Underlying mechanisms and efficacy of a suicide-focused psychological intervention for psychosis, the Cognitive Approaches to Combatting Suicidality (CARMS): a multicentre, assessor-masked, randomised controlled trial in the UK

Abstract

Background: There is a need for theoretically grounded and testable suicide-focused psychological therapies, especially in people with severe mental health problems, specifically non-affective psychosis. We aimed to test both the underlying mechanisms and efficacy of a suicide-focused therapy, cognitive behavioural suicide prevention for psychosis (CBSPp).

Methods: We did a multicentre, assessor-masked, randomised controlled trial conducted at four UK National Health Service (NHS) sites. Participants were eligible for enrolment if they were 18 years or older; met ICD-10 criteria for non-affective psychosis (F20-F29); had self-reported suicidal experiences in the 3 months before recruitment; were under the care of an NHS mental health services team; were not receiving a psychological therapy as part of a similar trial; had sufficient competency in the English language not to need an interpreter to participate; and were able to give informed consent. After screening for eligibility and completion of baseline assessments, participants were randomly allocated in a 1:1 ratio to either the treatment as usual group (control) or CBSPp therapy plus standard treatment group (treatment), with stratification by use of antidepressant medication and NHS site. Randomisation took place using an online Sealed Envelope randomisation procedure conducted and overseen by the Manchester Academic Health Sciences Centre Clinical Trials Unit (MCTU). This system was accessed only by specially allocated staff (eg, the CARMS Trial Manager or the MCTU Trial manager). When the system randomly allocated a participant, an unmasked email confirmation was sent to the unmasked coprincipal investigator and the CARMS Trial Manager (also unmasked). Assessors, the trial statistician, and one coprincipal investigator were masked to allocation group. Participants were informed of their randomised allocation group by an unmasked CARMS staff member. Experts-by-experience were involved in all stages of the research. For those in the treatment group, approximately 24 one-to-one therapy sessions were offered, usually weekly, for around 50 min per session. Assessments were conducted at baseline, month 6, and month 12. The primary outcome was suicide ideation severity measured by the 25-item Adult Suicide Ideation Questionnaire (ASIQ), assessed at month 6 relative to baseline. Outcome analyses used mixed models in the intention-to-treat population. Planned mediation indirect linear regression models examined appraisals of poor social support, emotional difficulties, interpersonal problem-solving difficulties, defeat, entrapment, and hopelessness as mediators at 6 months with allocation condition (treatment vs control) as the predictor variable and suicidal ideation severity (ASIQ) at 6 months as the outcome variable, whilst controlling for baseline levels of the mediator and outcome variables. The trial was registered before recruitment at ClinicalTrials.gov (NCT03114917) and ISRCTN (ISRCTN17776666) and is complete.

Findings: We recruited participants from four NHS sites from June 21, 2017, to Nov 25, 2020, with the final 12-month assessment completed on Jan 10, 2022. 479 participants were screened for eligibility, and of these 329 (69%) provided consent and were enrolled. After 69 participants were lost to follow-up, 292 participants were randomly allocated to the treatment and control groups (149 [51%] and 143 [49%], respectively). At baseline, 161 (55%) participants were male and 130 (45%) were female; one participant (<1%) in the treatment group was missing gender data. Mean age was 35·1 years (SD 13·2; range 18-69), and 247 (85%) were White or Caucasian. Severity of suicidal ideation was not statistically different between the treatment (n=136) and the control groups (n=116) at 6 months (p=0·07; Cohen's d -0·20 [95% CI -0·42 to 0·02]). A significant indirect mediation effect (-2·85, -7·00 to -0·23) showed that therapy strengthened social support appraisals which, in turn, reduced suicidal ideation severity at 6 months more in the treatment group than in the control group. Suicide attempts were the most frequent severe adverse event (132 attempts in 26 participants in the treatment group vs 91 attempts in 30 participants in the control group). There were four deaths in the CARMS study, none of which were by suicide or considered related to treatment.

Interpretation: CBSPp had no effect on suicidal ideation at 6 months, but it did have a significant indirect effect via improving appraisals of social support. It is important to explore how suicide-focused therapies can be developed and implemented in mental health services so that they maximally enhance appraisals of social support and social connection in ways that endure.

Funding: Efficacy and Mechanism Evaluation Programme, a Medical Research Council and National Institute for Health and Care Research partnership (reference number: 13/161/25).