The contribution of a process evaluation to explain the results of a cluster randomized controlled trial for people with depression in long-term physical conditions.

Talk Code: 

The problem

The COINCIDE trial showed that collaborative and integrated care leads to modest reductions in depression and improvements in self-management for those with depression and diabetes and heart disease (CHD). However, the treatment effects reported in COINCIDE did not extend to improvements in quality of life and functional outcomes, and adherence to the treatment model was limited - less than half of trial patients received the allotted number of therapy sessions and a third either withdrew after referral or did not attend any sessions. Questions therefore remain about how to improve the delivery of integrated care for people with multimorbidity.

The approach

Process evaluation as part of a cluster randomized controlled trial of collaborative care versus usual care for people with depression and diabetes and CHD. Within trial quantitative process data were collected at 4-months follow-up (patient assessment of chronic illness; PACIC), and in-depth interviews were conducted with patients (n=31) and health professionals (n=30) to identify mechanisms associated with delivery of the intervention and outcomes. Quantitative process data were analysed blind to treatment allocation and interview data was analysed before the result for the primary outcome was known.


387 patients were recruited from 36 general practices in the north west of England. 19 practices were randomized to collaborative care, and 20 practices to usual care. Patients in the collaborative care arm (n=191) were significantly less depressed than patients in the usual care arm (n=191); the treatment effect for reductions in depression was equivalent to a standardised mean difference of 0.3, which was less than the pre-specified effect of 0.4. Patients in the collaborative care arm reported being better self-managers, but there were no significant between group differences for self-efficacy, quality of life, functional outcomes, and social support. The mean total for the PACIC was 0.39 points higher (95% CI 0.16 to 0.62) in the collaborative care group, suggesting that positive treatment effects associated with COINCIDE were linked to improvements in patient centredness. However, this did not necessarily relate to integration of physical and mental health care as the qualitative data suggested that patients preferred a protected ‘space' away from routine primary care consultations to discuss mental health problems, and opportunities to focus on managing mood problems that were not explicitly linked to their physical health.


The COINCIDE model led to mental health gains and was regarded as patient centred. However, patients preferred to separate their mental health care from concerns about their physical health, and professionals maintained barriers around their expertise, also reducing opportunities for integration. Treatment effects may have been enhanced through greater therapeutic integration, but collaborative care needs to be sufficiently flexible to respond to the complex health needs of people with multimorbidity


  • Peter Coventry, University of Keele, Keele, UK
  • Sarah Knowles, University of Keele, Keele, UK
  • Isabel Adeyemi, University of Keele, Keele, UK
  • Nia Coupe, University of Keele, Keele, UK
  • Carolyn Chew-Graham, NIHR CLAHRC Greater Manchester, Manchester, UK