Implementing a ward-based programme to improve care for older inpatients: process evaluation of the cluster randomised CHERISH trial

Background: Older inpatients are at high risk of hospital-associated complications, particularly delirium and functional decline. These can be mitigated by consistent attention to age-friendly care practices such as early mobility, adequate nutrition and hydration, and meaningful cognitive and social activities. Eat Walk Engage is a ward-based improvement programme theoretically informed by the i-PARIHS framework which significantly reduced delirium in a four-hospital cluster trial. The objective of this process evaluation was to understand how Eat Walk Engage worked across trial sites. Methods: Prospective multi-method implementation evaluation on medical and surgical wards in four hospitals implementing Eat Walk Engage January 2016-May 2017. Using UK Medical Research Council guidance, this process evaluation assessed context, implementation (core components, implementation strategies and improvements) and mechanisms of impact (practice changes measured through older person interviews, structured mealtime observations and activity mapping) at each site. Results: The four wards had varied contextual barriers which altered dynamically with time. One ward with complex outer organisational barriers showed poorer implementation and fewer practice changes. Two experienced facilitators supported four novice site facilitators through interactive training and structured reflection as well as data management, networking and organisational influence. Novice site facilitators used many implementation strategies to facilitate 45 discrete improvements at individual, team and system level. Patient interviews (42 before and 38 after implementation) showed better communication about program goals in three sites. Observations of 283 meals before and 297 after implementation showed improvements in mealtime positioning and assistance in all sites. Activity mapping in 85 patients before and 111 patients after implementation showed improvements in cognitive and social engagement in three sites, but inconsistent changes in mobility. The improvements in mealtime care and cognitive and social engagement are plausible mediators of reduced delirium observed in the trial. The lack of consistent mobility improvements may explain why the trial did not show reduction in functional decline. Conclusions: A multi-level enabling facilitation approach supported adaptive implementation to varied contexts to support mechanisms of impact which partly achieved the programme goals. Contexts changed over time, suggesting the need for adequate time and continued facilitation to embed, enhance and sustain age-friendly practices on acute care wards and optimise outcomes. Trial registration: The CHERISH trial was prospectively registered with the ANZCTR (http://www.anzctr.org.au): ACTRN12615000879561.