Rapid Responses to:

RESEARCH: Marc Stone, Thomas Laughren, M Lisa Jones, Mark Levenson, P Chris Holland, Alice Hughes, Tarek A Hammad, Robert Temple, and George Rochester Risk of suicidality in clinical trials of antidepressants in adults: analysis of proprietary data submitted to US Food and Drug Administration BMJ 2009; 339: b2880 [Abstract] [Full text]

Rapid Responses: Submit a response to this article

Rapid Responses published:

Suicidal thoughts and actions versus completed suicide

Paula J Clayton, MD, New York, NY 10005   (20 August 2009)

depressants and suicide

Billy Levin   (20 August 2009)

Age and anti- depressants-the statistical dilemma

Zekria Ibrahimi   (21 August 2009)

Risk of Suicidality (Sic)

Derek HALL   (22 August 2009)

Beware of pharmaceutical company bias.

Sumeet Gupta   (24 August 2009)

Not the last word

David Healy   (26 August 2009)

Author's Reply

Marc B Stone, Thomas Laughren, Tarek A Hammad, P Chris Holland   (27 August 2009)

Author's reply

Marc B Stone, Thomas Laughren, Tarek A Hammad, P Chris Holland   (27 August 2009)

Not the last word

David Healy   (28 August 2009)

The safety issue - what about total mortality ?

Nir Tsabar   (31 August 2009)

Authors' reply

Marc B Stone   (2 September 2009)

Comment on “Risk of suicidality in clinical trials of antidepressants in adults: analysis of proprietary data submitted to US Food and Drug Administration”

Robert D. Gibbons, Sharon-Lise T. Normand - Harvard University, Joel B. Greenhouse - Carnegie Mellon University   (22 September 2009)

Suicide "Attempts"--So What?

Donald J. Farber   (26 September 2009)

Suicidal thoughts and actions versus completed suicide 20 August 2009
Paula J Clayton, MD, Medical Director, American Foundation for Suicide Prevention 120 Wall Street, 22nd Floor, New York, NY 10005 Send response to journal: Re: Suicidal thoughts and actions versus completed suicide	Dear Sir, It was surprising to see the recent article by Stone et al (1) in the August issue of the British Medical Journal. As the editorial by Geddes (2) so nicely questions, why, since the data have not been updated since the initial publication more than two years ago and the present report only selectively reports the fuller analysis, was it published in the BMJ now? I am concerned that the report and discussion of the selective data in this publication are misleading. While the authors do define the word suicidality as “suicidal thoughts and actions,” it is not a word that, to my knowledge, is defined in the dictionary. Furthermore, the manuscript fails to define for the reader the distinction between completed suicide and suicidal thoughts and attempts. Even when those who are seen in the emergency room for a suicide attempt are followed for the next 20 years, only five to ten percent of them actually die by suicide (3). Contrary to what this article implies, suicide attempts are not synonymous with completed suicide. In support of the authors, they do state on page 7 of 10 that, “this study can do little to resolve whether antidepressants affect the risk of death by suicide, even in a population of tens of thousands, there was only a handful of cases.” A major problem with this report is that there were eight completed suicides in the whole study of about 100,000, but the authors never tell us how the number is divided between the drug- and placebo-treated groups. It is fair to conclude that in their meta-analysis of 372 double-blind randomized short term placebo-controlled trials of antidepressants, the FDA showed no evidence of an increase in suicide in any age group. Their analysis shows a slight, but significant, increase in suicidal behavior or ideation among adolescents and young adults treated with antidepressants compared to those treated with a placebo. The reader should be told, however, that such attempts are especially common among this age group. For example, according to the 2006 Centers for Disease Control (CDC) data, self-harm reports for those 15 to 19 years old are 323/100000, while the suicide rate in this age group is 7.3/100000. This is in contrast to adults 25 and older, for whom the self-harm rates are 148/100000 and the suicide rate is 14.75. In other words, suicide attempts are more common among teens, but teens’ completed suicide rate is half that of adults’ –again, the relationship between the two is unclear. It is certainly important for clinicians and caretakers to be aware of this safety issue when antidepressants are prescribed. And it is essential for both groups to closely monitor such treatment in young people, especially at the beginning of treatment. It is also a good idea to implement other safety planning measures. However, it is equally as important that clinicians and caretakers not be frightened of the use of such medications when they are appropriately indicated. Depression is clearly an illness for which appropriate treatment with antidepressants is not only effective in resolving the condition, but may in fact be lifesaving. It is most important, however, that these data be reported in an informative, clear and balanced way in published research articles. I notice that the Editorial by Geddes, Barbui and Cipriani was commissioned by the BMJ. It may be worthy of the BMJ to consider commissioning a paper by the noted statistician Joel Greenhouse of the University of Pittsburgh, who has recently examined the meta-analysis of treatment data underlying the black box labeling of antidepressants. Such a paper would be a service to all. Thank you for considering the concerns I express to you in this letter. Sincerely, Paula J Clayton MD Medical Director, American Foundation for Suicide Prevention New York, New York References 1. Stone MB, Laughren T, Jones ML, Levenson M, Holland PC, Hughes A, et al. Risk of suicidality in clinical trials of antidepressants in adults: analysis of proprietary data submitted to US Food and Drug Administration. BMJ 2009; 339:b2880. 2. Geddes JR, Barbui C, Cipriani A. Editorial : Risk of suicidal behaviour in adults taking antidepressants. BMJ 2009;339:b3066. 3. Jenkins et al. Suicide rate 22 years after parasuicide: cohort study. BMJ 2002;325(7373):1155. Competing interests: None declared
depressants and suicide 20 August 2009
Billy Levin, med practice special interest field ADHD Box 14201, Farrarmere ,Benono S. Africa Send response to journal: Re: depressants and suicide	It is well known that many depressed patient on antidepressants claim they are better but very few claim they are fine.Very few doctors monitor the doses with a rating scale(Hamilton). If they did it would be obvious how few are on optimal treatment. These patients on an antidepressant who are not responding still have their depression fostering suicide concepts.It is not the antidepressant to be blamed. Many teenageers and young adults with non diagnosed ADHD become depressed. The condition has a built in "suicider!" Hasty impulsive emotional doing or saying without thinking.THey would perhaps have their depression recognised but not the cause of their depression which is ADHD. They would not respond to antidepressants alone. Giving them antidepressants and stimulants at the same time ,but in optimal doses, would give sometimes a dramatic positive responce! Optimal treatment of both conditions is essential to reduce the suicise tendency. Do not blame the innocent antidepressants which reduce suicide. When doctors accepted the FDA's black box warning and stopped using antidepressants the teenage suicise escalated exponentially! Competing interests: None declared
Age and anti- depressants-the statistical dilemma 21 August 2009
Zekria Ibrahimi, psychiatric patient Coombs Library, Southall, UB1 3EU Send response to journal: Re: Age and anti- depressants-the statistical dilemma	The controversial thrust of this article (1) is that the young are more prone to suicide ideation than the old as a result of anti- depressant prescription. Any statistical approach, however sophisticated it may seem, is fundamentally handicapped; it can point (always tentatively, given the lurking confounders) to a connection, but not to the actual medical cause. At the level of neuron and axon and neurotransmitter, why are anti-depressants able to disturb, not soothe, mood? Table 2 has an odds ratio for the under 25s that just misses Fisherian significance at 0.07, but the risk difference is just within the Fisherian threshold at 0.03. The statistical point is that the 5% significance level is only a guideline, and to stick to it strictly without medical interpretation is a blind way of proceeding. David Healey observed how there have been signals linking antidepressant to suicidality, even if the signals did not have precise statistical significance(2). The forest plot here (fig.2) is unusual in that it is not back transformed logarithmically, so that the odds ratio confidence intervals are skewed, not symmetrical. In my always inadequate view, there is an element of bad practice in not having a consistent x- axis (perhaps a log one of 0.1 to 10) for all forest plots in evidence based medicine. Different sorts of x- axis produce visual confusion when we try with difficulty to compare effect sizes across forest plots. Being young and taking anti- depressants can be dangerous. But why? The statisticians must now pass the baton on to the doctors and medical researchers- who have so far failed to find the cause in the brain for this age- related anomaly. As an older person, who is considerably grey and wrinkled, I am puzzled. Surely anti- depressants should be just as dangerous for us elderly ones? REFERENCES: (1) Risk of suicidality in clinical trials of antidepressants in adults: analysis of proprietary data submitted to US Food and Drug Administration. Marc Stone et al. BMJ 2009;339:b2880. (2) The antidepressant tale: figures signifying nothing? David Healy. Advances in Psychiatric Treatment (2006), vol. 12,320-328. Competing interests: None declared
Risk of Suicidality (Sic) 22 August 2009
Derek HALL, Retired Military Consultant Surgeon TS28 5HZ Send response to journal: Re: Risk of Suicidality (Sic)	If you mean "Risk of Suicide", please say so..... Leave Gerundised Nouns where they belong - in the Trash (Rubbish) Can (Bin). Have a Nice Day (Thank you). Good Job, man! (Well Done!). I feel like Suicidalising myself, know what I mean? Competing interests: None declared
Beware of pharmaceutical company bias. 24 August 2009
Sumeet Gupta, Consultant psychiatrist Hundens Lane Resource Centre,Darlington,DL3 8RR Send response to journal: Re: Beware of pharmaceutical company bias.	The meta-analysis by Stone et al further confirms the fact that antidepressant drugs are associated with increased suicidality in young persons(1).The concerns have been raised since 1990 but drug companies managed to keep the issue under the carpet until 2003 by selectively publishing the only positive data. Even when they did publish the data they tried to spin the conclusions in their favour (2, 3).Paroxetine (Seroxat), manufactured by GlaxoSmithKline dominated the headlines in 2003/04 as they did not reveal the data showing increased suicidal risks in children. In 2008 Turner et al compared the published trials of 12 antidepressant drugs submitted to FDA. They found that on one hand majority of the positive studies were published whilst on the other hand majority of the negative trials were not published. The above discrepancy led to overestimation of the effect size (4). This paper once again emphasised the importance of availability of all trials data irrespective of the outcomes. The negative, failed or unpublished trials are also as important as published, positive trials .Lastly; we should not take trials, done by pharmaceutical companies, at their face value. 1. Stone MB, Laughren T, Jones ML, Levenson M, and Holland PC, Hughes A, et al. Risk of suicidality in clinical trials of antidepressants in adults: analysis of proprietary data submitted to US Food and Drug Administration. BMJ 2009; 339:b2880. 2. Teicher MH, Glod C, Cole JO.Selective Publication of Antidepressant Trials and Its Influence on Apparent Efficacy. Am J Psychiatry. 1990 Feb;147(2):207-10 3. Healy D. The antidepressant tale: figures signifying nothing? Advances in Psychiatric Treatment (2006) 12: 320-327 4.Turner EH, Matthews AM, Linardatos E, Tell RA, Rosenthal R. Selective publication of antidepressant trials and its influence on apparent efficacy N Engl J Med. 2008 Jan 17;358(3):252-6 Competing interests: None declared
Not the last word 26 August 2009
David Healy, Professor of Psychiatry Cardiff University LL57 2PW Send response to journal: Re: Not the last word	Dear Sirs The analysis provided by Dr Stone and colleagues in the BMJ (1) is welcome but there are inconsistencies between this dataset and others. Taking the data for sertraline in the Stone et al paper laid out in Table 1, we can see there is a marked difference between this and the data submitted to the UK regulator 2 years earlier (2). While there is some scope for differences in suicidal acts based on the coding system specified by the regulator, it is difficult to account for differences in completed suicides this way. It can also be noted that the data provided by Stone and colleagues for suicides and suicidal acts on sertraline differs markedly from the data for all other drugs listed (1). Had the data submitted to the UK regulator, which was more in line with the data for other drugs listed in the Stone article, been used, the conclusions drawn might have been quite different. A further dataset has recently been published by Pfizer (3). This is laid out in Table 3 and again shows a substantial increase in the risk of suicide or suicidal acts on sertraline. But of further interest is the data for the elderly provided in this publication (Table 4), which is at odds with the data for the elderly in the Stone et al article. In this most recent dataset the risk for the elderly is in fact greater than for other age groups. It would accordingly perhaps be a mistake to think that the Stone et al article provides the last word on the influence of age on the risks posed by antidepressants. David Healy MD FRCPsych Professor of Psychiatry, Cardiff University. 1. Stone M, Laughren T, Jones ML, Levenson M, Holland PC, Hughes A, Hammad TA, Temple R, Rochester G (2009). Risk of suicidality in clinical trials of antidepressants in adults: analysis of proprietary data submitted to US Food and Drug Administration. BMJ doi 10,1136/bmj/h2880. 2. Expert Working Group on the Safety of Selective Serotonin Reuptake Antidepressants 2004. Medical & Healthcare Products Regulatory Agency, London. Available from: www.mhra.gov.uk/home/idcplg?IdcService=GET_FILE&dID=1391&noSaveAs=1&Rendition=WEB. 3. Vanderburg DG, Batzar E, Fogel I, Kremer CME (2009). A pooled analysis of suicidality in double-blind, placebo-controlled studies of sertraline in adults. J Clin Psychiatry 70, 674-683. Competing interests: DH has been an expert witness in legal cases involving SSRI antidepressants and suicide, commonly on the side of the plaintiff/claimant.
Author's Reply 27 August 2009
Marc B Stone, Medical Officer U.S. food and Drug Administration, Silver Spring, Maryland 20993, Thomas Laughren, Tarek A Hammad, P Chris Holland Send response to journal: Re: Author's Reply	Dr. Clayton cites Dr. Geddes’s question as to why the article was published but overlooks his answer: “Its objective is to make a summary of these important results more widely available in a way similar to the publication in the BMJ of summaries of Cochrane reviews.” The original report was neither concise, subject to external peer review nor available in standard searches of the medical literature. Our BMJ paper also presents numerous analyses not contained in the original report: risk differences in addition to odds ratios as well as presenting odds ratios for suicidal ideation or behavior, suicidal ideation alone and suicidal behavior alone treating age as a continuous variable. In regards to the use of the term “suicidality”, a search using Google Scholar identifies the term in approximately 53 000 scholarly articles. In Dr. Clayton’s own book, The Medical Basis of Psychiatry, the word appears 23 times, with one chapter identifying it as a “Keyword”. As Dr. Clayton notes, we were quite clear about our definition. Regarding the connection between suicide attempts and completed suicide, we clearly acknowledge this as a limitation of the analysis. We considered the question of whether antidepressants could raise the risk of completed suicide to be unanswerable with our data; our principal interest was in the effect of the drugs on suicidal thoughts and behavior, important concerns in and of themselves. Nevertheless, one has to be concerned there is some link between suicidal behaviors and completed suicide. As Dr. Clayton points out, suicide attempts are common in young people; a small increase in risk would result in a substantial increase in absolute numbers of suicide attempts. Should we be reassured that “only 5 to 10%” of those who are seen in ERs for a suicide attempt eventually kill themselves? Furthermore, there is no compelling reason to believe that suicide attempts occurring as an adverse reaction to antidepressant drugs should have the same low level of lethality as other attempts. As for her complaint that we do not state how many of the eight completed suicides in the study occurred in the antidepressant group, although the numbers are too small for a meaningful comparison, we do report the odds ratio (2.13) and the number of subjects in each group. Simple arithmetic would allow an interested reader to infer that at least six of the completed suicides occurred in the antidepressant group. Regarding the finding that suicidal behavior is more common in adolescents than in adults, even though completed suicides are less common, this is interesting, but not particularly relevant. We are looking at a signal based on drug-placebo differences within different age strata. In reference to her concerns about balance, we would note the language of the boxed warning: “must balance this risk of suicidality with the clinical need. Depression and certain other psychiatric disorders are themselves associated with increases in the risk of suicide.” In addition to discussing epidemiological data that relate a decline in suicide rates with increasing use of antidepressants, we also state in the last paragraph of our paper that “…depression is a serious illness that itself is a strong predictor of suicide.” We do not question the benefits of antidepressant treatment but that is a topic outside the scope of our paper. Competing interests: None declared
Author's reply 27 August 2009
Marc B Stone, Medical Officer U.S. Food and Drug Administration, Silver Spring, Maryland 20993, Thomas Laughren, Tarek A Hammad, P Chris Holland Send response to journal: Re: Author's reply	Dr. Levin suggests that the occurrence of suicidal thinking or behavior in subjects taking antidepressants is due to inadequate treatment effect. This does not explain why the rate would be higher in the drug- treated group than in the placebo group where all subjects received “inadequate” treatment (i.e., no drug treatment at all). It is conceivable that incomplete treatment of depression may cause some patients to become actively suicidal. We would note, however, that the strongest relative increase in suicidal behavior associated with antidepressant use found in our study was in subjects who had conditions other than depression. Dr. Levin’s comments about ADHD and depression are interesting. It is true that a substantial fraction of younger patients with ADHD may, at some time, have comorbid depression. Exactly how to treat both conditions simultaneously is a reasonable question, but is beyond the scope of our paper. Dr. Levin’s statement that “When doctors accepted the FDA's black box warning and stopped using antidepressants the teenage suicise [sic] escalated exponentially!” is not supported by the data. The rise in suicide rates in adolescents that occurred in 2004 preceded the warning which was issued at the very end of 2004 and did not appear in most labeling until 2005. Antidepressant use by children and adolescents in 2004 was essentially unchanged from 2003, and so could not explain the 2004 rise in suicide rates. Use did decline for children and adolescents in 2005 and later years, perhaps in response to the warning. According to the Centers for Disease Control and Prevention (National Vital Statistics Reports, Vol. 57, No. 14, April 17, 2009), there was a decline in suicide rates in 2005 relative to 2004 in the age 15-24 age group and a further decline in 2006 to levels equaling the levels seen in 2001 and 2002. Competing interests: None declared
Not the last word 28 August 2009
David Healy, Professor of Psychiatry Cardiff University LL57 2PW Send response to journal: Re: Not the last word	Dear Sirs The analysis provided by Dr Stone and colleagues in the BMJ (1) is welcome but there are inconsistencies between this dataset and others. Taking the data for sertraline in the Stone et al paper laid out in Table 1, we can see there is a marked difference between this and the data submitted to the UK regulator 2 years earlier (2). While there is some scope for differences in suicidal acts based on the coding system specified by the regulator, it is difficult to account for differences in completed suicides this way. It can also be noted that the data provided by Stone and colleagues for suicides and suicidal acts on sertraline differs markedly from the data for all other drugs listed (1). Had the data submitted to the UK regulator, which was more in line with the data for other drugs listed in the Stone article, been used, the conclusions drawn might have been quite different. Table 1: Sertraline-Placebo Controlled Trial Data submitted to FDA 2006 Table 2: Sertraline-Placebo Controlled Trial Data submitted to MHRA 2004 A further dataset has recently been published by Pfizer (3). This is laid out in Table 3 and again shows a substantial increase in the risk of suicide or suicidal acts on sertraline. But of further interest is the data for the elderly provided in this publication (Table 4), which is at odds with the data for the elderly in the Stone et al article. In this most recent dataset the risk for the elderly is in fact greater than for other age groups. Table 3: Sertraline-Placebo Controlled Trial Data 2009 Table 4: Sertraline-Placebo Controlled Trial Data for Elderly 2009 It would accordingly perhaps be a mistake to think that the Stone et al article provides the last word on the influence of age on the risks posed by antidepressants. David Healy MD FRCPsych Professor of Psychiatry, Cardiff University. 1. Stone M, Laughren T, Jones ML, Levenson M, Holland PC, Hughes A, Hammad TA, Temple R, Rochester G (2009). Risk of suicidality in clinical trials of antidepressants in adults: analysis of proprietary data submitted to US Food and Drug Administration. BMJ doi 10,1136/bmj/h2880. 2. Expert Working Group on the Safety of Selective Serotonin Reuptake Antidepressants 2004. Medical & Healthcare Products Regulatory Agency, London. Available from: www.mhra.gov.uk/home/idcplg?IdcService=GET_FILE&dID=1391&noSaveAs=1&Rendition=WEB. 3. Vanderburg DG, Batzar E, Fogel I, Kremer CME (2009). A pooled analysis of suicidality in double-blind, placebo-controlled studies of sertraline in adults. J Clin Psychiatry 70, 674-683. Competing interests: DH has been an expert witness in legal cases involving SSRI antidepressants and suicide, commonly on the side of the plaintiff/claimant.
The safety issue - what about total mortality ? 31 August 2009
Nir Tsabar, Geriatric Intern, Pardes-Hana Pardes-Hana, Israel Send response to journal: Re: The safety issue - what about total mortality ?	Since death was adjudicated only by the sponsors, and since all death cases are relevant if the main issue is safety, I believe a report of total mortality risk should be mandatory. Sincerely, Nir Tsabar, MD/PhD Competing interests: None declared
Authors' reply 2 September 2009
Marc B Stone, Medical Officer U.S. Food and Drug Administration, Silver Spring, Maryland 20993 Send response to journal: Re: Authors' reply	We agree with Dr. Tsabar. To acquire some additional assurance that failure to identify deaths as suicide did not affect the results, we required sponsors to identify all subjects who died from any cause within 90 days of beginning study treatment. We found no anomalies. Competing interests: None declared
Comment on “Risk of suicidality in clinical trials of antidepressants in adults: analysis of proprietary data submitted to US Food and Drug Administration” 22 September 2009
Robert D. Gibbons, Professor and Director of the Center for Health Statistics University of Illinois at Chicago, 1601 W. Taylor, Chicago IL 60614, Sharon-Lise T. Normand - Harvard University, Joel B. Greenhouse - Carnegie Mellon University Send response to journal: Re: Comment on “Risk of suicidality in clinical trials of antidepressants in adults: analysis of proprietary data submitted to US Food and Drug Administration”	In their recent paper (1) Stone and colleagues report on their previous meta-analysis of 372 double blind placebo randomized controlled trials (RCTs) of antidepressant medication and suicidality. While the overall effect of antidepressant medication on the primary endpoint of suicidal ideation or behavior was protective (OR=0.85, CI=0.71-1.02, p<0.08), a secondary analysis based on age stratification revealed protective effects for participants 25 years of age and older (OR=0.74, CI=0.60-0.90, p<0.003), and in the direction of increased risk for participants under 25 years of age (OR=1.62, CI=0.97-2.71, p<0.07). On the basis of this finding, further analysis of suicidal behavior alone, and their previous meta-analysis of pediatric studies (2) they conclude that “risk of suicidality associated with use of antidepressants is strongly age dependent.” As further support of this conclusion they refer to case-control studies (3,4) which found increased risk of suicidality in patients aged 18 and younger who were treated with antidepressants, but note that these studies were subject to confounding based on differential severity of illness between cases (i.e., patients with suicidality) and controls. They also note that ecological studies have found inverse associations between antidepressant treatment and completed suicide, but are limited by the absence of person-level data. In discussing the results of their study, Stone and colleagues overlooked the largest person-level study that has ever been conducted in this area (5), which examined the association between antidepressant treatment and suicide attempts in 226,866 U.S. veterans with new episodes of major depressive disorder. This study found an overall decreased risk of suicide attempts in patients receiving selective serotonin reuptake inhibitors (SSRI) monotherapy relative to those patients who did not receive antidepressant treatment (OR=0.37, CI=0.29-0.47, p<0.0001). When age stratified, no evidence of a relationship between differential risk of suicide attempt with treatment and age was observed (ages 18-25 OR=0.35, CI=0.14-0.85, p<0.021; ages 26-45 OR=0.44, CI=0.29-0.65, p<0.0001; ages 45-65 OR=0.42, CI=0.30-0.59, p<0.0001; ages >65 OR=0.38, CI=0.16-0.91, p<0.036). Similar results were obtained for non -SSRI and tricyclic antidepressants and also for within-subject comparisons of suicide attempt rates before and after initiation of antidepressant treatment. Finally, within-subject comparisons showed similar decreases in suicide attempt rates with antidepressant treatment, both in the VA data (5) and in a large-scale medical claims database (6,7) that is more representative of the general population (i.e., includes women and children) than the VA population. These findings raise serious questions regarding the generalizability of FDA’s findings. As noted in the editorial that accompanied their paper (8), patients who are actively suicidal are not enrolled in the RCTs studied by FDA. A more careful examination of the Stone findings and those from the veterans study is informative. In terms of rates of suicidal behavior in treated patients, they are actually quite similar in young adults for the FDA RCTs 670/100,000 and the VA study 477/100,000. What is different, is the placebo rate of suicidal behavior observed for young adults by the FDA (305/100,000) versus the VA study (1368/100,000). Apparently, receiving placebo in an RCT and receiving no pharmacologic treatment in a real-world setting are associated with quite different rates of suicidal behavior, and can have a profound effect on the conclusions of the study (9). Additional inspection of the FDA study results raises further questions regarding potential reporting bias. Among young adults randomized to placebo, the rate of suicidal ideation was 495/100,000 whereas the rate of suicidal behavior was 305/100,000. This is a ratio of only 1.6 to 1, which seems remarkably low. Suicidal thoughts should be 5 to 10 times higher than suicidal behavior (10). This discrepancy suggests that either there is bias in reporting suicidality in RCTs that were not designed to study suicidality or the patient samples are simply not representative of the general depressed patient population. Interestingly, in adults aged 25 and over, where protective effects of treatment were observed, the ratio of suicidal ideation to suicidal behavior was 3.3 to 1 (i.e., 550/100,000 for ideation versus 167/100,000 for suicidal behavior). This finding suggests that the bias in reporting and or lack of representativeness may be more profound for young adults. It is also important to note that FDA’s finding of increased risk of suicidality with antidepressant treatment was based solely on spontaneous reports of suicidality both in pediatric, young adult, and adult samples. In their pediatric analysis, prospective ratings of suicidality were available, and showed no significant difference between treated and placebo control subjects. In the young adult and adult studies, the prospective ratings of suicidality were apparently not even requested. This disconnect between the prospective ratings and retrospective review of spontaneous adverse event reports raised considerable debate among the original scientific advisory board, and at least in part led to several of the members of the committee voting against the black box warning. So which study is correct? The strength of the FDA study is that it is based on randomization, so within each individual study, ignoring important predictors of suicide will only lead to increased uncertainty in the treatment effect, but should not lead to biased conclusions. The weakness of the FDA study is that it is not representative of routine practice, is restricted to a population that may have little resemblance to a routine clinical sample where heightened levels of “suicidality” may exist, and may have excluded other studies that may not have been published. Note, however, that meta-analysis of multiple RCTs is not equivalent to a single well controlled RCT with suicidal behavior as a predefined endpoint. In meta-analysis, study-level differences in patient populations, indications and types of treatment can lead to biased conclusions and it is for this reason that meta-analysis should not be relied upon to derive causal inferences as it is an observational study of “studies.”. By contrast, the strength of the VA study is its generalizability, at least to the adult male population with similar demographic characteristics to men who have served in our nation’s military. The weakness of the study is that it is observational in nature and ignoring potential confounders not only leads to increased uncertainty in the treatment effect estimate, but can also lead to biased conclusions due to confounding by indication (i.e., severity). We note, however, that it is reasonable to assume that patients not receiving pharmacologic treatment would be less severely ill and if anything would have lower suicide attempt rates than patients treated pharmacologically, if the treatment did not have a significant benefit. We observed just the opposite. 1. Stone MB, Laughren T, Jones ML, Levenson M, Holland PC, Hughes A, et al. Risk of suicidality in clinical trials of antidepressants in adults: analysis of proprietary data submitted to US Food and Drug Administration. BMJ 2009;339:b2880. 2. Hammad T, Laughren T, Racoosin JA. Suicidality in pediatric patients treated with antidepressant drugs. Arch Gen Psychiatry 2006;63:332-9. 3. Olfson M, Marcus SC, Shaffer D. Antidepressant drug therapy and suicide in severely depressed children and adults. Arch Gen Psychiatry 2006;63:865-72. 4. Martinez C, Rietbrock S, Wise L, Ashby D, Chick J, Moseley J, et al. Antidepressant treatment and the risk of fatal and non-fatal self harm in first episode depression: nested case-control study. BMJ 2005;330:389- 93. 5. Gibbons RD, Brown CH, Hur K, Marcus SM, Bhaumik DK, Mann JJ. The relationship between antidepressants and suicide: results of analysis of the veterans health administration datasets. American Journal of Psychiatry 2007; 164:1044–1049. 6. Simon GE, Savarino J, Operskalski B, Wang PS: Suicide risk during antidepressant treatment. Am J Psychiatry 2006; 163:41–47. 7. Simon GE, Savarino J: Suicide attempts among patients starting depression treatment with medications or psychotherapy. Am J Psychiatry 2007; 164:1029–1034 8. Geddes, J. R., Barbui, C., Cipriani, A. (2009). Risk of suicidal behaviour in adults taking antidepressants. BMJ 2009;339: b3066-b3066. 9. Weisberg, HI, Hayden VC, and Pontes VP. Selection criteria and generalizability within the counterfactual framework: explaining the paradox of antidepressant-induced suicidality. Clinical Trials 2009; 6:109 -118. 10. Goldsmith SK, Pellmar TC, Kleinman AM, Bunney WE. Reducing Suicide: A National Imperative. Washington, DC: National Academies Press; 2002:1-516. Robert D. Gibbons Ph.D. University of Illinois at Chicago Sharon-Lise T. Normand Ph.D. Harvard University Joel B. Greenhouse Ph.D. Carnegie Mellon University Competing interests: Dr Gibbons reports having served as an expert witness for the US Department of Justice, Wyeth and Pfizer Pharmaceuticals in cases involving antidepressants and suicide. His work in this area is supported by grants R01 MH8012201 from the National Institute of Mental Health and U18HS016973 from the Agency for Healthcare Research and Quality.
Suicide "Attempts"--So What? 26 September 2009
Donald J. Farber, Attorney at Law 94903 Send response to journal: Re: Suicide "Attempts"--So What?	Two Points Might be Taken: (1) I have been, and remain, a critic of the FDA's managament of antidepressants from the earliest days of Prozac until 2004. Since 2004, however, I laud the FDA for its catch up work. Dr. Stone's (et al) article was, in my view, the state of the data and to that end the commentary was valuable. But its past time, now, that the FDA twisted industry arms to do prospective suicide testing that the FDA's Dr. Paul Leber failed to get Eli Lilly to do 20 years ago. Industry's excuse of "ethical" barriers is getting old--as are the unresolved antidepressant suicide questions that continue to linger. (2) Dr. Clayton's point on "completed suicides" as the true barometer of risk has been the repititive theme of antidepressant advocates since the bottomn fell out of industry and mainstream psychiatry's 15 year argument denying the risk. I take the "completed suicide" data as fact-- but its a red herring, for two reasons. In Washington, who could question the ridicule a tire manufacturer would receive from Congress, the media, and the public if, in testimony before Congress, the executive acknowledged his company's tires were involved in the most highway accidents, but insisted nonetheless his tires were "safe" because most of the accident victims survived. The "completed suicide" focus in the antidepressant debate is no less disengenuous. Secondly, and most critically here, reliance on public Coroner's reports, a common argument made by antidepressant advocates to exonerate antidepressants in suicide cases, e.g. "negative" findings, is more than unreliable, it is misleading. This is not an indictment of medical examiners or toxicologists, but recogniation that normal testing, at least in the U.S., is bureaucratically driven to detect drugs of abuse, and not antidepressants, especially traces at less than "therapeutic" levels. Most of my cases where private testing was conducted, and we knew the decedent had likely taken the antidepressant, results were positive for the antidepressant at tracings less than the designated "therapeutic" range. Competing interests: I have represented, in over 10 years, approximately 35 plaintiffs in legal actions involving suicide and antidepressants.