The book series

follow the money · widespread use

Mania: A Short History of Bipolar Disorder

David Healy · Johns Hopkins University Press, 2008 (audiobook Tantor, 2023)

the diagnosis grew fastest in the years the drugs for it were being sold hardest. the numbers back that. they do not back the idea that bipolar disorder is invented.

the book says

• the publisher: a “provocative history of bipolar disorder” that “illuminates how perceptions of illness, if not the illnesses themselves, are mutable over time,” and “examines the rise of psychotropic treatments and pharmacological marketing over the past four decades.”
• healy’s own statement of the thesis, in a journal: “the term ‘mood stabilizer’ had barely been heard of before 1995 when Abbott Laboratories got a license for using the anticonvulsant sodium valproate (Depakote) for treating acute mania.” after that, he writes, the academic community “still has not come to a consensus on what the term ‘mood stabilizer’ means,” and the lack of consensus did not slow the marketing.³
• a clinical reviewer’s summary: manic-depressive illness “was supplanted with bipolar disorder, a concept that lends itself better to diagnostic elasticity … then it was just a short step away to the bipolar teen, the bipolar child and finally the bipolar toddler.” he called the book “an excellent exposé.” another reviewer found it “probing and challenging” but thin on the long history.^10,11

what the record shows

• office visits where a child or teenager was given a bipolar diagnosis went from 25 per 100,000 people in 1994-95 to 1,003 per 100,000 in 2002-03. that is about forty times as many in eight years. adult visits went from 905 to 1,679, less than double. these are visits, not people, and a third of the youth visits also carried an ADHD diagnosis. the authors called for studies of whether the diagnoses were accurate.¹
• hospital discharges tell the same story: children discharged with bipolar disorder went from 1.3 to 7.3 per 10,000 between 1996 and 2004. the authors offered two explanations and did not choose: doctors recognising it better, or “upcoding” for reimbursement.²
• the concept itself: a review co-written by healy found “a more ambiguous picture as regards pharmacotherapeutic specificity to bipolar disorders than is commonly conceded.”⁴
• the money. the makers of two of the antipsychotics sold as mood drugs in that era later settled with the US government over marketing for unapproved uses: Eli Lilly paid $1.415 billion in 2009 over Zyprexa (olanzapine), including a misdemeanor plea about promotion for dementia in elderly patients, and AstraZeneca paid $520 million in 2010 over Seroquel (quetiapine), promoted for unapproved uses to primary care doctors, pediatricians and geriatricians. together, about $1.9 billion. neither case was about child bipolar disorder, and the AstraZeneca settlement admitted no liability.^5,6

the other direction

• bipolar disorder is real and it is not rare. the national comorbidity survey put lifetime bipolar I at 1.0% and bipolar II at 1.1% of US adults, and found that “use of antimanic medication is uncommon” among them. in adults the bigger problem in that survey was too little treatment, not too much.⁷
• it is one of the most heritable conditions in psychiatry: a twin study put heritability at 85%.⁸
• the oldest mood drug works on the thing that matters most. across 48 trials, lithium cut suicides compared with placebo (odds ratio 0.13), though the events were few and the interval is wide. lithium is a cheap, unpatented salt, which is its own point about which drugs get marketed.⁹

the author

• healy discloses having been a speaker, consultant or trialist for Lilly, Janssen, SmithKline Beecham, Pfizer, AstraZeneca and others, and an expert witness in cases about antidepressants and suicide or homicide.³
• he has been a paid expert for plaintiffs suing drugmakers. in one case a federal appeals court upheld the exclusion of his testimony after court-appointed experts called his analysis unscientific.^13,12
• he co-founded RxISK, a drug-harm reporting site, and writes for Mad in America. we do not cite either.

our take

the strongest version of healy’s case is not “bipolar is fake.” it is that psychiatry works backwards: a drug gets a license for one effect, the effect gets a name (“mood stabilizer”), and the name widens the diagnosis that the drug is sold for. a forty-fold rise in eight years among children is not biology. it is how the diagnosis was being made. the counterweight is just as real: bipolar I is heritable, dangerous and undertreated in adults, and the cheapest drug for it is one nobody markets. both can be true, and the system rewards only one of them.

the number: forty-fold: child and teen office visits with a bipolar diagnosis, 25 to 1,003 per 100,000, 1994-95 to 2002-03.

go deeper: olanzapine: the full legal record · quetiapine: the full legal record · what the fraud settlements add up to

widespread use · oversight gaps

Unshrunk: A Story of Psychiatric Treatment Resistance

Laura Delano · Viking, 2025 (UK: Octopus)

one woman’s thirteen years and nineteen drugs. her story is not the average outcome, but the gap it exposes is: nobody planned the way off.

the book says

• the publisher: at fourteen delano saw her first psychiatrist, was diagnosed with bipolar disorder and started on a mood stabilizer and an antidepressant; over thirteen years she accumulated “a long list of diagnoses and a prescription cascade of nineteen drugs,” until doctors called her “treatment resistant.” the book asks: “what if her life was falling apart not in spite of her treatment, but because of it?”¹⁴
• the only journal review we found (argentina’s Vertex) is sympathetic but clinical. it lists what her care was dealing with, including anorexia, alcohol use and suicide attempts, notes that many medication changes were decisions she made herself, and praises the book for showing how hard it is to tell apart a treatment that is not working, poor adherence, the illness’s own course and side effects.¹⁵

what the record shows

• the scale. one in four americans taking an antidepressant has been taking it for ten years or more, and 68% for two years or more (2011-2014). in 2015-2018, 13.2% of US adults had used one in the past month.^16,17
• how common withdrawal is depends on who you ask. a 2024 meta-analysis of mostly short trials found symptoms in 31% of people stopping an antidepressant and 17% of people stopping a placebo, with severe symptoms in 2.8%; it put the drug’s own share at about 15%. a 2019 review built mostly on people who had stopped in real life found 56%, and nearly half of those rated it at the most severe level offered.^18,19
• the disagreement is live. critics re-pooled the studies that measured withdrawal systematically and got 55%; a 2025 JAMA Psychiatry analysis of 50 trials found about one extra symptom in the first week, “below the threshold for clinically significant discontinuation syndrome.” both sides disclose interests: tapering clinics and book royalties on one side, drug company ties on the other.^20,21
• the guidance moved late. in 2019 the Royal College of Psychiatrists said symptoms can be “more severe” and last “much longer” for some people. the UK guideline (NICE, 2022) now says withdrawal usually passes in one to two weeks but can last “several weeks, and occasionally several months,” and to reduce in proportional steps. a 2019 review found the standard two- to four-week tapers “show minimal benefits over abrupt discontinuation.”^23,24,22
• the US label for the most prescribed antidepressant, Zoloft (sertraline), lists discontinuation symptoms from nausea to “electric shock sensations” and seizures, and recommends a gradual reduction. it gives no number for how often they happen or for how long.²⁵
• children: office visits where a child got drugs from two or more psychiatric classes rose from 14.3% to 20.2% of psychotropic visits between 1996-99 and 2004-07, and the authors wrote that “little is known about the safety and efficacy” of those combinations.²⁷

the other direction

• the best trial on staying on: 478 people who felt well enough to stop were split. within a year 56% of those who tapered to placebo relapsed, against 39% who stayed on. for many people the drug is doing a job.²⁶
• the book’s own evidence cuts both ways. her illness was severe and mixed, and the one journal review reads it as a portrait of how hard it is to know what is causing what, not as proof that the drugs did it.¹⁵

the author

• delano founded and leads the Inner Compass Initiative, a nonprofit on psychiatric drug withdrawal, and works as a paid consultant to people coming off psychiatric drugs. she wrote a blog series for Mad in America from 2012 to 2015. we do not cite it.

our take

unshrunk is a memoir, and one person’s thirteen years are not a trial. but the question it forces is a systems question: tens of millions of people were started on these drugs with a plan to begin and no plan to end. one in four users are past ten years, the regulator’s label still gives no rate for what stopping feels like, and the UK only wrote down “months” in 2022. the answer is not to stop. it is to ask for the exit plan at the start, and never to change anything on your own.

the number: one in four: share of US antidepressant users who had taken them for ten years or more (NCHS, 2011-2014).

go deeper: staying on your antidepressant: both sides · how many adults take psychiatric medication · when an antidepressant stops working

follow the money · widespread use

Crazy Like Us: The Globalization of the American Psyche

Ethan Watters · Free Press, 2010

the japan chapter holds up best: a slogan, then a six-fold jump in sales. the romantic idea that poor countries do better without western care holds up worst.

the book says

• watters, as quoted by reviewers: “we are engaged in the grand project of Americanizing the world’s understanding of the human mind.” four case studies: anorexia in hong kong, trauma counselling in sri lanka after the 2004 tsunami, schizophrenia in zanzibar, and depression in japan.
• a psychiatrist reviewing it in the British Journal of Psychiatry found the sri lanka account of “rival research groups scheming to recruit to their own specific brand of post-traumatic therapy” alarming, and the book overall “more polemical than analytical.”³⁹

what the record shows

• japan. a japanese psychiatrist documented it in a journal: until the late 1990s depression carried stigma in japan; after the first SSRI arrived in 1999, drug companies ran “educational campaigns” and coined the phrase kokoro no kaze, “a cold of the soul.” antidepressant sales went up six-fold, from ¥14.5 billion in 1998 to ¥87 billion in 2006. his objection: a cold is short, and the phrase hid how long treatment lasts. the first SSRI in japan was fluvoxamine (Luvox, Fujisawa; Depromel, Meiji Seika), and his sales figures come from GlaxoSmithKline statistics.^28,29
• hong kong. in a 1993 study of 70 chinese patients with anorexia, 41 (58.6%) had no fear of fatness at all. by 2007 the clinic’s cases had doubled and looked increasingly like the western, fat-phobic pattern.^31,32
• sri lanka. single-session trauma debriefing, the kind of intervention that flooded disaster zones, was found by a Cochrane review not to prevent PTSD; in one trial it raised the risk at a year. a 1999 critique called imported trauma programmes a way that “aggrandise the Western agencies and their ‘experts’.”^36,38

the other direction

• the distress was real. twenty months after the tsunami, a random household survey in sri lanka found clinically significant PTSD in 21%, depression in 16% and anxiety in 30%, and people named a western-style hospital among the things that helped.³⁷
• the famous finding that schizophrenia does better in poorer countries held up in the WHO follow-up, but a later review of 23 studies in 11 countries found it “cannot be assumed”: “lack of care is associated with relatively poor outcomes,” and long-term death rates reached 30.7% at one indian site.^34,35
• lee’s own hong kong data show anorexia kills whatever it looks like: three of 88 patients died, ten times the expected rate.³³
• care built locally works. in zimbabwe, lay health workers on a “friendship bench” cut depression symptoms to 13.7% against 49.9% in usual care. and the WHO says only a third of people with depression get any formal care.^40,41
• in japan, suicide fell in years when SSRI prescribing rose. that is ecological data and proves nothing on its own, but it is the other side of the ledger.³⁰

the author

• watters is a journalist. we found no financial conflict. the book is journalism and is not itself a source on this site; the journal papers above are.

our take

the problem watters found is not that western ideas travel. it is that they travel with a sales plan and without an outcome audit. in japan, the phrase came from the people selling the fix, and sales rose six times over. in sri lanka, the counselling arrived before anyone checked whether it worked. where the book goes too far is in romanticising the alternative: untreated illness is not freedom. the test is the same everywhere: who wrote the definition, who sells the fix, and who measured whether people got better.

the number: six-fold: antidepressant sales in japan, ¥14.5 billion (1998) to ¥87 billion (2006), after the “cold of the soul” campaign.

go deeper: how psychiatric drug marketing feeds itself

widespread use · oversight gaps

The Age of Diagnosis: How Our Obsession with Medical Labels Is Making Us Sicker

Suzanne O’Sullivan · Thesis (Penguin Random House), 2025; UK: Hodder, subtitled Sickness, Health and Why Medicine Has Gone Too Far

overdiagnosis is real and the people who widen the definitions are mostly paid by industry. underdiagnosis is also real, in different people, at the same time.

the book says

• the publisher: “conditions like ADHD and autism are on the rapid rise,” and “our obsession with diagnosis is harming more than helping” because it “too often pathologizes difference, increases our anxiety.” o’sullivan, a london neurologist, calls for “new and better vocabularies for suffering.”⁴²
• a GP living with long covid reviewed it in the British Journal of General Practice and said it “ignores the huge research literature” on long covid, and warned that calling conditions overdiagnosed makes it easy “to justify putting less funding into those areas.”⁵¹

what the record shows

• who widens the definitions. of 16 recent US guideline publications on common conditions, 10 widened the definition of a disease and one narrowed it. none rigorously assessed the harms of widening. on average 75% of panel members had ties to drug or device companies, and 12 of 14 panels were chaired by someone with ties.⁴³
• psychiatry’s own panel: 69% of the DSM-5 task force reported industry ties, up from 57% for DSM-IV.⁴⁴
• ADHD: a review of 334 studies found evidence of overdiagnosis and overtreatment in children, and only 5 studies had checked whether the extra, milder cases benefit; those suggested the harms may outweigh the benefits.⁴⁵
• adults: an estimated 15.5 million US adults (6.0%) had a current ADHD diagnosis in 2023, about half diagnosed at 18 or older. FDA staff counted a 30% rise in new stimulant prescriptions among adults aged 20 to 39 from before to during the pandemic, and wrote that research is needed to tell “unmet need vs overprescribing.”^46,47
• the mechanism o’sullivan points at has a name in the literature, the prevalence inflation hypothesis: awareness campaigns help people report real symptoms, and may also lead some to read ordinary distress as disorder, which can then become self-fulfilling. its authors present it as a hypothesis to test, not a finding.⁴⁸

the other direction

• in the UK only about one in nine adults with ADHD has been diagnosed, and diagnosed adults lose roughly seven to nine years of life expectancy.⁴⁹
• in the US, about half of adults with a mental illness got no treatment in 2025.⁵⁰

the author

• o’sullivan is a london neurologist whose work is epilepsy and functional disorders. we found no financial conflict.

our take

the book is right that a label is not free. but the systemic problem is sharper than “we diagnose too much.” a diagnosis is the key that unlocks school support, insurance payment and a prescription, so the system keeps cutting keys, and the committees that cut them are mostly paid by the companies that sell what the key opens. that is the part worth being angry about. what the book underplays is the person on the other side of the same system who never got a key at all.

the number: seventy-five percent: average share of guideline panel members with industry ties, on panels where 10 of 16 widened a disease definition and none rigorously assessed harms.

go deeper: how childhood ADHD got counted · the DSM

follow the money · widespread use

Dopamine Nation: Finding Balance in the Age of Indulgence

Anna Lembke · Dutton, 2021

a book about phones and pleasure, written by a psychiatrist whose real expertise is the opioid epidemic. the opioid record is the part that holds.

the book says

• the publisher: “this book is about pleasure. it’s also about pain,” and “the relentless pursuit of pleasure leads to pain.” it calls the smartphone “the modern-day hypodermic needle.” the core model is a balance: every pleasure is followed by an opposite pull toward pain, so chasing more pleasure leaves you worse off.⁵²
• lembke is a stanford psychiatrist who specialises in addiction. her earlier book, Drug Dealer, MD (2016), is about how doctors were led to overprescribe opioids.

what the record shows

• the scale she lived through: US opioid prescriptions peaked in 2012 at more than 255 million, 81.3 for every 100 americans. by 2020 the rate had fallen to 43.3. in 2023, 105,007 americans died of a drug overdose.^60,61
• the money. in may 2007 Purdue Frederick, the maker of OxyContin, and three executives pleaded guilty to misbranding it, and paid $634.5 million; the company admitted falsely claiming it was “less addictive, less subject to abuse, and less likely to cause withdrawal symptoms.” in 2020 Purdue Pharma pleaded guilty to three felony conspiracies and admitted that “from May 2007 through at least March 2017” it conspired to defraud the United States. the crimes it admitted began the same month as the first plea. it was sentenced in april 2026: a $3.544 billion fine and $2 billion forfeiture; the Sackler family paid $225 million separately.^58,59
• the dopamine science is more careful than the cover. dopamine drives wanting, not pleasure: in the leading model, “liking” is “not dependent on dopamine.” the withdrawal part of her balance is supported for drugs of addiction; stretching it to phones is her extrapolation.^53,54

the other direction

• on phones: across 355,358 adolescents, digital technology use explained “at most 0.4% of the variation in well-being.”⁵⁵
• the courts are split on who pays for the opioid epidemic. ohio’s supreme court threw out public-nuisance claims against pharmacies in 2024; a federal appeals court revived the huntington, west virginia case against distributors in 2025.^63,62

the author

• lembke discloses that she has been “retained as a medical expert witness in federal, state, and county opioid litigation (plaintiff side).”⁵⁶
• a federal judge found her qualified on addiction science but not on whether marketing caused sales, and excluded parts of her testimony about pharmacies.⁵⁷

our take

read dopamine nation for the clinic stories and for the idea that relief you chase can deepen the hole. but the book underplays its own best evidence. the clearest case of pleasure sold as medicine was not the phone. it was a pill marketed as less addictive, by a company that pleaded guilty to that lie in 2007 and kept going for ten more years. when the fine is smaller than the business, a guilty plea is a cost, not a stop.

the number: 255 million: US opioid prescriptions at the 2012 peak, five years after the maker of the best-known one pleaded guilty to lying about addiction.

go deeper: the overdose the prescription was part of

widespread use · volume-driven care

Bad Therapy: Why the Kids Aren’t Growing Up

Abigail Shrier · Sentinel, 2024

right about blanket classroom programs, which the best trial found did nothing. wrong if you read it as “therapy does not work for kids.”

the book says

• the publisher: “an investigation into a mental health industry that is harming, not healing, American children.” its claims: talk therapy “can induce rumination,” social emotional learning “handicaps our most vulnerable children,” and gentle parenting “can encourage emotional turbulence.” and a concession: mental health care “can be lifesaving when properly applied to children with severe needs, but for the typical child, the cure can be worse than the disease.” the book is built on interviews, not a review of trials.⁶⁴

what the record shows

• the biggest trial of a classroom program: 84 schools and 8,376 students aged 11 to 14 in the UK, half given ten lessons of mindfulness. at a year, “no evidence” it beat normal teaching on depression risk, wellbeing or behaviour. it was run by the people who lead the Oxford Mindfulness Centre, and they published the null result.⁶⁵
• in the same trial, students already at risk scored slightly worse after the program; the authors said it may be contraindicated for them.⁶⁶
• across 137 studies and 56,620 young people, there was “little evidence” that school programs aimed only at preventing depression or anxiety work. a second review found small effects that faded.^67,68
• talking about a trauma straight after it can backfire: single-session debriefing did not prevent PTSD, one trial found higher risk at a year, and the UK guideline says not to offer it.^36,70
• the idea that awareness itself can inflate symptoms has a name, prevalence inflation, and its authors say it still needs testing.⁴⁸

the other direction

• therapy for children who actually have a disorder works. across 447 studies and 30,431 young people, the average effect was 0.46; a treated child had a 63% chance of doing better than an untreated one. it was strongest for anxiety.⁶⁹
• the pattern is targeted versus universal, not therapy versus none. grief counselling shows the same thing: little effect for everyone, real effect for people who were struggling.⁷¹

the author

• shrier is a journalist and runs a paid newsletter. we found no financial tie to the claims in this book.

our take

the systemic point is not that therapy is bad. it is that schools bought blanket programs for every child before anyone tested them, and the best test found nothing. that money comes out of the budget for the children who need real treatment, where the evidence is good. the question for a parent is simple and fair to ask: what does this program measure, and what did it find.

the number: 8,376 students in 84 schools: the largest trial of classroom mindfulness, and no benefit at one year.

go deeper: how long should therapy take

philosophy

Zen in the Art of Archery

Eugen Herrigel · German 1948; English 1953 (Pantheon; trans. R.F.C. Hull, intro D.T. Suzuki)

the most famous zen story in the west was built on a translation gap. the master called the miracle shot a coincidence.

the book says

• herrigel, a german philosophy professor who taught in sendai in the 1920s, describes his years in japan learning archery from the master Awa Kenzō: the arrow must be released by “It,” not “I,” and one night the master hit the target in the dark, then split the first arrow with the second.

what the record shows

• a 2001 study in the Japanese Journal of Religious Studies went back to the japanese sources. awa “never practiced Zen even once in his life”; his “great doctrine” was a new religion of archery he founded in 1927. the interpreter at the lessons says he was not there the night of the famous shot, and that he often smoothed over things awa said.⁷²
• asked about the shot, awa said: “that was just a coincidence! I had no special intention to demonstrate such a thing.” in traditional japanese archery, splitting your own arrow is a failure, not a miracle. yamada suggests “It shoots” may have been a mishearing of ordinary praise: “that’s it.”⁷²

the other direction

• yamada does not call herrigel a fraud; he argues the myth grew from language gaps and herrigel’s own expectations. awa did say “bow and Zen are one,” and an early archery school’s records show real zen influence.⁷²

the author

• herrigel died in 1955. yamada notes he was a Nazi sympathiser who took part in party activities after returning from japan, and later became rector at Erlangen.⁷²

our take

the lesson survives the myth. effort that loosens its grip is a real idea, and it is not zen’s alone. but the story is also a warning about how a good feeling becomes doctrine: one man, a language he did not speak, an interpreter who was absent, and seventy years of readers who never checked. the same thing happens with a diagnosis.

the number: the master’s word for the famous shot: “coincidence.”

philosophy

Psycho-Cybernetics

Maxwell Maltz · Prentice-Hall, 1960 (updated edition with Matt Furey, 2015)

the source of “21 days to build a habit.” maltz never quite said it, and the best evidence says about 66.

the book says

• maltz, a plastic surgeon, wrote that it “usually requires a minimum of about 21 days to effect any perceptible change in a mental image,” pointing to how long patients take to get used to a new face. the book’s larger claim is that self-image steers behaviour like a guidance system, and that vividly imagined practice trains it.

what the record shows

• habits: in a study of real people building a daily habit, the time to reach 95% of peak automaticity ranged from 18 to 254 days, and missing a day did not reset it. a 2024 review of 20 studies put the median at 59 to 66 days, with individuals ranging from 4 to 335, and said the results “clearly refute the popular notion that habits can be formed in approximately 21 days.” the 66-day median comes from the first study.^74,73
• imagined practice: a meta-analysis found “mental practice has a positive and significant effect on performance.” maltz’s stronger claim, that the nervous system “cannot tell the difference” between imagined and real experience, is not what the evidence shows.⁷⁵

the other direction

• the core of the book, that how you see yourself shapes what you attempt, is close to ideas that later became mainstream cognitive therapy. the number was wrong; the direction was not.

the author

• maltz died in 1975. the brand now belongs to the Psycho-Cybernetics Foundation, whose president co-wrote and narrates the updated edition and sells coaching built on it.

our take

a surgeon’s aside about new faces became a law of behaviour, and a whole industry now sells 21-day challenges. the real number is about three times longer, and the most useful finding in the research is the kind one: missing a day does not undo it.

the number: sixty-six days: median time to a habit in the study usually cited (range 18 to 254), against the 21 in the popular version.

philosophy

The Obstacle Is the Way: The Timeless Art of Turning Trials into Triumph

Ryan Holiday · Portfolio/Penguin, 2014

stoicism is a real ancestor of CBT. “stoicism works” has almost no trials, and the grin-and-bear-it version tracks worse wellbeing.

the book says

• holiday retells marcus aurelius and epictetus in three movements: see the obstacle clearly, act on it, and accept what cannot be changed.

what the record shows

• the lineage is real: “both Albert Ellis and Aaron Beck … explicitly acknowledged the role of Stoicism as the philosophical precursor of their treatment approaches.” epictetus’s line that people are disturbed not by things but by their views of things is the seed of cognitive therapy.⁷⁶
• but the folk version, staying quiet and unbothered, was linked to lower wellbeing in samples from new zealand, norway and the US.⁷⁷

the other direction

• cognitive therapy, the descendant, has decades of trials. the ancestor, taken on its own, has almost none.

the author

• holiday’s Daily Stoic brand sells a paid membership, courses, medallions and a 21-day challenge (the same number as maltz).

our take

take the stoics for what they are: a way to think about what is yours to control. the evidence for the method lives in CBT, not in the medallions.