Psychosis treatment usually combines medication, a coordinated early-care team and, for some people, talking therapy, and the evidence is strongest for the first two. A 2018 meta-analysis of 10 randomised trials found that early intervention services did better than usual care on every outcome that could be pooled, [correll-2018] while large reviews of antipsychotic drugs find real benefit alongside marked differences in side-effects. [leucht-2013] This guide covers what each review found, where they disagree, what they could not show and what to ask when treatment is being planned. Our overview of psychosis covers what the experience involves and why early assessment matters.
Do antipsychotics work, and is one better than the rest?
On average they work, and the differences between them are smaller than the differences in what they do to the body. A 2013 meta-analysis pooled 212 studies and 43,049 participants to compare 15 drugs. It found small but robust differences in efficacy, and that antipsychotics differed substantially in side-effects. The authors argued that the findings challenge the straightforward split into first-generation and second-generation drugs. [leucht-2013]
A 2019 network meta-analysis extended this to 32 drugs across 402 studies and 53,463 participants with acute symptoms. All of the drugs reduced overall symptoms more than placebo, although the result was not statistically significant for six of them. The standardised mean difference, a measure of effect size, ranged from -0.89 for clozapine to -0.03 for levomepromazine. [huhn-2019]
The authors’ summary is the useful part: there are some efficacy differences between antipsychotics, but most are gradual rather than discrete, while differences in side-effects are more marked. Compared with placebo, the average weight gain ranged from a loss of 0.16 kg (ziprasidone) to a gain of 3.21 kg (zotepine). [huhn-2019] That is why the choice of drug is often a conversation about which side-effects a person can live with.
Do early intervention services help?
Yes, in the early phase of psychosis, according to the best pooled evidence we could verify. Early intervention services are multidisciplinary teams that deliver several psychosocial and medication treatments together from one team, instead of sending a person to separate providers. A 2018 meta-analysis of 10 randomised trials with 2,176 patients (mean age 27.5 years, 62.3% male) found they were associated with better outcomes than treatment as usual on all 13 outcomes that could be pooled. [correll-2018]
Two of those outcomes were the main ones. All-cause treatment discontinuation was lower with early intervention services (risk ratio 0.70, 95% confidence interval 0.61 to 0.80), and so was at least one psychiatric hospitalisation (risk ratio 0.74, 0.61 to 0.90). Involvement in school or work was higher (risk ratio 1.13). [correll-2018] A risk ratio below 1 on a bad outcome means fewer of them, so 0.70 is about 30% fewer dropouts and 0.74 about 26% fewer hospitalisations.
Standardised mean differences for early intervention services versus treatment as usual at the end of treatment, Correll et al. (2018), 10 randomised trials, 2,176 patients. The reported values are negative (-0.32, -0.22, -0.28); bars show their size, and a larger bar means milder symptoms with early intervention.
The advantage was evident at 6, 9 to 12, and 18 to 24 months of treatment, except for general symptom severity and depressive symptom severity at 18 to 24 months. [correll-2018] The authors concluded that the results support funding and using these services for people with early-phase psychosis.
Does talking therapy help with psychosis?
It may, though the evidence is thinner and more mixed than for the other two. The review we could verify looked at one setting only: people with psychosis admitted to psychiatric hospital. It found 23 reports of 18 randomised trials of treatments such as cognitive behavioural therapy for psychosis (CBTp), Acceptance and Commitment Therapy and Metacognitive Therapy. [wood-2020]
Cognitive behavioural therapy-informed treatments were found not to be effective in reducing positive symptoms, the main outcome, after therapy or at follow-up. When one study was removed, a positive effect appeared at both time points. They did show a significant favourable effect on negative symptoms (after therapy), on total symptoms and functioning (after therapy and at follow-up) and on readmission (at follow-up). [wood-2020]
In the authors’ words the findings are equivocal: the treatments have an effect on some symptom measures but not others, and inpatient-adapted versions need further study. The lead author of the review was also a lead author on one of the included trials, which the paper declares. [wood-2020]
| Treatment | What the review found | Certainty |
|---|---|---|
| Antipsychotic drugs (32 drugs, 402 studies) | All reduced overall symptoms more than placebo, to very different degrees | Often low or very low |
| Early intervention services (10 trials) | Better than usual care on all 13 pooled outcomes | Trials of 9 to 24 months |
| Cognitive behavioural therapy-informed treatment (18 trials, inpatients) | Mixed: no clear effect on positive symptoms, benefit on negative and total symptoms | Authors call it equivocal |
What can this research not show?
It cannot show which treatment will suit one person, or what happens over many years. Five limits matter most:
- The antipsychotic evidence is rated low. The 2019 authors reported that confidence in the evidence was often low or very low, even where the numbers look clear. [huhn-2019]
- Some people were left out. That analysis excluded people on a first episode, people with treatment-resistant illness and people with mainly negative or depressive symptoms, so it does not describe them. [huhn-2019]
- The earlier analysis is older. The 2013 comparison included studies reported between October 1955 and September 2012, and many trial reports did not give full details of how people were randomised. [leucht-2013]
- Early intervention services are a package. The trials tested multimodal programmes, so the pooled result cannot say which part (medication, family support, work help) did the work, and the trials ran for 9 to 24 months at most. [correll-2018]
- The therapy review covers inpatients only. Its finding says nothing firm about talking therapy for people living at home, and the review’s own authors call the results equivocal. [wood-2020]
What should you ask when treatment is being planned?
Ask what the plan is for medication, for support beyond medication and for what happens if the first drug does not suit. This section is our suggestion and is not something the reviews tested.
Useful questions are which side-effects to expect and how they will be monitored, whether an early intervention service exists locally and whether you or your relative can be referred, and what help with work or study, family involvement and talking therapy is available. Do not stop or change a prescribed antipsychotic without speaking to the prescriber first. If you are supporting someone, our guide to early signs of psychosis explains why assessment matters, and our article on hearing voices covers one common experience in detail.
No screener on this site covers psychosis. The self-assessment hub lists the seven we do have, and none is built for this question, which also matters because reduced insight is part of the picture.
When to seek help
Speak to a doctor promptly if you or someone close to you is hearing or seeing things others do not, holds beliefs that are frightening or fixed, or is withdrawing and not making sense to others. A first episode should be assessed and medically investigated. If anyone is at immediate risk or feels unable to stay safe, contact your local emergency services or a crisis helpline.
How MyFreud can help
Between appointments, a record helps a clinician see what is changing. Daily mood tracking in MyFreud shows how your days run over weeks, and the notepad is somewhere to note sleep, side-effects and anything you want to raise. It does not replace a doctor or a treatment team.
Download MyFreud and start today: App Store or Google Play.