Suicide is one of the leading causes of early death worldwide. In Brazil, suicide rates have been rising steadily over the past two decades, and most suicides occur in low- and middle-income countries where access to specialized care is limited. There is an urgent need for fast-acting, practical interventions that can be delivered in public emergency settings. Two promising approaches have emerged: esketamine, a medication that can rapidly reduce suicidal thoughts within hours, and Crisis Response Planning (CRP), a brief session in which a trained clinician works with the person to create a personalized written plan for managing future suicidal crises.

The goal of this clinical trial is to learn if esketamine or Crisis Response Planning (CRP), each added to enhanced treatment as usual (eTAU), can prevent future suicide-related events compared to eTAU alone in adolescents and adults aged 14 years or older who recently attempted suicide or have severe suicidal thoughts. The main questions it aims to answer are:

Does a single esketamine infusion plus eTAU lower the risk of a new suicide-related event compared to eTAU alone over 12 months? Does a single session of Crisis Response Planning plus eTAU lower the risk of a new suicide-related event compared to eTAU alone over 12 months? Which approach leads to faster or more lasting improvements in suicidal thoughts, depression, anxiety, sleep, well-being, hopelessness, and quality of life? Are these interventions feasible, acceptable, and cost-effective within a public health system?

Researchers will compare three groups to see which approach works best to prevent suicide attempts, suicide-related hospitalizations, and suicide deaths over one year.

A total of 468 participants will be randomly assigned in equal numbers (156 per group) to one of three groups:

Esketamine group: receive a single intravenous esketamine infusion (0.50 mg/kg given over 40 minutes) in a monitored medical setting with continuous heart rate, blood pressure, and oxygen monitoring, plus eTAU. A physician will be present throughout. Participants will be observed for up to 24 hours before discharge.

Crisis Response Planning group: complete one 20-to-45-minute session with a trained clinician to build a personal crisis plan that includes warning signs, coping strategies, reasons for living, support contacts, and emergency resources. Participants will leave with a written and digital copy of their plan, plus eTAU.

Enhanced treatment as usual (eTAU) group: receive standard emergency care, safety counseling about access to lethal means, connection to the local mental health network (including Psychosocial Care Centers and primary care), and a scheduled psychiatric follow-up appointment within 7 days.

Participants will:

Complete health questionnaires about suicidal thoughts, depression, anxiety, sleep, well-being, hopelessness, and quality of life at 11 time points over one year (at enrollment, 24 hours, 7 days, 2 weeks, 4 weeks, 8 weeks, 16 weeks, 24 weeks, 32 weeks, 40 weeks, and 1 year) Provide a blood sample at the start of the study for exploratory analyses of biological markers that may be related to treatment response Use a smartphone app to report their mood, thoughts, and emotions four times a day for four weeks after enrollment Be monitored for safety and adverse events throughout the entire study period

The study takes place in the public emergency network of Indaiatuba, São Paulo, Brazil (population approximately 256,000), and is designed to reflect real-world clinical conditions. Participants who experience a new suicide-related event during the study will be offered an open-label rescue treatment combining esketamine and Crisis Response Planning, and will continue to be followed for the remainder of the year. The study also includes an evaluation of how well these interventions can be adopted and sustained within Brazil's public mental health system, including assessments of acceptability, feasibility, and cost-effectiveness.

STUDY DESIGN AND CONCEPTUAL FRAMEWORK The SAVE Study (Esketamine Versus Crisis Response Planning Versus Enhanced Treatment as Usual for Suicide Prevention) is a single-municipality, pragmatic, three-arm, parallel-group, superiority randomized controlled trial with 1:1:1 allocation and a Hybrid Type 1 effectiveness-implementation approach. The pragmatic design embeds interventions within routine emergency-department workflows, uses broad inclusion criteria, flexible delivery, and minimal exclusion criteria to maximize real-world generalizability and scalability across low- and middle-income country (LMIC) public health networks. The protocol adheres to the SPIRIT 2025 guidelines (Chan et al., 2025), and reporting will follow the CONSORT 2025 Statement (Hopewell et al., 2025). An independent Data and Safety Monitoring Board (DSMB) oversees safety throughout the trial.

SETTING The trial is conducted in Indaiatuba, São Paulo State, Brazil (population approximately 256,000), which integrates urgent and emergency units (UPA/UEU), a reference hospital emergency department (ED), a Medical Specialties Center (CEEM), a Psychosocial Care Center (CAPS), and primary care units (UBS) within the municipal Psychosocial Care Network (RAPS). Randomization and interventions occur in the ED/UEU; linkage to CAPS and primary care ensures continuity of treatment. This setting was selected because it represents a typical mid-sized Brazilian municipality with a functional but resource-constrained public mental health infrastructure, enhancing the transferability of findings to similar contexts nationally and internationally.

STUDY FLOW Once clinical stabilization is achieved, potentially eligible patients-based on clinical and psychiatric evaluations conducted by attending physicians in the emergency setting-are referred to study researchers for further evaluation. The presence of a primary psychotic disorder, acute psychosis, or acute mania is assessed by the attending physician before referral to the research team. Study researchers then visit the site to conduct additional assessments and invite eligible individuals to participate. Those who provide informed consent (or assent, for minors aged 14-17, with guardian consent) complete further evaluations to confirm eligibility, including sociodemographic, clinical, and psychiatric assessments, as well as the ASSIST for substance use disorder screening, which also serves as an exclusion criterion. Female participants additionally undergo a rapid pregnancy test within the emergency setting. Eligible participants proceed to baseline assessments (T0) and are randomized in a 1:1:1 ratio. Interventions are delivered immediately in the ED/UEU.

RANDOMIZATION, CONCEALMENT, AND BLINDING A statistician not involved in outcome assessments generates a computer-randomized sequence with permuted, variable block sizes and stratification by three factors: (1) age group (≥18 vs. 14-17 years), (2) sex (male vs. female/other), and (3) index presentation (suicide attempt [actual, interrupted, or aborted] vs. severe suicidal ideation vs. non-suicidal self-injury). Allocation is implemented via REDCap's automated randomization module, which features role-based access and audit-trail logging, ensuring concealment.

While participants and treating clinicians are necessarily unblinded due to the nature of the interventions, outcome assessors and data analysts are kept blinded to treatment allocation. Assessors are trained not to inquire about treatment received and to terminate assessments if participants voluntarily reveal their allocation, with reassignment to an alternative blinded assessor. Data analysts receive coded datasets (Groups A, B, and C) without treatment identification until completion of the primary analyses. Emergency unblinding is limited to serious safety concerns requiring knowledge of allocation for appropriate clinical management, following a pre-specified protocol approved by the Ethics Committee. All unblinding events are documented with detailed justification and reported to the Ethics Committee.

INTERVENTION DETAILS Arm A - Enhanced Treatment as Usual (eTAU, control): Routine ED care augmented by a structured linkage across RAPS, booked psychiatric follow-up within 7 days (CAPS/CEEM) after the index event, and a lethal means safety counseling intervention delivered at the emergency department by a research team member. The lethal means counseling consists of developing a plan to limit the participant's access to potentially lethal suicide attempt methods. Risk assessment and emergency procedures are standardized and identical across all three arms. Researchers have no interference with eTAU clinical decisions.

Arm B - Esketamine + eTAU: One intravenous esketamine infusion at 0.50 mg/kg over 40 minutes, administered in a monitored clinical setting with continuous pulse oximetry, non-invasive blood pressure, and ECG monitoring. A physician is present throughout; an ACLS-capable team is immediately available. Structured symptom ratings occur during infusion and for 240 minutes afterward. Discharge can occur after 24 hours if clinically stable. Pre-specified safety thresholds trigger intervention, including: sustained systolic blood pressure ≥180 or ≤90 mmHg, diastolic blood pressure ≥105 or ≤50 mmHg, blood pressure shift of ±25 mmHg, or clinically significant dissociation or agitation. The esketamine infusion protocol follows established safety guidelines from long-term extension studies.

Arm C - Crisis Response Planning (CRP) + eTAU: A 20-to-45-minute single-session Crisis Response Plan delivered by trained clinicians, comprising collaborative identification of personal warning signs, internal self-regulation strategies, reasons for living, social support contacts, and professional and emergency contacts. Participants leave with a written plan (printed card plus digital photograph). Therapists complete a standardized 3-hour training program, which includes didactic content, role-plays, certification, and weekly supervision. Sessions utilize fidelity checklists, and 10-20% of sessions are audio-recorded for adherence and competence coding, with a target inter-rater reliability of κ ≥ 0.80 across fidelity audits.

RESCUE TREATMENT FOR RECURRENT EVENTS Participants who experience a recurrent suicide-related event (i.e., meet the primary endpoint) during follow-up are offered an open-label rescue treatment combining esketamine and CRP while remaining in scheduled follow-up assessments. The rescue phase is analyzed descriptively and does not contribute to the primary intention-to-treat (ITT) analysis of the endpoint.

ASSESSMENT SCHEDULE Follow-up assessments are conducted at 11 time points: baseline (T0), 24 hours (T1), 7 days (T2), 2 weeks (T3), 4 weeks (T4), 8 weeks (T5), 16 weeks (T6), 24 weeks (T7), 32 weeks (T8), 40 weeks (T9), and 1 year (T10), with monthly event surveillance between scheduled visits. Core instruments include the Columbia-Suicide Severity Rating Scale (C-SSRS), Beck Scale for Suicide Ideation (BSI), Patient Health Questionnaire-9 (PHQ-9), Montgomery-Åsberg Depression Rating Scale (MADRS), Generalized Anxiety Disorder-7 (GAD-7), Snaith-Hamilton Pleasure Scale-Brazilian version (SHAPS-BR), Pittsburgh Sleep Quality Index abbreviated version (Short PSQI), 12-Item Short Form Health Survey (SF-12), Barratt Impulsiveness Scale Short Form (BIS-15), Beck Hopelessness Scale 7-item version (BHS-7), Short Warwick-Edinburgh Mental Well-Being Scale (SWEMWBS), Alcohol, Smoking and Substance Involvement Screening Test (ASSIST), McLean Screening Instrument for Borderline Personality Disorder (MSI-BPD), Duke University Religion Index (DUREL), Adult ADHD Self-Report Scale Short Form (ASRS-S), Altman Self-Rating Mania Scale (ASRM), and Clinical Global Impression scales (CGI-S and CGI-I). All psychometric instruments are validated Brazilian Portuguese versions. The assessment battery was streamlined following expert consultation to reduce participant burden in this high-acuity population.

Baseline-only measures serving as covariates and potential moderators include impulsivity (BIS-15), substance use (ASSIST), borderline personality features (MSI-BPD), religiosity/spirituality (DUREL), ADHD symptoms (ASRS-S), and manic symptoms (ASRM).

ECOLOGICAL MOMENTARY ASSESSMENT (EMA) A smartphone application (MindLogger) collects four daily ratings for four consecutive weeks following the index event. Each brief survey assesses momentary affect (e.g., happiness, sadness, loneliness, anger, anxiety, stress, hopefulness, calmness, boredom), self-evaluative states (e.g., self-worth, perceived purpose, forgiveness, hopelessness), and cognitive-emotional regulation strategies (e.g., distraction, rumination, reappraisal, emotional acceptance, seeking social support). Participants also report on contextual factors, including current company (e.g., alone, with family, partner, friends), interpersonal events (e.g., recent conflicts), and substance use within the past two hours (alcohol, cannabis, opioids, stimulants, sedatives, hallucinogens, nicotine).

The EMA further captures momentary suicidal processes through items evaluating the wish to live and to die, current suicidal ideation intensity, controllability over suicidal thoughts, perceived ability to resist acting on them, and recent self-harm behaviors. Responses are rated on brief ordinal scales reflecting real-time intensity or frequency. This approach enables high-frequency, real-world sampling of emotional, cognitive, behavioral, and contextual dynamics during the critical post-index period, enhancing temporal resolution and ecological validity for modeling short-term risk fluctuations.

BIOMARKERS AND LABORATORY PROCEDURES At baseline, venous blood samples (10 mL in EDTA tubes) are collected and immediately analyzed at the hospital laboratory. Samples are destroyed after report release (no biobank storage). The biomarker panel includes: high-sensitivity C-reactive protein (hs-CRP), homocysteine, comprehensive vitamin assessment (B1/thiamine, B2/riboflavin, B6/pyridoxine, B12/cobalamin, D/25-hydroxyvitamin D, folate), essential minerals (magnesium and zinc in erythrocytes, selenium), complete thyroid panel (TSH, free T3, free T4), and hormonal markers (testosterone, DHEA-S, prolactin, SHBG). Exploratory analyses will examine whether baseline inflammatory and nutritional markers moderate treatment response.

SAMPLE SIZE JUSTIFICATION Sample size determination followed an event-driven design appropriate for the time-to-first-event primary endpoint, analyzed using Cox proportional hazards models. Statistical power depends primarily on the number of observed events.

For the two co-primary pairwise comparisons (esketamine + eTAU vs. eTAU, and CRP + eTAU vs. eTAU), the family-wise type I error rate was controlled at αFWER = 0.05 using the Holm-Bonferroni sequential procedure, corresponding to α = 0.025 (two-sided) for each comparison. Required numbers of events were computed using Schoenfeld's proportional hazards method. Assuming equal allocation (p = 0.5), 90% power, and target hazard ratios of 0.50 for esketamine and 0.39 for CRP, this yielded approximately 103 and 56 required events, respectively.

The expected 12-month cumulative event risk in the control group was set at 30%, with a 24-month accrual period and 12-month follow-up period, accounting for 20% attrition/censoring. Under these assumptions, the corresponding two-arm totals were approximately 312 participants (156 per group) for the esketamine comparison and 182 participants (91 per group) for the CRP comparison. For the three-arm (1:1:1) design, the study was powered to the limiting contrast (esketamine vs. eTAU), yielding a total sample size of 468 participants (156 per arm). The CRP vs. esketamine comparison will be analyzed as a key secondary, exploratory outcome.

STATISTICAL ANALYSIS PLAN All analyses follow the intention-to-treat principle, using two-sided tests and 95% confidence intervals. The two co-primary comparisons are analyzed under Holm-Bonferroni family-wise error control, with α = 0.025 per contrast.

Primary endpoint analysis: Cox proportional-hazards models adjusted for prespecified covariates (sex, age group, index presentation, baseline suicidal ideation, PHQ-9 scores, and study center). Proportional-hazards assumptions will be tested using Schoenfeld residuals; when violated, time-varying coefficients or restricted mean survival time (RMST) estimates will be employed. Kaplan-Meier curves and adjusted log-rank tests will visualize survival distributions. The CRP vs. esketamine contrast will be treated as a key secondary exploratory analysis, reported with hazard ratios and 95% CIs, but not included in α-adjusted inference.

Supplementary count analysis: As a pre-specified supplementary analysis of the primary endpoint, negative-binomial regression models with person-time offsets will estimate incidence rate ratios (IRRs) and 95% CIs for total suicide-related events (including recurrences), adjusted for the same covariate set. This analysis captures cumulative event burden complementary to the time-to-first-event approach. This analysis will be reported alongside the primary survival model but will not be included in the family-wise error control.

Secondary endpoints: Trajectories of continuous symptom scales (PHQ-9, GAD-7, SHAPS-BR, Short PSQI, SWEMWBS, SF-12, MADRS) will be analyzed with linear mixed-effects models with random intercepts and fixed effects for group, time, and group × time interaction. Binary repeated outcomes will be analyzed with logit GEE models.

Cost-effectiveness analysis: Conducted from a societal perspective, using 3% annual discounting for costs and utilities (SF-6D derived from SF-12) and a willingness-to-pay threshold of less than 3× Brazilian GDP per capita per QALY gained.

Sensitivity analyses: Andersen-Gill recurrent-events Cox models, mean cumulative function estimation, and Fine-Gray competing-risk models for suicide deaths. Per-protocol analyses excluding major violations will test the robustness of results. Moderation and prediction analyses will assess treatment interactions with prior attempts, baseline severity, and biomarkers, reporting marginal effects with 95% CIs.

Missing data handling: Censoring at last contact for events, inverse probability weighting (IPW) for dropout-related missingness, and multiple imputation (MI) for baseline covariates. Additional sensitivity analyses will compare IPW, mixed-model estimation under missing at random (MAR), and pattern-mixture models for missing not at random (MNAR) mechanisms.

Analyses will be conducted in R (v4.4) and Stata 18, with a detailed statistical analysis plan (SAP) finalized before unblinding.

IMPLEMENTATION EVALUATION The implementation component is structured using the Implementation Research Development Tool (ImpRes-BR) to plan, specify, and evaluate strategies. The evaluation employs multiple frameworks, including CFIR 2.0 (Consolidated Framework for Implementation Research) to identify multilevel determinants, ERIC (Expert Recommendations for Implementing Change) for strategy specification, a Power/Interest matrix for stakeholder mapping, and AIM/IAM/FIM (Acceptability of Intervention Measure / Intervention Appropriateness Measure / Feasibility of Intervention Measure) for implementation outcomes. AIM/IAM/FIM will be administered to clinicians participating in CRP and esketamine implementation at baseline, 6 months, and 12 months. Qualitative data will be analyzed through thematic analysis using NVivo v14.

Core implementation strategies encompass: (i) clinician training and capacity building covering C-SSRS administration, CRP delivery, and IV esketamine monitoring procedures, with simulation-based practice, job aids, and fidelity checklists; (ii) stakeholder engagement and co-design with the Municipal Health Secretariat, ED leaders, and CAPS coordinators; (iii) audit and feedback via automated REDCap dashboards with quarterly site-level performance reports; (iv) contextual adaptation through focus groups and interviews to refine workflows and identify context-specific barriers; and (v) workflow integration with standardized order sets, safety checklists, and escalation algorithms, with assignment of sustainability champions.

The mixed-methods approach incorporates surveys, fidelity audits, and semi-structured interviews with purposive sampling of approximately 20 professionals per unit, adjusted until thematic saturation.

DATA MANAGEMENT AND CONFIDENTIALITY Data are captured in REDCap (Research Electronic Data Capture) with role-based permissions, range checks, and audit trails. Identifiers are stored separately from analytic datasets. All data are maintained in compliance with Brazilian data protection legislation (Lei Geral de Proteção de Dados, LGPD).

SAFETY MONITORING AND ADVERSE EVENT REPORTING All arms include standardized suicide risk assessment procedures at every contact. Esketamine monitoring adheres to predefined safety thresholds as described above; pregnancy testing is conducted when applicable. Adverse events (AEs) and serious adverse events (SAEs), including hospitalizations, suicide attempts, and deaths, are documented and reported following CONSORT Harms reporting guidelines. The independent DSMB reviews cumulative safety data quarterly or on an ad hoc basis as needed. All adverse events are reported to the responsible Ethics Committee in accordance with Brazilian regulatory requirements (Circular Letter No. 13/2020 and Law No. 14,874).

REGULATORY COMPLIANCE Esketamine administration in the research setting complies with Anvisa Resolution RDC 885/2024 governing investigational use of esketamine in clinical research. The study was approved by the Ethics Committee of the Hospital das Clínicas, Faculdade de Medicina, Universidade de São Paulo (HCFMUSP), under CAAE 89878225.0.0000.0068 (approval no. 7.723.561, 2025). Municipal authorizations from Indaiatuba RAPS were obtained. The study complies with Brazilian legislation (Law No. 14,874 and Resolution CNS 466/12) and the Declaration of Helsinki.

Inclusion Criteria:

1. Age 14 years or older
2. Presentation to a public emergency service (ED/UEU) within the study municipality
3. Recent suicide attempt within the past 30 days (actual, interrupted, or aborted attempt)
4. Current severe suicidal ideation, defined as endorsement of items 4 or 5 on the Columbia-Suicide Severity Rating Scale (C-SSRS) screening version, indicating active suicidal ideation with specific plan and/or method, or active suicidal ideation with intent to act
5. Residency in the catchment area of the study municipality (Indaiatuba, São Paulo, Brazil), enabling completion of follow-up assessments
6. Ability to provide written informed consent (for participants aged 18 years or older) or written assent with written informed consent from a parent or legal guardian (for participants aged 14-17 years)
7. No clinical decision for involuntary or voluntary hospitalization in a psychiatric inpatient unit following the index emergency department evaluation

Exclusion Criteria:

1. Contraindications to esketamine, including: aneurysmal vascular disease, arteriovenous malformation, history of intracerebral hemorrhage, or known hypersensitivity to esketamine or ketamine
2. Current pregnancy or breastfeeding (confirmed by rapid pregnancy test in the emergency setting for female participants of childbearing potential)
3. Medical instability requiring intensive care unit (ICU) admission without feasible study follow-up
4. Primary psychotic disorder (e.g., schizophrenia, schizoaffective disorder), current acute psychosis, or current acute manic episode precluding informed participation
5. Severe substance use disorder compromising capacity for treatment adherence, as assessed by the Alcohol, Smoking and Substance Involvement Screening Test (ASSIST)
6. Inability to maintain contact for follow-up assessments, including participants whose residential address is located outside the study municipality

Individual participant data will not be shared to protect participant privacy and confidentiality, particularly given the sensitive nature of suicide-related data and the vulnerable population being studied. The study involves detailed mental health assessments, suicide risk evaluations, and personal crisis information that requires strict confidentiality protections under Brazilian data protection laws and research ethics guidelines.

Aggregate, de-identified results and statistical analyses will be made available through peer-reviewed publications and conference presentations. Researchers interested in collaboration or secondary analyses may contact the principal investigator to discuss potential partnerships within appropriate ethical and regulatory frameworks.

The study protocol, statistical analysis plan, and aggregate results will be made publicly available to support scientific transparency while maintaining participant confidentiality.

This is a pragmatic trial in which participants and treating clinicians are necessarily unblinded due to the nature of the interventions (intravenous esketamine infusion, face-to-face Crisis Response Planning session, and enhanced treatment as usual). Outcome assessors are blinded to treatment allocation; they are trained not to inquire about treatment received and to terminate the assessment if a participant voluntarily reveals their assigned group, with reassignment to an alternative blinded assessor. Data analysts are also blinded: they receive coded datasets labeled Groups A, B, and C without treatment identification until completion of the primary analyses. Emergency unblinding is restricted to serious safety concerns requiring knowledge of allocation for appropriate clinical management, following a pre-specified protocol approved by the Ethics Committee. All unblinding events are documented with detailed justification and reported to the Ethics Committee.