Process evaluation embedded within a randomised trial of caregiver training after stroke

Observations of day-to-day stroke unit practice, multidisciplinary team and family meetings, together with instances of formal and informal caregiver training were undertaken in 10 of 36 TRACS trial centres (6 intervention and 4 control units). Documentary analysis supplemented the observational data. Semi-structured interviews were conducted with patients and caregivers three months after discharge (n = 37 pairs). Interviews with patients and caregivers explored their in-hospital experiences and their perceptions of receiving information and formal or informal preparation for caring for the patient at home following discharge.

Multidisciplinary team members from eight of the 10 stroke units (n = 53), participated in semi-structured interviews soon after recruitment to the TRACS cluster randomised controlled trial closed. These interviews explored team members’ experiences and perceptions of providing information about stroke and their participation in formal or informal caregiver training in hospital.

The TRACS cluster randomised controlled trial demonstrated no clinical or statistical differences between groups on the primary outcomes of functional independence (patients), or caregiver burden. Similarly, for the range of secondary outcomes measured at 6 and

12 months, no clinical or statistical differences were evident. Thus, the outcomes seen in the early single-centre trial were not replicated in TRACS. The process evaluation findings are important in contributing to an understanding of the possible reasons for the TRACS study outcomes.

The process evaluation found that contextual factors including the organisational history and team relationships within stroke units, external policy drivers, and service development initiatives, impinged on implementation of the caregiver training programme (the LSCTC)  in unintended ways that could not have been predicted through focus on how staff worked individually and collectively with caregivers and with patients at unit level. Factors that facilitated or impeded the effectiveness of the cascade training model used, whether and how stroke unit teams made sense of and engaged individually and collectively with a complex caregiver training intervention, and what impact these factors had on embedding the intervention in routine stroke unit practice were identified.

We concluded that, where implementation of complex interventions depends on multiple providers (in this case a wide range of stroke unit team members), time needs to be invested in reaching agreement on who will take responsibility for delivery of specific components and in determining how implementation and its effectiveness will be monitored. This goes beyond concern with intervention fidelity; explicit consideration also needs to be given to the implementation process in terms of how programme change can be effected at organisational, practice, and service delivery levels. In this study, Normalization Process

Theory’s constructs proved helpful in identifying vulnerable features of implementation processes in respect of the work involved in embedding complex interventions such as the LSCTC.

The process evaluation has provided insight into factors which influenced the design, implementation and delivery of the TRACS LSCTC training programme. Any future decision to develop and implement a TRACS-LSCTC like intervention on a national basis needs to take account of our findings, which go beyond reporting on the fidelity of intervention implementation. Incorporating patient and caregiver experiences and perceptions ensured that the research moved beyond reporting on objective measures of the effectiveness of the intervention.