Exploring mediators of the recovery process over time among mental health service users, using a mixed model regression analysis based on cluster RCT data

Personal recovery is associated with many significant health-related factors, but studies exploring associations between activity factors and personal recovery among service users are scarce. The aims of this study were hence to; 1) investigate if various aspects of activity may mediate change in recovery while also acknowledging clinical, sociodemographic and well-being factors; 2) explore the effects of two activity-based interventions, Balancing Everyday Life (BEL) or standard occupational therapy (SOT), on personal recovery among service users.

Two-hundred-and-twenty-six service users were included in a cluster RCT, 133 from BEL units and 93 from SOT units. Participants commonly had a diagnosis of mood disorder and the mean age was 40. Instruments used targeted activity, mastery and functioning. A mixed-model regression analysis was employed.

The model tested was whether selected variables could be used to mediate the change in recovery from the start to a six-month follow-up after intervention. Participants’ personal recovery increased after treatment and increased further at the follow-up. The general level of recovery was negatively related to a diagnosis of depression/anxiety, both before and after treatment, but depressed/anxious service users still increased their recovery. There were no significant relations between recovery and sex or age. The interactions between change in recovery and changes in depression/anxiety, satisfaction with activities, sex, and age were all non-significant. All possible treatment mediators included were related to change in recovery, the strongest being occupational engagement and mastery, followed by activity satisfaction and symptoms. Mediation was shown by the decrease in the effect of the time factor (from intervention start to completion) when the covariates were introduced. In all cases the time variable was still significant. When testing a model with all variables simultaneously as covariates, occupational engagement and mastery were strongly significant. There was no difference between interventions regarding recovery improvement.

The treatments were equally beneficial and were effective regardless of gender, age and diagnosis. Those who gained most from the treatment also gained in feelings of mastery and activity engagement. Activity engagement also moderated the level of recovery. To enhance recovery, interventions should facilitate meaningful activities and gaining control in life.

The study was registered with ClinicalTrials.​gov. Reg. No. NCT02619318. Retrospectively registered: December 2, 2015.