topics

98 long-form pages. each one takes a question people actually ask about psychiatric medication, diagnosis or the health system to the primary record: trials, labels, regulators, courts and federal files, both directions, with the sources listed and dated.

does it work?

one question per page, answered from trials, labels and regulators, both directions, including where the honest answer is that the evidence is thin.

do any supplements actually work? — seventeen substances people take instead of, alongside or without a prescription: regulatory status, the best trial for each, and the risks.
does ashwagandha work for anxiety? — the best independent estimate is between almost nothing and about 10 points on a clinical anxiety scale, and the headline meta-analysis grades its own evidence low.
does CBD help anxiety? — the signal exists only at single 300 to 600 mg pharmaceutical doses in lab challenge tests; the largest 2026 meta-analysis found no significant effect.
is kratom dangerous? — no controlled trial supports kratom for any psychiatric use, opioid-type dependence and withdrawal are documented, and the US legal status is a patchwork.
St John’s wort: the interaction list — it works for mild to moderate depression, which is exactly why the interactions matter, including with hormonal birth control.
does ERP work for OCD? — exposure and response prevention cut Y-BOCS by 14.5 points against placebo, against 4.7 for clomipramine and 3.5 for SSRIs, in a 54-trial network.
does changing what you eat treat depression? — SMILES, AMMEND and HELFIMED beside the 16-trial meta-analysis (g 0.16), PREDIMED’s null primary result and the 1,025-person MooDFOOD null.
when an antidepressant stops working — tachyphylaxis in 9 to 57 percent across trials, and each prior course cut the odds of responding to the next.
antidepressants and emotional blunting — 46 percent of 669 treated patients reported blunting, no difference by drug, and it tracks the depression score yet exceeds recovered controls.
staying on your antidepressant: what the column left out — a Men’s Health column checked claim by claim: the trial that backs staying on, the two withdrawal reviews that disagree, the guideline, and the author’s payment record.
do genetic tests for psychiatric medication work? — GeneSight, Genomind and the rest: five randomised trials by their primary outcome, what the guideline bodies and the FDA say, and the one genetic test everyone backs.
does psychiatric medication keep you out of hospital, or put you there? — readmission rates, the register studies, the dose-reduction trial, the measured risks of starting an antidepressant, and what nobody has studied.
the overdose the prescription was part of — benzodiazepine plus opioid death records, the 2016 boxed warning, polypharmacy risk, and the two findings that cut the other way.
two numbers nobody measured — two statistics about psychiatric medication and violence that point in opposite directions, traced to their sources.
does tms work for depression — what the trials actually show — what the largest comparison of brain-stimulation trials found for tms, why the clinic success rate is not the trial success rate, what the stanford accelerated protocol has and has not shown, and who tms has not been shown to help.
does ect work for depression — what the trials actually show — the highest odds of response of any brain stimulation in the 113-trial network, and the oldest sham trials in the field. what the memory evidence measures and what it does not, the relapse figure nobody quotes, and what nice and the fda actually say.
does ketamine work for depression — what the trials actually show — the pooled odds at 24 hours and how certain they are, racemic versus esketamine, how long it lasts, the licensing trial that failed, the critics in their own words, and three ketamine-versus-ect studies that disagree.
does trazodone work for sleep — what the evidence actually shows — approved for depression, prescribed for sleep: seven placebo trials pooled, the cochrane review, the sleep guideline that suggests against it, and the label's own drowsiness numbers.
does hydroxyzine work for anxiety — what the evidence actually shows — an antihistamine approved for anxiety with no data past four months: five trials pooled by cochrane, the two largest trials and their numbers, the 2019 network of 22 drugs, and the heart-rhythm warning with its dose limits.
is it ok to use chatgpt as a therapist? what the trial, the tests and the new laws actually say — the one randomised trial of a purpose-built chatbot, the two pooled estimates that disagree, the stanford and rand tests of what the general models do with risk, what fda, the apa and three state legislatures did in 2025, and where talking to people fits.
is depression a chemical imbalance? what the 2022 review, the replies, the scans and the labels actually say — where the serotonin idea came from, what the 2022 umbrella review found in its own words, the 2023 replies in the same journal, the pet and depletion studies, why the drugs can work without the theory, what the explanation does to people who hear it, and what the labels, nimh and the apa say today.
mental health help without going to the hospital: what exists, what the studies found, and when the hospital is the right call — what 988 does and when it calls 911, mobile crisis teams and the two studies that tested them, crisis stabilization units and the medicaid data on who uses them, the houston services by name, and when the hospital is the right call.

the medication record

what the approvals rested on, how long the trials ran, what each drug is approved for, and what changed on the label.

what each medication is approved for — trazodone for depression, not insomnia; lamotrigine for bipolar maintenance, not acute episodes; the approved and off-label uses of 21 psychiatric drugs.
the longest trial behind each approval — one number per drug, the longest pivotal trial behind its FDA approval, beside how many Americans take it.
how long psychiatric drugs were tested — of 72 randomised trials in the library, 57 state how long they ran, and the horizons are counted here.
how the drug was found — the discovery story behind every medication on the site, graded documented, legend or no source: the accidents that started each class and the designed drugs that followed.
what changed — a dated record of FDA label revisions to the medications we cover, checked nightly against the FDA’s own files.
ketamine vs esketamine: FDA status — esketamine (Spravato) is approved for two depression indications under a REMS; racemic ketamine is approved as an anaesthetic, and every psychiatric use of it is off-label.
psychedelic therapy: the FDA status table — where every psychedelic drug stands with the FDA, from the one approval to the applications, the designations and the rejections.
AI mental health tools: FDA status — the FDA lists more than 1,500 AI-enabled devices and none is generative AI for mental health; what is regulated, what is not, and the advisory committee record.
the 1991 hearing and the thirteen-year warning — what the FDA committee had in hand in 1991, why the 2004 boxed warning came from trials that did not exist yet, and the 2007 age gradient.
lithium and the promoted brands — the cheap salt with the best hospital record, its fall from 30.4 to 17.6 percent of bipolar visits, and the payment records on the brands that replaced it.
antipsychotics in nursing homes — the 2005 warning, 88 percent of 2007 claims for dementia, three dementia-promotion settlements, the CMS measure that cut use to 14.1 percent, and the diagnoses that hid residents from it.

legal and safety records

the litigation and regulator record per drug: settled is never adjudicated, verdicts carry their reductions, and verified absences state their window.

aripiprazole (Abilify) — the 2016 FDA warning on compulsive gambling, eating, shopping and sexual behaviour, and the state settlements that followed.
buprenorphine (Suboxone) — the largest opioid-drug resolution the DOJ has obtained: a subsidiary’s felony plea, a former CEO’s misdemeanour, and what the label says.
bupropion (Wellbutrin, Zyban) — a guilty plea, a generic the FDA withdrew, and a boxed warning that was removed.
citalopram and escitalopram (Celexa, Lexapro) — one page for two molecules because it was one case: Forest’s 2010 guilty plea and $313 million resolution, count by count.
duloxetine (Cymbalta) — the withdrawal litigation the plaintiffs lost: centralisation denied twice, class certification denied, and the label’s own discontinuation numbers.
fluoxetine (Prozac) — the 1994 verdict that turned out to be a settlement, and the rest of the record from the courts.
gabapentin (Neurontin) — the most prescribed drug in the library has no psychiatric approval; the off-label marketing campaign and the 2004 guilty plea behind its rise.
lamotrigine (Lamictal) — the boxed warning for serious rash quoted at the label’s own incidence figures, and the dispensing errors the FDA warned about.
lisdexamfetamine (Vyvanse) — Shire’s $56.5 million False Claims Act settlement over marketing it as less abusable, and the schedule II record.
modafinil (Provigil) — Cephalon’s 2008 guilty plea and $425 million over off-label promotion, and the pay-for-delay case that reached the Supreme Court.
olanzapine (Zyprexa) — the two Zyprexa litigations kept in separate columns: the 2009 DOJ resolution and the product-liability settlements.
paroxetine (Paxil) — Study 329, published 2001 and reanalysed 2015, and GSK’s $3 billion plea, told in dates.
quetiapine (Seroquel) — AstraZeneca’s $520 million civil settlement in 2010, with no criminal count, and what the label added since.
risperidone (Risperdal) — the $8 billion verdict and its reduction to $6.8 million ninety days later, sourced to the filing itself.
sertraline (Zoloft) — settled and adjudicated matters only, including what sertraline is verifiably not part of.
venlafaxine (Effexor) — a heart medicine in the bottle, and the price of a two-year delay.
zolpidem (Ambien) — a safety record written by the regulator rather than the courts: the 2013 dose cut for women and the 2019 boxed warning on complex sleep behaviours.
what the fraud settlements add up to — ten DOJ resolutions that named a psychiatric drug, about $12.6 billion and eight guilty pleas, on one table.

the system, counted

how the headline numbers of mental health are produced, prevalence to hospital stays, with every method break annotated.

how many Americans have a mental illness? — SAMHSA’s survey count by age, 2021 to 2025, beside AHRQ’s hospital stays, cost and length of stay.
how many adults take psychiatric medication? — every federal number with its definition attached, and why they disagree.
more treatment, same suicide rate — the CDC suicide rate from 1900 to 2023 beside treatment use, each on its own axis, and what the journals say about the gap.
what medical errors add up to — the cardiology-conference study, the contested US death-from-error estimates, and the medication-error counts.
psychiatric drugs for children on Medicaid — the three measures every state reports about children and psychiatric medication, with Minnesota’s own rows and four questions for an appointment.
LGBTQ+ mental health in the data — the published federal numbers with the measurement construct named on every figure, and what stopped being counted.
private equity in mental health care — acquisitions from 32 facilities in 2010 to 1,330 in 2021, what the studies found inside acquired hospitals and clinics, and the Senate youth residential report.
the mental health lexicon — twenty terms from social-media mental-health culture, each graded against the research.

follow the money

who funds the research, the education, the advocacy and the coverage, from federal filings and the organisations’ own disclosures.

who funds psychiatric drug research — a funding audit of 235 verified study records across 51 medications, substances and topics.
who funds doctors’ education — drug and device companies gave $841.9 million to accredited continuing medical education in 2025, from the accrediting body’s own report.
who funds mental health advocacy — pharmaceutical money to advocacy organisations and professional associations, from public records only.
who funds the nonprofits — eighteen US mental-health nonprofits sorted by disclosure quality rather than size, from their own annual reports.
pharma and the news — the ad money between the drug industry and the media, the coverage record, and the number nobody publishes.
how pharma makes and spends money — revenue, R&D, SG&A and psychiatric drug sales for eight manufacturers, from the 10-Ks.
what is a PBM? — the middlemen between your prescription and its price: three companies process nearly 80 percent of US prescriptions.
who profits from US health care — six sectors, ten years, 28 named SEC filers: margins and growth against CPI.
where the money goes — US health spending, prices and named-basket profits as time series, 1960 to today.
how psychiatric drug marketing feeds itself — the loop from short trial to speaker fee to prescription, what the best-paid psychiatrists prescribe, and what nobody measures long term.
what the best-paid psychiatrists publish about the drugs that pay them — 129 papers matched to four years of federal payment records and read one by one: 93 favourable, none concluding the drug does not work.
the psychiatrists industry paid most in 2024 — the top recipient lists of 38 psychiatric-manufacturer entities in CMS Open Payments 2024, summed in one place.
drug company payments to psychiatrists near you — enter a ZIP code and read the 2024 federal payment records for the psychiatrists in that area: who paid, how much, and for which products.
the depression questionnaire: who wrote it, who paid, and who gets paid on your score — the phq-9 was paid for by pfizer and the form says so. what a score of ten means, why screening alone changed nothing in sixteen trials, how the number became a medicare payment measure, and what the money actually is.
the other questionnaires: who wrote them, who paid, and which prescriptions they gate — twenty forms behind the prescriptions: who wrote each, who paid where the record says, which are free and which are sold, and where the fda and medicare wrote them into rules.
who pays for the fda: the budget, the user fees, and the share that comes from drug companies — in fy2026 user fees passed congressional appropriations for the first time. the drug review center is 73 percent fee funded, and drug makers pay about a third of the whole agency. the fda’s own operating plan, line by line.
who funds the organizations that run mental health care: the psychiatric and psychological apas, the ama, samhsa and nimh — the psychiatric apa sells the dsm and a meeting, the psychological apa licenses a database, the ama collects royalties on codes the law makes you use, samhsa kept its $7.4 billion and nimh sat flat. the filings, line by line.

where diagnoses come from

the manual, the votes, the counts and the theories, from the primary texts.

the DSM: the book, the money, the count — what the manual is, who writes it, and why it describes symptoms rather than explaining them.
what the DSM is for — the anatomy of a criteria set, the 227 symptom combinations that satisfy major depression, and whether it improves care.
did psychiatry work backwards from its drugs? — the dopamine and serotonin theories were fitted to drug effects; the imaging and genetic findings that cut the other way.
how homosexuality left the DSM — Westphal 1869 to the 1973 vote and the 1974 referendum, the replacement diagnoses that lasted until 1987, and the politics left in.
how transgender left the mental-disorders chapter — Hirschfeld 1910, DSM-III to DSM-5, and the WHO’s 2019 move of gender incongruence into the sexual-health chapter, beside the registry findings that cut the other way.
who was Thomas Szasz? — what he actually argued, quoted from the primary texts rather than the slogan, with both directions on every contested point.
how autism got counted — all 12 ADDM surveillance years with every method break annotated in the CDC’s own words.
how childhood ADHD got counted — the longest clean federal series on any US mental-health diagnosis, with every survey change annotated; DSM-5 barely moved it.

sex, fertility and pregnancy

the high-shame questions, answered from the trials and the registries.

antidepressants in pregnancy: what the brain scans show — six imaging studies 2016 to 2026 on one table, the 2026 fetal-MRI finding with its caveats, and the meta-analysis where fathers’ prescriptions predicted the same outcomes.
who warned about antidepressants in pregnancy — the FDA and MHRA notices in order, the numbers inside them, the 2011 walk-back, and the agencies we could not read listed as unverified.
sex, fertility, pregnancy, and mental health — the hub: what the research shows across sex, fertility, pregnancy and new parenthood — and where it runs out.
ssri sexual side effects — the numbers nobody gave you — how common sexual side effects are when someone actually asks, what the trials found for sildenafil and bupropion, and what the eu regulator made manufacturers put on the label about symptoms that persist.
infertility and depression — how common depression is during infertility, why "just relax" is not supported by the evidence, and what happens to people psychologically when treatment ends without a baby.
after a miscarriage — post-traumatic stress, anxiety and depression after early pregnancy loss — how long it lasts, what happens to partners, and what the trials of one-off counselling did and did not settle.
depression and anxiety in pregnancy — antenatal depression and anxiety: prevalence by trimester, what two meta-analyses found about birth outcomes, depression in fathers, and the trials of support offered to everyone rather than only the high-risk.
postpartum depression — prevalence that does not fade at six weeks, why "no history of depression" is weaker protection than people assume, and the two things with the strongest evidence behind them.

getting through it

therapy, loneliness, attention and the things with trial evidence that cost nothing.

what works that costs nothing — eight formulas from the Lancet, BMJ and JAMA Psychiatry anyone can start this week: the dose, the trial, the effect size and the catch.
how long should therapy take? — 4 to 26 sessions in routine care, 95 percent of responders by session 14, frequency beats length, and six questions to ask before session one.
how to find the right therapist — the therapist explains about 5 percent of outcome but recovery rates ran 23.5 to 95.6 percent across 119 therapists; what the matching quizzes cannot tell you.
male loneliness: what actually moves it — the support-seeking gap, the health data, and what moves it.
AI vs human support — chatbots measurably reduce depression symptoms in the short term; the one thing they cannot do, and why most people still want a human.
where your attention goes — the research on looking inward against looking outward, with the two kinds of inward attention kept apart.

the rest of the library: everything we publish, 37 medication pages, supplements and unapproved substances, comparisons, claims we went and checked, the book series and the glossary.