This is the second post in a series examining the idea that one in four people suffer from mental illness at some point in their lives. Those who read the first, and the comments below it, will know that this much-quoted statistic has no apparant basis in the scientific literature. In fact, I'm still not sure where it came from.
But there's more. One in four is a strikingly high figure. That's surely a large part of why it's so widely cited. Yet those studies which have attempted to estimate the lifetime prevalence of mental illness have all arrived at even highernumbers, from 1-in-3 to 2-in-3. 50% is typical. Compared to the actual data, one in four is rather conservative.
But is it really true that half of us suffer from mental illness at some point? Guess.
The 50%+ estimates come from population surveys which attempt to study a random sample of the population of a certain country. In order to establish whether each person is mentally ill or not, they use structured diagnostic interviews. These consists in asking the subject a fixed ("structured") series of questions, and declaring them to have a certain mental disorder if they answer "Yes" to a given number of them.
For example, the National Comorbidity Survey Replication (NCS-R), the most recent American study, used a stuctured diagnostic interview called the NCS-R Interview Schedule, which was based on the DSM-IV criteria for mental illness. You can download the Schedulehere. Here's the kind of thing it involves - the possible answers being Yes, Know, Don't Know, or Refuse to Say:
Have you ever in your life had a period lasting several daysor longer when most of the day you felt sad, empty ordepressed?
Have you ever had a period lasting several days or longer when most of the day you were very discouraged about how things were going in your life?And if you answered yes to these preliminaries, you got some more questions. A lot more:
Think of times lasting two weeks or longer when (this problem/these problems) with your mood (was/were) most severe and frequent. During those times, did your feelings of (sadness/or/discouragement/or/lack of interest) usually last less than one hour a day, between 1 and 3 hours, between 3 and 5 hours, or more than 5 hours?LESS THAN 1 HOUR..................................1BETWEEN 1 AND 3 HOURS......................2BETWEEN 3 AND 5 HOURS......................3MORE THAN 5 HOURS..............................4DON’T KNOW.............................................8REFUSED.....................................................9
And so on. The interviewer's instructions spell out exactly which questions to ask, in which order. They even specify how fast to say certain parts. It's about as close to getting a robot to do the interviewing as technology permits.
Such surveys are nothing if not rigorous - everyone gets asked the same questions. They're usually very large - the NCS-R included 10,000 people. And the diagnoses are well-defined; when the NCS-R talks about "Major Depressive Disorder", we know exactly what it's referring to. These are considerable strengths.
But this impressive diagnostic edifice rests on shaky foundations. For one thing, as Allen Hortwitz and Jerome Wakefield (authors of a flavor-of-the-month book) point out in a provocative article, it is built on the assumption that you can diagnose mental illness purely on the basis of the symptoms. Depression, say, consists in experiencing two weeks or more of sad mood, difficulty sleeping, difficulty concentrating, etc.
Yet this leaves no room for taking into account the context of the symptoms, something that quickly leads to absurd conclusions. Taken literally, the DSM-IV criteria for depression would deem most people to be depressed after a break-up, or even a bad bout of influenza. Of course in real life, no doctor does take them literally (at least, I would hope not.) But this is exactly what population surveys do.
A further problem with this approach is that it relies upon the interviewee to recognise their own psychiatric symptoms - without any training or experience in doing that. Anyone who has suffered from clinical depression will know how it stifles ones ability to take interest or pleasure in activities. And any reasonably experienced clinican, talking to a patient, will be able to recognise this symptom, "anhedonia". You know it when you see it - if you're experienced. But to present a lay person with a yes-no question such as
During the episodes of being sad, empty, or depressed, did you ever lose interest in most things like work, hobbies, and other things you usually enjoy?
And to take their answer as the truth is absurd. Anhedonia is much more than merely losing interest in your work or hobbies; that is a partial description of it, but it no more captures the essence of anhedonia than does the description of music as "a series of notes".
So while Hortwitz and Wakefield argue that population surveys overestimate the prevalence of mental illness, but they might equally well be underestimating it. All it takes is for someone to fail to recognise a certain symptom, and it wouldn't go down on paper. (Imagine asking people, "Are you annoying?" - the most annoying people are annoying precisely because they don't recognise their own annoyingness.) If someone is unwilling or unable to talk about a certain aspect of their lives, it won't get recorded either.
In the case of surveys which attempt to estimate lifetime prevalence, there is also the problem that human memory is not perfect. To ask a 50 year old person whether they have ever had a period lasting 2 weeks or longer when they felt discouraged for over 3 hours per day, and expect a meaningful answer, is - well - bonkers.
So, for all of their strengths, the results of these population surveys could be either overestimating or underestimating the "true" prevalence of mental illness. It seems most likely that they overestimate it, because the estimates are so high that there is little room above them. But equally, everyone in the nation could be secretly hiding immense inner anguish, and these surveys would not pick up on it. The true prevalence might be 100%, for all we know.
And ultimately, the notion of a single "true" estimate for mental illness is the most problematic thing of all. As I suggested in the previous post, you can change the prevalence of mental illness almost at will, simply by declaring some things to not count as mental illnesses. Is drinking too much a mental illness? What about a phobia of heights? What about mild autism? What about problem gambling? What about depression that's not so severe as to prevent people living a normal life? There are no right or wrong answers to these questions. But this means there is no right or wrong value for the prevalence of mental illness.
In the next post, I'll be looking at why the estimate of one in four has proven so popular.
Horwitz, A., & Wakefield, J. (2006). The epidemic in mental illness: Clinical fact or survey artifact? Contexts, 5 (1), 19-23 DOI: 10.1525/ctx.2006.5.1.19