Table 2 Optimising Pelvic Floor Exercises to Achieve Long-term benefits (OPAL)

Trial: The OPAL trial was a large multi-centre pragmatic randomised controlled trial of two active treatment arms delivered across 23 primary and secondary care sites for 600 women.

Intervention: OPAL aimed to determine the effectiveness of two active treatment arms: basic pelvic floor muscle training (PFMT) and basic pelvic floor muscle training (PFMT) with biofeedback mediated intensive for the treatment of stress or mixed (stress and urgency) female urinary incontinence.

Purpose: The OPAL trial has an embedded mixed methods process evaluation and a longitudinal qualitative case study, which aim to explain the trial outcomes. The longitudinal qualitative study aimed to investigate women’s experiences and adherence to the interventions.

Qualitative design: A two-tailed embedded multiple case study design utilising longitudinal interviews. Within a multiple case study design, Yin outlines a ‘two-tailed’ approach where cases are selected to represent two extremities in relation to phenomena; in this case, the extremities are the control and intervention groups [46]. Units of analysis were at the individual case (the participants) and the trial arms (intervention and control). There were two units of analysis to enable an in-depth exploration within each case but also at the trial arm level to explore commonalities and differences between the cases in each arm and between the arms.

Quantitative design: Prespecified analysis to explore fidelity, engagement and mediating factors using descriptive and interpretative statistics.

Sample: The two-tailed case study design means multiple cases (n = 20) from each trial arm were sampled to enable comparison between the trial arms. Cases were purposively sampled for variance in centre type (university hospital, district general hospital or community delivered service), therapist delivery type (physiotherapist/nurse), women’s type of UI (stress or mixed) and over time to reflect recruitment to the trial.

Data collection: Mirroring the trial data collection, the case study was longitudinal, with women interviewed four times (baseline, post-treatment, 12 months and 24 months post-randomisation). Twenty women per arm were recruited, and 24 women across both arms were interviewed 4 times. GF

Conceptual framework: Framework for process evaluation design [45].

Analysis: A case was built and summarised over 2 years, with four data points for each woman. Case summaries were written summarising women’s experiences. Theoretical propositions were developed to guide the analysis. All the cases for one trial arm were grouped and within arm consistencies/inconsistencies searched for. The experimental and comparator tails were compared to one another using the theoretical propositions. Thematic and theoretical saturation was achieved.

What did the case study process evaluation design reveal about context? Adherence to the interventions in the OPAL trial was hugely variable; in both trial arms, there were some women who had good adherence, some who were adherent at certain time points and some who did not adhere well at all. The case study was useful in illuminating the ways in which the context of the participants’ lives influenced adherence in both trial arms. The temporal nature of the data collection and case studies was useful in illustrating that the context of the women’s lives was dynamic, and thus, their engagement with the interventions was also dynamic. The in-depth case studies were useful in illustrating that based on each participants’ unique circumstances, different contextual factors and personal characteristics were important, such as their motivation to maintain engagement. Across-arm case comparison was able to illustrate that these factors were not related to the interventions but specific to the participants. Across-arm comparisons showed that although many participants had not maintained adherence, they felt more skilled in pelvic floor muscle training (PFMT) and able to restart PFMT exercise after a break [47, 48].