Background: 
Treatment fidelity refers to the strategies used to enhance and monitor the reliability and validity of interventions including the replication of studies. Fidelity is rarely reported in Speech and Language Therapy intervention studies. Peer-befriending is intended to improve psychosocial wellbeing for people with stroke and aphasia. It involves people with experience of the condition (i.e. befrienders) providing social and emotional support to those who have recently had a stroke and aphasia. This paper will report on the fidelity of peer befriending for people with aphasia post-stroke as part of a feasibility trial (SUPERB) underway. Aims are: (1) to investigate the adherence to protocol of peer befriending visits, and training and supervision of peer befrienders; and (2) to explore the inter-rater and intra-rater reliability of checklists designed to examine fidelity of peer befriending.

Methods & Procedures: 
The current study is a feasibility single-blinded, mixed-methods, parallel group phase II RCT comparing peer-befriending (n=30) vs. usual care (n=30) for people with aphasia post-stroke. Ten befrienders with mild-moderate aphasia at least one-year post stroke are recruited from the community. Befrienders will attend 5-6 hours of training (across 2-3 days) and monthly group supervision sessions with other befrienders and a facilitator. The 30 participants randomised to the peer-befriending arm of the trial will be befriended by a trained befriender. Each participant will get 6-visits (over 3-months) soon after discharge from hospital. To measure fidelity, three checklists were created for the intervention visits, training and supervision, based on the Health Behaviour Change Competency Framework (HBCCF) and Kagan’s supported conversation measures.  All training and supervision sessions and one (of six) visits for each befriender-befriendee pair were videotaped and rated for adherence to protocol using the checklists. Adherence is evaluated by calculating a per cent fidelity score. Inter-rater and intra-rater reliability are calculated using Kappa statistics. 

Outcomes & Results: 
Preliminary results are positive with high treatment fidelity scores for 2 training workshops (93.8%), 10 group supervision sessions (92.9%-100%) and 8 intervention visits (87.5%-100%). Inter- and intra-rater reliability could not be calculated on the training sessions as there was no variability in the data. Inter-rater reliability for the supervision sessions was excellent (k = 0.76-1.0) and fair-to-excellent for the intervention visits (k = 0.45-0.77). Intra-rater reliability was good-to-excellent for the supervision sessions (k = 0.61-1.0) and intervention visits (k = 0.60 – 1.00). 

Conclusions: 
Early fidelity results demonstrate that training and supervision of befrienders is being delivered as intended in the SUPERB trial. Variation in the reliability of the checklists to detect the presence (or absence) of behaviours suggest that further training and/or refinement of the checklists may be warranted. Calculating fidelity is an important element of intervention research to improve the validity of the study and future replication.

SUpporting wellbeing through PEeR-Befriending (SUPERB) trial: An exploration of fidelity in peer-befriending for people with aphasia