This was a realist-informed participatory research design [2]. There were two data collection and theory development stages – a rapid realist literature review (n=21 published studies) in consultation with an international continence-expert stakeholder group (n=5), followed by qualitative interviews with stroke survivors (n=7) and their partners (n=5).
The rapid realist review of published and grey literature drew on the standards for realist synthesis [3]. We invited five international continence-expert researchers to determine the review scope, identify key literature and researchers for follow-up. An information specialist provided guidance in the English language searches of Medline, Cinahl, Scopus, Proquest and Web of Science databases (all from inception to October 2018). We also searched grey literature (e.g. stroke guidelines) in five developed countries for clinical recommendations about post-stroke continence management based on best evidence syntheses. Study characteristics and data were extracted into an Excel spreadsheet and Word document respectively. Data relevance (the depth and descriptive detail about patient experiences) and rigour (study quality) were assessed and grade high, moderate or low. Both extraction and assessment were performed by one researcher and checked by a second. We inductively coded the included literature (n=21) in relation to intervention architecture, contexts, mechanisms, and outcomes using bespoke evidence tables in Word. In response to the emerging findings and consistent with realist literature review methods, we conducted additional targeted literature searches to optimise relevant theoretical conceptualisation. Specifically, we searched for concepts associated with self-efficacy, self-management, participation and community integration for people experiencing incontinence after stroke.
Next, we interviewed stroke survivors and their partners (providing unpaid care) to ensure that the explanatory theory was relevant to their experiences, the current and local service delivery context. In the interview we explored their experiences, including the various contexts and mechanisms that appeared to contribute (or not) to the successful negotiation of incontinence to enable social participation. We purposively selected stroke survivors who were either at the point of discharge from inpatient rehabilitation or at the point of discharge from formal community-based rehabilitation. Eligible participants had a stroke, were aged 18 years or over, had experienced urinary incontinence at some point post-stroke and lived in their own home. While the literature review informed the interview schedule, we remained open to the direction and priority given to interview topics by the different participants. Interviews were coded independently by two researchers. NVivo was used to gather the data under the following code headings: (a) intervention resources, actions or tasks, (b) contextual factors, (c) mechanisms, and (d) outcomes. To focus on chains of causality, we paid attention to data that provided evidence of links between these headings. The researchers compared and debated their coding decisions throughout the analytic process.
When synthesising the findings from the two data collection stages, we used context-mechanism-outcome (CMO) configurations as an analytic tool to provide a deeper understanding of relationships between the context in which continence management support is delivered, how these strategies or programs are intended to work (i.e., the mechanisms of effect) and participation-related outcomes that have occurred in response to receiving continence management support. Analysis was iterative, and conceptualisation and refinement of theory occurred as the emerging theory was presented to and ‘checked’ with members of the international continence-expert stakeholder group. The theory is, therefore, an analytical product and not merely a descriptive account. We developed a narrative overview of key CMO configurations, and a diagrammatic representation of these.