To Name :
To Email :
From Name :
From Email :
Comments :

Evidence-Based Reviews

Safe use of SSRIs in young adults: How strong is evidence for new suicide warning?

Pediatric suicide rates increased in 2003-04 after the black-box warning, which has now been extended to patients age 18 to 24.

Vol. 6, No. 11 / November 2007

CASE: Life is 'not worth it'

Mr. B, age 20, has taken a semester leave from college because of gradually worsening depressed mood. Over the past 2 months he has lost interest in jogging and playing piano—which he usually enjoys. He reports reduced libido, middle insomnia, loss of appetite, feeling as if his head is “full of cotton,” trouble concentrating, and waking in the morning with a sense of dread. His anxiety dissipates during the day, but he continues to feel sad and sometimes weepy, which is unusual for him.

Mr. B reports feeling hopeless at times and has had vague thoughts about life being “not worth it if I continue to feel like this” but denies specific suicide plans. Your initial impression is that Mr. B is in the midst of a major depressive episode and that a selective serotonin reuptake inhibitor (SSRI) is indicated. As you finish taking his history, you run through your mind the pros and cons of the recommendation you will make to him.

Do SSRIs raise or lower the risk for suicidal behavior in young adults such as Mr. B? The answer is complicated and goes beyond an “either/or” question, as the FDA acknowledged in May 2007 when it:

  • extended the black-box warning of increased suicidality risk with antidepressants to cover adults age 18 to 24 as well as children and adolescents
  • included language in the warning about the benefits of treating depression and the suicide risk associated with untreated depression, given concerns about declining antidepressant prescriptions and rising suicides among youth. 1
  • To help you make informed decisions when treating depression in adults, this article reviews the studies leading up to and following the FDA’s meta-analysis of antidepressant trial data in patients age 18 and older. Our goal is to provide a framework for clinical treatment of adults age 18 to 24 and those age ≥25.


FDA’s pediatric suicidality analysis: What the data showed

The FDA meta-analysis designed to investigate a reported association between antidepressants and suicidality in children and adolescents found contradictory results:

  • Pooled adverse event data from 24 pediatric antidepressant trials totaling >4,400 patients showed a higher risk of suicidal ideation or behavior (no suicides occurred) with antidepressants (4%) vs placebo (2%).
  • Systematically collected suicide-related item scores from 17 of the trials showed no evidence that antidepressants worsen suicidality or cause it to emerge.

One interpretation of these findings is that antidepressants’ effect on suicidality is small and therefore subject to measurement error.

Another is ascertainment bias; any side effect associated with active medication encourages discussion with the clinician and may distort the frequency of reported adverse events.

The FDA meta-analysis also found:

  • Relative risk for suicidality ranged 10-fold among agents, from 0.9 with fluoxetine to 8.8 with venlafaxine.
  • Most suicide-related events occurred in subjects having the highest baseline levels of suicidality.
  • Hostility and agitation emerged with SSRI use, particularly during the first month of treatment.
  • Patient age, sex, or history of suicide attempt/ideation did not affect the results.

Source: Reference 7

First hints of suicidality

SSRIs revolutionized depression treatment. From 1985 to 1999, annual U.S. antidepressant prescriptions quadrupled, with SSRIs accounting for 70% of the increase (see “Antidepressants and suicide risk, 1985 to 2007”). At the same time, the age-adjusted suicide rate:

  • dropped 22.5% for women (who account for twice as many antidepressant prescriptions as men)
  • dropped 12.8% for men (without change in the rank order of suicide methods). 2

For many patients, increased antidepressant use improved treatment of major depressive and other antidepressant-responsive disorders. In 1990, however, case reports suggested SSRIs might cause suicidal thoughts or behavior. 3 Hypothesized mechanisms included increased aggression 4 and akathisia. 5 An FDA review found no proof, and a meta-analysis of data from 17 double-blind, randomized, controlled trials found no association between fluoxetine and suicidal thoughts or behavior. 6

The debate rekindled in June 2003 when the British Committee on Safety of Medicines warned against using paroxetine or venlafaxine in children. After conducting its own meta-analysis, the FDA in 2004 ordered a black-box warning about suicidality and the use of antidepressants in children and adolescents ( Box ). 7

After the pediatric ‘black box.’ Antidepressant prescriptions for children and adolescents declined in the years 2003 to 2004, as did diagnosis of pediatric depression. 8-10 Antidepressant prescribing also showed signs of shifting from general practitioners to psychiatrists. 8 At the same time, the suicide rate among youth age <17 rose 11% from 1.26/100,000 to 1.4/100,000—after 3 consecutive years of decline—according to new data from the Centers for Disease Control and Prevention. 11 In patients age >60, SSRI prescriptions continued to rise and suicide rates fell,9 a pattern of change consistent with antidepressants protecting against suicide.

An independent meta-analysis by Bridge et al12 examined the pediatric trial data used in the FDA meta-analysis plus 7 additional studies. Its findings differ in 2 important ways from those of the FDA review:

  • Antidepressants—including others besides fluoxetine—showed efficacy in treating anxiety disorders and depression in children and adolescents.
  • The frequency of suicide-related adverse events (no trial patients committed suicide) was approximately 3% on active medication—25% lower than the FDA estimated rate—and 2% on placebo, similar to the FDA estimate.

The number needed to treat (NNT)— number of patients who must be treated to get a therapeutic response that would not have happened with placebo—ranged from 3 to 10. The number needed to harm (NNH)—number of patients who must be treated for 1 suicidal ideation/nonfatal attempt to occur that would not have happened with placebo—ranged from 112 to 200. The authors interpreted this as “indicating a favorable overall risk-to-benefit profile for antidepressants in the treatment of pediatric [major depressive disorder], [obsessive-compulsive disorder] (OCD), and non-OCD anxiety disorders.” 12 These findings appear to support the efficacy of antidepressants in pediatric patients and a favorable risk-benefit ratio.

Antidepressants and suicide risk, 1985 to 2007










Case reports suggest link between suicide and SSRI use

FDA analysis finds no association between SSRIs and increased suicide risk


UK agency warns of suicide-related events in children treated with paroxetine and venlafaxine

FDA conducts meta-analysis, requires black-box warnings of risk of suicidality in youth taking antidepressants

FDA meta-analysis finds age-dependent effect of antidepressants on suicidality risk in adults

FDA expands warning of increased suicidality risk with antidepressants to adults age <25

Antidepressant prescriptions quadruple; age-adjusted suicide rate drops 22.5% for women and 12.8% for men

Pediatric depression diagnoses and antidepressant prescriptions decline; suicides increase 11%

Bridge et al meta-analysis finds 25% lower rate of suicide-related events in youth than the FDA found

What about adults?

Overall effect. A subsequent FDA meta-analysis of antidepressant clinical trial data in adults13 found 8 suicides in 372 trials totaling nearly 100,000 persons. All occurred in the 295 trials with psychiatric indications. Among these psychiatric trials, 59% had a suicidal behavior/ideation event in either the test-drug or placebo arm, and 41% had none. Eleven antidepressants were included in the meta-analysis:

  • 6 SSRIs (citalopram, escitalopram, fluoxetine, fluvoxamine, paroxetine, and sertraline)
  • 2 SNRIs (duloxetine and venlafaxine)
  • 3 others (bupropion, mirtazapine, and nefazodone).

Overall, antidepressants showed a protective (antisuicidal) effect in adults as compared with placebo (odds ratio 0.85 [95% CI: 0.71 to 1.02, P=0.08]), with no difference in effect between SSRIs and non-SSRIs.

Age-specific findings. When the FDA analysis was stratified by age, however, antidepressants’ benefit appeared greater for patients age ≥25 than for those age 18 to 24. The data suggested:

  • elevated suicidality risk among adults age <25
  • neutral or possibly protective effect for adults age 25 to 64
  • protective effect in adults age ≥65 ( Table 1 ). 13

Table 1

FDA meta-analysis: Suicide rates by age in antidepressant trials

Age group (yr)

Suicide rate (%)(test drug/placebo)

Suicide attempt rate (%)(test drug/placebo)

18 to 24



25 to 30



31 to 64






Source: Reference 13

For 18-to 24-year-olds, the suicide rate was 0.03% (~1/4,000) in these mostly 8- to 12-week trials, and the suicide attempt rate was 0.55% (~1/200). For comparison, the lifetime prevalence of suicide was 2.2% to 8.6%—depending partly on illness severity—in a meta-analysis of patients with mood disorders. 14

The odds ratio for suicidal behavior (preparatory acts, attempt, or suicide) for subacts, attempt, or suicide) for subjects age 18 to 24 on test drug vs placebo was 2.31 (95% CI: 1.02, 5.64) [event rate/sample: 23/3810 vs 8/2604]. NNH was 333, which means 333 adults in this age group would need to be treated with an antidepressant for 1 to experience a suicidal behavior event that would not have happened with placebo.

Compare an NNH of 333 with the much lower NNH values associated with antiarrhythmic treatment of atrial fibrillation (AF), 15 an important cardiovascular cause of morbidity and mortality. A meta-analysis of 44 AF trials totalling 11,322 subjects found that—although “moderately effective” for maintaining sinus rhythm—all but 2 of the 10 drugs were pro-arrhythmic. Their NNH values of 17 to 119 are, at best, approximately one-third the NNH for antidepressants for suicidality in young adults based on adverse event reports. With NNH, higher is safer.

The age-related pattern the FDA found in its adult meta-analysis ( Table 2 ) 13 is consistent with its earlier pediatric analysis 7 but not with the more recent findings of Bridge et al 12 that included a larger data set.

Table 2

FDA meta-analysis: Risk of suicidality in adults taking antidepressants

Protective benefit appeared greater for dults age ≥25 than for those age 18 to 24

Elevated risk appeared to apply to adults age <25 with any psychiatric disorder, not just depression

A “slight suggestion” of increased risk was seen with SNRIs vs other classes, but no significant differences among drugs or drug classes

Risk was increased in adults who did not respond to active drug treatment

Risk was not affected by patient sex, race, geographic location, inpatient vs outpatient care, or treatment with SSRIs vs non-SSRIs

Risk may be lower with sertraline than with other antidepressants, although this trend could be a false-positive result related to multiple tests

A “sensitivity analysis” using alternate statistical methods to test the robustness of the findings yielded similar results

SNRIs: serotonin-norepinephrine reuptake inhibitors;
SSRIs: selective serotonin reuptake inhibitors
Source: Reference 13

Mixed evidence

Aside from the FDA meta-analysis, 13 what is the evidence that antidepressants—or specifically SSRIs—may cause suicidality in adults? Among 9 major published studies in adults of the relationship between SSRIs and suicidal behavior (deaths or attempts):

Table 3A

4 studies found no relationship
between SSRI use and suicide in adults

Study design

Main results


Meta-analysis of FDA data by Khan et al; 16 >48,000 patients in trials that included fluoxetine, sertraline, paroxetine, citalopram

No difference in suicide rates among SSRIs, other antidepressants, or placebo

Patients not representative of general clinical population
Trials mostly short-term; tended to exclude suicidal patients

Case-control study in UK primary care practice by Jick et al; 17 antidepressant users, 555 with suicidal behavior vs 2,062 without

Risk of suicidal ideation or behavior did not differ between SSRIs vs non-SSRIs

Observational study
Confounding by indication*

Case-control study in UK primary care practice by Martinez et al; 18 >146,000 persons, first antidepressant prescription for depression

No evidence that risk of suicide or nonfatal self-harm was greater with SSRIs than with tricyclics

Observational study
Confounding by indication*

Meta-analysis in UK by Gunnell et al; 19 477 clinical trials (N>40,000) of SSRIs vs placebo in depression

No evidence that SSRIs increased suicide risk
Weak evidence of increased self-harm risk

Lack of individual or trial-level data
Evidence suggests nonfatal suicidal ideation/self-harm events are underreported

SSRI: selective serotonin reuptake inhibitor; UK: United Kingdom

* Confounding by indication: Clinicians may preferentially prescribe SSRIs to patients thought to be at risk for suicide because of these drugs’ relative safety in overdose.

Table 3B

2 studies found SSRIs may increase suicide risk in adults

Study design

Main results


Meta-analysis (1967 to 2003) by Fergusson et al; 20 702 trials comparing SSRIs vs placebo or other drug for any condition (N>87,000)

Suicide attempts almost twice as likely with SSRIs vs placebo, but no difference in fatal attempts; no difference in risk between SSRIs and TCAs

Many methodologic limitations in clinical trials, such as underreporting of suicidal behavior
Limited to public data

Case-control study of geriatric population in Ontario by Juurlink et al; 21 1,138 suicides vs 4,552 controls matched for demographic characteristics and antecedent illness patterns

SSRIs were associated with higher risk of suicide than other antidepressants during first month of therapy (adjusted OR: 4.8, 95% CI=1.9–12.2)
Risk independent of recent diagnosis of depression or receipt of psychiatric care
Violent suicides more common with SSRIs than other antidepressants
No difference in suicide risk after first month
Absolute risk of suicide with antidepressants was low

Confounding by indication*
14% of suicide sample excluded from analysis

CI: confidence interval; OR: odds ratio; SSRIs: selective serotonin reuptake inhibitors; TCAs: tricyclic antidepressants; UK: United Kingdom

* Confounding by indication: Clinicians may preferentially prescribe SSRIs to patients thought to be at risk for suicide because of these drugs’ relative safety in overdose.

Table 3C

3 studies found SSRIs may lower suicide risk in adults

Study design

Main results


Case-control forensic toxicology by Isacsson et al; 22 14,857 suicides compared with 26,422 deaths by accident or natural causes in Sweden (1992 to 2000)

SSRIs less likely than other antidepressants to be detected in suicide victims

Naturalistic study
Possible residual confounding

Observational study by Simon et al;23 82,285 antidepressant treatment episodes among 65,103 health plan outpatients (1992 to 2003)

Suicide risk in acute-phase treatment ~1/3,000; risk of suicide attempt leading to hospitalization ~1/1,000
No increased risk of suicide or serious attempt suggested during first month of treatment
No greater risk seen with newer drugs (mostly SSRIs)
General decline in risk of suicide attempts after starting antidepressant treatment

Potential uncontrolled confounding
Geographically limited sample
Possible misclassification in computerized records
Lack of data on medication adherence
Death certificates may underestimate suicide rates

Observational study by Gibbons et al 24 of suicide attempts in veterans treated for depression; 226,866 patients treated with SSRI, other antidepressant, or no antidepressant (2003 to 2004)

SSRI treatment associated with ~1/3 lower risk of suicide attempts compared with no antidepressant treatment
Finding consistent in veterans age 18 to 25 and in older veterans

Sample of veterans, 92% male
Dataset did not include suicides, so results pertain only to suicide attempts

SSRIs: selective serotonin reuptake inhibitors

Evidence of protection

Epidemiologic studies. Suicide attempt rates in depressed youth and adults—including those age <25—are highest in the month preceding treatment and decline steadily after antidepressant treatment or psychotherapy begins, according to depression studies in a large group health plan. The pattern was the same whether a primary care physician or psychiatrist prescribed the antidepressant. 25

Did you miss this content?
Botulinum toxin for depression? An idea that’s raising some eyebrows