How can the Awareness and Beliefs About Cancer (ABACus) randomised controlled trial process evaluation help us to understand trial outcomes and associated implications? And, what can it teach us about community interventions and trials in the future?
Problem
Cancer outcomes are poor in the UK’s socioeconomically deprived communities, with low symptom awareness and fatalistic beliefs contributing to delayed help-seeking and advanced stage disease. The ABACus 3 trial tested the effectiveness of a theory-grounded lay advisor-facilitated cancer awareness intervention for adults (aged 40+) living in deprived communities in healthcare and community settings. The intervention entails completion of a touchscreen questionnaire (cancer symptoms, screening, risk factors), with personalised behavioural advice delivered by a trained lay advisor through specified behaviour change techniques (BCTs). In parallel to the trial, the process evaluation was conducted to assess intervention fidelity, dose, contamination and reach and to identify key mechanisms of change and contextual influences at the social and environmental level.
Approach
Semi-structured interviews were conducted with lay advisors post-training and post-intervention delivery to explore resource use, engagement, barriers and enablers, delivery of BCTs and perceived social and contextual influences. Purposefully sampled semi-structured participant interviews were conducted 2-4 weeks and 6 months post-randomisation to explore influences on knowledge, beliefs and behaviours, contamination, reach, actual and perceived barriers to behaviour change and contextual factors (10% across time-points and arms). Interviews were transcribed verbatim and analysed thematically supported by NVivo, 20% independently dual-coded. A purposive sample of intervention delivery sessions (20%; based on lay advisor, setting type, participant age and gender) were audio-recorded with 50% of these additionally observed. Audio-recordings and observations were used to quantitatively assess BCT delivery against a pre-defined coding matrix for personalised results to assess fidelity and dose using content analysis.
Findings
Six lay advisor (n=3) and 37 participant interviews were conducted (n=15 at 2-4 weeks, n=22 at 6 months). Twelve intervention delivery sessions were audio-recorded (7 observed) with an average of 68% of the theoretical maximum BCTs being delivered across the three lay advisors (range 23.7% - 94.9%). Personalisation was recognised as key to successful engagement; however, advisors were concerned that tailoring delivery to the participant’s interest may have diluted symptom awareness messages, highlighted in the range of BCTs delivered. Participants reported high symptom knowledge (across both arms) associated with prior experience and awareness of mass-media cancer awareness campaigns. Contamination occurred between trial arms, especially within community settings where participants discussed taking part and their results within their social networks. Fidelity and dose of the intervention were high even though the content of discussions, including providing additional information and signposting to resources, differed by session and advisor.
Consequences
High self-reported knowledge is consistent with baseline ceiling effects in the main trial primary outcome. Whilst personalisation increases participant engagement, it may dilute core intervention messages. Findings highlight critical learning about effective methods of engaging deprived populations in research and reinforce the importance of conducting robust process evaluations of cancer awareness interventions due to complexities in evaluation.