Suicide Risks and SSRIs: New Data Should Change the EquationWhat does it take to get the FDA to revisit a precedent? In light of accumulating data on antidepressants, the FDA should probably reconsider its black box warning on the dangers associated with the much-maligned class of antidepressant drugs known as SSRIs.
The FDA’s risk communication strategy for SSRIs needs retooling
Sally Satel, MD
Medical Progress Today
January 12, 2006
The problem is that the FDA used a black box warning for SSRIs in 2004 based on emerging data – but the data remain in flux, and tend to reinforce the drugs’ benefits. This is a paradigm case for why the FDA needs a flexible risk management strategy that focuses more on helping doctors and patients make nuanced decisions based on evolving science, rather than blanket warnings that can cloud the real issues at stake.
This month, the American Journal of Psychiatry published a much-cited report on the prevention of suicide with antidepressants. It is the latest installment in the controversy over whether drugs such as Paxil (the class of selective serotonin reuptake inhibitors, SSRI’s and their relatives) are too risky to give to patients under 18.
According to the Boston Globe, this report is "likely to reshape the debate over antidepressants." This may or may not be true but, at the very least, the study presents compelling evidence that these medications do not intensify, on average, the risk of suicidal preoccupation and attempts in patients starting treatment -- one of the claims promoted by critics of the drugs.
The issue of SSRI safety was most recently in the spotlight in June 2003 when Eliot Spitzer, New York Attorney General, sued GlaxoSmithKline, the maker of Paxil, for concealing evidence that the drug caused suicidal behavior in young patients. What followed was a series of FDA and Congressional hearings, memorable for heart-wrenching testimony of parents whose children committed suicide while taking the medication, and an ambitious re-analysis of data from 24 clinical pediatric trials. (Trials in which, it must be noted, none of 4400 children killed themselves.)
In October 2004, an FDA advisory group to the FDA voted 15 -8 in favor of putting a “black box” warning on all antidepressants, announcing the risk of suicidal thinking and behavior (the FDA accepted the recommendation). A black box is the strongest caution the FDA uses; the next step would be an outright ban on prescriptions a la Vioxx. Prescriptions for SSRI-type drugs went down about 25% percent in youth after the black box warning was even considered by the FDA, earlier in 2004. (The black box was applied to all antidepressants for all patients – adults as well as children and adolescents - even though the reanalysis was limited to SSRI-type drugs in a pediatric population).
It is too early to know the consequences of this decline – an optimistic reading is that the reduction occurred among the subgroup of physicians who were too ready to pass out the drugs in the first place. But Dr. Thomas Insel, head of the National Institute of Mental Health worries. “My expectation is that we may see an increase [in suicidality].” The black box decision, he said, “is entirely driven by a focus on risk and a neglect of benefit.”
The new study, “Suicide Risk During Antidepressant Treatment” by researchers at Group Health Cooperative health plan in Seattle was funded by the National Institutes of Health. The study examined records of over 65,000 patients who were prescribed an antidepressant; a small but unspecified number of patients were 18 and under. Ten years worth of data were collected on all patients who started an antidepressant and on their clinical status three months before they started the medication and for the next six months. When researchers compared a pre-treatment period to the six month follow-up, they found that the number of serious suicide attempts fell by half within the first month of treatment, and declined progressively thereafter. Adolescents and children exhibited the same declining pattern of suicide attempt.
Because there was no placebo control group in the study, we cannot know whether patients would have enjoyed spontaneous remissions even if they had not taken the drug. This caveat was not adequately appreciated by the press, which overstated the findings somewhat in headlines (e.g., “Antidepressant Risk of Suicide Now Called Low,” Boston Globe Jan. 1, 2006; “Antidepressants Seem to Cut Suicide Risk in Teenagers and Adults, Study Says,” New York Times Jan. 1, 2006). The study does show that antidepressants did not increase the rate of suicide or attempts in kids or adults. That is vital defensive armor in the lawsuits now pending against SSRI-makers. The observation that drugs were associated with a significant reduction in average suicide risk should calm physicians nervous about prescribing them, and parents too scared to allow children to take them.
In addition, the Group Health study comprised a large, real world sample, over ten-fold the number patients that the FDA analyzed for its assessment. Furthermore, these patients were treated under natural conditions for six months compared to subjects in short-duration drug company trial whose pediatric subjects suffer milder symptoms - which are notoriously less responsive to medications and may go a long way toward explaining why the FDA found such limited evidence of overall symptom improvement in the 24 trials it reviewed. Also, the “suicide attempts” among trial subjects were often dramatic attention-getting events (e.g. superficial wrist cutting). In the Group Health study, suicide attempts were defined as serious self-harm requiring medical or psychiatric hospitalization.
Finally, the Group Health study is also consistent with epidemiologic trends suggesting that antidepressants have helped prevent youth suicide. If many young people were committing suicide as a side effect of medication, it seems reasonable that we'd be seeing a spike in the adolescent suicide rate. Since the early 1990's, the incidence of suicidal ideation among adolescents is down over 40%, and actual suicides are down about a third, according to the Centers for Disease Control.
Along these lines, a study in the Archives of General Psychiatry in 2003 looked at the numbers of antidepressant prescriptions written for adolescents (by zip code) across the country and compared them with census data on youth suicide. The greater the increase in the prescriptions in particular geographic areas, the sharper the reduction in the teen suicide rate in those locations over the last decade. Whether substance abuse rates or gun laws differed across these areas and independently affected the volume of suicides is unknown.
The FDA says this study won't change its mind. The head of FDA's Office of Drug Safety concedes the study is “important,” but contends that "it doesn't alter our ongoing concern about children." An official from the medical policy division was reported by the Boston Globe as having told its reporter that “no clinical study has proven that antidepressants work in people under 18.”
Two recent studies call the statement about medication effectiveness into question. “The Treatment for Adolescent Depression Study,” published in Journal of the American Medical Association in 2004, measured symptoms in 439 depressed adolescents (pre-trial) before treatment with fluoxetine (Prozac). The team, led by John March of Duke University randomly assigned subjects to one of four groups: placebo, fluoxetine, placebo plus cognitive behavior therapy (CBT) and fluoxetine plus CBT. They were followed for 12 weeks. Of the group receiving CBT/fluoxetine, 71 percent were rated at 12 weeks as responders. Among those taking fluoxetine alone, 60.6 percent responded. For those who received CBT alone the response rate was 43.3 percent. And for those receiving a pill placebo, the response rate was 34.8 percent. Thus, the medication condition (with and without therapy) exceeded placebo and CBT alone performance. In addition, twenty nine percent of the entire sample expressed ideas of suicide before the trial began and at the end of the 12-week trial all subjects experienced a decline.
A second 2004 study on subjects under 18 appeared in Journal of the American Academy of Child & Adolescent Psychiatry, entitled “Fluoxetine Treatment for Prevention of Relapse of Depression in Children and Adolescents: A Double-Blind, Placebo-Controlled Study” by Graham Emslie of University of Texas Southwestern Medical Center and colleagues. The authors found that depressed kids in remission on fluoxetine were significantly more likely to maintain their improvement if they remained on the drug than if they were switched to placebo. In this random discontinuation study, those remaining on medication enjoyed remission for, on average, an added six months compared to 3.3 months in those who were put on placebo at the beginning of the trial. About a third in the fluoxetine group relapsed compared to about 56% in the placebo group.
The Emslie study reminds us of the big picture when treating depression. Suicide is tragic but it is rare and not easy to predict. Poor response to treatment is a far bigger problem than rare side effects. This is due, in large measure, to the poor follow-up of patients who are prescribed medications; sometimes because they drop out of care and sometimes because the physicians do not schedule visits with sufficient frequency.
The decision of the FDA to adopt the advisory panel decision about a black box warning was based on the panel’s weighing risks in the context of benefit. Because the efficacy of SSRI’s in the 24 studies examined by the FDA did not show impressive results, the panel felt that the risk to benefit ratio tilted far enough in the direction of risk to justify the warning. However, given the low rates of suicidality in the 24 trials, plus the additional literature showing reduction in depression with meds compared to placebo and the methodological/assignment problems with the 24 studies, it seems as though the panel underestimated the benefits of the medications.
The “reshaping” of the debate on antidepressants is indeed in order. We need better information on both old and new treatments for depression, and better ways to convey uncertainty to physicians and families without undue alarm. But, in the interim, we should emphasize that the risks of antidepressant therapy in adolescents with serious depression are surpassed by the benefits.
- This interpretation is complicated by the study’s finding that patients who took the older (tricyclic) antidepressants had smaller numbers of suicides and attempts than those taking the SSRI-type drugs. After all, spontaneous remissions should affect all subjects equally, no matter what drug they took – only if, that is, antidepressants did not help them in the first place. (Unfortunately, the SSRI-type drugs vs. older (tricyclic) comparison was not examined separately; probably because the sample sizes were too small for patients under 18. This limits the confidence with which researchers could say that SSRI-type drugs per se reduced suicide attempts in kids – the most newsworthy issue.) Is it possible, though, that the patients receiving tricyclics were somehow more suicide-prone than those prescribed SSRI-type drugs? This would surely be the case if non-responders were switched to tricyclics. But, given common clinical practice patterns, patients who fail one SSRI-type drug are typically switched to others in that class; not to the tricyclic class. Also, given the high risk of lethality in overdose, doctors who think patients are serious about self-harm, or very impulsive, are more likely to prescribe SSRI-types which are safer in an overdose situation.
- As far as suicide-related events, however, 9 subjects on Prozac (8%) reported "suicide-related events" compared to 4 (4%) with placebo. No child committed suicide. In fact the authors excluded subjects considered to be at high risk for self-harm. The interesting question is why subjects on placebo engaged in self-harm and suicide attempts less often than those on Prozac? No one can say with confidence, but a plausible explanation is that the activating side effects of SSRIs may offset its benefits (in some subgroup). Note, though, that in TADS, suicidal ideation was noted by self-report questionnaire while suicidal behaviors had to be spontaneously reported by a subject or his parent. One popular theory holds that on questionnaires, subjects are known to be more forthcoming but in spontaneous reporting one might expect a difference depending upon activation and engagement – both of which are often enhanced by medications. Thus, the rate of reporting of suicidal behavior was greater on drug compared to placebo, not the actual incidence of the thoughts.
- Trivedi M. et al Evaluation of Outcomes with Citalopram for Depression Using Measurement-based Care in STAR-D: Implications for Clinical Practice. Am J Psychiatry 2006; 163: 28-40 (In Trivedi’s study of 2876 patients treated with citalopram there were no suicides.); Simon G.E. et al Treatment Process and Outcomes for Managed Care Patients’ Receiving New Antidepressant Treatment from Psychiatrists and Primary Care Physicians. Arch Gen Psychiatry 2001; 58:395-401.