Copyright @ Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. Fluoxetine and Adult Suicidality Revisited An Updated Meta-Analysis Using Expanded Data Sources From Placebo-Controlled Trials Charles M. Beasley Jr, MD,* Susan G. Ball, PhD,*y Mary E. Nilsson, MS,* John Polzer, DVM, MS,* Sitra Tauscher-Wisniewski, MD,*z John Plewes, MD,* and Nayan Acharya, MBBS, MRCP, MFPM* Abstract: Selective serotonin reuptake inhibitor treatments have been suggested by some to induce emergence of suicidality (ideation and behaviors). The objective of this study was to assess suicidality emergence by adverse event and rating scale data in the largest available, adult, major depression, double-blind, placebo- controlled, fluoxetine trial database (18 trials). Adverse event reports and comments for patients (fluoxetine, n = 2200; placebo, n = 1551) were searched for suicide-related events that were then classified into Food and Drug Administration categories. For 16 trials, suicidality was also examined by Hamilton Depression Scale item 3 (suicide) scores, and these data were analyzed along with the combination of event-based data and scale-based data. Compar- isons between treatments were made for various estimates of wors- ening (risk) and improvement (benefit) of suicidality. Fluoxetine treatment did not result in greater worsening but was associated with greater improvement and faster resolution of ideation (P e 0.05 vs placebo). Data sources were differentially sensitive in detecting changes in suicidal ideation and behaviors. Fluoxetine treatment led to greater benefit rather than risk for suicidality. (J Clin Psychopharmacol 2007;27:682–686) W ith increasing use of selective serotonin reuptake in- hibitors (SSRIs) during 1980s and 1990s, some authors suggested that these antidepressants might induce suicidality, including both ideation and behaviors. 1,2 In a meta-analytic review, Beasley et al 3 found that fluoxetine was not asso- ciated with excess suicidality compared with placebo in adult patients treated for major depressive disorder (MDD). Depending upon the analytical method, conflicting results have been reported regarding SSRIs and increased risk of suicidality. 4–6 In 2004, after a meta-analysis of SSRI trials for psychiatric illnesses in pediatric patients, the US Food and Drug Administration (FDA) required a Bblack box warning[ regarding suicidality to be added to all SSRI labels, 7,8 and a subsequent analysis in young adults found similar results for patients between 18 and 24 years. 9 This study evaluated suicidality in the fluoxetine adult MDD database using adverse event (AE) data (events formally recorded as AEs or described in comments), the FDA method, rating scale data, and their combination. Various estimates of both worsening and improvement, based on these data sources, are presented. MATERIALS AND METHODS Data Sources The data set consisted of all MDD double-blind placebo-controlled trials of fluoxetine in adults, with at least 30 patients that were sponsored by Eli Lilly and Company (18 trials: 15 outpatient, 1 inpatient, and 2 inpatient/outpatient settings). The modal trial dosage was 20 mg/d of fluoxetine (range, 5–80 mg) across trials. Suicidality by AEs (All 18 Trials) Potential suicide-related AEs were identified using a computerized text-string search of investigator-recorded verbatim and dictionary-coded (eg, Coading Symbols for a Thesaurus of Adverse Reaction Terms) terms in AE modules of case report forms. Identical searches were performed on free-text narratives, and any incidents found within comments were included as AEs. The text strings for these searches were accident, attempt, cut, gas, hand, hung, injur, jump, mutilate, overdose, self damag, self harm, self-harm, self inflict, self- inflict, shoot, slash, suic, poison, asphyxiation, suffocation, firearm, burn, drown, gun, immolat, and monoxide. Potential events were reviewed by persons blinded to treatment. Events were initially evaluated to exclude false- positive events (eg, die in San Diego) and then reviewed by 2 additional Lilly health care personnel, at least one of whom was a physician. Reviewers independently assigned each event to 1 of 9 FDA codes, 10 based on the system developed by Posner et al, 11 and the following outcomes were determined: completed suicides, suicide attempts, suicidal acts (fatal and nonfatal attempts), preparatory suicidal behavior, suicide be- havior (acts plus preparatory behaviors), suicidal ideation, all suicidality (behavior plus ideation), and all possible suicidal- ity (sum of all suicidality and categories of self-injury, intent unknown; fatality, not enough information; and nonfatality, not enough information). A treatment-emergent AE was Brief Report Journal of Clinical Psychopharmacology  Volume 27, Number 6, December 2007 682 *Lilly Research Laboratories; yIndiana University School of Medicine, Indianapolis, IN; and zMedical University of Vienna, Vienna, Austria. Received March 6, 2007; accepted after revision September 10, 2007. Research was funded by Eli Lilly and Company. All authors are employees and/or shareholders of Eli Lilly and Company. Address correspondence and reprint requests to Charles M. Beasley Jr, MD, Drop Code 1730 Lilly Corporate Center, Indianapolis, IN. E-mail: cmbeasleyj@lilly.com. Copyright * 2007 by Lippincott Williams & Wilkins ISSN: 0271-0749/07/2706-0682 DOI: 10.1097/jcp.0b013e31815abf21