For Peer Review Only

BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

Effectiveness of continence promotion for older women via community organisations: a cluster randomised trial

ForJournal: peerBMJ Open review only Manuscript ID: bmjopen-2013-004135

Article Type: Research

Date Submitted by the Author: 27-Sep-2013

Complete List of Authors: Tannenbaum, Cara; Université de Montréal, Faculty of Medicine Agnew, Rona; Glasgow Caledonian University, Benedetti, Andrea; McGill University, Departments of Medicine and of Epidemiology, Biostatistics & Occupational Health Thomas, Doneal; McGill University, Departments of Medicine and of Epidemiology, Biostatistics & Occupational Health van den Heuvel, Eleanor; Brunel University,

Primary Subject Urology Heading:

Secondary Subject Heading: Geriatric medicine

Keywords: Urinary incontinences < UROLOGY, PUBLIC HEALTH, GERIATRIC MEDICINE

http://bmjopen.bmj.com/

on September 24, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 35 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 4 5 Effectiveness of continence promotion for older women via community organisations: 6 a cluster randomised trial 7 8 9 1 2 3 10 Cara Tannenbaum, MD, MSc Rona Agnew, PhD Andrea Benedetti, PhD Doneal 11 Thomas, MSc4 Eleanor van den Heuvel, PhD5 12 13 14 1 15 AssociateFor Professor, peer Faculty of Medicine, review Université de Montréal, only Quebec, Canada 16 2Clinical Research Fellow, Glasgow Caledonian University, Glasgow, UK 17 18 3Associate Professor, Departments of Medicine and of Epidemiology, Biostatistics & 19 20 Occupational Health, McGill University, Quebec, Canada 21 4PhD candidate, Department of Epidemiology, Biostatistics & Occupational Health, 22 23 McGill University, Quebec, Canada 24 5 25 Director of Assistive Technology Programme, Brunel Institute for Ageing Studies, 26 Brunel University, Uxbridge, UK 27 28 29 30 Word count : 3182 31 32 33 Key words: Continence promotion, cluster trial, community organisations, older

34 http://bmjopen.bmj.com/ 35 women, urinary incontinence 36 37 38 Corresponding Author: Dr. Cara Tannenbaum, Centre de Recherche de l’Institut 39 40 universitaire de gériatrie de Montréal, 4545 Queen Mary Road, Montreal, Quebec, 41 Canada H3W 1W5. Tel (514) 340-2800. Fax : (514) 340-3530. 42 on September 24, 2021 by guest. Protected copyright. 43 Email: [email protected] 44 45 46 The Corresponding Author has the right to grant on behalf of all authors and does 47 48 grant on behalf of all authors, an exclusive licence on a worldwide basis to the BMJ 49 50 Publishing Group Ltd to permit this article (if accepted) to be published in BMJ Open 51 editions and any other BMJPGL products and sublicences such use and exploit all 52 53 subsidiary rights, as set out in the licence. 54 55 56 57 58 59 60 1 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 35 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 All authors have completed the Unified Competing Interest form at 4 5 www.icmje.org/coi_disclosure.pdf (available on request from the corresponding 6 author) and declare: no support from any organisation for the submitted work; no 7 8 financial relationships with any organisations that might have an interest in the 9 10 submitted work in the previous three years, and no other relationships or activities that 11 could appear to have influenced the submitted work. 12 13 14 15 The authorsFor retained peerfull independence review from the study sponsors only in the design, 16 implementation, analysis, interpretation and reporting of the findings. 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 24, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 2 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 35 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 ABSTRACT 4 5 6 Objectives: The primary objective of this cluster randomised controlled trial was to 7 8 compare the effectiveness of the three experimental continence promotion 9 10 interventions against a control intervention on urinary symptom improvement in older 11 women with untreated incontinence recruited from community organisations. A 12 13 second objective was to determine whether changes in incontinence-related 14 15 knowledge,For new uptake peer of risk-modifying review behaviours explained only these improvements. 16 17 18 Setting: 71 community organisations across the United Kingdom 19 20 21 Participants: 259 women aged 60 years with untreated incontinence entered the trial; 22 23 88% completed the 3-month follow-up. 24 25 26 Interventions: The three active interventions consisted of a single 60-minute group 27 28 workshop on 1) continence education (20 clusters, 64 women); 2) evidence-based 29 30 self-management (17 clusters, 70 women); or 3) combined education and self- 31 management (17 clusters, 61 women). The control intervention was a single 60- 32 33 minute educational group workshop on memory loss, polypharmacy and osteoporosis

34 http://bmjopen.bmj.com/ 35 (17 clusters, 64 women). 36 37 38 Primary and secondary outcome measures: The primary outcome was self-reported 39 40 improvement in incontinence 3 months post-intervention at the level of the individual. 41 Changes in incontinence-related knowledge and behaviours were also assessed. 42 on September 24, 2021 by guest. Protected copyright. 43 44 45 Results: The highest rate of urinary symptom improvement occurred in the combined 46 intervention group (66% vs 11% of the control group, prevalence difference 55%, 47 48 95% CI 43%-67%, intracluster correlation 0). Thirty percent versus 6% of participants 49 50 reported significant improvement respectively (prevalence difference 23%, 95% CI 51 10%-36%, intracluster correlation 0). The number-needed-to-treat was 2 to achieve 52 53 any improvement in incontinence symptoms, and 5 to attain significant improvement. 54 55 Changes in knowledge and self-reported risk-reduction behaviours paralleled rates of 56 improvement in all intervention arms. 57 58 59 60 3 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 35 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 Conclusion: Continence education combined with evidence-based self-management 4 5 improves symptoms of incontinence among untreated older women. Community 6 organisations represent an untapped vector for delivering effective continence 7 8 promotion interventions. 9 10 11 Trial registration: ClinicalTrials.gov ID number NCT01239836 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 24, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 4 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 35 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 Article Summary : Strengths and limitations of this study 4 5 6 • First study to provide Level 1 evidence that continence promotion is an 7 8 effective strategy for improving urinary symptoms among untreated 9 10 community-dwelling older women 11 12 13 • Participants were recruited via community organisations with representation 14 15 acrossFor diverse peer socio-economic review strata only 16 17 18 • Rates of knowledge acquisition and behaviour change provide an explanatory 19 20 mechanism for the observed improvements in incontinence in participants 21 22 receiving the education combined with the self-management strategy 23 24 25 • Only self-reported outcomes and crude dichotomous measures of behaviour 26 27 change were collected so results must be interpreted with caution 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 24, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 5 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 35 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 Introduction 4 5 6 Urinary incontinence is more frequent than breast cancer, heart disease or diabetes 7 among older women, but remains a stigmatized and untreated condition despite its 8 9 high prevalence. [1-4] In the United States, Canada, the U.K. and other European 10 11 countries, up to 40% of women aged 65 years and older experience involuntary urine 12 leakage, but little more than 15-30% seek care.[1-4] Even fewer physicians feel 13 14 competent evaluating or treating incontinence.[5-6] Urinary incontinence is associated 15 For peer review only 16 with obesity, cardiovascular disease, diabetes, depression, social isolation, decline in 17 function, falls, nursing home admission, and onerous out-of-pocket expenses.[6-10] 18 19 In many cases incontinence can be improved, and even cured, when evidence-based 20 21 diagnostic and treatment strategies are appropriately applied.[6,11-14] 22 23 24 It is a commonly held misconception that incontinence is a normal part of aging.[15] 25 26 Not-for-profit organisations seek to raise continence awareness worldwide and 27 promote treatment for incontinent individuals. Media campaigns, brochures and 28 29 public awareness lectures attempt to de-stigmatise incontinence and increase help- 30 31 seeking, but the effectiveness of these initiatives for reaching their target population 32 remains unknown.[15] Transmission of public health education via community 33

34 organisations is an unexplored strategy for improving urinary symptoms.[16] Data http://bmjopen.bmj.com/ 35 36 from randomised trials are needed to determine whether the delivery of an evidence- 37 based, continence intervention via community organisations is an effective method for 38 39 treating incontinence. 40 41 42 The primary objective of this cluster randomised controlled trial was to compare the on September 24, 2021 by guest. Protected copyright. 43 44 effectiveness of the three experimental continence promotion interventions against a 45 46 control intervention on urinary symptom improvement in older women with untreated 47 incontinence recruited from community organisations. A second objective was to 48 49 determine whether changes in incontinence-related knowledge, attitudes and new 50 51 uptake of risk-modifying behaviours explained improvements in incontinence. We 52 hypothesized that continence education combined with evidence-based self- 53 54 management would yield the greatest improvement in incontinence symptoms, 55 56 measured at the level of the individual, 3-months post-intervention. 57

58 59 60 6 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 35 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 4 5 METHODS 6 7 8 Study design and oversight 9 10 A 4 arm, parallel-group, controlled, cluster randomised trial was conducted. The study 11 design, recruitment methods and interventions have been reported.[16] Clustering 12 13 was at the level of the community organisation, from whence participants were 14 15 recruited.For The choice peer of a cluster design review served to prevent contaminationonly between 16 participants in the same community organisation. The trial was designed by two of the 17 18 authors and was overseen by the full investigator team, which had full access to the 19 20 data. The data were collected at community organisations across the United 21 Kingdom. All participants provided written informed consent. The protocol was 22 23 approved by the Brunel University Research Ethics Committee. 24 25 26 Study population and recruitment 27 28 Inclusion criteria for community organisations included any organisation throughout 29 30 the United Kingdom that consented to participate in the trial between November 2010 31 and September 2012. A community organisation was loosely defined as any not-for- 32 33 profit group of individuals with a shared interest. These included interest and charity

34 http://bmjopen.bmj.com/ 35 groups, seniors’ housing groups, women’s lobby groups and Asian caregiver 36 associations.[16] Organisations were contacted strategically by convenience 37 38 sampling, word of mouth and referral. A research coordinator approached community 39 40 organisations to join the trial by telephone, email and newspaper advertising. 41 42 on September 24, 2021 by guest. Protected copyright. 43 Inclusion criteria for participants were women aged 60 years and older who reported 44 45 urinary incontinence at least once weekly on the International Consultation on 46 Incontinence Questionnaire (ICIQ), and who were not under active treatment for 47 48 incontinence. For privacy reasons, many community organisations were 49 50 uncomfortable screening their members for incontinence in advance, so eligibility to 51 participate in the trial could only be ascertained by the research coordinator on the day 52 53 of delivery of the intervention.[16] Eligibility to participate in the trial was established 54 55 by asking all attendees at the workshop to complete a baseline screening questionnaire 56 upon arrival. At this time, a study information sheet and consent form were 57 58 distributed to all participants. All women, regardless of eligibility or desire to enrol in 59 60 7 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 35 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 the trial, were permitted to stay for the workshop. Only those women who wished to 4 5 enrol in the trial submitted the signed consent form to the workshop facilitator 6 following delivery of the intervention, however all attendees were encouraged to 7 8 submit the baseline screening questionnaire even if they were continent or did not 9 10 want to participate in the trial. 11 12 13 Interventions 14 15 For peer review only 16 The interventions were applied at the level of each cluster. The three experimental 17 18 interventions to be tested were continence education, self-management including the 19 20 distribution of an evidence-based risk factor reduction tool for incontinence, and a 21 combined intervention that included both components. The sham control intervention 22 23 was a lecture on health promotion for older women that addressed topics other than 24 25 incontinence. All interventions were delivered once in group format to 8-16 women 26 by the same facilitator at a venue of the organisation’s choosing, and lasted 60-90 27 28 minutes. A slide presentation with a pre-established script prepared for the facilitator 29 30 was delivered at each workshop. 31 32 33 The continence promotion intervention incorporated elements of constructivist

34 http://bmjopen.bmj.com/ 35 learning that challenged older adults’ erroneous beliefs about accepting incontinence 36 as a normal part of aging, and aimed to change attitudes and create new knowledge 37 38 about the different types, etiology, risk factors and treatment options for urine 39 40 loss.[16-17] The self-management workshop reviewed self-management theory in an 41 interactive format, and provided a customised evidence-based self-management 42 on September 24, 2021 by guest. Protected copyright. 43 program for risk factor modification for incontinence to each participant.[18-19] The 44 45 program targeted pelvic floor muscle weakness, obesity, consumption of caffeinated 46 beverages, smoking, vision loss and constipation, with instructions on how to keep a 47 48 bladder diary to help monitor symptoms. The content of the combined intervention 49 50 condensed elements from the continence promotion workshop along with self- 51 management theory, and provided the customized self-management tool to 52 53 participants. The control intervention addressed other non-bladder related aspects of 54 55 older women’s health such as memory problems, polypharmacy, osteoporosis, 56 nutrition, physical fitness and vision impairment. 57 58 59 60 8 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 35 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 4 5 Study Outcomes 6 The primary outcome was the participant’s global impression of improvement in 7 8 incontinence symptoms, measured at 3 months post-intervention by telephone 9 10 interview using the patient’s global impression of improvement (PGI-I) questionnaire. 11 The PGI-I is a validated, single-item global rating of change scale that asks the patient 12 13 to describe how their incontinence condition is now compared to how it was prior to 14 15 the interventionFor (very peer much better, muchreview better, a little bit better,only no change, a little 16 bit worse, much worse and very much worse).[20] The primary outcome, any 17 18 improvement, was defined as a rating of a little bit better, much better or very much 19 20 better. A secondary outcome, significant improvement, was defined as much better or 21 very much better. The ICIQ, which measures the frequency, severity and bother from 22 23 incontinence was used at baseline to screen participants for inclusion to the trial, and 24 25 was repeated at follow-up.[21] The ICIQ diagnostic item was used by participants to 26 describe the type of incontinence at baseline. A pre- and post-8-item questionnaire on 27 28 knowledge and attitudes towards incontinence was administered at baseline and at 3- 29 30 month follow-up, as were risk factors and behaviors related to incontinence.[17] Risk 31 factors and behaviors included performance of pelvic floor muscle exercises three 32 33 times weekly (yes, no), daily consumption of 1 cup or more of tea or coffee (yes, no),

34 http://bmjopen.bmj.com/ 35 fluid intake > 1.5 L/day (yes, no), weight and height (self-report) and smoking status 36 (yes, no). At three month follow-up participants were asked whether they had sought 37 38 treatment for urine leakage during the past 3 months. All follow-up interviews were 39 40 performed by the research coordinator, who was blinded to participant identification. 41 42 on September 24, 2021 by guest. Protected copyright. 43 44 45 Randomisation and allocation concealment 46 Group allocation occurred by non-stratified randomisation in blocked groups of 4 of 47 48 consenting organisations that agreed to host a workshop. An independent statistician 49 50 at a distant study site was responsible for randomisation using computer-generated 51 random digits. Community organisations were informed that one of four workshops 52 53 would be delivered on health topics of interest to older women, but not which one. In 54 55 this way, group allocation was concealed from both the clusters and the individual 56 participants, who were invited by the host organisation to attend a “Women’s Health 57 58 Worshop”. The research coordinator remained unaware of group allocation at the 59 60 9 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 35 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 time each community organisation was recruited to the trial because she was only 4 5 informed which workshop to prepare for each organisation several days before each 6 workshop. The trial is considered open-label because both the research facilitator who 7 8 delivered the intervention and the participants who received it were aware of which 9 10 intervention was being delivered. 11 12 13 Sample Size 14 15 The trial Forwas designed peer to detect a minimal review 35% difference inonly the number of 16 participants reporting any improvement (very much better, much better, a little bit 17 18 better on the PGI-I) between the experimental and control conditions, assuming a rate 19 20 of improvement in the control condition as high as 20%, with 80% power and alpha 21 0.05 two-sided (n=34). Using an inflation factor of 1.65 to account for an anticipated 22 23 maximum intracluster correlation (ICC) of 0.05 and unequal cluster size yielded a 24 25 recruitment target of 56 participants per group.[22] 26 27 28 Statistical Methods 29 30 Differences in baseline characteristics between the four groups were determined. To 31 assess the primary outcome we estimated the unadjusted risk difference (prevalence 32 33 of the outcome) and 95% confidence intervals (CI) via generalized estimating

34 http://bmjopen.bmj.com/ 35 equations (GEEs) for participants who reported any improvement on the PGI-I. We 36 repeated the same analysis for those who reported significant improvement. GEEs 37 38 with an identity link and an exchangeable correlation structure were used to account 39 40 for possible correlation between women in the same organisation.[23] To adjust for 41 the imbalance in potential confounders in the groups at baseline, additional analyses 42 on September 24, 2021 by guest. Protected copyright. 43 were conducted using multivariable logistic regression estimated via GEE with an 44 45 exchangeable correlation structure. Potential confounders included age and baseline 46 incontinence severity (ICIQ score) as continuous predictors, and living alone, 47 48 depression, heart disease, falls, arthritis, diabetes, high blood pressure, educational 49 50 status and general health perception as dichotomous predictors. Both intent-to-treat 51 (ITT) and per protocol (PP) analyses were performed. For the ITT analysis, 52 53 participants with missing data were assumed to have no change in incontinence status 54 55 at three-month follow-up. The number needed to treat was calculated as the inverse of 56 the difference in absolute event rates between the experimental and control 57 58 groups.[24] We report intra-cluster (intra-community organisation) correlation 59 60 10 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 35 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 coefficients (ICC) from the marginal model using GEE with assumed exchangeable 4 5 correlation structure and robust standard errors.[25] To estimate adjusted mean group 6 differences in ICIQ scores from baseline to 3-month follow up, we used GEE with a 7 8 Gaussian regression model for continuous outcomes and followed the same procedure 9 10 outlined above. 11 12 13 Improvements in incontinence-related knowledge by intervention type for proportions 14 15 of individualsFor responding peer correctly toreview each knowledge questionnaire only item at baseline 16 compared to 3-month follow-up were estimated using McNemar’s test for matched 17 18 pair analysis. Rates of improvement in self-reported risk modifying behaviours for 19 20 incontinence were calculated, along with 95% confidence intervals. Differences in 21 improvement rates between the intervention and the control group were compared 22 23 using Fisher’s Exact test using a per protocol analysis. A difference in response for 24 25 each health behavior item that indicated adoption of a new risk modifying behaviour 26 was defined as a positive change at 3-month follow-up compared to baseline. 27 28 Reduced coffee and tea intake refer to individuals who reduced their consumption to a 29 30 single cup per day or less. Weight loss was determined by a positive response to the 31 question, “Has your weight changed (yes, no) and if so, how much do you now 32 33 weigh?” and evidence of self-reported current weight lower than self-reported weight

34 http://bmjopen.bmj.com/ 35 at baseline. All statistical analyses were run using RStudio 0.97.310.0, an integrated 36 development environment for R. 37 38 39 40 RESULTS 41 42 on September 24, 2021 by guest. Protected copyright. 43 Study participants and follow-up 44 45 Four-hundred-and-twenty different community organisations were approached over 46 an 18-month period to participate in the trial. Of these, 17% consented and succeeded 47 48 in hosting an intervention, yielding 71 clusters that were randomised. Approximately 49 50 one quarter of the groups contacted refused; 2% expressed interest but were unable to 51 organise a workshop; and a little over half failed to give any response although most 52 53 of them had been followed up and had received extra information on the project.16 54 55 Figure 1 depicts the study flow of the clusters and participants through the trial. 56 Seven-hundred-and-sixty three women attended the workshops, of whom 322 (42%) 57 58 were known to be eligible for the trial. The mean number of participants recruited 59 60 11 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 35 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 from each cluster for the continence promotion group was 3 + 2, whereas it was 4 + 2 4 5 for the other 3 groups. Eighty-percent (259/322) of known eligible attendees to the 6 workshops consented to take part in the trial. Two-hundred-and-twenty-eight of these 7 8 (88%) were available for 3-month follow-up. Table 1 compares the baseline 9 10 characteristics of participants in each trial arm. 11 12 13 Primary outcomes 14 15 The highestFor rate of improvementpeer in incontinencereview occurred inonly the combined 16 intervention group, 66% compared to 11% in the control group (prevalence difference 17 18 55%, 95% CI 43%-67%), yielding a number-needed-to-treat of 2. Thirty percent of 19 20 the combined group reported significant improvement compared to 6% of controls 21 (prevalence difference 23%, 95% CI10%-36%, number-needed-to-treat of 5). In 22 23 adjusted analyses, the likelihood of achieving a significant improvement in urinary 24 25 symptoms from exposure to the combined intervention was five times greater than 26 exposure to the control intervention (OR 4.94, 95% CI 1.45-16.86). Compared to 27 28 controls, the participants in the combined intervention reported an adjusted mean 2.05 29 30 point (95% CI 0.87-3.24) greater improvement on the ICIQ from baseline to 3-month 31 follow up. The adjusted mean difference in ICIQ scores was also significantly higher 32 33 for the continence education group compared to the control group (1.33 point greater

34 http://bmjopen.bmj.com/ 35 improvement (95% CI 0.33-2.32)), but not for the self-management group. The per 36 protocol analysis for the primary outcome and the intra-cluster correlation coefficients 37 38 for each analysis are shown in Table 2. 39 40 41 Key secondary outcomes 42 on September 24, 2021 by guest. Protected copyright. 43 Table 3 shows the changes in incontinence-related knowledge attributable to receipt 44 45 of each intervention. Participants exposed to the combined intervention showed the 46 greatest acquisition in knowledge, exhibiting significant within-group improvement 47 48 on 6 out of 8 questionnaire items. Participants learned that incontinence is not an 49 50 inevitable or irreversible part of aging, that losing weight, changing the type of fluid 51 intake and performing pelvic floor muscle exercises can reduce urinary symptoms, 52 53 and that wearing undergarment protection is not always the best way to manage 54 55 incontinence. 56 57 58 59 60 12 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 35 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 The proportion of participants with modifiable risk factors for incontinence in each 4 5 group at baseline is shown in Table 1. The adoption of various risk-modifying 6 behaviours among participants occurred to a different degree as a result of exposure to 7 8 all three experimental but not the control intervention (Figure 2). At three-month 9 10 follow-up, the proportion of women reporting uptake of pelvic floor muscle exercises 11 and weight loss was significantly higher in the continence education group (46% and 12 13 20% respectively), the self-management group (34% and 20%) and the combined 14 15 interventionFor group (53% peer and 18%) comparedreview to controls (8% only and 3%). Many women 16 additionally reduced their coffee intake and total fluid intake. The proportion of 17 18 women who made an appointment to consult a health professional for urine leakage 19 20 was 19% in the continence promotion group, 7% in the self-management group, 16% 21 in the combined intervention group and 4% in the control group. 22 23 24 25 DISCUSSION 26 27 28 In this cluster-randomised trial testing the effectiveness of 3 different continence 29 30 promotion interventions, we found that health education combined with the delivery 31 of an evidence-based self-management tool via community organisations to untreated 32 33 older women yielded the highest rate of urinary symptom improvement in 66% of

34 http://bmjopen.bmj.com/ 35 recipients, half of whom reported significant improvements in incontinence. These 36 outcomes translate into a number-needed-to-treat of 2 and 5 respectively, a magnitude 37 38 of effect rarely achieved during public health interventions. Both new knowledge 39 40 acquisition and the adoption of risk-modifying behaviours such as exercise and 41 weight loss occurred as a result of community organisations’ involvement in reaching 42 on September 24, 2021 by guest. Protected copyright. 43 untreated incontinent women outside the health care system. 44 45 46 Strengths and weaknesses of the study 47 48 49 This is the first randomised trial to test the effectiveness of continence promotion 50 51 strategies through community outreach. Both explanatory mechanisms and final 52 health outcomes were assessed, and the use of a cluster randomised design was 53 54 chosen to avoid contamination of the control group.[26,27] Our choice of comparator 55 56 controlled for the placebo effect of participating in a group intervention. Breaches in 57 the fidelity and quality of implementation of the intervention were minimized by 58 59 60 13 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 35 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 having the same facilitator deliver each intervention. Improvements in urinary 4 5 symptoms were shown with two validated measures, the PGI-I and the ICIQ. We 6 believe the results have wide external validity as the community groups included 7 8 women with varied educational levels and wide socioeconomic status. 9 10 11 The results of this study confirm findings from previous randomised trials suggesting 12 13 a positive effect of continence education and self-monitoring strategies on urinary 14 15 symptomFor improvement peer in untreated incontinentreview individuals.[28-33] only However all 16 previous trials invited participants for clinical assessments prior to the delivery of the 17 18 intervention, or involved individualized education sessions. The current trial delivered 19 20 group continence interventions without medical or nursing evaluations, in a true 21 public health approach, to both continent and incontinent women as part of the regular 22 23 activities offered by each community organisation. Rates of improvement reported in 24 25 this trial on the PGI-I were similar to or exceeded those reported in other studies using 26 self-help booklets, in the range of 50%. Because of the nature of recruitment and 27 28 delivery of the intervention via community organisations, bladder diaries and pad tests 29 30 could not be collected at baseline. The results of our trial can therefore not be directly 31 compared to other trials that used more objective measures of symptom improvement. 32 33

34 http://bmjopen.bmj.com/ 35 Other limitations apply. Due to the nature of recruiting potential participants, 36 individuals could not be screened and enrolled in the trial prior to randomisation of 37 38 the clusters.[27] The result was an imbalance between groups, accounted for by 39 40 analyses that took into account group differences in age, health status and baseline 41 incontinence severity. The trial was not designed to measure the dose-response of 42 on September 24, 2021 by guest. Protected copyright. 43 knowledge acquisition and behaviour change on urinary symptom improvement. Thus 44 45 only crude, dichotomous self-reported measures of behaviour change were collected 46 and should be interpreted with caution. 47 48 49 50 Relevance to the discipline 51 The value of continence promotion interventions likely reflects the delivery method as 52 53 well as the quality of the content. Group interventions that deliver continence 54 55 education, self-management information, or a combination of the two will improve 56 incontinence symptoms in 59%, 41% and 66% of recipients, respectively. It is 57 58 surprising that the self-management intervention alone was not associated with 59 60 14 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 35 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 significant improvement compared to sham control in this trial. This can potentially 4 5 be explained by the fact that continence education was completely omitted from the 6 self-management workshop, whereas some information on bladder functioning was 7 8 provided to participants during the initial work that tested the self-management tool.19 9 10 11 Implications for practice 12 13 Implementation of community-based programs that promote behavioural techniques 14 15 as first-lineFor management peer for incontinence review support evidence only for the superior efficacy 16 and tolerability of conservative management approaches over pharmacological 17 18 treatment for incontinence.[34] As “silent sufferers” become better informed that 19 20 effective strategies exist for improving urinary symptoms, patient demand for care 21 will likely increase. Almost 20% of women made an appointment to discuss urine 22 23 leakage with a health professional in the three months following receipt of the 24 25 continence education intervention. Evidence-based guidelines exist for physicians to 26 evaluate and manage urinary incontinence when first-line behavioural strategies fail, 27 28 and will need to be more frequently applied.[5,11] 29 30 31 In conclusion, continence education combined with self-management delivered via 32 33 community organisations to untreated older women with incontinence leads to

34 http://bmjopen.bmj.com/ 35 symptom improvement in 1 out of every 2 recipients. At the current time, the majority 36 of older women with incontinence do not seek care, and either self-manage their 37 38 symptoms inappropriately or use protection to palliate urine leakage.[1-4] As 39 40 incontinence is associated with multimorbidity and other deleterious health effects, 41 results from this trial provide strong justification for public health outreach via 42 on September 24, 2021 by guest. Protected copyright. 43 community organisations to reduce urine leakage among untreated individuals. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 15 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 35 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 Acknowledgments 4 5 6 We wish to acknowledge the assistance of Mira Jabbour and Francine Giroux for their 7 8 expert help with database management and the data analyses. Nikki Cotterill, Adele 9 10 Long aided with recruitment in the Bristol area. The above-mentioned individuals 11 received financial compensation for their contribution to this work. We express 12 13 gratitude to all the participants who took part in this trial. Although not exhaustive, 14 15 particularFor thanks are offeredpeer to the following review organisations andonly groups for 16 participating in the trial: the Women’s Institute, AGE UK, the University of the Third 17 18 Age, Queens Nursing Institute (Scotland), Kinship Carers, the Women’s Guild, 19 20 Hanover Housing Association, Good Neighbours, Asian Carers Groups, Older 21 Peoples Forum throughout the UK. 22 23 24 25 Author Contribution Statement 26 27 28 Cara Tannenbaum designed the study and participated in the data analysis and 29 30 interpretation. She wrote the first draft of the manuscript. Rona Agnew was 31 responsible for data collection, participated in the data analysis and interpretation and 32 33 critically reviewed the manuscript. Andrea Benedetti was responsible for the data

34 http://bmjopen.bmj.com/ 35 analysis and interpretation and critically reviewed the manuscript. Doneal Thomas 36 conducted the analyses and reviewed the manuscript. Eleanor van den Heuvel helped 37 38 design the study, participated in the study implementation, helped interpret the 39 40 findings and critically reviewed the manuscript. 41 42 on September 24, 2021 by guest. Protected copyright. 43 Data Access and Sharing Statement 44 45 46 Cara Tannenbaum, Doneal Thomas and Andrea Benedetti had full access to all the 47 48 data in the study and take responsibility for the integrity of the data and the accuracy 49 50 of the data analysis. Patient level data and the full dataset is available upon request 51 from the authors. Consent for data sharing was not obtained but the presented data are 52 53 anonymised and the risk of identification is low. 54 55 56 Competing interests 57 58 59 60 16 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 17 of 35 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 Cara Tannenbaum declares having been an advisory board member or received 4 5 speaker honoraria from Pfizer, Watson, Astellas, Allergen and Ferring 6 pharmaceuticals in the past 3 years, but not in relation to this work. All other authors 7 8 declare: no support from any organisation for the submitted work; no financial 9 10 relationships with any organisations that might have an interest in the submitted work 11 in the previous three years, and no other relationships or activities that could appear to 12 13 have influenced the submitted work. 14 15 For peer review only 16 Funding and Sponsor’s Role 17 18 19 20 The trial was funded by a joint collaboration between the Canadian Institutes of 21 Health Research and the Economic and Social Research Council (UK). The authors 22 23 retained full independence from the study sponsors in the design and conduct of the 24 25 study; collection, management, analysis, and interpretation of the data; preparation, 26 review, and approval of the manuscript; and decision to submit the manuscript for 27 28 publication. 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 24, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 17 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 18 of 35 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 REFERENCES 4 5 6 1. Sexton CC, Coyne KS, Thompson C, Bavendam T, Chen CI, Markland A. 7 8 Prevalence and effect on health-related quality of life of overactive bladder in 9 10 older Americans: Results from the Epidemiology of Lower Urinary Tract 11 Symptoms Study. J Am Geriatr Soc 2011;59:1465–1470. 12 13 2. Tannenbaum C, Corcos J, Assalian P. The relationship between sexual activity 14 15 and urinaryFor incontinence peer in older reviewwomen. J Am Geriatr onlySoc 2006;54:1220-4. 16 3. O’Donnell M, Lose G, Sykes D, Voss S, Hunskaar S. Help-seeking behaviour 17 18 and associated factors among women with urinary incontinence in France, 19 20 Germany, Spain and the United Kingdom. Eur Urol 2005;47:385-92 21 4. Herschorn S, Gajewski J, Schulz J, Corcos J. A population-based study of urinary 22 23 symptoms and incontinence: the Canadian Urinary Bladder Survey. BJU Int 24 25 2008;101:52-8. 26 5. Bland DR et al. The effects of implementation of the Agency for Health Care 27 28 Policy and Research urinary incontinence guidelines in primary care. J Am Geriatr 29 30 Soc 2003;51:979–984. 31 32 6. Subak LL, Wagner TH. Talking about incontinence: the first step toward 33

34 prevention and treatment. JAMA 2010;303:2184-2185. http://bmjopen.bmj.com/ 35 7. Tannenbaum C, Gray M, Hoffstetter S, Cardozo L. Comorbidities associated with 36 37 bladder dysfunction. Int J Clin Pract 2013;67:105-13. 38 39 8. Farage MA, Miller KW, Berardesca E, Maibach HI. Psychosocial and societal 40 burden of incontinence in the aged population. Arch Gynecol Obstet 41 42 2008;277:285-290. on September 24, 2021 by guest. Protected copyright. 43 44 9. Holroyd-Leduc JM, Mehta KM, Covinsky KE. Urinary incontinence and its 45 association with death, nursing home admission, and functional decline. J Am 46 47 Geriatr Soc 2004;52:712-8. 48 49 10. Chiarelli PE, Mackenzie LA, Osmotherly PG. Urinary incontinence is associated 50 with an increase in falls: a systematic review. Aust J Physiother 2009;55:89-95. 51 52 11. Holroyd-Leduc JM, Tannenbaum C, Thorpe KE, Straus SE. What type of urinary 53 54 incontinence does this woman have? JAMA 2008;299:1446-56. 55 12. Goode PS, Burgio KL, Richter HE, Markland AD. Incontinence in older women. 56 57 JAMA 2010;303:2172-81. 58 59 60 18 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 19 of 35 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 13. Tannenbaum C, Bachand, G, DuBeau CE, Kuchel GA. Experience of an 4 5 incontinence clinic for older women : no apparent age limit for potential physical 6 and psychological benefits. J Women Health Gend Based Med 2001;10:751-56. 7 8 14. Tannenbaum C, Brouillette J, Corcos J. Rating improvements in urinary 9 10 incontinence: do patients and physicians agree? Age Ageing 2008;37:1-5 11 15. Newman DK, Buckley B, Gordon D et al. Continence promotion, education and 12 13 primary prevention. In Abrams P, Cardozo L, Khoury S, Wein A eds: 14 th 15 Incontinence.For 5 Internationalpeer Consultation review on Incontinence, only Paris, February 16 2012. Health Publication Ltd 2013. 17 18 16. Agnew R, van den Heuvel E, Tannenbaum C. Efficiency of using community 19 20 organisations as catalysts for recruitment to continence promotion trials. Clin 21 Trials 2013;10:151-9. 22 23 17. Tannenbaum C, Drali R, Holroyd-Leduc J Richard L. Lessons learned: Impact of 24 25 a continence promotion activity for older community dwelling women. Neurourol 26 Urodyn 2010;29:540-544. 27 28 18. Wilde MH, Bliss DZ, Booth J, Cheater F, Tannenbaum C. Self-Management of 29 30 Urinary and Fecal Incontinence. AJN 2013 in press. 31 19. Holroyd-Leduc J, Strauss S, Thorpe K, Davis D, Schmaltz H, Tannenbaum C. 32 33 Translation of evidence into a self-management tool for use by women with

34 http://bmjopen.bmj.com/ 35 urinary incontinence. Age Ageing 2011;40:227-33. 36 20. Yalcin I, Bump RC. Validation of two global impression questionnaires for 37 38 incontinence. Am J Obstet Gynecol 2003;189: 98–101. 39 40 21. Avery K, Donovan J, Peters TJ et al. ICIQ: a brief and robust measure for 41 evaluating the symptoms and impact of urinary incontinence. Neurourol Urodyn 42 on September 24, 2021 by guest. Protected copyright. 43 2004; 23: 322–30. 44 45 22. Eldridge SM; Ashby D; Kerry S. Sample size for cluster randomised trials: effect 46 of coefficient of variation of cluster size and analysis method. Int J Epidemiol 47 48 2006;35:1292-1300. 49 50 23. Ukoumunne OC, Forbes AB, Carlin JB, Gulliford MC. Comparison of the risk 51 difference, risk ratio and odds ratio scales for quantifying the unadjusted 52 53 intervention effect in cluster randomised trials. Stat Med 2008;27: 5143-5155. 54 55 24. Users’ guide to the medical literature. Integrating research evidence with the care 56 of the individual patient. JAMA 2000;283;2829-2836. 57 58 59 60 19 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 20 of 35 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 25. Hanley JA, Negassa A, Edwardes MD. GEE analysis of negatively correlated 4 5 binary responses: a caution. Stat Med 2000;19:715–722. 6 26. Craig P, Dieppe P, Macintyre S, Mitchie S, Nazareth I, Petticrew M. Developing 7 8 and evaluating complex interventions: the new Medical Research Council 9 10 guidance. BMJ 2008;337:979-983. 11 27. Campbell MK, Piaggio G, Elbourne DR, Altman DG. Consort 2010 statement: 12 13 extension to cluster randomised trials. BMJ 2012;345:e5661. 14 15 28. KincadeFor JE, Dougherty peer MC, Carlson review JR, Hunter GS, Busby-Whitehead only J. 16 Randomised clinical trial of efficacy of self-monitoring techniques to treat urinary 17 18 incontinence in women. Neurourol Urodyn 2007;26:507-11. 19 20 29. Wagg AR, Barron D, Kirby M, Stott D, Corlett K. A randomised partially 21 controlled trial to assess the impact of self-help vs. structured help from a 22 23 continence nurse specialist in women with undiagnosed urinary problems in 24 25 primary care. Int J Clin Pract 2007;61:1863-73. 26 30. Dougherty MC, Dwyer JW, Pendergast JF et al. A randomised trial of behavioral 27 28 management for continence with older rural women. Res Nurs Health 2002;25:3- 29 30 13. 31 31. McFall SL, Yerkes AM, Cowan LD. Outcomes of a small group educational 32 33 intervention for urinary incontinence: episodes of incontinence and other urinary

34 http://bmjopen.bmj.com/ 35 symptoms. J Aging Health 2000;12:250-67. 36 32. Milne J. The impact of information on health behaviors of older adults with 37 38 urinary incontinence. Clin Nurs Res 2000;9:161-76. 39 40 33. Goode PS, Burgio KL, Locher JL et al. Effect of behavioral training with or 41 without pelvic floor electrical stimulation on stress incontinence in women: a 42 on September 24, 2021 by guest. Protected copyright. 43 randomized controlled trial. JAMA 2003;290(3):345-52. 44 45 34. Shamliyan T, Wyman J, Kane RL. Nonsurgical treatments for urinary 46 incontinence in adult women: diagnosis and comparative effectiveness. 47 48 Comparative Effectiveness Review No. 36. (Prepared by the University of 49 50 Minnesota Evidence-based Practice Center under Contract No. HHSA 290-2007- 51 10064-I.) AHRQ Publication No. 11(12)-EHC074-EF. Rockville, MD. Agency 52 53 for Healthcare Research and Quality. April 2012. Available at: 54 55 www.effectivehealthcare.ahrq.gov/reports/final.cfm. 56 57 58 59 60 20 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 21 of 35 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 4 Table 1: Baseline characteristics and distribution of modifiable risk factors of 5 participants 6 7

8 9 Continence Self- Combined Control 10 education management intervention intervention 11 (n=64) (n=70) (n=61) (n=64) 12 Mean ± SD 13 Age 70.8 ± 7.9 71.0 ± 6.8 70.4 ± 6.7 74.1 ± 8.1 14 Mean ICIQ score* 8.5 ± 4.4 6.8 ± 3 7.3 ± 5.6 6.7 ± 3.4 15 For peer review(% yes)only 16 17 Lives alone 48.4 40.0 37.7 59.4 18 Education 19 University degree or 20 equivalent 31.2 45.7 37.7 19.0 21 22 General health perception 23 Good, very good, excellent 53.1 85.7 80.3 75.0 24 25 Fair/poor 45.3 14.3 16.4 25.0 26 Depression 48.4 35.7 32.8 20.3 27 Heart disease 35.9 25.7 16.4 21.0 28 Falls 45.3 31.4 18.0 18.8 29 Arthritis 78.1 52.9 44.3 57.8 30 Diabetes 39.1 24.3 18.0 20.3 31 32 High blood pressure 59.0 40.0 45.9 55.6 33

34 Type of incontinence http://bmjopen.bmj.com/ 35 Stress only 15.6 12.9 14.8 33.3 36 Urgency only 32.8 35.7 29.5 20.6 37 Mixed 45.3 42.9 55.7 39.7 38 39 40 Modifiable risk factors 41 Performs pelvic floor 42 muscle exercises 3 times on September 24, 2021 by guest. Protected copyright. 43 per week 18.8 15.7 11.9 15.6 44 Self-reported body mass 45 index > 27 kg/m2 ** 53.2 53.0 42.4 49.2 46 Drinks more than 1.5 litres 47 of fluid/day 43.8 44.3 54.1 37.5 48 49 Drinks one cup of tea or 50 more/day 85.9 84.3 73.8 84.4 51 Drinks one cup of coffee or 52 more/day 46.9 62.9 65.6 64.1 53 Smokes 6.3 4.3 4.9 6.2 54 55 *ICIQ = International Consultation on Incontinence Questionnaire, used to measure 56 57 the severity and bother from urinary incontinence. Scores range from 0 to 21, with 58 higher scores representing worse incontinence. 59 60 21 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 22 of 35 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 4 **Self-reported body mass index: calculated as weight (kg)/height2 (m) based on 5 participant’s self-reported height and weight at baseline 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 24, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 22 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

0 0 4.94 23 23 23.27 17.63 (1.45- (5.91- (5.09- 16.86) 91.59) 61.13) vs control vs Combined Combined

SM SM 1.81 0.14 0.06 3.46 2.71 6.60) 8.41) (0.50- (1.08- (0.87- 11.03) vs control vs Adjusted**

0 0 2.83 9.14 10.40 nence (0.59- (3.05- (3.05- Conti- 13.66) 35.48) 27.37) vs control vs

0 0 .11) 6.3 17.7) (2.2- 17.14 15.51 (6.51- (6.50- 45 37.01) Odds ratio (95% CI) ratio (95% Odds vs control vs Combined Combined

vs vs SM SM 3.8 3.8 0.25 0.18 4.64 5.16 control (1.48- (1.73- 15.37) 14.56) Crude (1.2-12.2)

4.2 0.03 0.24 11.72 11.45 11.45 (4.27- (4.54- 30.67) 30.21)

vs control vs Continence (1.4-13.0)

0 0 .55 0.23 0.59 0.36) 0.74) 0.67) (0.10- (0.45- (0.43- vs control vs Combined Combined BMJ Open

Any improvement Any http://bmjopen.bmj.com/

0.14 0.25 0.18 0.29 0.28 0.27) 0.51) 0.48) (0.01- (0.07- (0.08- SM vs vs SM CI)* control

Very much better or much better or much better much Very

0.16 0.03 0.02 0.51 0.48 0.29) 0.67) (0.03- (0.34- Prevalence difference (95% difference Prevalence (0.33-0.64 vs control vs Continence

on September 24, 2021 by guest. Protected copyright. - - 0.06 0.13 0.11 Control

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

- - 0.30 0.73 0.66

For peer review only Combined

- - 0.21 0.47 0.41 SM

- - Prevalence at 3-month follow-up at 3-month Prevalence 0.22 0.64 0.59 nence Conti-

Intention-to- treat

ICC per per ICC ICC Intention-to- treat Per protocol Per Intention-to- treat

protocol Table 2: Prevalence, risk difference and odds ratios for self-reported improvements in incontinence at 3-months at 3-months in incontinence improvements self-reported for andratios difference odds Tablerisk Prevalence, 2: Page 23 of 35 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

0 0 Page 24 of 35 5.32 24 24 (1.39- 20.34)

0 0.02 2.28 8.24) (0.63-

0 0 2.51 (0.48- 13.18)

0 0 5.8 (2.0-17.4)

3.5 3.5 0.08 0.06 (1.06- 11.31)

3.7 0.01 0.02 11.3) (1.2-

0 0.02 0.26 0.42) (0.10- BMJ Open

http://bmjopen.bmj.com/ 0.08 0.06 0.15 0.29) (0.00-

0.01 0.02 0.16 0.30) (0.02-

on September 24, 2021 by guest. Protected copyright. - - 0.08 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

- - 0.33 For peer review only

- - 0.24

incontinence severity incontinenceseverity score SM = Self-management;SM ICC = intra-cluster correlationcorrelation were*95% CI’s calculated robust using standard errors **Adjusted for age, livingalone, depression, heart disease, falls, arthritis, diabetes, high blood pressure, educational status, health perception,general baseline and

ICC per ICC ICC Intention-to- treat Per protocol Per protocol 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from 25 25

1 1 68.6 66.1 69.2 67.2 0.63 90.4 86.9 51.9 50.8 0.73 82.7 79.7

n=64 Control

7.3 89.1 68.3 27.3 52.5 0.73 92.7 88.3 36.1 38.2 77.0 0.001 <0.001 <0.001 n=61 Combined Combined intervention

75.8 66.7 0.11 33.9 40.6 0.77 88.7 92.8 0.06 17.7 32.9 0.02 63.9 79.7 Self- n=70 management

77.6 64.5 0.03 36.8 57.1 93.1 88.9 0.12 29.3 41.9 36.2 73.0 <0.001 BMJ Open n=64 education Continence Continence http://bmjopen.bmj.com/

on September 24, 2021 by guest. Protected copyright. For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

3 month follow up % % agreement follow up 3 month Baseline % agreement Baseline 3 month follow up % % agreement follow up 3 month Baseline % agreement Baseline 3 month follow up % % agreement follow up 3 month Baseline % agreement Baseline 3 month follow up % % agreement up follow 3 month Baseline % agreement Baseline 3 month follow up % % agreement follow up 3 month Baseline % agreement Baseline

p value for change p value p value for change p value p value for change p value For peer for change* p value review only

5. What you drink can drinkcan you What 5. leakage urine to contribute 4. Wearing pads or diapers is ordiapers pads Wearing 4. urinary manage to way best the incontinence 3. Urine leakage can be caused caused be can leakage Urine 3. 2. Once people start to leak to start leak people Once 2. to able never are they urine, again urine their control 1. Urinary incontinence is a incontinence Urinary 1. ageing of part normal

by many different different things many by knowledge Table in incontinence-related Change 3: Page 25 of 35 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from Page 26 of 35 26 26

75 1.0 98.1 96.9 0.48 80.8 0.79 65.4 71.4 0.77

n=64 Control

0.02 98.2 85.2 90.9 66.1 0.36 76.4 67.2 0.01 <0.001 n=61 Combined Combined intervention

0.13 96.7 88.6 0.69 77.4 74.3 0.65 65.2 0.36 Self- n=70 management

0.04 96.5 85.5 0.08 77.6 61.3 0.63 77.6 72.6 0.26 BMJ Open n=64 education Continence Continence http://bmjopen.bmj.com/

on September 24, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

p value for change p value p value for change p value p value for change p value p value for change p value

For peer review only

muscles can help control urine urine control help can muscles 8. Exercising pelvic floor pelvic Exercising 8. leakage 7. Losing weight can lead to lead to can weight 7. Losing incontinence in improvement 6. How much you drink can drink can you much How 6. leakage urine to contribute

* McNemar's test for matched-pairs data matched-pairs for test McNemar's * 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 27 of 35 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 Figure 1: Flow of participants through the trial 4 5 6 420 organisations 7 contacted 8 9 10 11 71 clusters 12 randomised 100 did not complete eligibility 763 participants 13 questionnaire 14 341 participants ineligible 15 63 eligible butFor no consent peer review only Enrolled 16 17 n=259 18 19 20 21 Continence Self- Combined Control 22 workshop Management intervention Intervention 23 Clusters : 20 Clusters : 17 Clusters : 17 Clusters :17 24 Cluster size: 1-7 Cluster size :1-11 Cluster size : 1-9 Cluster size : 1-8 25 Individuals : 64 Individuals : 70 Individuals: 61 Individuals : 64 26 27 28 29 30 3 month follow 3 month 3 month 3 month follow 31 up follow up follow up up 32 Completed: 59 Completed: 62 Completed: 55 Completed: 52 33 Lost-to-follow- Lost-to-follow- Lost-to-follow- Lost-to-follow- up: 2 up: 0 up: 1 up: 1

34 http://bmjopen.bmj.com/ Withdrawals: 3 Withdrawals: 8 Withdrawals: 5 Withdrawals: 11 35 36 37 38 39 40 41 42 on September 24, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from Page 28 of 35

BMJ Open http://bmjopen.bmj.com/ on September 24, 2021 by guest. Protected copyright. For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml For peer review only *Significantly different from the control group (p<0.05) using Fisher’s Exact test *Significantly per the (p<0.05) protocol from in Fisher’s test analysis using control Exact group different Error barsconfidence intervals 95% represent

Figure 2: Change in risk-modifying behaviours at 3-month follow-up follow-up 3-month at behaviours in risk-modifying 2: Change Figure 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 29 of 35 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 Table 1: CONSORT 2010 checklist of information to include when reporting a cluster 4 randomised trial 5 6 Section/Topic Item Standard Checklist item Extension for cluster Page 7 No designs No * 8 9 Title and abstract 10 11 1a Identification as a Identification as a cluster 1 12 randomised trial in the title randomised trial in the title 13 14 1b Structured summary of trial See table 2 3-4 15 For design,peer methods, results, review and only 16 conclusions (for specific 17 guidance see CONSORT for 18 1,2 abstracts) 19 20 Introduction 21 22 Background and 2a Scientific background and Rationale for using a cluster 6 23 objectives explanation of rationale design 24 25 2b Specific objectives or Whether objectives pertain to the 5 26 hypotheses the cluster level, the individual 27 participant level or both 28 29 Methods 30 31 Trial design 3a Description of trial design Definition of cluster and 6 32 (such as parallel, factorial) description of how the design 33 including allocation ratio features apply to the clusters

34 http://bmjopen.bmj.com/ 35 36 3b Important changes to - 37 methods after trial 38 commencement (such as 39 eligibility criteria), with 40 reasons 41 42 Participants 4a Eligibility criteria for Eligibility criteria for clusters 6-7 on September 24, 2021 by guest. Protected copyright. 43 participants 44 45 4b Settings and locations where 6 46 the data were collected 47 48 Interventions 5 The interventions for each Whether interventions pertain to 7 49 group with sufficient details the cluster level, the individual 50 to allow replication, participant level or both 51 including how and when they 52 were actually administered 53 54 Outcomes 6a Completely defined pre- Whether outcome measures 8 55 specified primary and pertain to the cluster level, the 56 secondary outcome individual participant level or both 57 measures, including how and 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 30 of 35 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 when they were assessed 4 5 6b Any changes to trial - 6 outcomes after the trial 7 commenced, with reasons 8 9 Sample size 7a How sample size was Method of calculation, number of 9,26 10 determined clusters(s) (and whether equal or 11 unequal cluster sizes are 12 assumed), cluster size, a 13 coefficient of intracluster 14 correlation (ICC or k), and an 15 For peer reviewindication of its onlyuncertainty 16 17 7b When applicable, - 18 explanation of any interim 19 20 analyses and stopping 21 guidelines 22 23 Randomisation: 24 25 Sequence 8a Method used to generate the 8 26 generation random allocation sequence 27 28 8b Type of randomisation; Details of stratification or 8 29 details of any restriction matching if used 30 (such as blocking and block 31 size) 32 33 Allocation 9 Mechanism used to Specification that allocation was 8

34 concealment implement the random based on clusters rather than http://bmjopen.bmj.com/ 35 mechanism allocation sequence (such as individuals and whether allocation 36 sequentially numbered concealment (if any) was at the 37 containers), describing any cluster level, the individual 38 steps taken to conceal the participant level or both 39 sequence until interventions 40 were assigned 41 42 Implementation 10 Who generated the random Replace by 10a, 10b and 10c - on September 24, 2021 by guest. Protected copyright. 43 allocation sequence, who 44 enrolled participants, and 45 who assigned participants to 46 47 interventions 48 10a Who generated the random 8 49 allocation sequence, who enrolled 50 clusters, and who assigned 51 clusters to interventions 52 53 54 55 10b Mechanism by which individual 6 56 participants were included in 57 clusters for the purposes of the 58 trial (such as complete 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 31 of 35 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 enumeration, random sampling) 4 5 10c From whom consent was sought 6 6 (representatives of the cluster, or 7 individual cluster members, or 8 both), and whether consent was 9 sought before or after 10 randomisation 11 12 13 14 15 For peer review only 16 Blinding 11a If done, who was blinded 8 17 after assignment to 18 interventions (for example, 19 participants, care providers, 20 those assessing outcomes) 21 and how 22 23 11b If relevant, description of the 8 24 similarity of interventions 25 26 Statistical methods 12a Statistical methods used to How clustering was taken into 9-10 27 compare groups for primary account 28 and secondary outcomes 29 30 12b Methods for additional 9-10 31 analyses, such as subgroup 32 33 analyses and adjusted analyses

34 http://bmjopen.bmj.com/ 35 36 Results 37 38 Participant flow (a 13a For each group, the numbers For each group, the numbers of 26 39 diagram is strongly of participants who were clusters that were randomly 40 recommended) randomly assigned, received assigned, received intended 41 intended treatment, and treatment, and were analysed for 42 were analysed for the the primary outcome on September 24, 2021 by guest. Protected copyright. 43 primary outcome 44 45 13b For each group, losses and For each group, losses and 26 46 exclusions after exclusions for both clusters and 47 randomisation, together with individual cluster members 48 reasons 49 50 Recruitment 14a Dates defining the periods of 6 51 recruitment and follow-up 52 53 14b Why the trial ended or was - 54 stopped 55 56 Baseline data 15 A table showing baseline Baseline characteristics for the 20 57 demographic and clinical individual and cluster levels as 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 32 of 35 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 characteristics for each applicable for each group 4 group 5 6 Numbers analysed 16 For each group, number of For each group, number of 26 7 participants (denominator) clusters included in each analysis 8 included in each analysis and 9 whether the analysis was by 10 original assigned groups 11 12 Outcomes and 17a For each primary and Results at the individual or cluster 11,22-25 13 estimation secondary outcome, results level as applicable and a 14 for each group, and the coefficient of intracluster 15 For estimatedpeer effect size review and its correlation (ICC onlyor k) for each 16 17 precision (such as 95% primary outcome 18 confidence interval) 19 20 17b For binary outcomes, 11,22-25 21 presentation of both 22 absolute and relative effect 23 sizes is recommended 24 25 Ancillary analyses 18 Results of any other analyses 11,22-25 26 performed, including 27 subgroup analyses and 28 adjusted analyses, 29 distinguishing pre-specified 30 from exploratory 31 32 Harms 19 All important harms or - 33 unintended effects in each

34 group (for specific guidance http://bmjopen.bmj.com/ 35 see CONSORT for harms3) 36 37 Discussion 38 39 Limitations 20 Trial limitations, addressing 12 40 sources of potential bias, 41 imprecision, and, if relevant, 42 multiplicity of analyses on September 24, 2021 by guest. Protected copyright. 43 44 Generalisability 21 Generalisability (external Generalisability to clusters and/or 13-14 45 validity, applicability) of the individual participants (as 46 47 trial findings relevant) 48 Interpretation 22 Interpretation consistent 13-14 49 50 with results, balancing 51 benefits and harms, and 52 considering other relevant 53 evidence 54 55 Other information 56 57 Registration 23 Registration number and 4 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 33 of 35 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 name of trial registry 4 5 Protocol 24 Where the full trial protocol - 6 can be accessed, if available 7 8 Funding 25 Sources of funding and other 4,16 9 support (such as supply of 10 drugs), role of funders 11 12 13 * Note: page numbers optional depending on journal requirements 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 24, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 34 of 35 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 4 5 Table 2: Extension of CONSORT for abstracts1,2 to reports of cluster randomised 6 trials 7 8 9 10 Item Standard Checklist item Extension for cluster trials 11 Title Identification of study as randomised Identification of study as cluster 12 randomised 13 14 Trial design Description of the trial design (e.g. parallel, 15 Forcluster, peer non-inferiority) review only 16 Methods 17 Participants Eligibility criteria for participants and the Eligibility criteria for clusters 18 settings where the data were collected 19 Interventions Interventions intended for each group 20 21 Objective Specific objective or hypothesis Whether objective or hypothesis pertains 22 to the cluster level, the individual 23 participant level or both 24 Outcome Clearly defined primary outcome for this Whether the primary outcome pertains to 25 report the cluster level, the individual participant 26 level or both 27 Randomization How participants were allocated to How clusters were allocated to 28 29 interventions interventions 30 Blinding (masking) Whether or not participants, care givers, 31 and those assessing the outcomes were 32 blinded to group assignment 33 Results

34 http://bmjopen.bmj.com/ Numbers randomized Number of participants randomized to Number of clusters randomized to each 35 each group group 36 1 37 Recruitment Trial status 38 Numbers analysed Number of participants analysed in each Number of clusters analysed in each 39 group group 40 Outcome For the primary outcome, a result for each Results at the cluster or individual 41 group and the estimated effect size and its participant level as applicable for each 42 precision primary outcome on September 24, 2021 by guest. Protected copyright. 43 44 Harms Important adverse events or side effects 45 Conclusions General interpretation of the results 46 Trial registration Registration number and name of trial 47 register 48 Funding Source of funding 49 50 51 52 53 54 55 56 57 58 1 Relevant to Conference Abstracts 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 35 of 35 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 REFERENCES 4 5 6 1 Hopewell S, Clarke M, Moher D, Wager E, Middleton P, Altman DG, et al. CONSORT 7 for reporting randomised trials in journal and conference abstracts. Lancet 2008, 8 371:281-283 9 2 Hopewell S, Clarke M, Moher D, Wager E, Middleton P, Altman DG at al (2008) 10 CONSORT for reporting randomized controlled trials in journal and conference 11 abstracts: explanation and elaboration. PLoS Med 5(1): e20 12 3 13 Ioannidis JP, Evans SJ, Gotzsche PC, O'Neill RT, Altman DG, Schulz K, Moher D. 14 Better reporting of harms in randomized trials: an extension of the CONSORT 15 statement.For Ann Intern peer Med 2004; 141(10):781-788.review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 24, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

Effectiveness of continence promotion for older women via community organisations: a cluster randomised trial

ForJournal: peerBMJ Open review only Manuscript ID: bmjopen-2013-004135.R1

Article Type: Research

Date Submitted by the Author: 05-Nov-2013

Primary Subject Urology Heading:

Secondary Subject Heading: Geriatric medicine

Keywords: Urinary incontinences < UROLOGY, PUBLIC HEALTH, GERIATRIC MEDICINE

http://bmjopen.bmj.com/

on September 24, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ 35 women, urinary incontinence 36 37 38 Corresponding Author: Dr. Cara Tannenbaum, Centre de Recherche de l’Institut 39 40 universitaire de gériatrie de Montréal, 4545 Queen Mary Road, Montreal, Quebec, 41 Canada H3W 1W5. Tel (514) 340-2800. Fax : (514) 340-3530. 42 on September 24, 2021 by guest. Protected copyright. 43 Email: [email protected] 44 45 46 The Corresponding Author has the right to grant on behalf of all authors and does 47 48 grant on behalf of all authors, an exclusive licence on a worldwide basis to the BMJ 49 50 Publishing Group Ltd to permit this article (if accepted) to be published in BMJ Open 51 editions and any other BMJPGL products and sublicences such use and exploit all 52 53 subsidiary rights, as set out in the licence. 54 55 56 57 58 59 60 1 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 24, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 2 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 ABSTRACT 4 5 6 Objectives: The primary objective of this cluster randomised controlled trial was to 7 8 compare the effectiveness of the three experimental continence promotion 9 10 interventions against a control intervention on urinary symptom improvement in older 11 women with untreated incontinence recruited from community organisations. A 12 13 second objective was to determine whether changes in incontinence-related 14 15 knowledgeFor and new uptakepeer of risk-modifying review behaviours explainonly these improvements. 16 17 18 Setting: 71 community organisations across the United Kingdom 19 20 21 Participants: 259 women aged 60 years and older with untreated incontinence 22 23 entered the trial; 88% completed the 3-month follow-up. 24 25 26 Interventions: The three active interventions consisted of a single 60-minute group 27 28 workshop on 1) continence education (20 clusters, 64 women); 2) evidence-based 29 30 self-management (17 clusters, 70 women); or 3) combined continence education and 31 self-management (17 clusters, 61 women). The control intervention was a single 60- 32 33 minute educational group workshop on memory loss, polypharmacy and osteoporosis

34 http://bmjopen.bmj.com/ 35 (17 clusters, 64 women). 36 37 38 Primary and secondary outcome measures: The primary outcome was self-reported 39 40 improvement in incontinence 3 months post-intervention at the level of the individual. 41 The secondary outcome was change in the International Consultation on Incontinence 42 on September 24, 2021 by guest. Protected copyright. 43 Questionnaire (ICIQ) from baseline to 3-month follow-up. Changes in incontinence- 44 45 related knowledge and behaviours were also assessed. 46 47 48 Results: The highest rate of urinary symptom improvement occurred in the combined 49 50 intervention group (66% vs 11% of the control group, prevalence difference 55%, 51 95% CI 43%-67%, intracluster correlation 0). Thirty percent versus 6% of participants 52 53 reported significant improvement respectively (prevalence difference 23%, 95% CI 54 55 10%-36%, intracluster correlation 0). The number-needed-to-treat was 2 to achieve 56 any improvement in incontinence symptoms, and 5 to attain significant improvement. 57 58 Compared to controls, participants in the combined intervention reported an adjusted 59 60 3 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 mean 2.05 point (95% CI 0.87-3.24) greater improvement on the ICIQ from baseline 4 5 to 3-month follow up. Changes in knowledge and self-reported risk-reduction 6 behaviours paralleled rates of improvement in all intervention arms. 7 8 9 10 Conclusion: Continence education combined with evidence-based self-management 11 improves symptoms of incontinence among untreated older women. Community 12 13 organisations represent an untapped vector for delivering effective continence 14 15 promotionFor interventions. peer review only 16 17 18 Trial registration: ClinicalTrials.gov ID number NCT01239836 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 24, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 4 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 Article Summary : Strengths and limitations of this study 4 5 6 • First study to provide Level 1 evidence that continence promotion is an 7 8 effective strategy for improving urinary symptoms among untreated 9 10 community-dwelling older women 11 12 13 • Participants were recruited via community organisations with representation 14 15 acrossFor diverse peer socio-economic review strata only 16 17 18 • Rates of knowledge acquisition and behaviour change provide an explanatory 19 20 mechanism for the observed improvements in incontinence in participants 21 22 receiving the combined education plus self-management strategy 23 24 25 • Only self-reported outcomes and crude dichotomous measures of behaviour 26 27 change were collected so results must be interpreted with caution 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 24, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 5 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 organisations is an unexplored strategy for improving urinary symptoms.[16] Data http://bmjopen.bmj.com/ 35 36 from randomised trials are needed to determine whether the delivery of an evidence- 37 based continence intervention via community organisations is an effective method for 38 39 treating incontinence. 40 41 42 The primary objective of this cluster randomised controlled trial was to compare the on September 24, 2021 by guest. Protected copyright. 43 44 effectiveness of the three experimental continence promotion interventions against a 45 46 control intervention on urinary symptom improvement in older women with untreated 47 incontinence recruited from community organisations. A second objective was to 48 49 determine whether changes in incontinence-related knowledge, attitudes and new 50 51 uptake of risk-modifying behaviours explain improvements in incontinence. We 52 hypothesized that continence education combined with evidence-based self- 53 54 management would yield the greatest improvement in incontinence symptoms, 55 56 measured at the level of the individual, 3-months post-intervention. 57

58 59 60 6 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ 35 groups, seniors’ housing groups, women’s lobby groups and Asian caregiver 36 associations.[16] Organisations were contacted strategically by convenience 37 38 sampling, word of mouth and referral. A research coordinator approached community 39 40 organisations to join the trial by telephone, email and newspaper advertising. 41 42 on September 24, 2021 by guest. Protected copyright. 43 Inclusion criteria for participants were women aged 60 years and older who reported 44 45 urinary incontinence at least once weekly on the International Consultation on 46 Incontinence Questionnaire (ICIQ), and who were not under active treatment for 47 48 incontinence. For privacy reasons, many community organisations were 49 50 uncomfortable screening their members for incontinence in advance, so eligibility to 51 participate in the trial could only be ascertained by the research coordinator on the day 52 53 of delivery of the intervention.[16] Eligibility to participate in the trial was established 54 55 by asking all attendees at the workshop to complete a baseline screening questionnaire 56 upon arrival. At this time, a study information sheet and consent form were 57 58 distributed to all participants. All women, regardless of eligibility or desire to enrol in 59 60 7 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 the trial, were permitted to stay for the workshop. Only those women who wished to 4 5 enrol in the trial submitted the signed consent form to the workshop facilitator 6 following delivery of the intervention, however all attendees were encouraged to 7 8 submit the baseline screening questionnaire even if they were continent or did not 9 10 wish to participate in the trial. 11 12 13 Interventions 14 15 For peer review only 16 The interventions were applied at the level of each cluster. The three experimental 17 18 interventions to be tested were continence education, self-management including the 19 20 distribution of an evidence-based risk factor reduction tool for incontinence, and a 21 combined intervention that included both components. The sham control intervention 22 23 was a lecture on health promotion for older women that addressed topics other than 24 25 incontinence. All interventions were delivered once in group format to 8-16 women 26 by the same facilitator at a venue of the organisation’s choosing, and lasted 60-90 27 28 minutes. A slide presentation with a pre-established script prepared for the facilitator 29 30 was delivered at each workshop. 31 32 33 The continence promotion intervention incorporated elements of constructivist

34 http://bmjopen.bmj.com/ 35 learning that challenged older adults’ erroneous beliefs about accepting incontinence 36 as a normal part of aging, and aimed to change attitudes and create new knowledge 37 38 about the different types, etiology, risk factors and treatment options for urine 39 40 loss.[16-17] The self-management workshop reviewed self-management theory in an 41 interactive format, and provided a customised evidence-based self-management 42 on September 24, 2021 by guest. Protected copyright. 43 program for risk factor modification for incontinence to each participant.[18-19] The 44 45 program targeted pelvic floor muscle weakness, obesity, consumption of caffeinated 46 beverages, smoking, vision loss and constipation, with instructions on how to keep a 47 48 bladder diary to help monitor symptoms. The content of the combined intervention 49 50 condensed elements from the continence promotion workshop along with self- 51 management theory, and provided the customized self-management tool to 52 53 participants. The control intervention addressed other non-bladder related aspects of 54 55 older women’s health such as memory problems, polypharmacy, osteoporosis, 56 nutrition, physical fitness and vision impairment. 57 58 59 60 8 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ 35 fluid intake > 1.5 L/day (yes, no), weight and height (self-report) and smoking status 36 (yes, no). At three month follow-up participants were asked whether they had sought 37 38 treatment for urine leakage during the past 3 months. All follow-up interviews were 39 40 performed by the research coordinator, who was blinded to participant identification. 41 42 on September 24, 2021 by guest. Protected copyright. 43 The original study protocol sought to examine reductions in urinary frequency as 44 45 measured on a bladder diary and reductions in the cost of pad use as primary and 46 secondary outcomes respectively. However as soon as recruitment for the trial 47 48 commenced it became apparent that distribution of bladder diaries and objective 49 50 measurement of pad use pre-intervention would not be possible. This occurred as a 51 result of privacy concerns expressed by participating community organisations for 52 53 revealing and sharing their members’ names and contact information with the 54 55 research team prior to the delivery of the workshops.[16] The PGI-I was therefore 56 used as the revised primary measure of effectiveness from the onset of the trial. Data 57 58 on self-efficacy for managing incontinence was also collected, but is not reported in 59 60 9 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 this paper due to problems with comprehension of the visual analogue response scale 4 5 during the 3-month telephone follow-up that occurred non-differentially among 6 participants in all arms of the study. 7 8 9 10 Randomisation and allocation concealment 11 Group allocation occurred by non-stratified randomisation in blocked groups of 4 of 12 13 consenting organisations that agreed to host a workshop. An independent statistician 14 15 at a distantFor study site peerwas responsible review for randomisation using only computer-generated 16 random digits. Community organisations were informed that one of four workshops 17 18 would be delivered on health topics of interest to older women, but not which one. In 19 20 this way, group allocation was concealed from both the clusters and the individual 21 participants, who were invited by the host organisation to attend a “Women’s Health 22 23 Worshop”. The research coordinator remained unaware of group allocation at the 24 25 time each community organisation was recruited to the trial because she was only 26 informed which workshop to prepare for each organisation several days before each 27 28 workshop. The trial is considered open-label because both the research facilitator who 29 30 delivered the intervention and the participants who received it were aware of which 31 intervention was being delivered. 32 33

34 http://bmjopen.bmj.com/ 35 Sample Size 36 The trial was designed to detect a minimal 35% difference in the number of 37 38 participants reporting any improvement (very much better, much better, a little bit 39 40 better on the PGI-I) between the experimental and control conditions, assuming a rate 41 of improvement in the control condition as high as 20%, with 80% power and alpha 42 on September 24, 2021 by guest. Protected copyright. 43 0.05 two-sided (n=34). Using an inflation factor of 1.65 to account for an anticipated 44 45 maximum intracluster correlation (ICC) of 0.05 and unequal cluster size yielded a 46 recruitment target of 56 participants per group.[22] 47 48 49 50 Statistical Methods 51 Differences in baseline characteristics between the four groups were determined. To 52 53 assess the primary outcome we estimated the unadjusted risk difference (prevalence 54 55 of the outcome) and 95% confidence intervals (CI) via generalized estimating 56 equations (GEEs) for participants who reported any improvement on the PGI-I. We 57 58 repeated the same analysis for those who reported significant improvement. GEEs 59 60 10 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 with an identity link and an exchangeable correlation structure were used to account 4 5 for possible correlation between women in the same organisation.[23] To adjust for 6 the imbalance in potential confounders in the groups at baseline, additional analyses 7 8 were conducted using multivariable logistic regression estimated via GEE with an 9 10 exchangeable correlation structure. Potential confounders included age and baseline 11 incontinence severity (ICIQ score) as continuous predictors, and living alone, 12 13 depression, heart disease, falls, arthritis, diabetes, high blood pressure, educational 14 15 status andFor general health peer perception asreview dichotomous predictors. only Both intent-to-treat 16 (ITT) and per protocol (PP) analyses were performed. For the ITT analysis, 17 18 participants with missing data were assumed to have no change in incontinence status 19 20 at three-month follow-up. The number needed to treat was calculated as the inverse of 21 the difference in absolute event rates between the experimental and control 22 23 groups.[24] We report intra-cluster (intra-community organisation) correlation 24 25 coefficients (ICC) from the marginal model using GEE with assumed exchangeable 26 correlation structure and robust standard errors.[25] In cases where an ICC<0 was 27 28 detected, we assumed a correlation structure of independence, but still used the 29 30 robust variance estimator. The robust variance estimator is robust to misspecification 31 of the correlation structure, so standard errors, confidence intervals and p values are 32 33 still correct. To estimate adjusted mean group differences in ICIQ scores from

34 http://bmjopen.bmj.com/ 35 baseline to 3-month follow up, we used GEE with a Gaussian regression model for 36 continuous outcomes and followed the same procedure outlined above. 37 38 39 40 Improvements in incontinence-related knowledge by intervention type for proportions 41 of individuals responding correctly to each knowledge questionnaire item at baseline 42 on September 24, 2021 by guest. Protected copyright. 43 compared to 3-month follow-up were estimated using McNemar’s test for matched 44 45 pair analysis. Rates of improvement in self-reported risk modifying behaviours for 46 incontinence were calculated, along with 95% confidence intervals. Differences in 47 48 improvement rates between the intervention and the control group were compared 49 50 using Fisher’s Exact test using a per protocol analysis. A difference in response for 51 each health behavior item that indicated adoption of a new risk modifying behaviour 52 53 was defined as a positive change at 3-month follow-up compared to baseline. 54 55 Reduced coffee and tea intake refer to individuals who reduced their consumption to a 56 single cup per day or less. Weight loss was determined by a positive response to the 57 58 question, “Has your weight changed (yes, no) and if so, how much do you now 59 60 11 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 weigh?” and evidence of self-reported current weight lower than self-reported weight 4 5 at baseline. All statistical analyses were run using RStudio 0.97.310.0, an integrated 6 development environment for R. 7 8 9 10 RESULTS 11 12 13 Study participants and follow-up 14 15 Four-hundred-and-twentyFor peer different communityreview organisations only were approached over 16 an 18-month period to participate in the trial. Of these, 17% consented and succeeded 17 18 in hosting an intervention, yielding 71 clusters that were randomised. Approximately 19 20 one quarter of the groups contacted refused; 2% expressed interest but were unable to 21 organise a workshop; and a little over half failed to give any response although most 22 23 of them had been followed up and had received extra information on the project.16 24 25 Figure 1 depicts the study flow of the clusters and participants through the trial. 26 Seven-hundred-and-sixty three women attended the workshops, of whom 322 (42%) 27 28 were known to be eligible for the trial. The mean number of participants recruited 29 30 from each cluster for the continence promotion group was 3 + 2, whereas it was 4 + 2 31 for the other 3 groups. Eighty-percent (259/322) of known eligible attendees to the 32 33 workshops consented to take part in the trial. Two-hundred-and-twenty-eight of these

34 http://bmjopen.bmj.com/ 35 (88%) were available for 3-month follow-up. Table 1 compares the baseline 36 characteristics of participants in each trial arm. 37 38 39 40 Primary and secondary outcomes 41 The highest rate of improvement in incontinence occurred in the combined 42 on September 24, 2021 by guest. Protected copyright. 43 intervention group, 66% compared to 11% in the control group (prevalence difference 44 45 55%, 95% CI 43%-67%), yielding a number-needed-to-treat of 2. Thirty percent of 46 the combined group reported significant improvement compared to 6% of controls 47 48 (prevalence difference 23%, 95% CI10%-36%, number-needed-to-treat of 5). In 49 50 adjusted analyses, the likelihood of achieving a significant improvement in urinary 51 symptoms from exposure to the combined intervention was five times greater than 52 53 exposure to the control intervention (OR 4.94, 95% CI 1.45-16.86). Compared to 54 55 controls, the participants in the combined intervention reported an adjusted mean 2.05 56 point (95% CI 0.87-3.24) greater improvement on the ICIQ from baseline to 3-month 57 58 follow up. The adjusted mean difference in ICIQ scores was also significantly higher 59 60 12 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 for the continence education group compared to the control group (1.33 point greater 4 5 improvement (95% CI 0.33-2.32)), but not for the self-management group. The per 6 protocol analysis for the primary outcome and the intra-cluster correlation coefficients 7 8 for each analysis are shown in Table 2. 9 10 11 Other outcomes 12 13 Table 3 shows the changes in incontinence-related knowledge attributable to receipt 14 15 of each intervention.For peerParticipants exposed review to the combined interventiononly showed the 16 greatest acquisition in knowledge, exhibiting significant within-group improvement 17 18 on 6 out of 8 questionnaire items. Participants learned that incontinence is not an 19 20 inevitable or irreversible part of aging, that losing weight, changing the type of fluid 21 intake and performing pelvic floor muscle exercises can reduce urinary symptoms, 22 23 and that wearing undergarment protection is not always the best way to manage 24 25 incontinence. 26 27 28 The proportion of participants with modifiable risk factors for incontinence in each 29 30 group at baseline is shown in Table 1. The adoption of various risk-modifying 31 behaviours among participants occurred to a different degree as a result of exposure to 32 33 all three experimental but not the control intervention (Figure 2). At three-month

34 http://bmjopen.bmj.com/ 35 follow-up, the proportion of women reporting uptake of pelvic floor muscle exercises 36 and weight loss was significantly higher in the continence education group (46% and 37 38 20% respectively), the self-management group (34% and 20%) and the combined 39 40 intervention group (53% and 18%) compared to controls (8% and 3%). Many women 41 additionally reduced their coffee intake and total fluid intake. The proportion of 42 on September 24, 2021 by guest. Protected copyright. 43 women who made an appointment to consult a health professional for urine leakage 44 45 was 19% in the continence promotion group, 7% in the self-management group, 16% 46 in the combined intervention group and 4% in the control group. 47 48 49 50 DISCUSSION 51 52 53 In this cluster-randomised trial testing the effectiveness of 3 different continence 54 55 promotion interventions, we found that health education combined with the delivery 56 of an evidence-based self-management tool via community organisations to untreated 57 58 older women yielded the highest rate of urinary symptom improvement in 66% of 59 60 13 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 recipients, half of whom reported significant improvements in incontinence. These 4 5 outcomes translate into a number-needed-to-treat of 2 and 5 respectively, a magnitude 6 of effect rarely achieved during public health interventions. Both new knowledge 7 8 acquisition and the adoption of risk-modifying behaviours such as exercise and 9 10 weight loss occurred as a result of community organisations’ involvement in reaching 11 untreated incontinent women outside the health care system. 12 13 14 15 StrengthsFor and weaknesses peer of the study review only 16 17 This is the first randomised trial to test the effectiveness of continence promotion 18 19 strategies through community outreach. Both explanatory mechanisms and final 20 21 health outcomes were assessed, and the use of a cluster randomised design was 22 chosen to avoid contamination of the control group.[26,27] Our choice of comparator 23 24 controlled for the placebo effect of participating in a group intervention. Breaches in 25 26 the fidelity and quality of implementation of the intervention were minimized by 27 having the same facilitator deliver each intervention. Improvements in urinary 28 29 symptoms were shown with two validated measures, the PGI-I and the ICIQ. We 30 31 believe the results have wide external validity as the community groups included 32 women with varied educational levels and wide socioeconomic status. 33

34 http://bmjopen.bmj.com/ 35 36 The results of this study confirm findings from previous randomised trials suggesting 37 a positive effect of continence education and self-monitoring strategies on urinary 38 39 symptom improvement in untreated incontinent individuals.[28-33] However all 40 41 previous trials invited participants for clinical assessments prior to the delivery of the 42 intervention, or involved individualized education sessions. The current trial delivered on September 24, 2021 by guest. Protected copyright. 43 44 group continence interventions without medical or nursing evaluations, in a true 45 46 public health approach, to both continent and incontinent women as part of the regular 47 activities offered by each community organisation. Rates of improvement reported in 48 49 this trial on the PGI-I were similar to or exceeded those reported in other studies using 50 51 self-help booklets, in the range of 50%. Because of the nature of recruitment and 52 delivery of the intervention via community organisations, bladder diaries and pad tests 53 54 could not be collected pre-intervention. The results of our trial can therefore not be 55 56 directly compared to other trials that used more objective measures of symptom 57 improvement. 58 59 60 14 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 4 5 Other limitations apply. Due to the nature of recruiting potential participants, 6 individuals could not be screened and enrolled in the trial prior to randomisation of 7 8 the clusters.[27] The result was an imbalance between groups, accounted for by 9 10 analyses that took into account group differences in age, health status and baseline 11 incontinence severity. The trial was not designed to measure the dose-response of 12 13 knowledge acquisition and behaviour change on urinary symptom improvement. Thus 14 15 only crude,For dichotomous peer self-reported review measures of behaviour only change were collected 16 and should be interpreted with caution. 17 18 19 20 Relevance to the discipline 21 The value of continence promotion interventions likely reflects the delivery method as 22 23 well as the quality of the content. Group interventions that deliver continence 24 25 education, self-management information, or a combination of the two will improve 26 incontinence symptoms in 59%, 41% and 66% of recipients, respectively. It is 27 28 surprising that the self-management intervention alone was not associated with 29 30 significant improvement compared to sham control in this trial. This can potentially 31 be explained by the fact that continence education was completely omitted from the 32 33 self-management workshop, whereas some information on bladder functioning was

34 19 http://bmjopen.bmj.com/ 35 provided to participants during the initial work that tested the self-management tool. 36 37 38 Implications for practice 39 40 Implementation of community-based programs that promote behavioural techniques 41 as first-line management for incontinence support evidence for the superior efficacy 42 on September 24, 2021 by guest. Protected copyright. 43 and tolerability of conservative management approaches over pharmacological 44 45 treatment for incontinence.[34] As “silent sufferers” become better informed that 46 effective strategies exist for improving urinary symptoms, patient demand for care 47 48 will likely increase. Almost 20% of women made an appointment to discuss urine 49 50 leakage with a health professional in the three months following receipt of the 51 continence education intervention. Evidence-based guidelines exist for physicians to 52 53 evaluate and manage urinary incontinence when first-line behavioural strategies fail, 54 55 and will need to be more frequently applied.[5,11] 56 57 58 59 60 15 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 In conclusion, continence education combined with self-management delivered via 4 5 community organisations to untreated older women with incontinence leads to 6 symptom improvement in 1 out of every 2 recipients. At the current time, the majority 7 8 of older women with incontinence do not seek care, and either self-manage their 9 10 symptoms inappropriately or use protection to palliate urine leakage.[1-4] As 11 incontinence is associated with multimorbidity and other deleterious health effects, 12 13 results from this trial provide strong justification for public health outreach via 14 15 communityFor organisations peer to reduce urinereview leakage among untreated only individuals. 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 24, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 16 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 17 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ 35 analysis and interpretation and critically reviewed the manuscript. Doneal Thomas 36 conducted the analyses and reviewed the manuscript. Eleanor van den Heuvel helped 37 38 design the study, participated in the study implementation, helped interpret the 39 40 findings and critically reviewed the manuscript. 41 42 on September 24, 2021 by guest. Protected copyright. 43 Data Access and Sharing Statement 44 45 46 Cara Tannenbaum, Doneal Thomas and Andrea Benedetti had full access to all the 47 48 data in the study and take responsibility for the integrity of the data and the accuracy 49 50 of the data analysis. Patient level data and the full dataset is available upon request 51 from the authors. Consent for data sharing was not obtained but the presented data are 52 53 anonymised and the risk of identification is low. 54 55 56 Competing interests 57 58 59 60 17 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 18 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 24, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 18 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 19 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 REFERENCES 4 5 6 1. Sexton CC, Coyne KS, Thompson C, et al. Prevalence and effect on health-related 7 8 quality of life of overactive bladder in older Americans: Results from the 9 10 Epidemiology of Lower Urinary Tract Symptoms Study. J Am Geriatr Soc 11 2011;59:1465–1470. 12 13 2. Tannenbaum C, Corcos J, Assalian P. The relationship between sexual activity 14 15 and urinaryFor incontinence peer in older reviewwomen. J Am Geriatr onlySoc 2006;54:1220-4. 16 3. O’Donnell M, Lose G, Sykes D, et al. Help-seeking behaviour and associated 17 18 factors among women with urinary incontinence in France, Germany, Spain and 19 20 the United Kingdom. Eur Urol 2005;47:385-92 21 4. Herschorn S, Gajewski J, Schulz J, et al. A population-based study of urinary 22 23 symptoms and incontinence: the Canadian Urinary Bladder Survey. BJU Int 24 25 2008;101:52-8. 26 5. Bland DR et al. The effects of implementation of the Agency for Health Care 27 28 Policy and Research urinary incontinence guidelines in primary care. J Am Geriatr 29 30 Soc 2003;51:979–984. 31 32 6. Subak LL, Wagner TH. Talking about incontinence: the first step toward 33

34 prevention and treatment. JAMA 2010;303:2184-2185. http://bmjopen.bmj.com/ 35 7. Tannenbaum C, Gray M, Hoffstetter S, et al. Comorbidities associated with 36 37 bladder dysfunction. Int J Clin Pract 2013;67:105-13. 38 39 8. Farage MA, Miller KW, Berardesca E, et al. Psychosocial and societal burden of 40 incontinence in the aged population. Arch Gynecol Obstet 2008;277:285-290. 41 42 9. Holroyd-Leduc JM, Mehta KM, Covinsky KE. Urinary incontinence and its on September 24, 2021 by guest. Protected copyright. 43 44 association with death, nursing home admission, and functional decline. J Am 45 Geriatr Soc 2004;52:712-8. 46 47 10. Chiarelli PE, Mackenzie LA, Osmotherly PG. Urinary incontinence is associated 48 49 with an increase in falls: a systematic review. Aust J Physiother 2009;55:89-95. 50 11. Holroyd-Leduc JM, Tannenbaum C, Thorpe KE, et al. What type of urinary 51 52 incontinence does this woman have? JAMA 2008;299:1446-56. 53 54 12. Goode PS, Burgio KL, Richter HE, et al. Incontinence in older women. JAMA 55 2010;303:2172-81. 56 57 58 59 60 19 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 20 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 13. Tannenbaum C, Bachand, G, DuBeau CE, et al. Experience of an incontinence 4 5 clinic for older women : no apparent age limit for potential physical and 6 psychological benefits. J Women Health Gend Based Med 2001;10:751-56. 7 8 14. Tannenbaum C, Brouillette J, Corcos J. Rating improvements in urinary 9 10 incontinence: do patients and physicians agree? Age Ageing 2008;37:1-5 11 15. Newman DK, Buckley B, Gordon D et al. Continence promotion, education and 12 13 primary prevention. In Abrams P, Cardozo L, Khoury S, Wein A eds: 14 th 15 Incontinence.For 5 Internationalpeer Consultation review on Incontinence, only Paris, February 16 2012. Health Publication Ltd 2013. 17 18 16. Agnew R, van den Heuvel E, Tannenbaum C. Efficiency of using community 19 20 organisations as catalysts for recruitment to continence promotion trials. Clin 21 Trials 2013;10:151-9. 22 23 17. Tannenbaum C, Drali R, Holroyd-Leduc J Richard L. Lessons learned: Impact of 24 25 a continence promotion activity for older community dwelling women. Neurourol 26 Urodyn 2010;29:540-544. 27 28 18. Wilde MH, Bliss DZ, Booth J, et al. Self-Management of Urinary and Fecal 29 30 Incontinence. AJN 2013 in press. 31 19. Holroyd-Leduc J, Strauss S, Thorpe K, et al. Translation of evidence into a self- 32 33 management tool for use by women with urinary incontinence. Age Ageing

34 http://bmjopen.bmj.com/ 35 2011;40:227-33. 36 20. Yalcin I, Bump RC. Validation of two global impression questionnaires for 37 38 incontinence. Am J Obstet Gynecol 2003;189: 98–101. 39 40 21. Avery K, Donovan J, Peters TJ et al. ICIQ: a brief and robust measure for 41 evaluating the symptoms and impact of urinary incontinence. Neurourol Urodyn 42 on September 24, 2021 by guest. Protected copyright. 43 2004; 23: 322–30. 44 45 22. Eldridge SM; Ashby D; Kerry S. Sample size for cluster randomised trials: effect 46 of coefficient of variation of cluster size and analysis method. Int J Epidemiol 47 48 2006;35:1292-1300. 49 50 23. Ukoumunne OC, Forbes AB, Carlin JB, et al. Comparison of the risk difference, 51 risk ratio and odds ratio scales for quantifying the unadjusted intervention effect in 52 53 cluster randomised trials. Stat Med 2008;27: 5143-5155. 54 55 24. Users’ guide to the medical literature. Integrating research evidence with the care 56 of the individual patient. JAMA 2000;283;2829-2836. 57 58 59 60 20 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 21 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 25. Hanley JA, Negassa A, Edwardes MD. GEE analysis of negatively correlated 4 5 binary responses: a caution. Stat Med 2000;19:715–722. 6 26. Craig P, Dieppe P, Macintyre S, et al. Developing and evaluating complex 7 8 interventions: the new Medical Research Council guidance. BMJ 2008;337:979- 9 10 983. 11 27. Campbell MK, Piaggio G, Elbourne DR, et al. Consort 2010 statement: extension 12 13 to cluster randomised trials. BMJ 2012;345:e5661. 14 15 28. KincadeFor JE, Dougherty peer MC, Carlson review JR, et al. Randomised only clinical trial of 16 efficacy of self-monitoring techniques to treat urinary incontinence in women. 17 18 Neurourol Urodyn 2007;26:507-11. 19 20 29. Wagg AR, Barron D, Kirby M, et al. A randomised partially controlled trial to 21 assess the impact of self-help vs. structured help from a continence nurse 22 23 specialist in women with undiagnosed urinary problems in primary care. Int J Clin 24 25 Pract 2007;61:1863-73. 26 30. Dougherty MC, Dwyer JW, Pendergast JF et al. A randomised trial of behavioral 27 28 management for continence with older rural women. Res Nurs Health 2002;25:3- 29 30 13. 31 31. McFall SL, Yerkes AM, Cowan LD. Outcomes of a small group educational 32 33 intervention for urinary incontinence: episodes of incontinence and other urinary

34 http://bmjopen.bmj.com/ 35 symptoms. J Aging Health 2000;12:250-67. 36 32. Milne J. The impact of information on health behaviors of older adults with 37 38 urinary incontinence. Clin Nurs Res 2000;9:161-76. 39 40 33. Goode PS, Burgio KL, Locher JL et al. Effect of behavioral training with or 41 without pelvic floor electrical stimulation on stress incontinence in women: a 42 on September 24, 2021 by guest. Protected copyright. 43 randomized controlled trial. JAMA 2003;290(3):345-52. 44 45 34. Shamliyan T, Wyman J, Kane RL. Nonsurgical treatments for urinary 46 incontinence in adult women: diagnosis and comparative effectiveness. 47 48 Comparative Effectiveness Review No. 36. (Prepared by the University of 49 50 Minnesota Evidence-based Practice Center under Contract No. HHSA 290-2007- 51 10064-I.) AHRQ Publication No. 11(12)-EHC074-EF. Rockville, MD. Agency 52 53 for Healthcare Research and Quality. April 2012. Available at: 54 55 www.effectivehealthcare.ahrq.gov/reports/final.cfm. 56 57 58 59 60 21 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 22 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 4 Table 1: Baseline characteristics and distribution of modifiable risk factors of 5 participants 6 7

34 Type of incontinence http://bmjopen.bmj.com/ 35 Stress only 15.6 12.9 14.8 33.3 36 Urgency only 32.8 35.7 29.5 20.6 37 Mixed 45.3 42.9 55.7 39.7 38 39 40 Modifiable risk factors 41 Performs pelvic floor 42 muscle exercises 3 times on September 24, 2021 by guest. Protected copyright. 43 per week 18.8 15.7 11.9 15.6 44 Self-reported body mass 45 index > 27 kg/m2 ** 53.2 53.0 42.4 49.2 46 Drinks more than 1.5 litres 47 of fluid/day 43.8 44.3 54.1 37.5 48 49 Drinks one cup of tea or 50 more/day 85.9 84.3 73.8 84.4 51 Drinks one cup of coffee or 52 more/day 46.9 62.9 65.6 64.1 53 Smokes 6.3 4.3 4.9 6.2 54 55 *ICIQ = International Consultation on Incontinence Questionnaire, used to measure 56 57 the severity and bother from urinary incontinence. Scores range from 0 to 21, with 58 higher scores representing worse incontinence. 59 60 22 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 23 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 24, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 23 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

0 0 Page 24 of 63 4.94 24 24 23.27 17.63 (1.45- (5.91- (5.09- 16.86) 91.59) 61.13) vs control vs Combined Combined

SM SM 1.81 0.14 0.06 3.46 2.71 6.60) 8.41) (0.50- (1.08- (0.87- 11.03) vs control vs Adjusted**

0 0 2.83 9.14 10.40 nence (0.59- (3.05- (3.05- Conti- 13.66) 35.48) 27.37) vs control vs

0 0 6.3 17.7) (2.2- 17.14 15.51 (6.51- (6.50- 45.11) 37.01) Odds ratio (95% CI) ratio (95% Odds vs control vs Combined Combined

vs vs SM SM 3.8 3.8 0.25 0.18 4.64 5.16 control (1.48- (1.73- 15.37) 14.56) Crude (1.2-12.2)

4.2 0.03 0.24 11.72 11.45 11.45 (4.27- (4.54- 30.67) 30.21)

vs control vs Continence (1.4-13.0)

0 0 .55 0.23 0.59 0.36) 0.74) 0.67) (0.10- (0.45- (0.43- vs control vs Combined Combined BMJ Open

Any improvement Any http://bmjopen.bmj.com/

0.14 0.25 0.18 0.29 0.28 0.27) 0.51) 0.48) (0.01- (0.07- (0.08- SM vs vs SM CI)* control

Very much better or much better or much better much Very

0.16 0.03 0.02 0.51 0.48 0.29) 0.67) (0.03- (0.34- Prevalence difference (95% difference Prevalence (0.33-0.64 vs control vs Continence

on September 24, 2021 by guest. Protected copyright. - - 0.06 0.13 0.11 Control

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

- - 0.30 0.73 0.66

For peer review only Combined

- - 0.21 0.47 0.41 SM

- - Prevalence at 3-month follow-up at 3-month Prevalence 0.22 0.64 0.59 nence Conti-

Intention-to- treat

ICC per per ICC ICC Intention-to- treat Per protocol Per Intention-to- treat

protocol Table 2: Prevalence, risk difference and odds ratios for self-reported improvements in incontinence at 3-months at 3-months in incontinence improvements self-reported for andratios difference odds Tablerisk Prevalence, 2: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

0 0 5.32 25 25 (1.39- 20.34)

0 0.02 2.28 8.24) (0.63-

0 0 2.51 (0.48- 13.18)

0 0 5.8 (2.0-17.4)

3.5 3.5 0.08 0.06 (1.06- 11.31)

3.7 0.01 0.02 11.3) (1.2-

0 0.02 0.26 0.42) (0.10- BMJ Open

http://bmjopen.bmj.com/ 0.08 0.06 0.15 0.29) (0.00-

0.01 0.02 0.16 0.30) (0.02-

on September 24, 2021 by guest. Protected copyright. - - 0.08 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

- - 0.33 For peer review only

- - 0.24

ICC per ICC ICC Intention-to- treat Per protocol Per protocol Page 25 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from Page 26 of 63 26 26

1 1 68.6 66.1 69.2 67.2 0.63 90.4 86.9 51.9 50.8 0.73 82.7 79.7

n=64 Control

7.3 89.1 68.3 27.3 52.5 0.73 92.7 88.3 36.1 38.2 77.0 0.001 <0.001 <0.001 n=61 Combined Combined intervention

75.8 66.7 0.11 33.9 40.6 0.77 88.7 92.8 0.06 17.7 32.9 0.02 63.9 79.7 Self- n=70 management

77.6 64.5 0.03 36.8 57.1 93.1 88.9 0.12 29.3 41.9 36.2 73.0 <0.001 BMJ Open n=64 education Continence Continence http://bmjopen.bmj.com/

on September 24, 2021 by guest. Protected copyright. For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

3 month follow up % % agreement follow up 3 month Baseline % agreement Baseline 3 month follow up % % agreement follow up 3 month Baseline % agreement Baseline 3 month follow up % % agreement follow up 3 month Baseline % agreement Baseline 3 month follow up % % agreement follow up 3 month Baseline % agreement Baseline 3 month follow up % % agreement follow up 3 month Baseline % agreement Baseline

p value for change p value p value for change p value p value for change p value For peer for change* p value review only

by many different different things many by knowledge Table in incontinence-related Change 3: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from 27 27

75 1.0 98.1 96.9 0.48 80.8 0.79 65.4 71.4 0.77

n=64 Control

0.02 98.2 85.2 90.9 66.1 0.36 76.4 67.2 0.01 <0.001 n=61 Combined Combined intervention

0.13 96.7 88.6 0.69 77.4 74.3 0.65 65.2 0.36 Self- n=70 management

0.04 96.5 85.5 0.08 77.6 61.3 0.63 77.6 72.6 0.26 BMJ Open n=64 education Continence Continence http://bmjopen.bmj.com/

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

p value for change p value p value for change p value p value for change p value p value for change p value

For peer review only

uscles can help control urine urine control help can uscles

8. Exercising pelvic floor pelvic Exercising 8. m leakage 7. Losing weight can lead to lead to can weight 7. Losing incontinence in improvement 6. How much you drink can drink can you much How 6. leakage urine to contribute

* McNemar's test for matched-pairs data matched-pairs for test McNemar's * Page 27 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open Page 28 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ 35 women, urinary incontinence 36 37 38 Corresponding Author: Dr. Cara Tannenbaum, Centre de Recherche de l’Institut 39 40 universitaire de gériatrie de Montréal, 4545 Queen Mary Road, Montreal, Quebec, 41 Canada H3W 1W5. Tel (514) 340-2800. Fax : (514) 340-3530. 42 on September 24, 2021 by guest. Protected copyright. 43 Email: [email protected] 44 45 46 The Corresponding Author has the right to grant on behalf of all authors and does 47 48 grant on behalf of all authors, an exclusive licence on a worldwide basis to the BMJ 49 50 Publishing Group Ltd to permit this article (if accepted) to be published in BMJ Open 51 editions and any other BMJPGL products and sublicences such use and exploit all 52 53 subsidiary rights, as set out in the licence. 54 55 56 57 58 59 60 1 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 29 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 24, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 2 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 30 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 ABSTRACT 4 5 6 Objectives: The primary objective of this cluster randomised controlled trial was to 7 8 compare the effectiveness of the three experimental continence promotion 9 10 interventions against a control intervention on urinary symptom improvement in older 11 women with untreated incontinence recruited from community organisations. A 12 13 second objective was to determine whether changes in incontinence-related 14 15 knowledgeFor and new uptakepeer of risk-modifying review behaviours explainonly these improvements. 16 17 18 Setting: 71 community organisations across the United Kingdom 19 20 21 Participants: 259 women aged 60 years and older with untreated incontinence 22 23 entered the trial; 88% completed the 3-month follow-up. 24 25 26 Interventions: The three active interventions consisted of a single 60-minute group 27 28 workshop on 1) continence education (20 clusters, 64 women); 2) evidence-based 29 30 self-management (17 clusters, 70 women); or 3) combined continence education and 31 self-management (17 clusters, 61 women). The control intervention was a single 60- 32 33 minute educational group workshop on memory loss, polypharmacy and osteoporosis

34 http://bmjopen.bmj.com/ 35 (17 clusters, 64 women). 36 37 38 Primary and secondary outcome measures: The primary outcome was self-reported 39 40 improvement in incontinence 3 months post-intervention at the level of the individual. 41 The secondary outcome was change in the International Consultation on Incontinence 42 on September 24, 2021 by guest. Protected copyright. 43 Questionnaire (ICIQ) from baseline to 3-month follow-up. Changes in incontinence- 44 45 related knowledge and behaviours were also assessed. 46 47 48 Results: The highest rate of urinary symptom improvement occurred in the combined 49 50 intervention group (66% vs 11% of the control group, prevalence difference 55%, 51 95% CI 43%-67%, intracluster correlation 0). Thirty percent versus 6% of participants 52 53 reported significant improvement respectively (prevalence difference 23%, 95% CI 54 55 10%-36%, intracluster correlation 0). The number-needed-to-treat was 2 to achieve 56 any improvement in incontinence symptoms, and 5 to attain significant improvement. 57 58 Compared to controls, participants in the combined intervention reported an adjusted 59 60 3 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 31 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 24, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 4 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 32 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 Article Summary : Strengths and limitations of this study 4 5 6 • First study to provide Level 1 evidence that continence promotion is an 7 8 effective strategy for improving urinary symptoms among untreated 9 10 community-dwelling older women 11 12 13 • Participants were recruited via community organisations with representation 14 15 acrossFor diverse peer socio-economic review strata only 16 17 18 • Rates of knowledge acquisition and behaviour change provide an explanatory 19 20 mechanism for the observed improvements in incontinence in participants 21 22 receiving the combined education plus self-management strategy 23 24 25 • Only self-reported outcomes and crude dichotomous measures of behaviour 26 27 change were collected so results must be interpreted with caution 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 24, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 5 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 33 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 organisations is an unexplored strategy for improving urinary symptoms.[16] Data http://bmjopen.bmj.com/ 35 36 from randomised trials are needed to determine whether the delivery of an evidence- 37 based continence intervention via community organisations is an effective method for 38 39 treating incontinence. 40 41 42 The primary objective of this cluster randomised controlled trial was to compare the on September 24, 2021 by guest. Protected copyright. 43 44 effectiveness of the three experimental continence promotion interventions against a 45 46 control intervention on urinary symptom improvement in older women with untreated 47 incontinence recruited from community organisations. A second objective was to 48 49 determine whether changes in incontinence-related knowledge, attitudes and new 50 51 uptake of risk-modifying behaviours explain improvements in incontinence. We 52 hypothesized that continence education combined with evidence-based self- 53 54 management would yield the greatest improvement in incontinence symptoms, 55 56 measured at the level of the individual, 3-months post-intervention. 57

58 59 60 6 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 34 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ 35 groups, seniors’ housing groups, women’s lobby groups and Asian caregiver 36 associations.[16] Organisations were contacted strategically by convenience 37 38 sampling, word of mouth and referral. A research coordinator approached community 39 40 organisations to join the trial by telephone, email and newspaper advertising. 41 42 on September 24, 2021 by guest. Protected copyright. 43 Inclusion criteria for participants were women aged 60 years and older who reported 44 45 urinary incontinence at least once weekly on the International Consultation on 46 Incontinence Questionnaire (ICIQ), and who were not under active treatment for 47 48 incontinence. For privacy reasons, many community organisations were 49 50 uncomfortable screening their members for incontinence in advance, so eligibility to 51 participate in the trial could only be ascertained by the research coordinator on the day 52 53 of delivery of the intervention.[16] Eligibility to participate in the trial was established 54 55 by asking all attendees at the workshop to complete a baseline screening questionnaire 56 upon arrival. At this time, a study information sheet and consent form were 57 58 distributed to all participants. All women, regardless of eligibility or desire to enrol in 59 60 7 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 35 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 the trial, were permitted to stay for the workshop. Only those women who wished to 4 5 enrol in the trial submitted the signed consent form to the workshop facilitator 6 following delivery of the intervention, however all attendees were encouraged to 7 8 submit the baseline screening questionnaire even if they were continent or did not 9 10 wish to participate in the trial. 11 12 13 Interventions 14 15 For peer review only 16 The interventions were applied at the level of each cluster. The three experimental 17 18 interventions to be tested were continence education, self-management including the 19 20 distribution of an evidence-based risk factor reduction tool for incontinence, and a 21 combined intervention that included both components. The sham control intervention 22 23 was a lecture on health promotion for older women that addressed topics other than 24 25 incontinence. All interventions were delivered once in group format to 8-16 women 26 by the same facilitator at a venue of the organisation’s choosing, and lasted 60-90 27 28 minutes. A slide presentation with a pre-established script prepared for the facilitator 29 30 was delivered at each workshop. 31 32 33 The continence promotion intervention incorporated elements of constructivist

34 http://bmjopen.bmj.com/ 35 learning that challenged older adults’ erroneous beliefs about accepting incontinence 36 as a normal part of aging, and aimed to change attitudes and create new knowledge 37 38 about the different types, etiology, risk factors and treatment options for urine 39 40 loss.[16-17] The self-management workshop reviewed self-management theory in an 41 interactive format, and provided a customised evidence-based self-management 42 on September 24, 2021 by guest. Protected copyright. 43 program for risk factor modification for incontinence to each participant.[18-19] The 44 45 program targeted pelvic floor muscle weakness, obesity, consumption of caffeinated 46 beverages, smoking, vision loss and constipation, with instructions on how to keep a 47 48 bladder diary to help monitor symptoms. The content of the combined intervention 49 50 condensed elements from the continence promotion workshop along with self- 51 management theory, and provided the customized self-management tool to 52 53 participants. The control intervention addressed other non-bladder related aspects of 54 55 older women’s health such as memory problems, polypharmacy, osteoporosis, 56 nutrition, physical fitness and vision impairment. 57 58 59 60 8 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 36 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ 35 fluid intake > 1.5 L/day (yes, no), weight and height (self-report) and smoking status 36 (yes, no). At three month follow-up participants were asked whether they had sought 37 38 treatment for urine leakage during the past 3 months. All follow-up interviews were 39 40 performed by the research coordinator, who was blinded to participant identification. 41 42 on September 24, 2021 by guest. Protected copyright. 43 The original study protocol sought to examine reductions in urinary frequency as 44 45 measured on a bladder diary and reductions in the cost of pad use as primary and 46 secondary outcomes respectively. However as soon as recruitment for the trial 47 48 commenced it became apparent that distribution of bladder diaries and objective 49 50 measurement of pad use pre-intervention would not be possible. This occurred as a 51 result of privacy concerns expressed by participating community organisations for 52 53 revealing and sharing their members’ names and contact information with the 54 55 research team prior to the delivery of the workshops.[16] The PGI-I was therefore 56 used as the revised primary measure of effectiveness from the onset of the trial. Data 57 58 on self-efficacy for managing incontinence was also collected, but is not reported in 59 60 9 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 37 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ 35 Sample Size 36 The trial was designed to detect a minimal 35% difference in the number of 37 38 participants reporting any improvement (very much better, much better, a little bit 39 40 better on the PGI-I) between the experimental and control conditions, assuming a rate 41 of improvement in the control condition as high as 20%, with 80% power and alpha 42 on September 24, 2021 by guest. Protected copyright. 43 0.05 two-sided (n=34). Using an inflation factor of 1.65 to account for an anticipated 44 45 maximum intracluster correlation (ICC) of 0.05 and unequal cluster size yielded a 46 recruitment target of 56 participants per group.[22] 47 48 49 50 Statistical Methods 51 Differences in baseline characteristics between the four groups were determined. To 52 53 assess the primary outcome we estimated the unadjusted risk difference (prevalence 54 55 of the outcome) and 95% confidence intervals (CI) via generalized estimating 56 equations (GEEs) for participants who reported any improvement on the PGI-I. We 57 58 repeated the same analysis for those who reported significant improvement. GEEs 59 60 10 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 38 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ 35 baseline to 3-month follow up, we used GEE with a Gaussian regression model for 36 continuous outcomes and followed the same procedure outlined above. 37 38 39 40 Improvements in incontinence-related knowledge by intervention type for proportions 41 of individuals responding correctly to each knowledge questionnaire item at baseline 42 on September 24, 2021 by guest. Protected copyright. 43 compared to 3-month follow-up were estimated using McNemar’s test for matched 44 45 pair analysis. Rates of improvement in self-reported risk modifying behaviours for 46 incontinence were calculated, along with 95% confidence intervals. Differences in 47 48 improvement rates between the intervention and the control group were compared 49 50 using Fisher’s Exact test using a per protocol analysis. A difference in response for 51 each health behavior item that indicated adoption of a new risk modifying behaviour 52 53 was defined as a positive change at 3-month follow-up compared to baseline. 54 55 Reduced coffee and tea intake refer to individuals who reduced their consumption to a 56 single cup per day or less. Weight loss was determined by a positive response to the 57 58 question, “Has your weight changed (yes, no) and if so, how much do you now 59 60 11 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 39 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ 35 (88%) were available for 3-month follow-up. Table 1 compares the baseline 36 characteristics of participants in each trial arm. 37 38 39 40 Primary and secondary outcomes 41 The highest rate of improvement in incontinence occurred in the combined 42 on September 24, 2021 by guest. Protected copyright. 43 intervention group, 66% compared to 11% in the control group (prevalence difference 44 45 55%, 95% CI 43%-67%), yielding a number-needed-to-treat of 2. Thirty percent of 46 the combined group reported significant improvement compared to 6% of controls 47 48 (prevalence difference 23%, 95% CI10%-36%, number-needed-to-treat of 5). In 49 50 adjusted analyses, the likelihood of achieving a significant improvement in urinary 51 symptoms from exposure to the combined intervention was five times greater than 52 53 exposure to the control intervention (OR 4.94, 95% CI 1.45-16.86). Compared to 54 55 controls, the participants in the combined intervention reported an adjusted mean 2.05 56 point (95% CI 0.87-3.24) greater improvement on the ICIQ from baseline to 3-month 57 58 follow up. The adjusted mean difference in ICIQ scores was also significantly higher 59 60 12 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 40 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ 35 follow-up, the proportion of women reporting uptake of pelvic floor muscle exercises 36 and weight loss was significantly higher in the continence education group (46% and 37 38 20% respectively), the self-management group (34% and 20%) and the combined 39 40 intervention group (53% and 18%) compared to controls (8% and 3%). Many women 41 additionally reduced their coffee intake and total fluid intake. The proportion of 42 on September 24, 2021 by guest. Protected copyright. 43 women who made an appointment to consult a health professional for urine leakage 44 45 was 19% in the continence promotion group, 7% in the self-management group, 16% 46 in the combined intervention group and 4% in the control group. 47 48 49 50 DISCUSSION 51 52 53 In this cluster-randomised trial testing the effectiveness of 3 different continence 54 55 promotion interventions, we found that health education combined with the delivery 56 of an evidence-based self-management tool via community organisations to untreated 57 58 older women yielded the highest rate of urinary symptom improvement in 66% of 59 60 13 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 41 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ 35 36 The results of this study confirm findings from previous randomised trials suggesting 37 a positive effect of continence education and self-monitoring strategies on urinary 38 39 symptom improvement in untreated incontinent individuals.[28-33] However all 40 41 previous trials invited participants for clinical assessments prior to the delivery of the 42 intervention, or involved individualized education sessions. The current trial delivered on September 24, 2021 by guest. Protected copyright. 43 44 group continence interventions without medical or nursing evaluations, in a true 45 46 public health approach, to both continent and incontinent women as part of the regular 47 activities offered by each community organisation. Rates of improvement reported in 48 49 this trial on the PGI-I were similar to or exceeded those reported in other studies using 50 51 self-help booklets, in the range of 50%. Because of the nature of recruitment and 52 delivery of the intervention via community organisations, bladder diaries and pad tests 53 54 could not be collected pre-intervention. The results of our trial can therefore not be 55 56 directly compared to other trials that used more objective measures of symptom 57 improvement. 58 59 60 14 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 42 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 19 http://bmjopen.bmj.com/ 35 provided to participants during the initial work that tested the self-management tool. 36 37 38 Implications for practice 39 40 Implementation of community-based programs that promote behavioural techniques 41 as first-line management for incontinence support evidence for the superior efficacy 42 on September 24, 2021 by guest. Protected copyright. 43 and tolerability of conservative management approaches over pharmacological 44 45 treatment for incontinence.[34] As “silent sufferers” become better informed that 46 effective strategies exist for improving urinary symptoms, patient demand for care 47 48 will likely increase. Almost 20% of women made an appointment to discuss urine 49 50 leakage with a health professional in the three months following receipt of the 51 continence education intervention. Evidence-based guidelines exist for physicians to 52 53 evaluate and manage urinary incontinence when first-line behavioural strategies fail, 54 55 and will need to be more frequently applied.[5,11] 56 57 58 59 60 15 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 43 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 24, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 16 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 44 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ 35 analysis and interpretation and critically reviewed the manuscript. Doneal Thomas 36 conducted the analyses and reviewed the manuscript. Eleanor van den Heuvel helped 37 38 design the study, participated in the study implementation, helped interpret the 39 40 findings and critically reviewed the manuscript. 41 42 on September 24, 2021 by guest. Protected copyright. 43 Data Access and Sharing Statement 44 45 46 Cara Tannenbaum, Doneal Thomas and Andrea Benedetti had full access to all the 47 48 data in the study and take responsibility for the integrity of the data and the accuracy 49 50 of the data analysis. Patient level data and the full dataset is available upon request 51 from the authors. Consent for data sharing was not obtained but the presented data are 52 53 anonymised and the risk of identification is low. 54 55 56 Competing interests 57 58 59 60 17 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 45 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 24, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 18 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 46 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 prevention and treatment. JAMA 2010;303:2184-2185. http://bmjopen.bmj.com/ 35 7. Tannenbaum C, Gray M, Hoffstetter S, Cardozo L. Comorbidities associated with 36 37 bladder dysfunction. Int J Clin Pract 2013;67:105-13. 38 39 8. Farage MA, Miller KW, Berardesca E, Maibach HI. Psychosocial and societal 40 burden of incontinence in the aged population. Arch Gynecol Obstet 41 42 2008;277:285-290. on September 24, 2021 by guest. Protected copyright. 43 44 9. Holroyd-Leduc JM, Mehta KM, Covinsky KE. Urinary incontinence and its 45 association with death, nursing home admission, and functional decline. J Am 46 47 Geriatr Soc 2004;52:712-8. 48 49 10. Chiarelli PE, Mackenzie LA, Osmotherly PG. Urinary incontinence is associated 50 with an increase in falls: a systematic review. Aust J Physiother 2009;55:89-95. 51 52 11. Holroyd-Leduc JM, Tannenbaum C, Thorpe KE, Straus SE. What type of urinary 53 54 incontinence does this woman have? JAMA 2008;299:1446-56. 55 12. Goode PS, Burgio KL, Richter HE, Markland AD. Incontinence in older women. 56 57 JAMA 2010;303:2172-81. 58 59 60 19 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 47 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ 35 urinary incontinence. Age Ageing 2011;40:227-33. 36 20. Yalcin I, Bump RC. Validation of two global impression questionnaires for 37 38 incontinence. Am J Obstet Gynecol 2003;189: 98–101. 39 40 21. Avery K, Donovan J, Peters TJ et al. ICIQ: a brief and robust measure for 41 evaluating the symptoms and impact of urinary incontinence. Neurourol Urodyn 42 on September 24, 2021 by guest. Protected copyright. 43 2004; 23: 322–30. 44 45 22. Eldridge SM; Ashby D; Kerry S. Sample size for cluster randomised trials: effect 46 of coefficient of variation of cluster size and analysis method. Int J Epidemiol 47 48 2006;35:1292-1300. 49 50 23. Ukoumunne OC, Forbes AB, Carlin JB, Gulliford MC. Comparison of the risk 51 difference, risk ratio and odds ratio scales for quantifying the unadjusted 52 53 intervention effect in cluster randomised trials. Stat Med 2008;27: 5143-5155. 54 55 24. Users’ guide to the medical literature. Integrating research evidence with the care 56 of the individual patient. JAMA 2000;283;2829-2836. 57 58 59 60 20 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 48 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ 35 symptoms. J Aging Health 2000;12:250-67. 36 32. Milne J. The impact of information on health behaviors of older adults with 37 38 urinary incontinence. Clin Nurs Res 2000;9:161-76. 39 40 33. Goode PS, Burgio KL, Locher JL et al. Effect of behavioral training with or 41 without pelvic floor electrical stimulation on stress incontinence in women: a 42 on September 24, 2021 by guest. Protected copyright. 43 randomized controlled trial. JAMA 2003;290(3):345-52. 44 45 34. Shamliyan T, Wyman J, Kane RL. Nonsurgical treatments for urinary 46 incontinence in adult women: diagnosis and comparative effectiveness. 47 48 Comparative Effectiveness Review No. 36. (Prepared by the University of 49 50 Minnesota Evidence-based Practice Center under Contract No. HHSA 290-2007- 51 10064-I.) AHRQ Publication No. 11(12)-EHC074-EF. Rockville, MD. Agency 52 53 for Healthcare Research and Quality. April 2012. Available at: 54 55 www.effectivehealthcare.ahrq.gov/reports/final.cfm. 56 57 58 59 60 21 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 49 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 4 Table 1: Baseline characteristics and distribution of modifiable risk factors of 5 participants 6 7

34 Type of incontinence http://bmjopen.bmj.com/ 35 Stress only 15.6 12.9 14.8 33.3 36 Urgency only 32.8 35.7 29.5 20.6 37 Mixed 45.3 42.9 55.7 39.7 38 39 40 Modifiable risk factors 41 Performs pelvic floor 42 muscle exercises 3 times on September 24, 2021 by guest. Protected copyright. 43 per week 18.8 15.7 11.9 15.6 44 Self-reported body mass 45 index > 27 kg/m2 ** 53.2 53.0 42.4 49.2 46 Drinks more than 1.5 litres 47 of fluid/day 43.8 44.3 54.1 37.5 48 49 Drinks one cup of tea or 50 more/day 85.9 84.3 73.8 84.4 51 Drinks one cup of coffee or 52 more/day 46.9 62.9 65.6 64.1 53 Smokes 6.3 4.3 4.9 6.2 54 55 *ICIQ = International Consultation on Incontinence Questionnaire, used to measure 56 57 the severity and bother from urinary incontinence. Scores range from 0 to 21, with 58 higher scores representing worse incontinence. 59 60 22 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 50 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

0 0 4.94 24 24 23.27 17.63 (1.45- (5.91- (5.09- 16.86) 91.59) 61.13) vs control vs Combined Combined

SM SM 1.81 0.14 0.06 3.46 2.71 6.60) 8.41) (0.50- (1.08- (0.87- 11.03) vs control vs Adjusted**

0 0 2.83 9.14 10.40 nence (0.59- (3.05- (3.05- Conti- 13.66) 35.48) 27.37) vs contro vs

0 0 6.3 17.7) (2.2- 17.14 15.51 (6.51- (6.50- 45.11) 37.01) Odds ratio (95% CI) ratio (95% Odds vs control vs Combined Combined

vs vs SM SM 3.8 3.8 0.25 0.18 4.64 5.16 control (1.48- (1.73- 15.37) 14.56) Crude (1.2-12.2)

4.2 0.03 0.24 11.72 11.45 11.45 (4.27- (4.54- 30.67) 30.21)

vs control vs Continence (1.4-13.0)

0 0 .55 0.23 0.59 0.36) 0.74) 0.67) (0.10- (0.45- (0.43- vs control vs Combined Combined BMJ Open

Any improvement Any http://bmjopen.bmj.com/

0.14 0.25 0.18 0.29 0.28 0.27) 0.51) 0.48) (0.01- (0.07- (0.08- SM vs vs SM CI)* control

Very much better or much better or much better much Very

0.16 0.03 0.02 0.51 0.48 0.29) 0.67) (0.03- (0.34- Prevalence difference (95% difference Prevalence (0.33-0.64 vs control vs Continence

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

- - 0.30 0.73 0.66

For peer review only Combined

- - 0.21 0.47 0.41 SM

- - Prevalence at 3-month follow-up at 3-month Prevalence 0.22 0.64 0.59 nence Conti-

Intention-to- treat

ICC per per ICC ICC Intention-to- treat Per protocol Per Intention-to- treat

protocol Table 2: Prevalence, risk difference and odds ratios for self-reported improvements in incontinence at 3-months at 3-months in incontinence improvements self-reported for andratios difference odds Tablerisk Prevalence, 2: Page 51 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

0 0 Page 52 of 63 5.32 25 25 (1.39- 20.34)

0 0.02 2.28 8.24) (0.63-

0 0 2.51 (0.48- 13.18)

0 0 5.8 (2.0-17.4)

3.5 3.5 0.08 0.06 (1.06- 11.31)

3.7 0.01 0.02 11.3) (1.2-

0 0.02 0.26 0.42) (0.10- BMJ Open

http://bmjopen.bmj.com/ 0.08 0.06 0.15 0.29) (0.00-

0.01 0.02 0.16 0.30) (0.02-

on September 24, 2021 by guest. Protected copyright. - - 0.08 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

- - 0.33 For peer review only

- - 0.24

1 1 68.6 66.1 69.2 67.2 0.63 90.4 86.9 51.9 50.8 0.73 82.7 79.7

n=64 Control

7.3 89.1 68.3 27.3 52.5 0.73 92.7 88.3 36.1 38.2 77.0 0.001 <0.001 <0.001 n=61 Combined Combined intervention

75.8 66.7 0.11 33.9 40.6 0.77 88.7 92.8 0.06 17.7 32.9 0.02 63.9 79.7 Self- n=70 management

77.6 64.5 0.03 36.8 57.1 93.1 88.9 0.12 29.3 41.9 36.2 73.0 <0.001 BMJ Open n=64 education Continence Continence http://bmjopen.bmj.com/

on September 24, 2021 by guest. Protected copyright. For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

3 month follow up % % agreement follow up 3 month Baseline % agreement Baseline 3 month follow up % % agreement follow up 3 month Baseline % agreement Baseline 3 month follow up % % agreement follow up 3 month Baseline % agreement Baseline 3 month follow up % % agreement follow up 3 month Baseline % agreement Baseline 3 month follow up % % agreement follow up 3 month Baseline % agreement Baseline

p value for change p value p value for change p value p value for change p value For peer for change* p value review only

by many different different things many by knowledge Table in incontinence-related Change 3: Page 53 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from Page 54 of 63 27 27

75 1.0 98.1 96.9 0.48 80.8 0.79 65.4 71.4 0.77

n=64 Control

0.02 98.2 85.2 90.9 66.1 0.36 76.4 67.2 0.01 <0.001 n=61 Combined Combined intervention

0.13 96.7 88.6 0.69 77.4 74.3 0.65 65.2 0.36 Self- n=70 management

0.04 96.5 85.5 0.08 77.6 61.3 0.63 77.6 72.6 0.26 BMJ Open n=64 education Continence Continence http://bmjopen.bmj.com/

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

p value for change p value p value for change p value p value for change p value p value for change p value

For peer review only

* McNemar's test for matched-pairs data matched-pairs for test McNemar's * 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 55 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 Figure 1: Flow of participants through the trial

34 235x186mm (300 x 300 DPI) http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 24, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 56 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 Figure 2: Change in risk-modifying behaviours at 3-month follow-up 29 297x186mm (300 x 300 DPI) 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 24, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 57 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ 35 36 3b Important changes to - 37 methods after trial 38 commencement (such as 39 eligibility criteria), with 40 reasons 41 42 Participants 4a Eligibility criteria for Eligibility criteria for clusters 6-7 on September 24, 2021 by guest. Protected copyright. 43 participants 44 45 4b Settings and locations where 6 46 the data were collected 47 48 Interventions 5 The interventions for each Whether interventions pertain to 7 49 group with sufficient details the cluster level, the individual 50 to allow replication, participant level or both 51 including how and when they 52 were actually administered 53 54 Outcomes 6a Completely defined pre- Whether outcome measures 8 55 specified primary and pertain to the cluster level, the 56 secondary outcome individual participant level or both 57 measures, including how and 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 58 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 concealment implement the random based on clusters rather than http://bmjopen.bmj.com/ 35 mechanism allocation sequence (such as individuals and whether allocation 36 sequentially numbered concealment (if any) was at the 37 containers), describing any cluster level, the individual 38 steps taken to conceal the participant level or both 39 sequence until interventions 40 were assigned 41 42 Implementation 10 Who generated the random Replace by 10a, 10b and 10c - on September 24, 2021 by guest. Protected copyright. 43 allocation sequence, who 44 enrolled participants, and 45 who assigned participants to 46 47 interventions 48 10a Who generated the random 8 49 allocation sequence, who enrolled 50 clusters, and who assigned 51 clusters to interventions 52 53 54 55 10b Mechanism by which individual 6 56 participants were included in 57 clusters for the purposes of the 58 trial (such as complete 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 59 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ 35 36 Results 37 38 Participant flow (a 13a For each group, the numbers For each group, the numbers of 26 39 diagram is strongly of participants who were clusters that were randomly 40 recommended) randomly assigned, received assigned, received intended 41 intended treatment, and treatment, and were analysed for 42 were analysed for the the primary outcome on September 24, 2021 by guest. Protected copyright. 43 primary outcome 44 45 13b For each group, losses and For each group, losses and 26 46 exclusions after exclusions for both clusters and 47 randomisation, together with individual cluster members 48 reasons 49 50 Recruitment 14a Dates defining the periods of 6 51 recruitment and follow-up 52 53 14b Why the trial ended or was - 54 stopped 55 56 Baseline data 15 A table showing baseline Baseline characteristics for the 20 57 demographic and clinical individual and cluster levels as 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 60 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 group (for specific guidance http://bmjopen.bmj.com/ 35 see CONSORT for harms3) 36 37 Discussion 38 39 Limitations 20 Trial limitations, addressing 12 40 sources of potential bias, 41 imprecision, and, if relevant, 42 multiplicity of analyses on September 24, 2021 by guest. Protected copyright. 43 44 Generalisability 21 Generalisability (external Generalisability to clusters and/or 13-14 45 validity, applicability) of the individual participants (as 46 47 trial findings relevant) 48 Interpretation 22 Interpretation consistent 13-14 49 50 with results, balancing 51 benefits and harms, and 52 considering other relevant 53 evidence 54 55 Other information 56 57 Registration 23 Registration number and 4 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 61 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 24, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 62 of 63 BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from

34 http://bmjopen.bmj.com/ Numbers randomized Number of participants randomized to Number of clusters randomized to each 35 each group group 36 1 37 Recruitment Trial status 38 Numbers analysed Number of participants analysed in each Number of clusters analysed in each 39 group group 40 Outcome For the primary outcome, a result for each Results at the cluster or individual 41 group and the estimated effect size and its participant level as applicable for each 42 precision primary outcome on September 24, 2021 by guest. Protected copyright. 43 44 Harms Important adverse events or side effects 45 Conclusions General interpretation of the results 46 Trial registration Registration number and name of trial 47 register 48 Funding Source of funding 49 50 51 52 53 54 55 56 57 58 1 Relevant to Conference Abstracts 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 63 of 63 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-004135 on 10 December 2013. Downloaded from