As I explained in parts 1 and 2, "one in four" has no basis in the scientific literature, although given how dubious the literature is, this is not necessarily a bad thing. It's not clear where the one in four meme originally came from, although most of the recent uses probably trace back to a 2001 WHO report which quoted it.
But why has one in four proven so popular? The simple answer is that it's high, but not too high. Were someone to say that one in every two people suffer from mental illness, most of us just wouldn't believe it. That's actually what most published studies have found, but it fails the laugh test. One in four is low enough to be believable but high enough to be striking, attention-grabbing, and memorable.
Why are so many people quoting the highest estimate they can get away with? After all, if we've (unconsciously) decided that one in two is "too high", as we seem to have done, we could equally well decide that one in four is too high. We could redefine "mental illness" such that the prevalence of it was, say, one in ten, or one in fifty. Those would be no more and no less valid than one in four. But we haven't, so why not? There are some obvious, but wrong, explanations -
- Most obvious of all is Big Pharma. It's a well known fact that pharmaceutical companies are pure evil and that Satan sits on the board of directors of most them. And clearly, pharma do have a financial interest here. The more people who are deemed mentally ill, the more who might buy their pills. However, to think that pharma are primarily responsible for the spread of one in four is simplistic. Cynics that we are, no-one takes what pharma says seriously. At the least, they would need some accomplices to help convince people of the idea.
- People sometimes accuse governments of talking up prevalence estimates. The Szaszian phrase "the therapeutic state" still crops up, with the implication that the government wants to use mental illness as an excuse to implement authoritarian policies. Well. This conspiracy theory doesn't seem all that realistic, given that the present British government, at least, couldn't conspire to get drunk in a brewery. Indeed if anything, modern governments generally treat the mentally ill as a financial burden. They require often-expensive treatment, and maybe also welfare payments. At least in Britain, the government currently trying to minimize the numbers officially considered mentally ill, to save money.
- Finally, "Psychiatry" is said to be expanding its power by defining everyone as mentally ill. Anyone who has been to a psychiatry conference will find it hard to take the idea of such a grand plot seriously, but more fundamentally, this assumes that psychiatrists like treating people just for the sake of it. Why would they? In some countries, true, they do have a financial incentive to treat as many people as possible, although no more than any other medical professional. But in Britain and other countries with nationalized health-care, psychiatrists are paid a salary and every additional patient is just more work. And more chance of getting called up in the middle of a romantic dinner and having to talk down some suicidal person for three hours.
Mental health charities and other advocates for the mentally ill like one in four because it's a great way of fighting the stigma attached to mental illness. One in four represents hard, scientific proof that mental illness is not rare, weird, and freakish - i.e. that the mentally ill are "just like everyone else" and have the same rights. Which is a perfectly good message, and remains one despite the fact that one in four is rubbish.
Academics like one in four - or rather, like high estimates of the prevalence of mental illness - because it gives their work an air of importance. Almost every research paper about depression, for example, starts with a paragraph of formulaic boilerplate to the effect that "Depression is really common". Here's the first paragraph of the first depression paper I plucked from PubMed at random (honestly) -
Major depression is a substantial public health problem, ranking first among the causes of worldwide disability. According to a study by the World Health Organization, depression was estimated as fourth leading cause of disability-adjusted life years (DALYs), a measure of disease burden, in all age groups and the second leading cause in people aged 15–44 years in 2001 (World Health Report, 2001). Lifetime incidence and prevalence estimates are not available for the UK or Scotland (Paykel et al., 2005). The one year prevalence for major depressive disorder (MDD) in Europe has been estimated at 3.9%, 5.0% in women and 2.6% in men (...). A large proportion of these patients remain untreated despite a number of campaigns aiming at increased awareness of depression.This doesn't quote one in four as such (academics rarely do), but you can see why these authors and everyone else are not exactly lining up to debunk it. I'm not accusing the authors of this paper of being deliberately disingenuous; I'm sure they believe what they wrote. But they wrote it and made it their first paragraph for a reason. It's a running joke in modern science that when you're applying for grant money or trying to get work published, you have to talk up the "practical implications" of your research. In medicine a crucial aspect of this is talking up the seriousness and importance of the disease you're studying.
So, one in four persists because it's in no-one's interest not to say it. There's no conspiracy, just the collective action of various groups all of whom benefit from the idea that mental illness is extremely common. Fascinatingly, the very same mental health charities (and pop psychologists) who are traditionally opposed to academic psychiatry and Big Pharma have ended up promoting statistics which perfectly serve the interests of those groups. Not on purpose, of course. No-one is trying to be deceptive, everyone is just doing what they think is the right thing - but the end result is that this profoundly dubious statistic has become almost universally accepted.
[BPSDB]
8 comments:
This has been an interesting series. The variability in estimated prevalence across studies is quite striking. What figure do you suggest mental health organizations should use in their literature, or should they drop the quantitative pretense entirely?
Thanks.
The problem isn't really with the numbers. It's with the concept of "mental illness", which is both a) very broad and b) very vague.
What you define as mental illness determines the figure you get. For example we can be pretty certain that no more than about 5% of people will ever experience mania or florid psychosis. (I'd guess more like 3%). If you define mental illness as being manic or psychotic, that would be your figure.
That's roughly what happened 100 years ago, when insane asylums were places you put people who were too manic or psychotic to live in society, and anxiety, depression, etc. were seen as "nervous disorders" - literally disorders of the nerves - treated by an entirely different medical speciality (neurologists, as opposed to "alienists" or asylum-doctors, who were much less presitgious). (I know this because I've read Edward Shorter's History of Psychiatry).
Whereas, if you define having mental illness as a self-reported DSM-IV disorder, it's 50%.
We as a society seem to have collectively decided that "mental illness" has an extremely broad definition. That's what I want to question, really. 1 in 4 is just the statistical tip of a conceptual iceberg...
Finally, "Psychiatry" is said to be expanding its power by defining everyone as mentally ill. ...this assumes that psychiatrists like treating people just for the sake of it. Why would they? ... every additional patient is just more work. And more chance of getting called up in the middle of a romantic dinner and having to talk down some suicidal person for three hours.
If only the most severely mentally ill are treated by psychiatrists, then they will have a higher probability of those after-hours calls and a lower probability of positive outcomes (plus more difficult visits). It's the psych equivalent of being a trauma surgeon. Maybe some psychs would rather be the equivalent of family practitioners who have a lot of lower-stress patients mixed into the caseload.
Still, I agree with your primary reason for the persistence of this particular urban legend.
That's a good point actually! but most of the psychiatrists I know find the not-really-very-ill-if-at-all patients to be quite frustrating to work with. Although in Britain, psychiatrists don't really see such people very often (family doctors on the other hand get plenty).
"We as a society seem to have collectively decided that "mental illness" has an extremely broad definition."
But in this case, 1 of 4 might still not be the wrongest of predictions?
well it's no worse than any other number. although as I said in Part I, no study has actually found a lifetime prevalence of 1 in 4. They generally report 1 in 3 to 1 in 2. But you could design a study that would find 1 in 4, if that was what you wanted.
From what I can see, the major papers (including the WHO paper) were not peer reviewed by a statistician, usually a major failing in many journals relating to any medical field. Medicos & psychologists are not statisticians. Some may know some statistics but that is not their profession. Likewise as a statistician I don't profess to being a psychologist just because I know about some parts of it. As such it is not highlighted clearly enough that the major driver for confusion is the definitions used and their blatant misinterpretation both by the authors, and media and other groups such as mental health organisations. For the authors there is the desire to publish something that shows up something significant otherwise it's hard to justify getting more money. For the latter they have a hard battle in justifying more money is needed to treat mental health. Whilst their intentions are noble, they are achieving this through a carefully crafted lie. People trust statistics, particularly those published in a journal. However, who can remember Andrew Wakefield and his "statistics"?
As for the comment by Neurocritic regarding the variability in estimated prevalence, I could not find (and I admit I didn't take the time to look that hard) any confidence intervals surrounding these prevalence. It must be stressed that these are sample based estimates and as such there is a degree of uncertainty surrounding them. Upon examination of this uncertainty we may find that the estimates are not that dissimilar at all.
The numbers presented by the papers should not be ignored. If their research has been done correctly then they are valid. What is needed are careful notes on what they relate to. For instance, they are the prevalence of "self diagnosed incidents within a year" for an example. These caveats are critical in understanding the value and appropriateness of the estimates and how they can be interpreted, particularly if any inference is to be made from them. Again, this is where a professional statistician should come in and provide advice.
It should be pointed out that mental health and other medical areas are not alone in this issue. Most areas of research have these issues. A recent example includes climate change where the panel that looked into the "Climategate" affair at the University of East Anglia's Climatic Research Unit. Whilst the panel concluded there was no scientific malpractice it remarked the following: "We cannot help remarking that it is very surprising that research in an area that depends so heavily on statistical methods has not been carried out in close collaboration with professional statisticians." Professor David Hand was on the panel and he is the current president of the Royal Statistical Society.
Alex: Those are all good points. In particular I think you're right that these numbers shouldn't be ignored even if we don't accept them "at face value" - and maybe I've been guilty of this myself - they are measuring something, even if we might not accept that it's "illness". The fact that a large % of people self-report symptoms like low mood is interesting, and maybe worrying, although equally it might be reassuring since moods are meant to go up and down according to circumstances - someone who was cheerful in every situation wouldn't necessarily benefit from it in the long run.
Post a Comment