In a working paper released recently, economists looked at the efficacy of antidepressants in a novel manner — how do SSRI antidepressant sales correspond to suicide rates? Examining data from 26 countries over the past 25 years, they found that an increase in SSRI sales of 1 pill per capita is associated with a decline in suicide mortality of around 5 percent.
Their discussion pretty much sums up their findings better than I can:
Our estimates suggest that on balance SSRIs may be a very cost-effective means for saving lives, which is important in part because data from the National Comorbidity Survey from 2001-3 suggest that only around 40 percent of people with severe mental health disorders were receiving any treatment [Kessler, Demler et al., 2005]. Commonly used SSRIs can currently be obtained in the United States for around $0.10 per pill. Our estimates thus imply that each additional $20,000 spent on SSRIs will avert one suicide completion, far below the cost per life saved from most other public health, regulatory, or other forms of government intervention.
But using this estimate in a more formal benefit-cost analysis raises difficult conceptual and normative questions about the appropriate way to value the life of someone who subjectively prefers death (at least at the time of the intervention). If SSRIs reduce the risk of suicide by reducing access to a method of self harm, then the suicidal person may or may not experience a switch to SSRIs as a net benefit, depending on the transience or permanence of their state of pain. On the other hand if SSRIs reduce the risk of suicide by improving the subjective utility of life, then persons at risk for suicide and their loved ones may have a considerable willingness to pay for such an intervention. Of course SSRIs also generate other benefits beyond their net effects on mortality that should be counted in this calculus, including improvements in mood, health status, functioning, and productivity.
Of course their study can’t answer questions about clinical efficacy and suicide, but they do provide an interesting public health view of this issue — that, in general, SSRI antidepressants decrease suicide in the population.
Hat Tip: Greg Mankiw’s Blog: Anti-depressants and Suicide
Working Paper: Anti-depressants and Suicide (available in PDF)
(3 votes, average: 3.67 out of 5)
Loading ...You can leave a response, or trackback from your own site.
9 Comments to
“Economists Examine the Link Between Antidepressants and Suicide”
Oh come on. You will be as aware as anyone how this data has been spun, an for whom.
A nice example of such spinning of so called sales data relationships here:
http://ahrp.blogspot.com/2007/02/pharma-spin-and-child-suicide-rates.html
After getting through all the Roche et al., adverts to get to this point of commenting I think we have to wonder what this blog is all about.
You might do better to comment on the fiddling of the suicide data in the actual trials - that would enhance your credibility.
Mark Redhead
You might do better if you read the study the blog entry was commenting on instead of making wild conjectures based upon your own bias or opinion.
I have read it. Studies of this sort are by and large nonsense. You could correlate suicide rates with global warming, number of hamburgers eaten - it means nothing at all.
More importantly - have you make any comment on your blog about cheating in SSRI drug trials. If not, I would have to query your motives. And if you feel inclined to promote nonsensical reserch you should balance it by discussing the similar nonsensical research where the conclusions are spun (see last reference I gave).
I note that a large part of the aim of this website seems to be to recruit patients into tials.
I’m sure the economists duly note your opinion.
You’d have to actually do a search here in order to answer your own question.
Yes, we have a section (one out of dozens) where we encourage people to consider enrolling in a research trial. Without research trials and participants (these trials include *all types* of treatments, including non-pharmaceutical treatments), how would researchers do future research on mental health concerns and their treatment?
I absolutely encourage people to consider enrolling in a research study if one is available for their concern. This is the main way we gain new knowledge within the field.
Except where there results are dodgy.
Knowledge and psychiatry are increasingly uneasy bedfellows.
I tried the search
grohol suicide
grohol gsk
grohol glaxo
gsk suicide
glaxo suicide
all yielded results of 0
I’m a little confused mister redhead. This is a paper written for the NBER, one the most prestigious centers of economic research in the country. This paper may be wrong, or misguided, but it isn’t spin. Spin gets published in fringe think tanks, not the NBER (which isn’t to say that the NBER is the word of god or even close to it).
Interesting discussion. From a friend’s FDA Testimony (Dr. Irving Kirsch and Dr. David Antonuccio)…
There are a total of 12 published randomized clinical trials in the entire world literature (see studies marked with an * in the reference list). Eight of these 12 trials failed to find any significant benefit of medication over inert placebo. Only 4 of the RCTs claimed significant differences between drug and placebo, and these did so only on clinician rated measures, not patient rated measures.
Of the 12 published randomized trials, 4 assessed SSRIs, 7 assessed tricyclics, and one assessed both SSRIs and tricyclics. Four of the five SSRI-placebo comparisons indicated significant differences. None of the TCA-placebo comparisons showed significant differences.
Three of the clinical trials did not report means and/or standard deviations, leaving 9 for meta-analysis. When these nine studies are combined, the placebo response is 87% of the drug response. It is 75% of the SSRI response and 97% of the tricyclic response.
Thus, the meta-analysis indicates that tricyclics have no significant pharmacological effect on depression in children. The effect of SSRIs is statistically significant, but it is not clinically significant. Overall, the effects of antidepressant medication are weaker in children than in adults (cf. Kirsch & Sapirstein, 1997; Kirsch et al. 2002). These conclusions are consistent with those found in all 7 prior reviews of the effects of antidepressants in depressed children (Ambrosini et al., 1992; Dujovne et al., 1995; Fisher & Fisher, 1996; Hazell et al., 1995; Kastelic et al., 2000; Michael & Crowley, 2002; Sommers-Flanagan & Sommers-Flanagan, 1996).
These results were drawn from studies with design flaws that typically favor the study drug. For example, they frequently exclude placebo responders before random assignment, rely on ratings by clinician’s who have a vested interest in the outcome, and are likely to be unblinded by medication side effects (Antonuccio et al., 1999; Antonuccio et al. 2002). Furthermore, these results are drawn from the published literature, which is subject to publication bias and ?file drawer? problems, meaning that many studies with negative results do not to get published. Adding unpublished studies, most of which have negative results, will surely shrink the difference between antidepressants and placebo even further.
Likewise,
Kirsch, I., The Emperor’s New Drugs: An Analysis of Antidepressant Medicatin Data, In Prevention & Treatment, Volume 5, Article 23, posted July 15, 2002: Abstract:This article reports an analysis of the efficacy data submitted to the U.S. Food and Drug Administration for approval of the 6 most widely prescribed antidepressants approved between 1987 and 1999. Approximately 80% of the response to medication was duplicated in placebo control groups
Finally, remember that RATES of antidepressant prescription do not equal NUMBERS of PATIENTS TREATED. In other words, you might have 100 prescriptions that go to Mr. Jones in 2006, and 50 prescriptions go to Mr. Smith.
Mr. Smith commits suicide in 2006, but Mr. Jones does not.
The “number of prescriptions” will make it look like the drugs are far safer than they really are, because the SURVIVING patients will simply OBSCURE the data from those patients who commit suicide.
Remember the placebo controlled trials on antidepressants and kids trump all these population studies showing correlational results. Correlational results do not prove causation while controlled studies do. The same thing happened with hormone replacement therapy (i.e. the populaiton studies said it was wonderful, the controlled trials proved it was the opposite.)
Also, the controlled trials show there is a relative risk problem (i.e., the antidepressants cause 2 EXTRA suicidal kids for every 100 treated COMPARED to every 100 treated by placebo- 4 suicidal kids for with an antidepressant relative to 2 suicidal kids treated by placebo). So it is possible that treating with an antidepressant might have a lower relative risk compared with no treatment at all, but nobody is suggesting that! But we don’t know for sure because there has never been a trial of depressed kids comparing an antidepressant with no treatment at all. It could be possible that the attention of a prescriber/professional might lower risk in kids who get these prescriptions. If that is the case, of course it would not necessarily imply it has anything to do with the chemical. What the controlled trials show is that the chemical results in a higher relative risk of suicidality than placebo.
Major confounds to the interpretation. First, high prescription rates may be a reflection of the fact that parents in a particular area are more attentive to the health and well-being of their children which would seem to be an innoculation against suicide.
Second, it may be true that the people who live in areas with higher prescription rates have higher incomes than those in counties with lower prescription rates (i.e. an inverse association between income and “mental health” and suicide. other factors could be education level and divorce rates, for example?
Did the researchers separate the prescription rates and suicide rates by age group or just assume that higher prescription rates overall and
higher suicide rates overall implies higher rates for children and adolescents?
What about the effect of dosages? Was 10 mg daily of an SSRI treated the same as 60 mg?
In places with a large retired population, such as parts of Arizona and Florida, there will be more prescriptions per capita, and less children
per capita. The elderly have a lower than average suicide rate.
Suicide rates were also highest in less densely populated parts of the western United States, areas that often include Indian reservations. Suicide rates were lowest in large metropolitan areas such as Chicago, New York, Boston, Los Angeles, San Diego, Seattle and Miami.
All of these varibles can be manipulated and are routinely to support whatever agenda or cause the researcher has.
Warmly,
Dr. Watson
According to Dr. Irving Kirsch in Prevention & Treatment, “there is now unanimous agreement that the mean difference between response to SSRI antidepressant drugs and response to inert placebo is very small. It is so small that, despite sample sizes involving hundreds of participants, 57% of the SSRI trials funded by the pharmaceutical industry failed to show a significant difference between drug and placebo. Most of these negative data were not published and were accessible only by gaining access to US Food and Drug Administration (FDA) documents.
Various methods were used to manipulate the results of SSRI drug studies to insure a favorable outcome:
1) Responders to the placebo are eliminated at the beginning of the study. (Placebo washout)
2) Benzodiazepine sedatives were given to mask the SSRI induced agitation.
3) Unfavorable drug studies are buried in the file cabinet and not disclosed to the public.
4) Miscoding suicidal events as “emotional lability”, and homicidal events as “aggression” to hide suicidal events from regulators.
5) False attribution of suicide to the placebo arm.
6) Hiring ghost writers to make the medical articles more favorable.
7) Cash settlements for SSRI drug litigants which seals records and withholds unfavorable drug studies from the public.
For more information and links see my Paxil, Prozac, and SSRI Induced Suicide Newsletter
Join the Conversation! Post a Comment:
Last reviewed: By John M. Grohol, Psy.D. on 22 Feb 2007
