Disclaimer
What you may do with what is published here, what you may not, and where the journal stops.
Last updated: 13 September 2026
Wherever you are, dial your local emergency number. If you, or someone near you, are in immediate danger, call. The numbers above are the European emergency number and two French lines: 3114 is the French national suicide prevention line, free, open 24 hours a day, seven days a week, from France.
This site answers no one. There is no duty rota, no monitored inbox, and nobody behind the page at the moment you are reading it.
Psychiatry Evidence Base reads the psychiatric literature, appraises it and publishes that appraisal. That is all this site does, and it is better said at the outset: nothing published here treats, diagnoses, or decides in anyone’s place.
01What this journal does
PEB is a digital scientific journal. It selects publications, appraises them against a critical reading grid, gives them a double score, and draws from them content intended for health professionals and for informed readers. The method is set out in plain terms on the method pages.
A PEB analysis is about a study, not about a patient. It says what that work shows, how solidly, and what it does not show. Moving from that reading to a course of action belongs to the clinician who has the case in front of them, and to no one else.
No content on this site replaces medical advice, a clinical examination, a guideline in force, or the judgement of the professional who decides.
02Reading this site creates no clinical relationship
Opening a page, using a scale, following a decision tree or writing to the contact address does not open any form of care. There is no consultation here, no teleconsultation, no personalised advice, no second opinion, no referral, no prescription.
No clinical question about an identified person will receive an answer, through any channel: email, comment, private message on a social network. This is not a matter of availability, it is a matter of medical ethics. Medical advice is given after an examination, not in writing, at a distance, on the strength of a summary.
03What is shown, what is suggested, what is opinion
The three are not read the same way, and PEB undertakes to tell them apart explicitly in every analysis.
Concordant controlled trials, good-quality meta-analyses, an effect size known with its confidence interval. The finding holds outside the setting that produced it.
A single result, a small sample, a surrogate outcome, observational data, a subgroup analysis. Interesting, not enough to change a practice.
A reasoned extrapolation, a position of the journal, a clinician’s view. Written as such, never presented as a result.
When a source does not report a value, PEB writes that it does not report it. No confidence interval reconstructed from a standard error, no I² inferred, no numerical threshold added to make a sentence more concrete. An empty cell is information, and it is published as such.
04Content ages, and it carries its date
Every page states when it was published. An analysis that is right in 2026 can be wrong in 2029, because a larger trial has appeared, because a pharmacovigilance alert has been issued, or because the guideline it discussed has been replaced.
Nothing here is updated automatically. Before relying on a page, check its date and the status of the guideline in force. When an analysis is corrected, the correction is visible on the page, with its date. Errors are not quietly erased.
05Medicines, doses, monitoring
The molecules, doses, titration schedules and monitoring tests cited in the analyses come from the sources studied. They describe what a study did or what a guideline recommends. They are not a prescription.
- Check the product informationThe summary of product characteristics and the marketing authorisation in your country of practice are what count, not an analysis published here.
- Off-label use is flaggedWhen a source deals with use outside the marketing authorisation, PEB says so. Responsibility for the prescription remains entire and personal.
- Interactions are not exhaustiveAn analysis discusses the interactions the study documented. It does not stand in for an interaction check.
06Scales, decision trees, simulators
The site hosts interactive tools. They are aids to reasoning and to teaching. None is a medical device, and none has been validated as clinical decision support in the regulatory sense.
They serve for detection and follow-up. A score is not a diagnosis, and a scale administered without an interview is worth nothing. The thresholds cited are those of the validation studies, with the populations in which they were established.
A tree renders the logic of one guideline, never a synthesis of several. It shows what the guideline says and what it leaves open. It knows neither your patient, nor their comorbidities, nor what they have already received.
They illustrate a line of reasoning, a kinetic profile, a monitoring schedule. The values displayed are for teaching. They do not size a course of action for a real person.
The scales, trees and simulators run entirely in your browser. Nothing is sent to the server, nothing is recorded, nothing is kept once the tab is closed. If you need a record, make it yourself, in the patient’s file, with the safeguards that apply to health data. The detail is in the privacy policy.
07Guideline appraisals and the AGREE II score
PEB appraises institutional guidelines with the AGREE II instrument, domain by domain, item by item, on the full text. The score is about the development method of the guideline: its rigour, its transparency, stakeholder involvement, the handling of declared interests.
A low score does not mean the guideline is wrong on substance. It means it documents poorly how it got there. The distinction has practical consequences: a guideline that is binding in your country of practice remains so, whatever methodological judgement is passed here.
08Social media, video, short formats
A post, or a thirty-second format, necessarily says less than a full analysis. That is the nature of the format, not a flaw, but it changes what the content can carry.
A reasoned reading, published for professionals. It points back to the full analysis, which remains the reference.
A visual way in. A carousel draws attention to a question, it does not settle it.
Teaching content based on the literature available on the day of filming. A video does not update itself, the page on the site does.
The shortest format, therefore the most reductive. No clinical decision should rest on it, and it says so itself.
In comments as in private messages, PEB gives no view on an individual situation and never comments on a treatment in progress. Comments posted by third parties bind only their authors.
09What software tools do, and what they do not
Literature scanning, formatting and technical proofreading rely on software tools, including language models. Better to say it than to leave the question hanging.
- What stays humanThe selection of studies, the critical appraisal, the score, the clinical interpretation, the decision to publish, and the signature.
- What is checked at the sourceEvery reference, before publication: DOI, PMID, full text where it is accessible.
- What is never delegatedA figure published by PEB comes from the source, word for word, or it is not published. No tool is allowed to fill an empty cell.
10Independence, funding, declared interests
The journal is written and published by one person, with no paid editorial board, no funding from the pharmaceutical industry, no advertising and no sponsored content. No company reviews or approves what is published.
Declared interests, where there are any, are stated on the page concerned. The detail of publishing and funding is on the legal notice page, and the author’s background on his professional page.
11Links to other sites
PEB links out to PubMed, to journals, to health agencies and to learned societies. That content is published by third parties, it changes, it can disappear or shift in meaning. A link is there so you can check for yourself, it is not an endorsement.
12Reporting an error
A wrong figure, a misattributed reference, an out-of-date guideline, a typo that changes the meaning: all of it can be reported, and corrected. The procedure, what to put in the message and the response times are on the contact page.
A well-founded report leads to a dated correction on the page, and to a note of what was corrected. It is the only quality control mechanism that does not depend on the author, so it is taken seriously.
13If you are not a health professional
The site is written for clinicians, and the vocabulary shows it. You may read it, no one is stopping you, and understanding how evidence is built is useful. But read it knowing two things.
- An analysis is not a diagnosisRecognising your symptoms in an article says nothing about what you have. Clinical pictures resemble one another, and that is most of what makes the work difficult.
- Never change a treatment on the strength of a page on this siteNot the dose, not the schedule, not stopping it. Some psychotropic drugs expose patients to withdrawal or to relapse when stopped abruptly. Talk to the doctor who follows you.
14What falls to you
The clinical decision belongs to the professional who takes it, within the legal and ethical framework of their country of practice. PEB cannot be held liable for a decision based on its content, for a partial reading of an analysis, for use of an interactive tool outside what it is intended for, or for third-party content reached through an outbound link.
This page supplements the legal notice, which describes the publisher, the hosting and the rights in the content. Where the two appear to conflict, the legal notice prevails.
For any question about this page: write to the editors. The frequently asked questions already answer a good many of them.
Analysis from Psychiatry Evidence Base, evidence-based psychiatry, explained with rigour.