Non-suicidal self-injury (NSSI) is a prevalent, clinically important behaviour, particularly among adolescents and young adults. NSSI is defined as deliberate, direct, self-inflicted destruction of body tissue without suicidal intent and for purposes not socially sanctioned. Common methods include cutting, hitting, scratching, burning, stabbing, and biting, with cutting frequently reported as the most prevalent method and emerging evidence of increased scratching in some clinical samples.
Although conceptually distinct from suicide attempts—principally by intent—NSSI and suicidal behaviours often co-occur across clinical and community populations. A substantial proportion of individuals with lifetime NSSI report suicide attempts, prompting interest in whether NSSI not only co-occurs but also precedes and predicts later suicidal behaviour. Longitudinal and retrospective studies suggest NSSI frequently has an earlier onset than suicide attempts, indicating a temporal gap that may be clinically meaningful.
Theoretical models provide mechanisms by which NSSI could increase risk for later suicidal behaviour. The Interpersonal Theory of Suicide posits that repeated exposure to painful and provocative experiences, including NSSI, can increase an acquired capability for suicide via habituation to pain and reduced fear of death. The Integrated Motivational–Volitional (IMV) Model differentiates factors that generate suicidal ideation from those that facilitate enactment of suicidal behaviour. Empirical work has linked NSSI characteristics—such as frequency, repetition, and engaging in multiple methods—to subsequent attempts and more severe outcomes in some studies.
Prior reviews have established associations between NSSI and suicidal behaviours but have important limitations. Many syntheses combined heterogeneous outcomes (suicidal ideation, plans, attempts, death), which can obscure clinically meaningful temporal pathways and severity gradients. Other reviews focused on co-occurrence or on the longitudinal course of NSSI itself rather than its role as a predictor of suicidal outcomes. To date, there is no comprehensive systematic review that simultaneously: defines NSSI as the exposure; restricts outcomes to temporally ordered suicide attempts or death; and examines indicators of method lethality or clinical severity.
This protocol describes a review designed to address those gaps by synthesising temporally ordered observational evidence on whether prior NSSI predicts later suicide attempts or suicide death, and whether characteristics of NSSI influence the risk or severity of later suicidal behaviour.
Primary review question
Among individuals with a history of non-suicidal self-injury (NSSI), what is the evidence for a temporal relationship between NSSI and subsequent suicide attempts or death based on prospective or retrospective studies?
Secondary review questions
Within studies demonstrating temporal ordering, to what extent do characteristics of NSSI (for example, frequency, repetition, number of methods, or recency) influence the risk of subsequent suicide attempts or death?
Is prior NSSI associated with escalation in the lethality or clinical severity of subsequent suicide attempts, as reflected by indicators such as method lethality, degree of medical injury, or need for clinical intervention?
This protocol follows the Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols (PRISMA-P) 2015 guidance and the completed review will be reported in line with PRISMA 2020. The review protocol was prospectively registered with the Open Science Framework (OSF) Registries (DOI provided in the source). Any amendments to the protocol will be documented in OSF with rationale and date.
Eligibility is structured using a Population–Exposure–Comparator–Outcome–Study design (PECO) framework. The review will include observational studies—both prospective and retrospective—that demonstrate temporal ordering with NSSI assessed prior to the suicidal outcome. Studies must report behavioural outcomes (suicide attempts or suicide death) rather than only suicidal ideation or planning.
Eligible studies may include individuals of any age drawn from clinical samples, community samples, school-based cohorts, emergency department presentations, or population-based datasets. Studies that focus exclusively on suicidal ideation or planning without behavioural self-injury will be excluded.
The exposure of interest is non-suicidal self-injury, operationalised as deliberate self-inflicted injury without suicidal intent as specified by original study authors. Eligible exposure measurements include NSSI assessed at baseline in prospective designs, or history of NSSI (lifetime or past-year) in retrospective designs. Studies using broader labels such as “self-harm” or “deliberate self-harm” will be included only when suicidal intent was explicitly assessed and reported as absent or when outcome data for non-suicidal acts are separable.
Comparators will vary by study design but should permit assessment of whether prior NSSI is associated with later suicide attempts or suicide death. Outcomes are restricted to temporally ordered, behaviourally enacted suicidal outcomes: suicide attempts and suicide death. The review will also examine indicators of attempt severity and lethality where reported (for example, method lethality, medical consequences, or requirement for clinical intervention).
The authors plan a comprehensive search of MEDLINE, Embase, PsycINFO, CINAHL, and Web of Science. The protocol reports registration on OSF and adherence to PRISMA-P standards. Specific search strings and date limits were described in the full protocol (referenced in the original source); any protocol amendments will be updated on OSF.
Study screening, selection, and data extraction will be conducted independently by two reviewers. Risk-of-bias assessment will use Joanna Briggs Institute tools appropriate to the included observational designs. The protocol specifies independent extraction and assessment procedures to minimise reviewer bias. Details such as conflict resolution procedures and data items to be extracted are outlined in the protocol.
Given anticipated heterogeneity in populations, measures, and outcomes, the authors plan a structured narrative synthesis rather than prespecified meta-analysis. Synthesis will focus on temporally ordered evidence connecting NSSI to subsequent suicide attempts or death, and on whether NSSI characteristics predict increased risk or greater lethality/clinical severity. The review will highlight the strength of temporal evidence and identify gaps in prospective data.
By restricting to temporally ordered observational evidence and to behaviourally enacted suicidal outcomes, this review aims to clarify whether and how prior NSSI functions as a predictor of later suicide attempts or death and whether specific NSSI characteristics signal elevated risk or more severe outcomes. The findings are expected to inform clinical formulation, safety planning, and future suicide prevention research by improving understanding of factors associated with subsequent suicidal behaviour. The authors note that contemporary clinical guidance recommends comprehensive psychosocial assessment, formulation, and safety planning rather than categorical risk stratification, and that evidence from this review could contribute to those practices.
The protocol reports no specific funding and no competing interests. It was peer-reviewed and published in PLoS One with registration and adherence to established reporting guidelines.