Developing models of how cognitive improvements change functioning: Mediation, moderation and moderated mediation

BACKGROUND Cognitive remediation (CRT) affects functioning but the extent and type of cognitive improvements necessary are unknown. AIM To develop and test models of how cognitive improvement transfers to work behaviour using the data from a current service. METHOD Participants (N49) with a support worker and a paid or voluntary job were offered CRT in a Phase 2 single group design with three assessments: baseline, post therapy and follow-up. Working memory, cognitive flexibility, planning and work outcomes were assessed. RESULTS Three models were tested (mediation - cognitive improvements drive functioning improvement; moderation - post treatment cognitive level affects the impact of CRT on functioning; moderated mediation - cognition drives functioning improvements only after a certain level is achieved). There was evidence of mediation (planning improvement associated with improved work quality). There was no evidence that cognitive flexibility (total Wisconsin Card Sorting Test errors) and working memory (Wechsler Adult Intelligence Scale III digit span) mediated work functioning despite significant effects. There was some evidence of moderated mediation for planning improvement if participants had poorer memory and/or made fewer WCST errors. The total CRT effect on work quality was d=0.55, but the indirect (planning-mediated CRT effect) was d=0.082 CONCLUSION Planning improvements led to better work quality but only accounted for a small proportion of the total effect on work outcome. Other specific and non-specific effects of CRT and the work programme are likely to account for some of the remaining effect. This is the first time complex models have been tested and future Phase 3 studies need to further test mediation and moderated mediation models.