Abstract
BACKGROUND: A theory-driven cognitive therapy (Feeling Safe) has produced much better outcomes for patients with persecutory delusions. There are four distinct response classes: very high delusion conviction with large improvement, very high delusion conviction with no response, high delusion conviction with large improvement, and high delusion conviction with modest improvement. Our objective was to apply principal trajectories analysis, a novel statistical method, to original trial data to estimate whether these groups may have responded differently to a different intervention: befriending. DESIGN: One hundred and thirty patients with persistent persecutory delusions were randomised to six months of Feeling Safe or befriending. Baseline assessments were used to assign patients allocated to befriending (who did not receive Feeling Safe) into the four Feeling Safe response classes. The treatment effect, including on potential mediators, was then estimated for these classes. RESULTS: Patients in two treatment response classes (Very high conviction/large improvement, High conviction/large improvement) benefited more from Feeling Safe, patients in one group (Very high conviction/no improvement) benefited more from befriending, and patients in the remaining group (High conviction/moderate improvement) benefited equally from the interventions. Mechanism differences were detected when Feeling Safe was superior to befriending, but not when befriending was superior. CONCLUSIONS: There may be patients with psychosis who benefit more from one type of therapy than another, likely due to different change mechanisms. The application of principal trajectories has generated testable hypotheses and a potential step toward personalised treatment. We recommend an investigation of whether sequential provision of the treatment types could enhance patient outcomes. Keywords: persecutory, delusions, outcome trajectories, psychosis, cognitive therapy.