in BMJ, Thinking about psychiatry

“One in four”: the anatomy of a statistic

(From a bus stop Archway – if you look carefully you can see the reflection of me and my bike)

This written by me and Jamie Horder published this week in the BMJ

Despite a lack of supporting evidence, the claim that one in four people will have a mental health problem at some point in their lives is a popular one. Where does this figure come from, and why does it persist, ask Stephen Ginn and Jamie Horder

“It’s time to talk” is a campaign currently being promoted by Time to Change, a charity whose aim is to change attitudes to people with mental ill health. On the charity’s website a banner tells us:

“1 in 4 of us will experience a mental health problem at some point in our lives, but we still don’t talk about it. What are we afraid of?”

This “one in four” figure has also appeared in government speeches(1) and NHS publications.(2) It is the name of a short film and the title of a mental health magazine.

Yet it is not always clear to what the figure refers. Time to Change seems to be referring to lifetime prevalence, while a 2010 advertising campaign by Islington Primary Care Trust stated, “One in four people will experience mental health problems each year.” A statement on the Royal College of Psychiatrists’ website reads, “One in four people has a mental health problem,” implying point prevalence.

The evidence base

The number’s origin is unclear. When one of us (SG) contacted a selection of organisations that use “one in four” in their literature, they cited a number of different sources. The earliest seems to be a 2001 World Health Organization report, Mental Health: New Understanding New Hope, which stated, “During their entire lifetime, more than 25% of individuals develop one or more mental or behavioural disorders (Regier et al 1988; Wells et al 1989; Almeida-Filho et al 1997).”(3)

However, none of the three papers cited contains an estimate of 25% lifetime risk. One did not report on lifetime prevalence at all,(4) and the two that did provide a lifetime figure of rather more than 25% (66% for “all [mental] disorders” in New Zealand and 31-51% in Brazil).(5, 6)

Lifetime prevalence of mental disorder seems never to have been estimated in the United Kingdom. In 2007 the annual psychiatric morbidity survey (APMS) estimated a UK prevalence of 23% in the past week.(7) In numerous other countries lifetime estimates are reported as being in the region of 50%.(8)

We are unaware of any evidence that straightforwardly supports a UK lifetime prevalence of 25%. The APMS past week prevalence most robustly supports one in four as a statement of the UK’s 12 month prevalence,(7) but in this case the UK lifetime prevalence would be expected to be much higher.

Counting cases

A 2005 meta-analysis estimated a yearly prevalence of 27% for the European Union (including the UK),(9) but a 2010 update of this work revised this to 38% a year,10 as a result of including more disorders such as insomnia and attention-deficit/hyperactivity disorder. This highlights the fact that over the years the consensus on what constitutes mental disorder has often changed.

Different population surveys adopt different definitions, and there is no agreement about whether to treat, for example, a phobia such as arachnophobia as “mental illness.” No major study has considered nicotine dependence or male erectile disorder in their calculations, despite these disorders being widespread and listed in the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV). Nicotine dependence is perhaps responsible for more deaths than any other psychiatric disorder.

Furthermore, surveys such as the APMS establish diagnosis in a very different way from how it is discerned clinically. In the clinic, a doctor works from a patient’s presenting complaint, through their history, and on to mental state examination. By contrast the APMS recruited a large representative sample and used a structured diagnostic interview to screen each participant for a range of disorders. Structured interviews involve a patient answering a fixed series of questions taken from published criteria.

Systematic checking of a symptom inventory in this way lacks the benefit of clinical judgment and simultaneously creates a risk of both over-diagnosis and under-diagnosis. Taken literally, the DSM-IV criteria for major depressive disorder would deem many people depressed after bereavement or the end of a relationship. Conversely, a patient’s imperfect recall or lack of insight into their own psychopathology could lead to under-reporting.

The popularity of “one in four”

Despite these drawbacks, why has this figure proved so popular? We would like to suggest some reasons.

Demonstrating relevance: For journalists, quoting a high prevalence of mental disorder helps illustrate the newsworthiness of stories about mental health.

Fighting stigma: The one in four statistic has been used extensively by charities to advocate the interests of people with mental illness. Much of their recent campaigning has focused on attempting to combat stigma and prejudice through providing a more inclusive vision of mental disorder—one in which it is nothing unusual and a threat to everyone.

Not too big, not too small: If the intent is to raise awareness of the burden of mental illness, why do organisations not cite the even higher, and better supported, figures of one in three or one in two lifetime prevalence? We suggest that one in four is high enough to gain people’s attention but not so high that it provokes incredulity, as claims that over 50% of people have had a mental illness indeed have.


The one in four figure for mental illness prevalence is widely quoted, related variously to lifetime, yearly, or point prevalence. The evidence indicates that it is best supported as an estimate of yearly prevalence. However, estimates of the population prevalence of mental disorder should be approached with caution, as the methods used often have shortcomings. It is important that people know that mental illness is common and that treatment of mental disorder is essential, but it is not clear that championing a poorly supported prevalence figure is the way to achieve this.


  1. Johnson A. Psychological therapies in the NHS: science, practice and policy (speech to the New Savoy Partnership Annual Conference). Department of Health, 2008.
  2. Tavistock and Portman NHS Foundation Trust. Mental health myths. 2011.
  3. World Health Organization. Mental health: new understanding, new hope. WHO, 2001:23.
  4. Regier DA, Boyd JH, Burke JD Jr, Rae DS, Myers JK, Kramer M. One-month prevalence of mental disorders in the United States. Based on five epidemiologic catchment area sites. Arch Gen Psychiatry1988;45:977-86.
  5. Wells JE, Bushnell JA, Hornblow AR, Joyce PR, Oakley-Browne MA. Christchurch psychiatric epidemiology study, part I: methodology and lifetime prevalence for specific psychiatric disorders. Aust N Z J Psychiatry1989;23:315-26.
  6. Almeida-Filho, Mari Jde J, Coutinho E, França JF, Fernandes J, Andreoli SB, et al. Brazilian multicentric study of psychiatric morbidity: methodological features and prevalence estimates. Br J Psychiatry1997;171:524-9.
  7. Weich S, Brugha T, King M, McManus S, Bebbington P, Jenkins R, et al. Mental well-being and mental illness: findings from the adult psychiatric morbidity survey for England 2007. Br J Psychiatry2011;199:23-8.
  8. Kessler, Chiu WT, Demler O, Merikangas KR, Walters EE. Prevalence, severity, and comorbidity of 12-month DSM-IV disorders in the national comorbidity survey replication. Arch Gen Psychiatry2005;62:617-27.
  9. Wittchen HU, Jacobi F. Size and burden of mental disorders in Europe: a critical review and appraisal of 27 studies. Eur Neuropsychopharmacol2005;15:357-76.
  10. Wittchen HU, Jacobi F, Rehm J, Gustavsson A, Svensson M, Jönsson B, et al. The size and burden of mental disorders and other disorders of the brain in Europe 2010. Eur Neuropsychopharmacol2011;21:655-79.

Write a Comment


  1. I don’t think that it’s easy to answer this question – the answer depends on how you choose to define mental disorder.

    Taylor and Chave (1980 or so, and quoted in Clare’s Psychiatry in Dissent)

    The size of the catch depends on the size of the mesh of the net that is used; mental institutions find the least, community studies more, and direct interviews find the most. Indeed, the over-enthusiastic psychiatric diagnostician can find evidence of psychiatric ill-health in most human beings; such findings tell us more about the observer than the observed

  2. In a related “statistical frenzy” story:

    Since I have worked in psychopharmacology, I’ve been fielding questions about the “Ambien causes cancer” study. Some of my concerns about the data are mirrored in the published reviews – since the article was released pre-publication – and some of my own concerns are related to the diverse nature of temazepam and zolpidem, the only two drugs for which actual data was provided. Specifically, temazepam being a benzodiazepine, a class of drugs that has not been historically associated with a higher incidence in cancer mortality rates – or any mortality rates, really, aside from polypharmacy, whereas zolpidem, being a “z-like benzodiazepine hypnotic” is not a true comparative drug. While they examined multiple sedative-hypnotics, that information was not included in the publication. They also apparently examined barbiturates but did not provide data based on drug class, and it is well-known that barbiturates alone have an extremely high mortality rate.

    Moreover, there’s something else that’s “off” about the statistical analyses – aside from not taking polypharmacy into account or (especially) the lack of information on specific drug class as well as on psychiatric diagnoses and associated pharmacotherapy – but my statistical capabilities are weaker than they once were when I worked in research so I can’t pinpoint precisely what it is about this data that makes me feel uneasy.

    I was hoping you could take a look and provide some perspective on these findings, which have led to some mass hysteria on the part of my friends/family members.

    I have also sent this to two other psychiatry blogs and am really hoping that – whether the conclusions are valid or not – someone will examine the data more closely than the media has done. If it is a valid analysis, then I’m more curious about the nature of the mortality link than “is Ambien going to kill me?” but I’m having trouble accepting this study as valid.

  3. Can you tell anyone there is a one in four chance of them getting something that’s impossible to diagnose accurately?
    Is is in the patient’s best interests?
    Who gains?

  4. Thank you for this! I find it incredibly annoying. Partly because it seeks to normalise something to make it acceptable; as if only something normal should be.

  5. I think this sort of campaign that uses the one in four figure (or any other prevalence figure) is difficult because in making mental illness seem ‘normal’ they somehow also make people with severe mental illness seem like they are making a fuss about nothing or being scroungers if they can’t just pull themselves together and get on with life, so I think simplistic campaigns about mental illness can do more harm than good.

  6. Totally agree with you David.
    It also doesn’t address the biggest source of discrimination – central government