- Design
- Systematic review and random-effects meta-analysis of 23 RCTs
- Population
- 2,016 adults with schizophrenia-spectrum disorders and auditory hallucinations
- Primary outcome
- Severity of auditory hallucinations at end of treatment
- Effect
- SMD −0.18 (95% CI −0.31 to −0.05), NNT 16, low certainty; avatar therapy SMD −0.37 (−0.51 to −0.22)
A systematic review and meta-analysis pooled 23 randomised trials of psychological and psychosocial interventions aimed specifically at auditory hallucinations in 2,016 adults with schizophrenia-spectrum disorders. Published in JAMA Psychiatry on 1 October 2026.
Compared with control conditions, targeted interventions reduced hallucination severity at the end of treatment (SMD −0.18, 95% CI −0.31 to −0.05; number needed to treat 16), rated as low-certainty evidence. Benefits extended to the frequency and distress of voices, positive symptoms, depression and anxiety. Avatar therapy, in which the person talks to a computer-generated representation of their main voice, showed the largest effect (SMD −0.37, −0.51 to −0.22). Dropouts and reported adverse events were low.
The overall effect is small, and the five approaches differ considerably, so the pooled figure blurs a stronger result for some and a weaker one for others. Avatar therapy is not widely available, especially in India. The practical message is that persistent, distressing voices despite adequate antipsychotic treatment warrant a referral for a voice-focused psychological approach where one exists, alongside rather than instead of medication.
- Consider referral for voice-focused psychological therapy when distressing voices persist despite adequate antipsychotic treatment
- Keep antipsychotic treatment in place; these therapies were tested as additions
- Ask local services whether avatar therapy or CBT for psychosis is offered
- Measure voice severity and distress before referral to judge any benefit
Why it matters
It gives a measurable, if modest, option for patients whose voices persist after medication has done what it can.
Don't overread it
The overall effect is small and of low certainty, and the avatar result comes from a subgroup.
The statistics, in plain English
A standardised mean difference of −0.18 is a small effect, and a number needed to treat of 16 means about 16 people would need the therapy for one to benefit beyond control. The avatar subgroup's −0.37 is roughly double, but subgroup comparisons are less reliable than the main result.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for psychiatry, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free