Now imagine that, for some reason, people decided that 10% of the population need to be taking this drug, instead of 1%. So sales of the drug sky-rocket. Eventually some clever person comes along and asks "This is one of the biggest selling drugs in the world - but does it work?" They look into it, and find that it doesn't work very well at all. For about 9 out of 10 people, it's completely useless! What a crap drug.Of course the drug hasn't changed, and what's crap was the decision to prescribe it to so many people.
*
Back to reality. According to accepted DSM-IV diagnostic criteria, close to 50% of people suffer from a mental illness at some point; a large fraction of this being depression. 10% of Americans took antidepressants last year according to the best estimates.
Guess what? Clever people have started asking "Antidepressants are amongst the biggest selling drugs in the world - but do they work?" And their answer is - not very well. The latest such claim came from Fournier et al and appeared in JAMA a couple of weeks ago: Antidepressant Drug Effects and Depression Severity.
These researchers re-analysed the data from six clinical trials testing antidepressants against placebo pills. The drugs were the tricyclic imipramine and the newer SSRI paroxetine. The total sample size was a respectable 718, and most trials lasted 8 weeks, which is longer than average for this kind of study. Here's what they found -
Grey circles are people on antidepressants, white circles people on placebo. What this shows is that the more severe the patient's depression, the more they get better - when they're given either drugs or placebos. However, because the improvement on antidepressants rises more steeply, the benefit of antidepressants versus placebos correlates with severity. The thin blue line marks the minimum severity for which the average effect of the drugs over placebo was "clinically significant" according to NICE criteria (although these are arbitrary).
Second, the patients in this analysis spanned a wide range of severity scores, from 10 points on the Hamilton Scale to nearly 40. In Kirsch et al almost all the trials had average severities in the narrow range of 22 to 29. Finally, none of the trials in the new paper used a placebo run-in period. These are meant to exclude people from the trial if they improve "too well" during an initial week or so of placebo pills. In theory, they bias trials against finding large placebo effects; it's not clear they actually work, but either way, it's good to know it wasn't a factor.
These researchers re-analysed the data from six clinical trials testing antidepressants against placebo pills. The drugs were the tricyclic imipramine and the newer SSRI paroxetine. The total sample size was a respectable 718, and most trials lasted 8 weeks, which is longer than average for this kind of study. Here's what they found -
Grey circles are people on antidepressants, white circles people on placebo. What this shows is that the more severe the patient's depression, the more they get better - when they're given either drugs or placebos. However, because the improvement on antidepressants rises more steeply, the benefit of antidepressants versus placebos correlates with severity. The thin blue line marks the minimum severity for which the average effect of the drugs over placebo was "clinically significant" according to NICE criteria (although these are arbitrary).*
So, this study says that antidepressants work better in more severe depression. This is not a new claim - Kirsch et al (2008) famously found the same thing, and long before that so did Khan et al (2002). However this new analysis has some advantages over previous ones. First, Fournier et al looked at what happened to each patient individually, whereas the previous studies found that in trials where the patients were more severely depressed, on average, antidepressants worked better.Second, the patients in this analysis spanned a wide range of severity scores, from 10 points on the Hamilton Scale to nearly 40. In Kirsch et al almost all the trials had average severities in the narrow range of 22 to 29. Finally, none of the trials in the new paper used a placebo run-in period. These are meant to exclude people from the trial if they improve "too well" during an initial week or so of placebo pills. In theory, they bias trials against finding large placebo effects; it's not clear they actually work, but either way, it's good to know it wasn't a factor.
*
Overall, the evidence all seems to point to the idea that people with more serious clinical depression respond better to antidepressants vs. placebos in clinical trials. The exact details are debatable, there's the issue of whether antidepressant clinical trials are realistic, and the question of how clinically effective antidepressants are is also controversial, but I'm not aware of any studies which have contradicted this central claim.
But when you start to think about it, this is a very odd result. Fournier et al say that
and refer to a couple of papers from the 1960s. But I must admit that I do find this very surprising. We don't wait until someone's nearly dead from a bacterial infection before we give them antibiotics, we give them early, when the disease is still mild. Doctors unfortunately don't tell people "Good news! You've got advanced-stage cancer - just the kind where drugs work best." Why is depression so different?
Look a little closer, and a possible answer emerges. Severity, in all of these studies, was measured using the Hamilton Rating Scale for Depression (HAMD). The HAMD has 17 items, and each asks whether you're suffering from certain symptoms; the more symptoms you have, and the more pronounced they are, the higher your total score. You get 1 point if you have "occasional difficulty falling asleep", 2 points for "nightly difficulty falling asleep", 4 points for "Hand wringing, nail biting, hair-pulling, biting of lips". Here's the whole thing.
The HAMD was designed in 1960 by a psychiatrist, Max Hamilton, and it was originally intended for use by staff at psychiatric hospitals for use on depressed inpatients. So it's not a measure of severity per se: it's a measure of how well your symptoms match those considered to be characteristic of severe depression in 1960.
Psychiatry's concept of depression - not to mention the wider culture's - has changed greatly since then. 1960 was a full 20 years before the DSM-III criteria of depression were published, which form the basis for today's DSM-IV criteria. A quick comparison of the DSM-IV alongside the HAMD reveals a lot of differences. It's quite possible to meet DSM-IV criteria for "Major Depressive Disorder" yet score low on the HAMD.
Which brings us back to the imaginary scenario at the start of this post. My personal interpretation of results like those of Fournier et al is this: antidepressants treat classical clinical depression, of the kind that psychiatrists in 1960 would have recognized. This is the kind of depression that they were originally used for, after all, because the first antidepressants arrived in 1953, and modern antidepressants like Prozac target the same neurotransmitter systems.
Yet in recent years "clinical depression" has become a much broader term. Many people attribute this to marketing on the part of pharmaceutical companies. Whatever the cause, it's almost certain that many people are now being prescribed antidepressants for emotional and personal issues which wouldn't have been considered medical illnesses until quite recently. (Antidepressants also have a long history of use for other conditions, like OCD, but this is a separate issue.)
My imaginary story used made up numbers: I'm not saying that only 10% of the people on antidepressants have "classic" depression. I don't know what the % is. But apart from that, in my opinion (and I don't think I'm alone), it's far from fantasy.
But when you start to think about it, this is a very odd result. Fournier et al say that
The general pattern of results reported in this work is not surprising. As early as the 1950s, researchers conducting controlled investigations of treatments for a wide variety of medical and psychiatric conditions described a phenomenon whereby patients with higher levels of severity showed greater differential (i.e., specific) benefit from the active treatments.
Look a little closer, and a possible answer emerges. Severity, in all of these studies, was measured using the Hamilton Rating Scale for Depression (HAMD). The HAMD has 17 items, and each asks whether you're suffering from certain symptoms; the more symptoms you have, and the more pronounced they are, the higher your total score. You get 1 point if you have "occasional difficulty falling asleep", 2 points for "nightly difficulty falling asleep", 4 points for "Hand wringing, nail biting, hair-pulling, biting of lips". Here's the whole thing.
The HAMD was designed in 1960 by a psychiatrist, Max Hamilton, and it was originally intended for use by staff at psychiatric hospitals for use on depressed inpatients. So it's not a measure of severity per se: it's a measure of how well your symptoms match those considered to be characteristic of severe depression in 1960.
Psychiatry's concept of depression - not to mention the wider culture's - has changed greatly since then. 1960 was a full 20 years before the DSM-III criteria of depression were published, which form the basis for today's DSM-IV criteria. A quick comparison of the DSM-IV alongside the HAMD reveals a lot of differences. It's quite possible to meet DSM-IV criteria for "Major Depressive Disorder" yet score low on the HAMD.
Which brings us back to the imaginary scenario at the start of this post. My personal interpretation of results like those of Fournier et al is this: antidepressants treat classical clinical depression, of the kind that psychiatrists in 1960 would have recognized. This is the kind of depression that they were originally used for, after all, because the first antidepressants arrived in 1953, and modern antidepressants like Prozac target the same neurotransmitter systems.
Yet in recent years "clinical depression" has become a much broader term. Many people attribute this to marketing on the part of pharmaceutical companies. Whatever the cause, it's almost certain that many people are now being prescribed antidepressants for emotional and personal issues which wouldn't have been considered medical illnesses until quite recently. (Antidepressants also have a long history of use for other conditions, like OCD, but this is a separate issue.)
My imaginary story used made up numbers: I'm not saying that only 10% of the people on antidepressants have "classic" depression. I don't know what the % is. But apart from that, in my opinion (and I don't think I'm alone), it's far from fantasy.
Fournier, J., DeRubeis, R., Hollon, S., Dimidjian, S., Amsterdam, J., Shelton, R., & Fawcett, J. (2010). Antidepressant Drug Effects and Depression Severity: A Patient-Level Meta-analysis JAMA: The Journal of the American Medical Association, 303 (1), 47-53 DOI: 10.1001/jama.2009.1943
