.:[Double Click To][Close]:.

Saturday, May 23, 2009

Do Antidepressants Help in Mild Depression?

Yes! says the BBC, reporting on the results of a new trial -
Drugs 'can help mild depression'
Not so fast. Read this before you reach for the Prozac.
It was about this time last year that Irving Kirsch and colleagues released Initial Severity and Antidepressant Benefits. This bombshell of a meta-analysis concluded, notoriously, that the benefits of antidepressants over and above placebo are in general pretty small. Moreover, it claimed that the benefits are even smaller - indeed pretty much zero - in people whose depression is not very severe to begin with.

However, Neuroskeptic readers will know that antidepressant trials are not all they're cracked up to be (1,2). On top of which Kirsch et al. were a little "creative" with their statistics, as bloggers P J Leonard and Robert Waldmann aptly demonstrated. So, the claim that antidepressants don't work in mild depression rests on shaky foundations.

But that doesn't mean that they do work. In fact, there have been very few studies looking at the effectiveness of drugs in mild to moderate depression. That's a shame, because mild depression is the most common reason why people are given antidepressants in real life.

Now a new clinical trial, run by the British National Health Service, has appeared. It was (drumroll) a Randomised controlled trial to determine the clinical effectiveness and cost-effectiveness of selective serotonin reuptake inhibitors plus supportive care, versus supportive care alone, for mild to moderate depression with somatic symptoms in primary care.

The researchers enlisted GPs (family doctors) from across the UK, and got them to refer suitable patients to the study. Patients could be included if their doctors considered that they were depressed and had been for at least 8 weeks. They also had to be aged 18 or over, and they had to be rated between 12 and 19 on the HAMD, a scale used to measure the severity of depression. (Slightly oddly, they were also required to show at least some evidence of "somatic" symptoms - aches, pains, indigestion, that kind of thing. I'm not sure why.) Patients were excluded if they "expressed suicidal intent" or if they admitted to drug or alcohol misuse.

A total of 602 patients were referred to the trial, but of these only 220 actually took part; the rest either didn't want to do it or were unsuitable for whatever reason. It took the researchers nearly 4 years and heroic efforts to recruit those 220 people, including reimbursing doctors £45 for each patient referred. This kind of research is frustrating. This is probably why there's so little of it.The volunteers were randomly assigned to get supportive care alone, or supportive care plus the doctor's choice of SSRI antidepressant. "Supportive care" is basically a euphemism for "doing sweet F. A.". The GPs were meant to see the patients 5 times over a 12 week period; given that a typical GP consultation in the UK lasts about 10 minutes, the idea that this constitutes any kind of "care", supportive or not, is a bit of a joke.

What happened? Well, to cut a very long story short, the patients assigned to SSRIs did better than the ones assigned to supportive care alone. Hurrah! But they only did slightly better. After 12 weeks they had a mean HAMD score of 8.7 compared to 11.2 in the supportive care group. The SSRI group also did a bit better on some other measures of health, well-being and general satisfaction. The difference on the BDI, a self-reported measure of depression, was not significant however (13.0 vs. 15.1)

So does that mean antidepressants "work" in mild depression? Maybe. Maybe not. The most obvious issue, of course, is that there was no placebo group in this trial. So any benefit of the pills could have just been psychological. Gettingly randomly assigned to "supportive care" and condemned to twiddle your thumbs for 12 weeks is not going to make anyone feel better. Starting on antidepressants, on the other hand, feels like a fresh start. It gives hope. It's change you can believe in.

But if giving people pills makes them feel better, isn't that good enough reason to do it? Who cares if it's all the placebo effect? Well, there's some truth to that, but the problem is that patients included in this trial were a rather unusual bunch. In particular, they were people who agreed to be randomized to get antidepressants or not, i.e. they had no strong preference either for or against pills.

Given that an awful lot of people do have such a preference, we can't assume that these results apply to the average patient in the clinic. As the authors note (page 59, emphasis mine):
The tallies of surgery logs completed by a number of the study GPs at various points during the study showed that only around 1 in 10 patients with a new episode of depression were referred into the study, mainly because the rest did not fulfil the inclusion criteria, particularly in terms of a lack of equipoise about the benefits of drug treatment on the part of the doctor or patient or both.
And of those 602 referred, only about a third actually took part, as mentioned above. So what we have here is a study on an unusual 3% of patients. What about the other 97%? We don't know. Still.

Or don't we? Well, it depends who "we" are. I suspect that a moderately competent doctor with experience treating depression probably does have a good idea of who is likely to benefit from drugs and who isn't. There's no substitute for real, hands-on clinical experience. There's more to life than trials...

ResearchBlogging.orgT Kendrick, J Chatwin, C Dowrick, A Tylee, R Morriss, R Peveler, M Leese, P McCrone, T Harris, M Moore, R Byng, G Brown, S Barthel, H Mander, A Ring, V Kelly, V Wallace, M Gabbay, T Craig and A Mann (2009). Randomised controlled trial to determine the clinical effectiveness and cost-effectiveness of selective serotonin reuptake inhibitors plus supportive care, versus supportive care alone, for mild to moderate depression with somatic symptoms in primary care Health Technology Assessment, 13 (22)

No comments:

Post a Comment