or (1b) a positive blood/urinary tract pus or tissue culture; and (2) a positive urine culture or, for catheterised patients or if a urine culture was not possible, a dipstick positive for nitrite or white blood cells.

4. Number of days IUC was in situ.

5. Functional independence, measured using the Barthel Index of Activities for Daily Living,52 with lower scores indicating greater disability.

6. QoL, measured using the Incontinence Quality of Life (IQoL) instrument53,54 and transformed to the scale 0–100. Higher IQoL scores indicate better QoL.

7. Number of falls. A fall was defined as ‘any fall requiring a medical/health professional examination such as physical examination, X-ray and/or an intervention such as suturing or surgery’.

8. Death.

9. Cost-effectiveness, based on QALYs gained, estimated from the responses to the EuroQol-5 Dimensions, five-level version (EQ-5D-5L),55,56 using the UK tariff value, and symptom-free days.

In addition, the following baseline characteristics of the patient were recorded following consent:

• information collected from case notes52 –

  • date of birth (age to be calculated)

  • sex

  • ethnicity

  • date of admission

  • date of stroke onset

  • NIHSS at baseline38,46 (scores closest to the date of baseline questionnaire completion were used)

  • side of body affected by stroke (left, right, neither or both)

  • type of stroke (e.g. cerebral infarct, cerebral haemorrhage)

  • stroke subtype (Oxford Community Stroke Project classification57,58)

  • pre-stroke modified Rankin scale (mRS59)

  • pre-stroke living circumstances

  • pre-stroke UI

  • IUC in situ

  • type of UI (urge, stress, mixed, ‘functional’ or unclear)

• information collected from the participant, consultee or clinical staff –

  • date baseline questionnaire was completed

  • cognitive ability, determined using the 6-CIT,47 with higher cognitive scores indicating greater cognitive impairment; the total cognitive score was also dichotomised,60 with the participant deemed to have cognitive impairment if their cognitive score was ≥ 8

  • ICIQ-UI-SF

  • LUSQ

  • Barthel Index

  • the health-related quality-of-life measure the EQ-5D-5L.

Sample size

Based on the 3-month ICIQ-UI-SF total scores, 818 participants were required to provide ≥ 90% power to detect a 1.89-point between-group difference using an independent-samples t-test (alpha = 5%), assuming that ≤ 25% of the true effect would be lost to contamination. It was based on a minimal clinically important difference (MCID) of 2.5261 and a common standard deviation (SD) of 8.32, computed from data collected for the ICONS-I feasibility trial. 26 The target for randomised participants was 1024 to allow for 20% attrition. 26,36

Randomisation and blinding

Randomisation (in a 1 : 1 ratio) was stratified by baseline continence category [catheterised, slight (ICIQ-UI-SF score of 1–5), moderate (ICIQ-UI-SF score of 6–12) or severe/very severe (ICIQ-UI-SF score of 13–21)] using blocks of random length, which we found to be prognostic of outcome in our ICONS-I feasibility trial,26 and by site. The allocation procedure was delivered using the secure remote web-based system provided by Sealed Envelope Ltd (London, UK).

The information required to perform the randomisation was entered into the web-based system by the research nurse who obtained consent; the nurse was required to confirm that they had checked the eligibility criteria prior to the allocation being made. The system provided information on the allocated group to the research nurse in the stroke unit. The research nurse recorded the allocated group on the participant registration form, and informed the stroke unit ICONS-II trial staff, who delivered the intervention to patients randomised to the intervention group.

Blinding of health-care staff and patients was not possible. Data co-ordinators at Lancashire Clinical Trials Unit (CTU) were also not blinded, as they were handling data from the process evaluation in addition to baseline and outcome data. However, the trial statistician in Lancashire CTU, who was responsible for the SAP and the analysis of the effectiveness data, remained blinded to the identity of the group codes until after the effectiveness analysis had been performed. An unblinded statistician independently performed the analysis of the internal pilot data and the process evaluation data at the end of the trial.

Economic evaluation

The aim of the health economics evaluation was to determine the cost-effectiveness of the SVP compared with the usual care provided after stroke in secondary care. The economic analysis was focused on the EQ-5D-5L scores and their changes during follow-up, that is at baseline (discharge from the stroke unit) and at 3 and 6 months post randomisation.

Internal pilot study

An internal pilot examined the feasibility of participant recruitment and the success of strategies for minimising contamination in usual-care patients; it was initially intended that this would run for a total of 72 site-months of recruitment from 18 sites over a 6-month period (project months 3–8), with a staggered start (six sites in project month 3, six sites in project month 5 and six sites in project month 7). The target was approximately five patients per site per month, with an overall target of 355 participants by the end of project month 8. The following progression criteria were used initially:

• Continue – possible to recruit to time and target – recruitment ≥ 80% of target (minimum of 284 participants). If recruitment was ≥ 80% but < 100% at this stage, we would consider modifications to the recruitment procedure to ensure that our target for the whole trial was met.

• Continue but modify protocol – it may be possible to recruit to time and target with the implementation of contingency plans (e.g. increased number of sites) – recruitment ≥ 60% but < 80% of target (213–283 participants).

• Pause trial – it may not be possible to recruit to time and target – recruitment < 60% of target (≤ 212 participants).

Owing to difficulties with sites opening, in June 2019 the internal pilot recruitment targets were changed. The revised schedule for the recruitment criterion was 45 site-months of recruitment, reduced from 72 months. The target was set at five to seven patients per site per month, with an overall target of a minimum of 225 participants during the 4-month recruiting period (4 June to 30 September 2019). The criteria were revised to the following:

• Continue – possible to recruit to time and target – recruitment ≥ 80% of target (minimum of 180 participants). If recruitment was ≥ 80% but < 100% at this stage, we would consider modifications to the recruitment procedure to ensure that our target for the whole trial was met.

• Continue but modify protocol – it may be possible to recruit to time and target with the implementation of contingency plans – recruitment ≥ 60% but < 80% of target (135–179 participants).

• Pause trial – it may not be possible to recruit to time and target – recruitment < 60% of target (≤ 134 participants).

The retention rates, reflecting the completeness of follow-up, were not included in the pilot progression criteria as few primary outcomes were expected to have been received by that time.

However, the trial continued to experience major issues with recruitment and site opening, and, after the Monitoring Hub meeting with the NIHR HTA programme, held on 21 November 2019, the new recruitment targets for the internal pilot were set up by the HTA programme in its decision letter of 9 December 2019. The team was given a further 3 months, with effect from 1 January 2020, to determine if the viability of the trial could be achieved. By 31 March 2020. The seven mature sites (i.e. the seven sites listed as open since July 2018) were required to have achieved a minimum average rate of three participants per site per month, with a total of 63 participants for those sites during that period.

The fact that the majority of stroke patients in the ward were not in the ICONS-II trial should have given a reasonable degree of flexibility of staffing, even with the restriction that the intervention group participants should receive toileting care from the ICONS-II trial-trained (nursing) staff only, and that the usual-care group patients should receive toileting care from (nursing) staff who were not trained on the ICONS-II trial intervention only.

In addition, we aimed to check whether or not the trained staff were providing toileting assistance to usual-care participants to any extent. We felt that this could also be an indicator of a lack of fidelity to usual care. We felt that asking staff to record who performed toileting assistance would deter trained staff from substantial delivering toileting assistance to usual-care participants.

The contamination of usual care was examined as above by looking at (a) the extent to which ICONS-II trial staff provided toileting assistance to usual-care participants and (b) the extent to which usual-care participants received the intervention:

1. Extent to which ICONS-II trial staff provided toileting assistance to usual-care participants –

   • Stroke unit staff were asked to record brief details of toileting assistance (including the signatures of staff providing assistance) on fluid balance charts for patients receiving usual care. All fluid balance charts for eligible participants included in the data extraction were included in the analysis. This enabled us to monitor whether or not and how often ICONS-II trial staff delivered toileting assistance to usual-care patients.

   • Outcomes were set as the following:

     • Continue without modification: ≤ 25% of usual-care participants receive toileting assistance from ICONS-II trial staff on ≥ 50% occasions.

     • Continue with modification: > 25% of usual-care participants receive toileting assistance from ICONS-II trial staff on ≥ 50% occasions.

2. Extent to which usual-care participants received the intervention –

   • We conducted a review of case notes for all usual-care participants at discharge, focusing on the three key elements of the SVP:

     1. – Presence of a strategy for minimising IUC in the acute phase unless IUC is clinically justifiable (to relieve urinary retention or when fluid balance is critical). If IUC is clinically justifiable, presence of a strategy for review and removal (including a TWOC).

     2. – Presence of a comprehensive continence assessment [including continence history, diagnosis of UI (urge, stress, mixed or other), pattern of UI and assessment of relevant comorbid conditions].

     3. – Presence of a tailored treatment plan including behavioural approaches (specifically, a tailored voiding interval and evidence of review and adjustment).

   • Outcomes were set as the following:

     • Continue without modification: ≥ 75% of usual-care participants do not receive all three key elements of the intervention (avoidance of indwelling urinary catheterisation, comprehensive continence assessment and a tailored treatment plan).

     • Pause trial: > 25% of usual-care participants receive all three key elements of the intervention.

Effectiveness analysis

Process evaluation methods

The process evaluation was conducted to assess fidelity to (1) the intervention and (2) usual care, or, conversely, the potential level of contamination, repeating the interim analysis performed for the internal pilot on the full data set. As a result, for the assessment of fidelity to usual care, only the results from the final analysis are presented in this report.

Public and patient involvement

Service user involvement was an important part of the ICONS-II trial. Service users contributed to the development and progression of the trial throughout its course. During the trial, the study team had regular meetings with the PCPI group, with five meetings taking place between January and October 2019. At the meetings, the trial’s progress and difficulties with recruitment were discussed, including recommendations of ways to address the recruitment issues. Members of the PCPI group were engaged in discussion and reviewing the poster to advertise the trial in participating stroke units. The PCPI group also discussed problems or issues that patients might encounter with the SVP, in particular bladder training and prompted voiding. The group emphasised that it is beneficial for patients to be part of a programme.

The group felt that a visit from PCPI members would help increase the motivation of stroke unit staff to progress the trial and recruitment. Unfortunately, the planned site visits were not implemented owing to the trial’s pause and then its early stop.

The study team was involved in active communication and collaboration with the Speakeasy group. The Speakeasy PCPI group developed aphasia-friendly patient information leaflets and consent forms in the ICONS-I feasibility trial; these were used in the ICONS-II trial for patients with aphasia. Members of the PCPI group provided feedback on the study design and participant documents, and attended trial meetings on a regular basis. The TSC included members of the public. Our co-investigators Cliff Panton and, during the early stages of the trial, David Britt took an active part in the trial, including being involved in discussions at the trial development and set-up meetings, in addition to Trial Management Group meetings.

Approaches to maximising retention and patient follow-up

To maximise recruitment and improve retention, a number of approaches were implemented as a response to the low recruitment rate observed during the internal pilot.

The recruitment challenges were as follows:

• The number of patients identified as having an episode of UI within the required time after admission to the stroke unit was smaller than anticipated.

• Some eligible patients thought that they were at the end of their life and did not want to consider participating in research.

• Sites were not recruiting participants classified as palliative.

• Sites found it difficult to obtain consent/assent within the suggested time window, especially when personal consultees were involved, with additional difficulties in co-ordinating meetings of study staff and potential personal consultees.

• Sites were unclear about when it was appropriate to use a nominated consultee if no personal consultee was available.

• Although several intervention training sessions could take place at a site in 1 day, not enough stroke unit staff were trained to enable recruitment to begin immediately after training.

The changes to the methods in response to challenges were as follows:

• To improve the likelihood of identifying patients as having an episode of UI within the first 72 hours after admission to the stroke unit and to make sure that each eligible patient had the opportunity to take part in the study, stroke unit staff were asked to have a sensitive conversation with patients, including asking them if they were affected by UI and explaining that they may be eligible for the study.

• End-of-life and palliative care patients were eligible, but they might not have been considering participating in research. To improve the likelihood of participation, it was suggested to sites that, although most patients should be recruited within the 72-hour period, it was acceptable for end-of-life and palliative care patients to be consented after this period to allow them more time to come to terms with their situation.

• The inclusion criterion regarding level of consciousness was amended to allow patients who improved within the first 14 days after admission to the stroke unit to be considered and approached. These patients were still required to meet the remaining inclusion criteria within 72 hours of admission to the stroke unit.

• Follow-up was to have been mainly postal, but the protocol1 included the option for this to be conducted over the telephone if the patient preferred this or when postal questionnaires were not returned. It became apparent that most of this patient group fell into the vulnerable category for COVID-19, were shielding and would have difficulty with postal returns. All participants were contacted by CTU staff who gave them the option of completing the questionnaire over the telephone. The order of the questionnaire was also adapted so that core outcome data questions and the 3-month ICIQ-UI-SF, LUSQ, Barthel Index and EQ-5D-5L were asked first when the questionnaire was completed over the telephone.

Recruitment and participant flow

A total of 3492 participants were screened for eligibility across the 17 sites; of those screened, 562 (16%) were deemed eligible. Of those eligible, 161 (29%) were consented and 157 (28%) were randomised, with 78 (50%) of those randomised allocated to usual care and 79 (50%) allocated to the intervention. The details of the patients who were screened but not eligible and of those who were eligible but not consented are given in Table 1.

For full details, see the Consolidated Standards of Reporting Trials (CONSORT) diagram (Figure 1).

FIGURE 1.

The CONSORT diagram. a, ICIQ-UI-SF score of 1–12 = slight/moderate, and score of 13–21 = severe/very severe.

Details of recruitment by site are given in Table 2, which provides the number of patients screened, eligible, consented and randomised by each site.

After the internal pilot revealed the difficulties with recruitment, we planned to increase the number of sites to compensate for slow recruitment. However, as the trial was paused because of COVID-19, these plans were not fully implemented. The last patient was recruited on 16 March 2020, the last 3-month outcomes were received on 16 June 2020, and the last 6-month outcomes were received on 22 September 2020. We had 17 sites contributing to the trial. Figure 2 shows the recruitment chart.

FIGURE 2.

Participant recruitment chart.

More details related to the recruitment, the attrition and the implications that recruitment had for the trial results are provided in Brief account of the trial.

Baseline characteristics

The usual-care and intervention group participants were similar in most characteristics and demographics at baseline, as shown in Table 3, with two notable exceptions: (1) the proportion of female patients was larger in the intervention group (54%) than in the usual-care group (42%) and (2) the proportion of participants living alone was larger in the intervention group (39%) than in the usual-care group (29%).

The usual-care group and intervention group participants were similar in most clinical and stroke-related characteristics at baseline, as shown in Table 4, with the notable exception of the median Barthel score, with the intervention group (median score of 6.5) being more functionally independent than the usual-care group (median score of 4).

The usual-care group and intervention group participants were similar in most incontinence characteristics at baseline, as shown in Table 5, apart from pre-stroke use of incontinence pads. A larger proportion of the intervention group (34%) than of the usual-care group (23%) were reliant on incontinence pads pre stroke.

Table 6 presents the number of participants with IUCs at baseline, where they were catheterised and the reason why they were catheterised. It is of note that a larger proportion of participants were catheterised because of retention of urine in the usual-care group (80%) than in the intervention group (64%).

Table 7 shows the ICIQ-UI-SF total scores at each time point they were collected, with the catheterised patients excluded, and then included with the maximum possible score. With catheterised participants excluded from the calculation of ICIQ-SF total, the usual-care and intervention groups were balanced at baseline. The score for each group decreased at each subsequent time point, with the usual-care group performing better than the intervention group at both the 3- and 6-month time points. In both scenarios, with catheterised participants excluded from the calculation of the ICIQ-UI-SF total, and with catheterised patients given the maximum score on the ICIQ-UI-SF total score, the intervention and usual-care groups were balanced at baseline, and the group with the better score switched from the intervention group at 3 months to the usual-care group at 6 months. The most improvement occurred between baseline and discharge, with reductions of > 4 points in all total scores from baseline and the intervention group showing the most improvement during the period.

Internal pilot

During the internal pilot, the target of a minimum of 225 participants recruited was not met, largely because only seven sites, rather than the planned 18, contributed participants during the period and because of difficulties with opening sites. By the end of September 2019, a total of 32 participants had been randomised. Following a monitoring meeting with the HTA programme in November 2019, rather than terminating the trial, the HTA programme allowed recruitment to continue for another 3 months (January–March 2020) to enable a clearer assessment of whether or not recruitment could be improved by implementing of a number of strategies in the seven established sites, as described in Brief account of the trial.

Two of the seven original established sites were closed, one on 4 December 2019 and one on 20 December 2019, because of difficulties identifying suitable participants and being able to deliver the intervention in a systematic way owing to staff shortages. The other five established sites continued to recruit, as did the later-opening sites.

In relation to contamination assessment, the analysis of both the usual-care group participants and the pre-implementation case-note review patients showed similar very low levels of structured voiding but some level of appropriate care (which included TWOCs). There was a strategy for minimising IUC in the acute phase unless clinically justifiable among a substantial percentage of the usual-care group participants (3/6; 50%), but, given the very small numbers, this was not inconsistent with the percentage observed among the pre-implementation case-note review patients (21/65; 32%). However, there was very little evidence of a comprehensive continence assessment being performed on usual-care participants, and it did not progress to a tailored treatment plan in any usual-care participants; this was consistent with the prevalence observed in the pre-implementation case-note review patients.

The conclusion of the internal pilot, therefore, was that there was no evidence of contamination of the usual-care group with the key components of the SVP intervention.

Trial findings: data completeness at follow-up

The results presented in this section and onwards relate to all follow-up data available at 3 or 6 months post randomisation.

At the primary time point of 3 months, of the 157 randomised participants, 94 (59%) returned a follow-up questionnaire, 18 (11%) had died and 16 (10%) had withdrawn from the trial. However, only 86 provided primary outcome data, so the percentage of missing primary outcome data was very high (45%).

When comparing the groups at 3 months, of the 94 returned questionnaires, 56 (60%) were from the usual-care group and 38 (40%) were from the intervention group. Four (5%) participants in the usual-care group and 12 (15%) participants in the intervention group withdrew, and there were 10 (13%) non-responders in the usual-care group and 19 (24%) non-responders in the intervention group at 3 months.

Overall, there was a substantial difference in missing data rates between the usual-care and intervention groups for the primary outcome (ICIQ-UI-SF at 3 months), with primary outcome data for only 35 out of 79 (44%) participants in the intervention group, compared with 51 out of 78 (65%) participants in the usual-care group. There was, therefore, strong evidence of differential percentages for missing primary outcome data in the two groups (p = 0.008; chi-squared test).

At 6 months, 46 (59%) participants from the usual-care group and 46 (58%) participants from the intervention group returned the questionnaire. Although the number of missing primary outcome data was more balanced between the groups at 6 months, it continued to be very large (Table 8).

Figures 3 and 4 show the number of participants who returned and the number of participants who did not return their questionnaires by the relevant follow-up time point, and the numbers of deaths and withdrawals from the study. Figure 3 covers the period between September 2019 and June 2020, and Figure 4 covers the period between December 2019 and September 2020. The last 3-month questionnaire was received on 16 June 2020 and the last 6-month questionnaire was received on 22 September 2020. However, in March 2020, new participant recruitment to the trial had to be paused because of the COVID-19 pandemic.

FIGURE 3.

Count in each follow-up category by 3-month questionnaire due date.

FIGURE 4.

Count in each follow-up category by 6-month questionnaire due date, excluding 3-month participant deaths/withdrawals.

Brief account of the trial

To provide context for the results on recruitment, participant flow and site opening, this section summarises the background of the trial and how it developed.

The initial trial protocol was written on 5 January 2018, and ethics and HRA approvals were granted by early April 2018 in preparation for the initiation of funding in May 2018 and for recruitment to start in July 2018. However, owing to internal delays, although the trial protocol was updated on 4 June 2018, ethics approval for changes was not obtained until late 2018.

Even after HRA and ethics approvals had been obtained, the trial had a difficult start, with aspects of the design being unpopular with prospective sites, which was evident from verbal feedback during site visits and from e-mail correspondence.

The trial opened to recruitment in June 2019, with the first participant recruited on 13 June 2019.

Implementing the trial with IPR required actions aimed at minimising the potential contamination, which was recognised as a risk because participants in both trial groups were treated in the same stroke unit. Specific points of concern and relevant mitigating actions are as follows:

• The number of patients identified as having an episode of UI within the first 72 hours after admission to the stroke unit was smaller than anticipated.

  Actions – sites were informed that patients who were self-caring and mobile often dealt with incontinence in secret because they feel embarrassed. Many of these patients resigned themselves to incontinence, thinking that it was inevitable following a stroke or because of their age. For these patients, it was not possible to detect the presence of UI from fluid balance charts or case notes, and active case finding was required. We asked stroke unit staff to have a sensitive conversation with these patients, including asking them if they were affected by UI.

• Issues around palliative care.

  Some eligible patients thought that they were at the end of their life and did not want to consider participating in research.

  Actions – the preference was for most patients to be recruited within the 96-hour period after admission. It was made clear to sites that patients could be consented after this period to give them more time to come to terms with their situation.

  Sites were not recruiting participants because they were classed as ‘palliative’.

  Actions – sites were advised to consider patients classed as ‘palliative’ very carefully; such patients could live for a substantial time, and improving continence might be a reasonable and achievable goal for these patients, with the potential to improve their QoL.

• Suggested 96-hour time window for consenting participants.

  Sites found it difficult to obtain consent/assent within the suggested 96-hour window, especially when personal consultees were involved.

  Actions – the related documentation (ethics application and protocol) was written with some flexibility and notes that ‘informed consent will be made within 96 hours of admission to the stroke unit in the majority of cases’. Sites were informed that there would be occasions when recruitment within 96 hours of admission would not be possible and so patients could be consented after this period.

  We amended the inclusion criterion for level of consciousness to allow patients who improved within the first 14 days after admission to the stroke unit to be considered and approached. The revised inclusion criterion (protocol inclusion criteria, section 4.2)1 was changed to ‘conscious: “Alert” or “Not alert but arousable” (NIHSS Point 1A score of 0 or 1) on the NIHSS within 14 days of admission to the stroke unit. Note that patients must be recruited within 72 hours of meeting this criterion wherever possible.’ These patients were still required to meet the remaining inclusion criteria within 72 hours of admission to the stroke unit. This change was submitted for approval on 30 August 2019, and Wales REC 5 approved this amendment on 27 September 2019.

• Using a personal/nominated consultee.

  In some cases, it was difficult to co-ordinate timing so that the study champion or CRN research nurse and potential personal consultees were at the unit at the same time.

  Actions – we advised sites to send the personal consultee information leaflet and consultee letter of invitation to potential personal consultees in advance to allow them time to consider these before they visited. If they were willing to provide consultee consent for their relative to take part, they could sign the consent form when they visited.

  Sites were unclear when it was appropriate to use a nominated consultee if no personal consultee (family member or friend) was available.

  Actions – we had approval from the REC to include a nominated consultee if no personal consultee (family member or friend) was available. Following consultation with the REC and our PCPI groups, we updated our advice to sites, asking them to consider this option if no personal consultee could be identified within 4 days of the patient being admitted to hospital.

• Staff training.

  We anticipated that sites would begin to recruit participants following the ‘intervention training’ session; however, although two or three intervention training sessions could take place at a site in 1 day, this was not enough to enable sufficient numbers of the stroke unit team to be trained to deliver the intervention immediately. We adopted a ‘train-the-trainer’ approach, and sites started to screen/recruit only once the study champion had undertaken further training sessions with stroke unit staff. This often took several weeks.

  Actions – we asked sites to ensure that all staff were trained within 4 weeks of the intervention training session.

• Low staffing levels.

  Some sites that initially had staff capacity to take part saw a reduction in staffing levels following the green-light letter. This required the study champion to identify and train further staff to deliver the intervention. In addition, there were low staffing levels in the summer months owing to holiday leave. Nine sites were unable to continue in the trial owing to inadequate staffing levels. This affected the original internal pilot period, but was less of an issue afterwards.

  Action – after summer 2019, some of these issues were resolved. Two sites that had particular issues with staff shortages were closed and replaced.

In total, 17 sites participated in the study: 14 sites recruited at least one participant and three sites were in the set-up stage only. Participant recruitment during the extension to the internal pilot progressed much better towards the target of 63 participants in 3 months in the original sites. Capacity and capability approval was obtained from 12 sites, and 11 of these recruited participants. Two of the original seven established sites had difficulties identifying suitable participants and being able to deliver the intervention in a systematic way owing to staff shortages and, in some instances, non-engagement with the delivery. One of sites was closed on 4 December 2019 and the other was closed on 20 December 2019. Recruitment continued in the remaining five established sites and in a further 10 sites that started recruitment later.

During March 2020, recruitment slowed because of the onset of the COVID-19 pandemic. In early March, it became very clear that COVID-19 was dominating resources, and sites started to report that they were unable to support further recruitment to and activities for the ICONS-II trial. Many staff designated as study champions were reallocated to frontline services, and CRN staff (as per HRA guidance) were prioritising recruitment to COVID-19 studies. The final patient was recruited on 16 March 2020. By 26 March 2020, none of the ICONS-II trial sites was conducting study activities. On this date, we contacted all sites with a formal notification of a pause to study activities due to the impingement to safety measures. At this point, 157 participants had been recruited.

In autumn 2020, we considered the possibility of restarting the trial and completed the reanalysis of the data on fidelity to usual care, assessing the degree of contamination. We concluded that the level of contamination of the usual-care group with the key components of the SVP intervention was much lower than the conservative 25% on which the original sample size had been based. Furthermore, we concluded that the impact of contamination of the usual-care group on the treatment effect was unlikely to be > 10%. However, recruiting the required number of participants (≥ 564 further participants, although the number varied depending on the assumptions made) after resuming the trial was unfeasible within the time frame and remaining budget and in the light of the ongoing COVID-19 situation. Moreover, the trial had high and potentially differential rates of missing primary outcome data between the intervention and usual-care groups (56% vs. 35%, respectively). It was also observed that the length of hospital stay in this study was much shorter than that in the cluster-randomised feasibility trial, ICONS-I. The overall median length of stay (IQR) in the stroke unit was 27 (16–45) days in the ICONS-II trial and 47 (30–68) days in the ICONS-I feasibility trial. It was concluded that a purely hospital-based intervention would not provide a patient with a sufficient amount of the SVP to be effective in managing incontinence; a minimum of 6 weeks is recommended for a bladder training programme. 62

Over the course of the study, the trial team identified various barriers that contributed to the difficulties experienced. Several were suspected potential barriers, but these suspicions were not supported by evidence until after recruitment started:

• Study champions were required at each site for implementation and support. There were delays in identifying and appointing these study champions.

• Staff were concerned that patients might see a difference in care, with concerns centred around intervention participants getting ‘better treatment’.

• Staff were concerned about the extra work/paperwork required (in the context of large numbers of vacancies).

• Managing staff rotation was more complicated when seeking to maintain levels of intervention staff.

• Staff had difficulty grasping issues of contamination, and the need for toileting of intervention participants to be undertaken by intervention-trained staff.

• There was poor completion of paperwork.

Trial findings: primary outcome analysis

The study did not recruit a sufficient number of participants, with only 157 (15%) participants recruited out of the target 1024. Only 11% (86 of the target 818) of the target primary outcome measures were available, and they were not balanced between the intervention and usual-care groups.

The results reported are for the intention-to-treat complete-case analysis. For the primary outcome, at 3 months, there were 51 observations in the usual-care group and 35 observations in the intervention group. The results of the primary outcome are shown in Table 9. Participants allocated to the intervention group had a lower ICIQ-UI-SF score than those allocated to the usual-care group at 3 months post randomisation: the mean (SD) ICIQ-UI-SF score at 3 months was 9.1 (7.8) in the usual-care group and 8.1 (7.4) in the intervention group (see Table 9). All comparisons must be interpreted with caution, as they are highly imprecise owing to the small numbers of participants.

The assumptions of the model were checked and considered to be reasonably satisfied. The model was also supported by Hausman’s specification test (p = 0.80).

Trial findings: secondary outcome analyses

Process evaluation

Eligibility criteria for entering a systematic voiding programme

We assumed that all of the participants who were incontinent at baseline, had the continence assessment performed and had the 3-day diary completed were eligible to enter a SVP, unless they were found to be continent.

Participants who were catheterised at baseline became eligible to enter a SVP if they were incontinent after a successful TWOC. Following IUC removal, participants had to complete a 3-day bladder diary to assess the pattern of UI and a continence assessment to determine the type of UI.

Of the 79 participants allocated to the intervention group, 51 (65%) were eligible to enter the SVP. However, only 35 out of 79 (44%) participants were put on a treatment regime (Figure 5).

FIGURE 5.

Flow of intervention group participants through the ICONS-II trial intervention.

The main documented reasons for ineligibility for a SVP in both the catheterised participants and those incontinent at baseline were participants being continent, catheterised or discharged early. Participants discharged within 4 days of being randomised were not put on a SVP.

A summary of the reasons why participants were ineligible for a SVP is presented in Table 15.

TABLE 15 - Eligible participants for SVP and reasons for ineligibility overall and by baseline status

Participant status	Incontinent participants, n/N (%)	Catheterised participants, n/N (%)	Total, n/N (%)
Baseline status	56/79 (71)	23/79 (29)	79/159 (50)
Eligible for SVP	44/56 (79)	7/23 (30)	51/79 (65)
Ineligible for SVP	12/56 (21)	16/23 (70)	28/79 (35)
Reason ineligible for SVP
Continent	1	3	4
Functional UI		1	1
Catheterised throughout		5	5
IUC reinserted	1		1
Discharged	5	2	7
COVID-19 pandemic	1		1
Unknown	4	5	9

A total of 35 (44%) of the intervention group participants and 35 (69%) participants eligible for a SVP were put on a treatment regime.

Of the 16 (31%) participants who did not commence a SVP, the majority (n = 15; 94%) had a recommended initial treatment plan documented. However, the reason why a treatment regime was not instituted was recorded for only 7 of these 15 patients. One participant was continent, three were discharged early and three remained in hospital when the trial was paused in March 2020 owing to the COVID-19 pandemic. Further details of these 15 participants can be found in Appendix 3, Table 32.

Just over half of the participants recorded as commencing a SVP were recorded as commencing the correct regime (n = 18; 51%), and only five (14%) participants were recorded as commencing a treatment regime different from that they had been initially recommended (Table 16). Documented reasons why participants changed from prompted voiding to bladder training were participants being aware of the need to void (n = 2), being able to have some control over their bladder (n = 1) and not having the 3-day diary properly completed (n = 1). Reasons why participants changed from bladder training to prompted voiding were not documented.

TABLE 16 - Eligible participants who entered a SVP

BT, bladder training; PV, prompted voiding.

Participant regime	n/N (%)
Eligible for SVP	51/79 (65)
SVP commenced	35/51 (69)
SVP not commenced	16/51 (31)
Initial treatment plan	15/16 (94)
Unknown	1/16 (6)
Allocated to correct regime	18/35 (51)
PV	12/18 (67)
BT	6/18 (33)
Allocated to incorrect regime	5/35 (14)
PV	1/5 (20)
BT	4/5 (80)
Unknown	12/35 (34)
Missing 6-CIT	11/12 (92)
Started on PV	8/11 (73)
Started on BT	3/11 (27)
Missing frequency of leak	1/12 (8)
Started on BT	1/12 (8)

For 12 (34%) participants, it was not possible to determine whether or not the correct regime had been started. Eleven of these participants could not communicate either verbally or in writing and, therefore, the 6-CIT could not be administered at baseline to determine whether or not they were cognitively impaired.

Problems starting and delivering the trial

The trial had a difficult start: verbal feedback during site visits and e-mail correspondence demonstrated that the design was unpopular with prospective sites, in particular in relation to IPR and the differences in care delivered to the two groups.

The intervention required 50% of staff to be trained in mechanisms of incontinence, the prompted voiding/bladder training programme for the intervention group, recording daily toileting logs and weekly assessment of progress. The trial funded site study champions for implementation and support.

The trial team identified various barriers contributing to the difficulties:

• Delays in identification and appointment of study champions.

• Staff concerns that patients would see differences in care, with concerns centred around intervention participants receiving ‘better treatment’.

• Staff concerns about the extra work/paperwork required (exacerbated by large numbers of staff vacancies).

• Managing staff rotation to maintain levels of intervention staff.

• Staff difficulty grasping issues of contamination and the need for the toileting of intervention participants to be undertaken by intervention-trained staff and the toileting of usual-care participants to be undertaken by staff who had not been trained in the intervention.

• Differences in stroke care settings and pathways between the ICONS-I feasibility trial and the ICONS-II trial [e.g. the estimated length of stay in a stroke unit substantially differed between the ICONS-I feasibility trial and the ICONS-II trial, with a median (IQRs) of 47 (30–68) and 27 (16–45) days, respectively]. Further research and work are needed to assess any unfavourable effect of the differences in care setting on the amount of intervention received, effectiveness of the intervention and attrition.

• Poor adherence and completion of paperwork by staff.

In December 2019, the team conducted an internal review following the monitoring hub meeting with the HTA programme on 22 November 2019. This review shaped radical adjustments to the study: closing underperforming sites, retraining site staff, adapting paperwork to reflect individual site patient records, sourcing new sites, setting performance targets by encouraging sites to provide complete paperwork sets rather than recruitment alone, and moving the focus from recruitment to intervention delivery. To improve site engagement in the trial, we emphasised that consent alone did not qualify for satisfactory participation in the study and that the introduced measures were linked to performance reporting within site research and development departments. Improvement in recruitment was steady until COVID-19 caused study suspension. Sites halted participation without completing data collection, and often before they had transferred outstanding data to the CTU.

From early March 2020, it became very clear that COVID-19 was dominating resources, and sites started to report that they were unable to support further recruitment and activities for the ICONS-II trial. Many staff designated as study champions were being reallocated to frontline services, and CRN staff (as per HRA guidance) were prioritising recruitment to COVID-19 studies. By 26 March 2020, none of the ICONS-II trial sites were conducting study activities. We contacted all sites with a formal notification of a halt to study activities. HRA guidance at this time was that no formal notification needed to be sent to HRA/REC, as this was countrywide.

We had hoped that we would be able to restart the trial, and discussed this with our sites and our funders; however, as time went on, it became apparent that the situation was not improving. We therefore conducted a full feasibility assessment of the study and reviewed the recruitment, attrition, data completeness, contamination, sample size and financial implications of restarting. Regretfully, it became apparent that a restart was not feasible either financially or in the current clinical and research landscape.

We recruited 157 participants, with an expectation of being able to carry out 3- and 6-month follow-up data collection. It was apparent during the COVID-19 pandemic that most of this patient group fell into the vulnerable category and were shielding and so had difficulty completing postal returns. We therefore contacted all participants, offering them the option of completing outcome data collection over the telephone. Although some participants enjoyed the chance to chat, some felt that the process was onerous. Many data collection items involved multiple-choice Likert scales, which were time-consuming to administer, and some participants found it difficult to remember the options. To assist participants, we adapted the order of the questions so that core outcome data questions were asked first.

Many participants reported anxiety and loneliness when contacted, and we experienced a high attrition rate. The questionnaire included questions on anxiety and depression and, in line with the protocol,1 any participant scoring above the protocol threshold (either by postal return or over the telephone) triggered a referral to their general practice. This involved a letter being faxed to their GP, informing them of the participant’s low mood/anxiety, and the requirement for the general practice surgery to acknowledge receipt as a transfer of responsibility (staff from the CTU called the surgery telling them that the fax/e-mail was coming and the general practice surgery faxed/e-mailed confirmation of receipt).

The number of missing data was problematic, as there was a high proportion of missing primary outcome data (attrition rate 46%), making the trial seriously underpowered.

Contamination estimated in the process evaluation analysis was lower than the worst-case scenario of 25% stated in the funding application. However, overall, it remains plausible that contamination of the usual-care group had an impact on the treatment effect. The process evaluation results show that this was limited to the component designed to minimise IUC in the acute phase, so the impact on the overall effect is likely to be far less than the conservative 25% on which the original sample size was based. The analysis assessing the degree of contamination in the usual-care group, together with the fact that the provision of toileting assistance by ICONS-II trial staff alone is unlikely to have had a substantial impact, at least not on the delivery of a SVP, has allowed us to conclude that the level of contamination of the usual-care group with the key components of the SVP intervention was unlikely to be > 10%. This would have allowed an adjustment to the relevant assumption and a reduction in sample size, although the high attrition rate implies that the overall missing data rate would be about 33% by the end of the trial, even with 20% target attrition after reopening, which would have an impact on validity. The trial would require another 564 participants, which would not be feasible within the time frame and remaining budget and in the light of the ongoing COVID-19 situation. All of this led to the conclusion that the ICONS-II trial was not viable.

Difference in attrition rates

There were concerns about the very high and differential rates of missing primary outcome data (ICIQ-UI at 3 months) between the intervention and usual-care groups (56% vs. 35%, respectively), which also suggested that a future trial would need to consider how this attrition might be reduced, otherwise a robust trial would not be possible. There was a substantial, statistically significant difference (p = 0.008) in attrition rate between the groups for the primary outcome, with primary outcome data for only 35 out of 79 (44%) participants in the intervention group and 51 out of 78 (65%) participants in the usual-care group (see Table 8). The attrition rate was more balanced at 6 months. Although the rate continued to be very high, there was an improvement in the intervention group (56% at 3 months vs. 47% at 6 months), with some participants who did not return the 3-month questionnaire returning their questionnaires at 6 months.

Considering the fact that COVID-19 had a much greater effect on the 6-month questionnaire return rate, including restricting access to the post room at University of Central Lancashire, the improvement in attrition rate could, potentially, be attributed to measures that the team applied to boost retention just before COVID-19 emerged. However, it can also be noted that the improvement could be a positive effect of COVID-19 (rather than a positive effect of the CTU measures), as participants may have changed their behaviours because of COVID-19 and introduced changes to their lifestyle.

Differences between the ICONS-I feasibility trial and the ICONS-II trial

The ICONS-II trial was informed by the ICONS-I feasibility trial. In the ICONS-I feasibility trial, 12 sites commenced recruitment between January 2011 and January 2012, with no sites dropping out. Recruitment started on 1 January 2011 and ceased at all sites on 31 July 2012. The parameter estimates obtained from the feasibility study were used in the ICONS-II trial sample size calculation and when planning intervention implementation. However, it became evident that there was a pronounced difference between the trials in the length of hospital stay. In the ICONS-I feasibility trial, the overall median (IQR) length of stay in the stroke unit was 47 (30–68) days; in the ICONS-II trial, it was 27 (16–45) days. This may have had a substantial impact on the amount of intervention patients received in the ICONS-II trial. It became apparent that a purely hospital-based intervention would be unlikely to provide a sufficient amount of the SVP to a patient to be effective in managing incontinence.

Other details of the ICONS-II trial length of stay trend during the recruitment period are presented in Appendix 4, Tables 33 and 34.

Study limitations

The main study limitation was the small sample size, which made the trial underpowered to detect intervention effects. The large number of missing data due to high attrition and problems with intervention fidelity seriously limited the usefulness of the study results. The availability of the data was also limited by poor adherence and completion of paperwork by staff.

A further limitation was that the voiding element of the intervention could not be delivered for as long as had been expected based on data from the ICONS-I feasibility trial. It may be that this part of the intervention, which was based entirely within the hospital setting, was not of sufficient duration to have a reasonable chance of having an impact, given the increasingly short length of stay following a stroke in today’s NHS.

As summarised in Problems starting and delivering the trial, a number of difficulties were related to hospital staff involvement in the trial that contributed to limitations affecting the fidelity of the intervention. These included staff concerns about visible differences in care between intervention and usual-care participants, concerns about extra work and excessive paperwork, problems with managing staff rotation to maintain levels of intervention staff and staff’s difficulty understanding issues of contamination between the two ICONS-II trial groups. All of these issues were reported ad hoc to the CTU team, and reflect the limitations of the trial process evaluation, which did not include this aspect.

Another limitation was the influence of COVID-19 on the study. The pandemic affected many aspects of the ICONS-II trial, as discussed in Chapter 3, Brief account of the trial including building staff pressures and public concern about going into hospital for several months, which interfered with the sites’ ability to participate in the study. It was not, however, possible to completely measure the impact of COVID-19 on the trial, which introduced a further limitation to the analysis performed.

Recommendations, lessons learnt and future study planning

Future studies should focus on interventions that are likely to be deliverable in the NHS. The intervention needs to fit with current practices and challenges in the NHS. The approach to stroke care has changed since the ICONS-I feasibility trial, with a more intense hyperacute phase, a considerably greater need for monitoring of vital signs and shorter patient stays. This is reflected in the substantially shorter length of stay in the ICONS-II trial than in the ICONS-I feasibility trial. This affected the opportunity for delivery and, hence, the efficiency of the SVP for UI. One of the main barriers to study start-up and completion was the time taken to implement the continence intervention in an already highly stressed NHS.

In planning a future study, we would recommend that testing a programme of catheter avoidance and removal of catheters is the most promising intervention to take forward. As there were fewer catheterised patients at all three time points in the intervention group in the ICONS-II trial, we have shown that the intervention is feasible, and observational data from other studies64,65 suggest that the presence of catheters increases mortality. A programme of catheter avoidance and catheter removal takes considerably less nurse time to implement than a continence programme, and is, therefore, more likely to be testable in a clinical study and more likely to be implemented once shown to be effective. This approach was also supported by the ICONS-I feasibility trial’s findings, which demonstrated some improvements in the intervention group compared with the usual-care group in terms of catheter removal (median 13 days, IQR 5–35 days, compared with median 20 days, IQR 8.75–35.25 days, respectively) and numbers of participants catheterised at discharge (intervention, n = 19; 15.2%; usual care, n = 35; 21.3%). 25

However, incontinence remains a prevalent and distressing problem, and future studies should not be avoided just because they are difficult. Experience from the ICONS-II trial suggests that such a complex and time-consuming intervention is not feasible in the hyperacute stroke environment. An alternative approach would be to consider a randomised controlled trial in which patients are assessed and recruited in hospital or early after discharge into the community, but the majority of the delivery of the intervention would be patient and carer led and take place in the community. Future studies should consider using a more appropriate intervention design and exploring the potential to reframe the research as an implementation study/trial, with a suitable design to evaluate effectiveness in parallel with the implementation. This would require more feasibility work. Both the cluster-randomised design and the individual-patient-randomised design proved to be difficult for clinical areas to implement.

As part of a future implementation trial, an embedded qualitative study, including an exploration of the views of patients, carers and community staff, would be useful for improving the quality of the evidence needed to fully understand the facilitators of and barriers to successful implementation of the intervention in this trial.

The use of study champions was a critical element of staff training in and staff engagement with the programme. The late appointment of some study champions had a large impact on trial recruitment. As COVID-19 arrived and NHS staff, including study champions, were diverted to COVID-19 duties, recruitment dropped rapidly. A future trial design should include study champions to support delivery and embedding of the intervention for maximum effectiveness. These study champions would not only help deliver the study, but also be instrumental for ensuring implementation of the findings.

Dedication

Dr David Britt, a member of the ICONS-II trial Patient and Public Involvement (PPI) group, died during the trial. He made a large contribution to this trial and the ICONS-I feasibility trial, and is greatly missed.

Contributions of authors

Caroline Watkins (https://orcid.org/0000-0002-9403-3772) (Professor of Stroke and Older People’s Care) co-led the original grant application, co-designed the trial, interpreted the study findings and co-led the drafting of the report. She became chief investigator when Lois Thomas retired.

Svetlana Tishkovskaya (https://orcid.org/0000-0003-3087-6380) (Senior Lecturer in Health Statistics, Statistician) co-wrote the SAP, conducted the data analysis, prepared the results for publication, interpreted the study findings and led the drafting of the report.

Chris Brown (https://orcid.org/0000-0002-0162-4988) (Senior Research Assistant, Statistics) co-wrote the SAP, analysed the study data, prepared the results for publication, interpreted the study findings and co-led the drafting of the report.

Chris Sutton (https://orcid.org/0000-0002-6406-1318) (Senior Lecturer in Clinical Trial Statistics) co-led the original grant application, co-designed the trial, co-wrote the SAP, provided statistical oversight, interpreted the study findings and assisted with the drafting of the report.

Yvonne Sylvestre Garcia (https://orcid.org/0000-0002-3288-9006) (Research Fellow in Clinical Trials Statistics) analysed the study data and assisted with the drafting of the report.

Denise Forshaw (https://orcid.org/0000-0001-5725-3736) (Principal Clinical Trials Manager) led on trial data management, contributed to data collection and assisted with the drafting of the report.

Gordon Prescott (https://orcid.org/0000-0002-9156-2361) (Reader in Medical Statistics) provided statistical oversight and co-led on the drafting of the report.

Lois Thomas (https://orcid.org/0000-0001-7177-0393) (Professor of Health Services Research) led on the original grant application, co-designed the trial, and contributed to data collection and drafting of the report.

Christine Roffe (https://orcid.org/0000-0002-5259-6649) (Professor of Stroke Medicine) interpreted the study findings and assisted with the drafting of the report.

Joanne Booth (https://orcid.org/0000-0002-7870-6391) (Professor of Rehabilitation Nursing) interpreted the study findings and assisted with the drafting of the report.

Kina Bennett (https://orcid.org/0000-0002-1427-688X) (Research Operations Manager) was part of the study management group via the TMG, provided advice and guidance as lead NHS site representative on research governance process and local NHS set-up processes, assisted with data collection, substantially contributed to the design of the work and approved the version to be published.

Brenda Roe (https://orcid.org/0000-0001-8227-0116) (Professor Emerita of Health Research) interpreted the study findings and assisted with the drafting of the report.

Bruce Hollingsworth (https://orcid.org/0000-0002-4314-6523) (Professor of Health Economics) contributed to the design of the health economics analysis, interpretation of the data and revision the work, and approved the version to be published.

Ceu Mateus (https://orcid.org/0000-0001-6219-219X) (Professor of Health Economics) conducted the health economics analysis and assisted with the drafting of the report.

David Britt (PPI member) contributed as a PPI representative for the trial.

Cliff Panton (PPI member) contributed as a PPI representative for the trial.

Publication

Thomas L, Roffe C, Booth J, Chapple C, Watkins C, Roe B, et al. ICONS II: Identifying Continence OptioNs after Stroke randomised controlled trial. Br J Neurosci Nurs 2018;14(Suppl. 2):S24–5.

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.

Stress incontinence

Table 22 shows the numbers of participants with stress incontinence at 3 and 6 months. The estimated odds of stress incontinence, adjusted for baseline stress incontinence type, were lower in the intervention group than in the usual-care group at both follow-up time points (3 months: OR 0.80, 95% CI 0.28 to 2.31; 6 months: OR 0.60, 95% CI 0.20 to 1.81).

Urge incontinence

Table 23 shows the numbers of participants with urge incontinence at 3 and 6 months. The estimated odds of urge incontinence, adjusted for baseline urge incontinence type, were higher in the intervention group than in the usual-care group at both follow-up time points (3 months: OR 1.69, 95% CI 0.37 to 7.78; 6 months: OR 1.83, 95% CI 0.43 to 7.76).

Number of urinary tract infections

For patients who had UTIs at discharge, and at 3 and 6 months, the distribution of the number of infections is presented as frequencies (%) by trial group in Tables 24–27. For example, Table 24 shows the number of UTIs by the time of discharge. In the usual-care group, 17 people experienced a single infection; in the intervention group, 12 people experienced a single infection and one person experienced two infections. Table 25 shows the number of people who experienced between one and five infections between discharge and 3 months. Table 26 shows the number of people who experienced between one and six infections in the previous 3 months; this is completed at 6 months but reported since discharge only.

As the numbers were small and regression analysis was not feasible, the Mann–Whitney U (Wilcoxon rank sum)-test was applied. It did not show a statistically significant difference in the number of UTIs between the two groups at each time point. (For details, see the summary table of the results, Table 31.)

Number of falls

The numbers of people who had experienced at least one fall by discharge and by 3 and 6 months are presented as frequencies by trial group in Table 28.

There had been fewer falls in the intervention group than in the usual-care group by discharge (4 vs. 13, respectively) and between discharge and 3 months (5 vs. 11, respectively). There were equal numbers of falls in both groups by 6 months (within the previous 3 months).

Total Barthel score

The summary statistics of total Barthel scores at 3 and 6 months, analysed using ordinal logit regression and examining relative cumulative odds between intervention groups, are given in Table 29 (lower scores indicate greater disability). The results are adjusted for the baseline total Barthel score and baseline continence category, with site included as a random effect.

For the intervention group, at 3 and 6 months, the adjusted odds of higher Barthel scores (indicating less disability) compared with lower Barthel scores were 0.43 and 0.48 times lower, respectively, than for the usual-care group.

Number of deaths

The numbers of deaths are presented as frequencies (%) by discharge and by 3 and 6 months for each trial group in Table 30.

Summary table of the results of secondary outcomes

Table 31 shows a summary of the results of the secondary outcomes, presented in this appendix, where inferential analysis was possible.

The estimated odds for stress incontinence were lower in the intervention group than in the usual-care group at both follow-up time points. The usual-care group had lower estimated odds of urge incontinence than the intervention group at both follow-up time points, and the odds of higher Barthel scores were higher in the usual-care group than in the intervention group.

Inclusion criteria

We revised the inclusion criterion for level of consciousness to:

• conscious: ‘Alert’ or ‘Not alert but arousable’ (NIHSS Point 1A score of 0 or 1) on the NIHSS3,46 within 72 hours of stroke onset.

The TSC suggested that people with an NIHSS Point 1A score of 2 would not be likely to benefit from the intervention.

We revised the inclusion criterion for UI to:

• UI, defined as ‘involuntary loss of urine’,38 within 72 hours of stroke onset or presence of IUC at the time of consent.

This brings the time period in line with that of the level of consciousness inclusion criterion.

Contamination monitoring: extent to which usual-care participants receive the intervention

We initially planned to conduct a retrospective review of case notes of 50% of patients admitted to each unit in the 3-month pre-intervention period to determine continence practice. However, as many of the sites that expressed an interest in participating were large (between 400 and 1300 admissions with stroke per annum), we believed that using this approach would generate too many data and waste research nurse time and resources. Our revised plan was as follows:

• To determine baseline continence practice, we will conduct a retrospective audit of case notes of 40 patients admitted to each unit in the 3-month pre-intervention period. Detailed review will be conducted of patients identified as incontinent. Audit data will be collected by the project-specific research nurse in each site.

Internal pilot trial continuation criteria

The TSC recommended changing the words ‘stop trial’ to ‘pause trial’ to allow the opportunity for rectifying causes of recruitment and contamination issues, should these arise (protocol section 3.4: internal pilot study).

Outcome measurement

We added an additional data point for the primary outcome, ICIQ-UI-SF: discharge from the stroke unit.

We added ‘number of urinary tract infections’ and ‘number of days indwelling urethral catheter in situ’ to our secondary outcomes. We will collect these at discharge from the stroke unit and at 3 and 6 months post randomisation.