What Do Patient-Level Studies Tell Us About Antidepressants and Suicide Risk?

As reviewed recently in this paper,¹ the use of antidepressants has been a rising source of controversy in this country, especially with respect to SSRI/SNRI agents. Some of the direst warnings have come from the current Secretary of Health and Human Services, who, in congressional testimony, inaccurately claimed that, “SSRIs have a black box warning, warning of suicidal, homicidal behavior.”² To be sure, there are legitimate concerns about the use of SSRIs and SNRIs in pediatric populations, and the FDA’s 2004 black box warning is often cited in this regard.³ While a review of that specific issue is beyond the scope of this focused commentary, we would refer readers to the recent systematic review by Soumerai et al, which found “significant unintended reductions in mental health care after the [FDA] warnings,” as well as “…marked increases in psychotropic drug poisonings and suicide deaths.”⁴

Focus on Patient Level Studies

Our review is focused primarily on patient-level studies between 2005 and 2024, including randomized controlled trials (RCTs); meta-analytical studies; and observational register-based studies Our selection of patient-level studies (PLSs) is predicated on several advantages PLSs have over other study designs. PLSs analyze data from each separate participant in a study, rather than relying on summarized, group-level data. Accordingly, PLSs can assess how specific traits—like age, gender, disease severity, or medication timing and dosage—affect treatment outcomes. PLSs also avoid the “ecological fallacy”; ie, fallaciously assuming a cause-and-effect relationship exists at the individual level just because a statistical correlation exists at the population level.⁵

Methodological Problems, Limitations, and Confounds

The literature investigating the relationship between antidepressant use and suicide risk is riddled with complex methodological problems, systematic limitations, and overlapping confounds. For example, because suicide is a rare event, statistical “noise” is high. Furthermore, suicidal thoughts and behaviors are integral aspects of severe major depressive disorder; hence, disentangling whether a suicide attempt was caused by the underlying illness or the pharmacological treatment is extremely difficult. Perhaps the thorniest problem is the risk of confounding by indication and severity. For example, a higher suicide rate in an antidepressant-exposed cohort may simply reflect that the group was sicker at baseline; and accordingly, was prescribed antidepressants.⁶

In theory, RCTs should eliminate confounding by indication through randomization. However, to minimize legal and medical liability, the vast majority of industry-funded phase 3 clinical trials strictly exclude patients with active suicidal ideation, producing a relatively low-risk population that does not reflect real-world psychiatric practice.

Finally, the literature is rife with the practice of pooling mild suicidal thoughts and lethal suicide attempts into a single metric (ambiguously termed, "suicidality"), which obscures the most serious risk. Indeed, there is a poor correlation between suicidal thoughts and suicide attempts or completed suicide. As McHugh et al have shown in a meta-analysis of over 70 studies, “…The utility of suicidal ideation as a test for later suicide is limited by a modest sensitivity and low PPV [positive predictive value].”⁷

Overview of the Principal Studies

Our focused review of 13 primarily patient‑level studies shows that the evidence regarding antidepressant treatment and suicide risk is broadly consistent across studies. For clarity and methodological coherence, we present the studies grouped by (1) RCTs and meta-analytical studies; and (2) observational register-based studies. Our analysis is summarized in the Table 1 and Table 2, which includes important limitations and confounders in these studies.

The research methodology across these 13 studies is quite varied. The data sources include large register-based studies in 6 studies; data from randomized controlled trials in 4; and data from health care systems in 3. Research design varies from very large samples to detect rare events to cohort studies that concentrate them in one sample. In one study, the researchers used registry data to construct a “pseudo-clinical trial” design. In all cases, antidepressant use or non-use was an independent variable.

The various study designs lead to diverse “hard” and “soft” outcome measures, including completed suicides; suicide attempts; and intensity of suicidal ideation, based on items on standard depression rating scales (HAM-D, MADRS, and others). Some studies relied on expert classification of events, and hospital or clinic contact for suicide attempts. The majority of studies (9/13) had “hard” outcomes like hospitalization or death. Three of 4 RCTs were able to report deaths by suicide. However, the studies as a whole confirm that completed suicide is a low frequency event, as noted previously. Two register-based studies also reported all-cause mortality and showed that it was lower in the antidepressant treated groups.

RCTs and Meta-Analyses

• Gibbons et al (2012), in a patient-level meta-analysis of all FDA-registered RCTs of fluoxetine and venlafaxine, found antidepressant treatment associated with significantly faster reduction in suicidal thoughts/behavior compared with placebo, in adults and the older adults.⁸ The study found no increased risk of suicidal thoughts or behaviors in any adult age group, while pediatric fluoxetine trials showed a non-significant trend toward increased risk.
• The Hammad et al study (2006) formed the formal statistical basis for the FDA black box warning, actually issued in October 2004.⁹ The study pooled data from 24 short-term pediatric RCTs involving over 4400 patients. The main finding was a 2-fold increased risk of “suicidality” (suicidal ideation and behaviors) in children taking antidepressants compared with those on a placebo; ie, 4% for those on antidepressants vs 2% for those on a placebo. Notably, there were zero completed suicides across all 24 trials.
• Stone et al (2009) evaluated 372 randomized RCTs comprising 99,231 adult participants.¹⁰ The study concluded that the risk of suicidality associated with antidepressant use is strongly age-dependent, with increased risk in young adults (under 25); no added risk for adults aged 25 to 64; and reduced risk for adults aged 65 and older, in whom antidepressant use was associated with a 63% decreased risk for suicidal behavior and ideation compared with placebo.¹¹
• Zisook et al (2024) performed a secondary analysis of the VAST-D trial (Veterans Affairs Augmentation and Switching Treatments for Improving Depression Outcomes).¹² This was a multi-site study of 1522 adult veteran outpatients with nonpsychotic major depressive disorder (MDD) who had an inadequate response to at least 1 adequate antidepressant trial. While the primary trial results were published in 2017, the 2024 secondary analysis focused on suicidal ideation outcomes during 12 weeks of treatment. It found that the overall prevalence of suicidal ideation dropped by more than half, falling from 46.5% at baseline to 21.1% at the end of treatment.

Observational Register Based Studies

• Gibbons et al (2007) carried out a retrospective cohort study (VHA data) of 226,866 veterans with incident depression who received care through the Veterans Health Administration (VHA) during fiscal year 2003.¹³ The study tracked suicide attempt rates among patients treated with SSRIs; newer non-SSRI antidepressants (eg, bupropion, mirtazapine, venlafaxine); tricyclic antidepressants (TCAs); or no antidepressant treatment. This observational study compared suicide attempt rates before vs. after initiating antidepressant therapy, adjusting for baseline psychiatric severity and demographics. Suicide attempt rates were significantly lower among patients treated with antidepressants compared to those who received no pharmacological treatment. Individual patient timelines showed that suicide attempt rates were significantly higher in the month immediately preceding the start of treatment than in the months following the initiation of medication. The Gibbons analysis found that SSRIs exerted a protective effect across all adult age groups, although in the 18 to 25 group, the effect did not reach statistical significance.
• Lagerberg et al (2022) studied 538,577 individuals aged 6 to 59 who initiated SSRI treatment between 2006 and 2013.¹⁴ The study utilized a nationwide, population-based cohort design combined with a within-individual analysis to control for time-invariant confounding variables. Suicide attempts and completed suicides were tracked before and after SSRI treatment milestones. The highest suicidal behavior risk occurred in the 30 days before SSRI initiation (IRR 7.35) and diminished after initiation (IRR 0.62 in first 30 days post-initiation vs. pre-initiation). These results were similar across age strata. The results did not suggest that SSRI-treatment increased the risk for suicidal behavior in either youths or adults; rather, it may reduce the risk.
• Lagerberg et al (2023) studied 162,267 individuals (aged 6-59 years) receiving a depression diagnosis during 2006–2018 in Stockholm County, Sweden, after at least 1 year without antidepressant dispensation.¹⁵This was a “target trial emulation” study, meaning that researchers use observational data (like medical records or insurance registries) to mimic the design of a RCT. Subjects were divided into SSRI initiators vs non-initiators and followed for 12 weeks, tracking suicide attempts and completed suicides. In contrast to their 2022 study, this smaller study found a statistically significant increased risk of suicidal behavior among SSRI initiators aged 6 to 24, with the highest risk in children and adolescents aged 6 to 17. There was no increased risk for adults aged 25 to 59, nor in the high-risk group of individuals with past suicidal behavior. This study raises the issue of “confounding by indication,” in that SSRI initiators may have been more severely ill than non-initiators.
• Martinez et al (2005) used a nested case control design to study 146,095 UK primary care patients with first-episode depression.¹⁶ The study compared the risk of non-fatal self-harm and suicide in patients taking SSRIs vs those taking tricyclic antidepressants. There was no comparison with “no drug treatment” subjects. The study found no significant difference in suicide or self-harm risk between SSRIs and older tricyclic antidepressants in adults but detected “weak evidence” of non-fatal self-harm in SSRI-treated patients aged 18 or younger (OR 1.59). The authors acknowledged that “preferential prescribing of SSRIs to patients at higher risk of suicidal behavior cannot be ruled out,” again raising the problem of confounding by indication. Notably, there were zero completed suicides among children aged 18 or younger taking antidepressants.
• Coupland et al (2015) studied 238,963 patients (aged 20-64) registered with UK general practices with a first diagnosis of depression between 2000 and 2011, followed up until 2012.¹⁷ Suicide/attempted suicide or self-harm were tracked. No significant difference was found in suicide/self-harm risk between SSRIs and TCAs, but higher rates were found for mirtazapine, venlafaxine, and trazodone. Risk was highest within the first 28 days of starting or stopping the drugs. The authors acknowledged that “Direct comparisons between groups receiving different treatments for the same condition are still subject to channeling bias, where the selection of a particular antidepressant depends on the characteristics of the patient. This could have occurred in this study if drugs such as mirtazapine, venlafaxine, and trazodone were more likely to be prescribed to patients at increased risk of suicide or attempted suicide.”
• Tiihonen et al (2006) used a nationwide (Finnish) cohort to study 5390 hospitalized suicide attempters over an average of 3.4 years.¹⁸ Participants were classified as antidepressant users or non-users. Current use of any antidepressant was linked to a 39% increased risk of attempted suicide compared with periods of non-use. Somewhat paradoxically, current antidepressant use was associated with a 32% decreased risk of completed suicide. Fluoxetine was linked to the lowest risk of completed suicide; venlafaxine, to the highest risk. The disconnect between suicide attempts and completed suicides may be due to a combination of faster rescues after suicide attempts, and less lethal means, such as impulsive pill overdoses rather than use of firearms. Current SSRI use was also linked to a 49% lower risk of all-cause mortality, probably driven by a reduction in cardiovascular and cerebrovascular deaths.
• Haukka et al (2009) tracked every individual living in Finland between 1999 and 2003 who started a prescribed antidepressant program for the first time.¹⁹ Current antidepressant (AD) use was linked to an 82% reduction in all-cause mortality compared with individuals who filled only a single prescription and then stopped. There was no statistically significant difference in suicide rates when looking at ongoing “general” antidepressant usage vs the one-prescription baseline group. However, current SSRI usage was associated with a 53% reduction in suicide risk when compared directly against both the one-prescription group and patients taking other non-SSRI classes of antidepressants.
• Osler et al (2019), tracked 2 national cohorts in Danish registries between 1997 and 2015: 997,911 military conscripts, and a clinical sample of 95,794 patients with a first-time affective disorder.²⁰ The study investigated how initiating antidepressant medication correlates with the timing of suicide, suicide attempts, and violent crime convictions. In both cohorts, the rates of attempted suicide were higher in the 28 days before initiation of antidepressant medication than in the period after initiation. The authors concluded that suicidal behavior immediately following prescription initiation probably reflects baseline disease severity and a delay in therapeutic mood response, rather than a harmful effect of the drug itself. In addition, the study did not find that younger individuals have a higher risk of suicidal behavior after antidepressant initiation. Rates of violent crime were similar before and after antidepressant initiation.
• Jakobsen et al (2023) used Danish register data and a validated cohort of 1842 suicide attempts to determine the association between redeemed antidepressant prescriptions and the risk of repeated suicide attempts (n=210).²¹ Whereas crude models showed all antidepressants to be significant risk factors (HR around 1.39) for repeated suicide attempts, all adjusted models showed all antidepressants to be insignificant risk factors. The authors concluded that when considering the individuals' baseline risk of repetition, there was no significant increased risk of repeated suicide attempts in individuals redeeming a prescription for any antidepressant (or only SSRIs).

Discussion

We have acknowledged several methodological limitations and confounds in the antidepressant-suicide risk studies we reviewed, particularly the thorny problem of confounding by indication. These problems make confident pronouncements about suicide risk in this context hazardous. Nevertheless, we believe that—taken together—the 13 studies reviewed converge on a consistent conclusion: antidepressant treatment does not increase suicide mortality in adults and may actually reduce risk of completed suicide in certain high‑severity populations. Indeed, no adult studies we reviewed found increased suicide deaths—ie, completed suicides—clearly attributable to antidepressant use.

In pediatric populations, the data are mostly reassuring but more complicated and nuanced. In general, patient‑level, randomized data show a modest increase in suicidal ideation and non‑fatal suicidal behavior with antidepressant treatment. However, completed suicides attributable to antidepressants are essentially absent in pediatric cohorts, and suicide attempts are too rare to permit firm conclusions. Observational cohorts suggest transient elevations in non-fatal self‑harm around treatment initiation in younger populations; however, these behaviors appear largely driven by underlying illness dynamics rather than by demonstrable, medication‑induced increases in suicidal behavior. Indeed, to put these data in a broader perspective, “…Emergent suicidality is a common occurrence in psychosocial treatment of adolescent depression, with rates similar to those reported recently in antidepressant trials.”²²

That said, in pediatric populations, increased suicidal ideation and behavior require careful monitoring and vigilance. Accordingly, we would recommend cautious and conservative use of antidepressants in children and adolescents. Vigilance is especially important during the first few weeks of antidepressant treatment. However, based on our review, we would by no means discourage use of antidepressants in cases of severe major depression in carefully monitored younger populations. Indeed, Bridge et al (2005) reported that youth treated with antidepressants were approximately six times more likely to benefit than to be harmed (NNT = 9; NNH = 59, respectively), reinforcing that modest increases in suicidal ideation and nonfatal behavior occur alongside clinically meaningful symptomatic improvement.²³

Concluding Thoughts

Currently available antidepressants are far from ideal medications and carry a significant side effect burden. Nevertheless, for a substantial subgroup of patients with major depressive illness, SSRIs and SNRIs provide important clinical benefits beyond that seen in placebo-treated cohorts.¹ With respect to completed suicides, the patient-level studies reviewed here do not point to a loud “suicide signal” associated with use of antidepressants in either adult or pediatric populations. On the contrary, most of the studies we reviewed point toward a reduction of suicide risk in adults, and no elevation of completed suicides in younger populations. This does not mean that antidepressant treatment never provokes suicidal thoughts or non-lethal behaviors. The first few weeks of treatment—particularly in younger populations—merit careful attention to the emergence of these problems, and to the emergence of agitation or behavioral activation. The latter problems sometimes point to the presence of undiagnosed bipolar spectrum disorders.²⁴ However, none of the risks discussed here warrants a “war on SSRIs” or the blanket condemnation of antidepressant treatment.²⁵ When used conservatively and monitored closely in a psychologically supportive milieu, antidepressants remain an important tool in the treatment of moderate-to-severe depression.

Dr Dawson is recently retired from clinical practice and is a freelance writer and researcher on psychiatric topics. He writes the Real Psychiatry Blog.

Dr Pies is Professor Emeritus of Psychiatry and Lecturer on Bioethics and Humanities, SUNY Upstate Medical University; Clinical Professor of Psychiatry, Tufts University School of Medicine; and Editor in Chief Emeritus of Psychiatric Times (2007-2010). Dr Pies is the author of several books. A collection of his works can be found on Amazon.

References

1. McIntyre RS, Pies RW. The top 10 misleading claims about antidepressants. Psychiatric Times. July 2, 2026. https://www.psychiatrictimes.com/view/the-top-10-misleading-claims-about-antidepressants

2. RFK Jr. Confirmation Hearing Day One. Rev. January 30, 2025. Accessed September 1, 2026. https://www.rev.com/transcripts/rfk-jr-confiramation-hearing-day-one

3. Ho D. Antidepressants and the FDA’s Black-Box warning: determining a rational public policy in the absence of sufficient evidence. AMA Journal of Ethics. 2012;14(6):483-488.

4. Soumerai SB, Koppel R, Naci H, et al. Intended and unintended outcomes after FDA pediatric antidepressant warnings: a systematic review. Health Aff (Millwood). 2024;43(10):1360-1369.

5. Greenland S. Ecological fallacies. International Encyclopedia of the Social & Behavioral Sciences. 2001;4094-4099.

6. Pompili M, Serafini G, Innamorati M, et al. Antidepressants and suicide risk: a comprehensive overview. Pharmaceuticals (Basel). 2010;3(9):2861-2883.

7. McHugh CM, Corderoy A, Ryan CJ, et al. Association between suicidal ideation and suicide: meta-analyses of odds ratios, sensitivity, specificity and positive predictive value. BJPsych Open. 2019;5(2):e18.

8. Gibbons RD, Brown CH, Hur K, et al. Suicidal thoughts and behavior with antidepressant treatment: reanalysis of the randomized placebo-controlled studies of fluoxetine and venlafaxine. Arch Gen Psychiatry. 2012;69(6):580-7.

9. Hammad TA, Laughren T, Racoosin J. Suicidality in pediatric patients treated with antidepressant drugs. Arch Gen Psychiatry. 2006;63(3):332-339.

10. Stone M, Laughren T, Jones ML, 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.

11. Cassels C. Antidepressant use, suicide risk in adults strongly linked to younger age. Medscape. August 13, 2009. Accessed September 1, 2026. https://www.medscape.com/viewarticle/707411

12. Zisook S, Moutier CY, Rush AJ, et al. Effect of next-step antidepressant treatment on suicidal ideation: findings from the VAST-D trial. Psychol Med. 2024;54(6):1172-1183.

13. Gibbons RD, Brown CH, Hur K, et al. Relationship between antidepressants and suicide attempts: an analysis of the Veterans Health Administration data sets. Am J Psychiatry. 2007;164(7):1044-1049.

14. Lagerberg T, Fazel S, Sjölander A, et al. Selective serotonin reuptake inhibitors and suicidal behaviour: a population-based cohort study. Neuropsychopharmacology. 2022;47(4):817-823.

15. Lagerberg T, Matthews AA, Zhu N, et al. Effect of selective serotonin reuptake inhibitor treatment following diagnosis of depression on suicidal behaviour risk: a target trial emulation. Neuropsychopharmacology. 2023;48(12):1760-1768.

16. Martinez C, Rietbrock S, Wise L, 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(7488):389.

17. Coupland C, Hill T, Morriss R, et al. Antidepressant use and risk of suicide and attempted suicide or self harm in people aged 20 to 64: cohort study using a primary care database. BMJ. 2015;350:h517.

18. Tiihonen J, Lönnqvist J, Wahlbeck K, et al. Antidepressants and the risk of suicide, attempted suicide, and overall mortality in a nationwide cohort. Arch Gen Psychiatry. 2006;63(12):1358-1367.

19. Haukka J, Arffman M, Partonen T, et al. Antidepressant use and mortality in Finland: a register-linkage study from a nationwide cohort. Eur J Clin Pharmacol. 2009;65(7):715-720.

20. Osler M, Wium-Andersen MK, Wium-Andersen IK, et al. Incidence of suicidal behaviour and violent crime following antidepressant medication: a Danish cohort study. Acta Psychiatr Scand. 2019;140(6):522-531.

21. Jakobsen SG, Larsen CP, Stenager E, Christiansen E. Risk of repeated suicide attempt after redeeming prescriptions for antidepressants: a register-based study in Denmark. Psychol Med. 2023;53(12):5510-5517.

22. Bridge JA, Barbe RP, Birmaher B, et al. Emergent suicidality in a clinical psychotherapy trial for adolescent depression. Am J Psychiatry. 2005;162(11):2173-2175.

23. Bridge JA, Salary CB, Birmaher B, et al. The risks and benefits of antidepressant treatment for youth depression. Ann Med. 2005;37(6):404-412.

24. Luft MJ, Lamy M, DelBello MP, et al. Antidepressant-induced activation in children and adolescents: risk, recognition and management. Curr Probl Pediatr Adolesc Health Care. 2018;48(2):50-62.

25. O’Connor K. The war on SSRIs. Psychiatry Online. 2026;61(7).