Suicidal ideation (SI) is a marker of severe stress and psychological burden. A history of psychological trauma and adverse childhood experiences (ACEs) is associated with experiencing SI (1). Such stressors are also strongly linked to the development of a range of psychiatric symptoms and disorders, including mood, anxiety, trauma, and personality disorders (2–5). Experiencing four or more types of ACEs is linked with a tenfold increased risk of SI (6). Individuals experiencing SI are likely to present with a wide range of mental health symptoms that may require psychiatric care.

There is a need to expand treatment options for SI, given the recognized limitations of current evidence-based, gold-standard non-pharmacological treatments, which include Cognitive Behavior Therapy (CBT), Dialectical Behavior Therapy (DBT), and Crisis Response Planning. These established interventions are often modestly effective for reducing SI and suicidal behaviors, can be highly resource-intensive, and suffer high client/patient attrition rates (7, 8). Rather than resolving underlying pathogenic memories, these interventions tend to focus on stabilization, cognitive strategies, and coping skills for times of emotional pain or crisis. While that approach fits with the generally accepted and conservative risk management approach for working with patients with complex mental health needs (9, 10), such methods may not aid in the resolution of long-standing, recurrent suicidal states.

Trauma focused psychotherapies (TFPs) hold promise for addressing suicidality (11–15). Interest in using TFPs to treat SI likely stems not only from the observation that trauma is a factor in the development of SI across a wide range of psychiatric disorders, but also because trauma treatment, even when not specifically targeting suicidality, may lead to a reduction in SI (16). PTSD treatment studies often report a reduction of SI along with decreased trauma and depression symptoms (16–18). A recent eye movement desensitization and reprocessing (EMDR) study that targeted trauma-related memories in those experiencing a mental health crisis with SI also demonstrated positive outcomes, with few adverse events (14). EMDR has also been studied prospectively for Major Depressive Disorder (MDD) (19–21), which is the most common mental illness suffered by those who complete suicide (22). Fereidouni et al. reported that nine EMDR sessions administered over 3 weeks significantly reduced SI in depressed inpatients compared to control (12). Such treatment, however, can evoke intense emotions, raising the possibility of worsening or treatment-emergent SI, driving clinician hesitancy to provide such treatments to patients already endorsing such thoughts (23–25). There is also a dearth of studies investigating the safety and efficacy of using remotely delivered EMDR in the context of SI, leaving clinicians with little empirically based guidance.

The Adaptive Information Processing (AIP) model of EMDR posits that traumatic or stressful experiences can overwhelm or otherwise interrupt the brain’s normal capacity to integrate information. When reactivated by trauma reminders, maladaptively stored information can present as psychiatric symptoms (26, 27). Altered states of consciousness that accompany extreme stress, dissociation, drug intoxication or physical illness can set the conditions for altered information processing during and following a stressful event (26–28). Implicit or explicit reminders of such content can reactivate the corresponding distressing affective, sensorimotor, and cognitive information stored in these memories. Survival responses, such as fight, flight, fawn, and freeze, are also believed to be embedded within such memories and can become elaborated over time as traumas accumulate. In such situations, modest stressors may act as reminders of these traumas and evoke a response that is much more intense than expected. The person may consequently re-experience states similar to those experienced at the time of the trauma, along with the same emotions, sensations, beliefs, defensive postures, urges, reflexes, and stress responses. As much of this information is stored in implicit, non-declarative memory, the person may not be aware that a memory is being activated; it is therefore experienced as happening now. Depression and trauma-related shame, for example, have been conceptualized as implicit memory networks constructed from previous experiences of shame, humiliation, and loss (29, 30). Thus, a mild perceived rejection might activate a network of unconscious memories, subjectively experienced as intense shame, physiological distress, and a strong urge to hide, attributed to the present situation. Access to healthy, adaptive information, held in other networks, can be simultaneously restricted (26). Resulting feelings and behaviors may be inconsistent with the person’s present developmental level or circumstances.

The AIP model postulates that EMDR addresses such maladaptively stored or pathogenic memories. It is hypothesized that EMDR reduces the negative valence of these memories and allows adaptive information to be integrated into the associated brain networks, thereby promoting a more balanced and constructive emotional response. Consequently, reminders no longer evoke distress or the previously rigidly held maladaptive beliefs (26, 30, 31). The AIP model aligns with neurobiological research indicating how alterations in memory formation, conditioning, learning, and memory reconsolidation processes contribute to the development of a wide range of psychiatric disorders (26, 32–35). Additionally, this perspective is consistent with studies demonstrating that brain areas impacted by early life stress during development are also those involved in memory, emotion and emotional regulation, self-referential processing (e.g., Default Mode Network), sensation and information integration (e.g., limbic system, prefrontal cortex, thalamus, insula, and brainstem) (28, 36–39), and that EMDR or bilateral eye movements influence many of these same brain areas (40–43).

The potential of using the AIP model to understand the roots of suicidal thinking emerges when comparing existing suicide theories against the AIP model ( Table 1 ). Suicide theories generally attribute SI to repetitive distressing states arising from aversive experiences and their consequences, such as the development of suicide-related beliefs. The Interpersonal Theory of Suicide and the Integrated Motivational-Volitional Model, for example, emphasize interpersonal experiences such as those arising from rejection, isolation, defeat, and the perception of being trapped or a burden on others (47, 48). Most theories suggest that suicide attempts occur once a person develops the capacity to act on suicidal thoughts, perhaps due to habituation to the aversion of acting on such impulses, perception of lack of alternative options, or a desire to escape from intolerable pain, loneliness, defeat, entrapment, burdensomeness, or hopelessness (49).

The Fluid Vulnerability Theory (FVT) most closely mirrors the AIP model. Originally based on Beck’s notion of “integrated cognitive-affective-behavioral networks” producing a “synchronous response to external demands”, the FVT suggests that past painful events lead to the development of suicidal states that are triggered by reminders (50, 51). Such states are conceptualized to include not only emotional and cognitive components, but also sensorimotor and physiological information (e.g., urges, sensations, arousal), echoing the AIP model of unprocessed, negatively valenced neural networks. As such, we hypothesized that targeting experiences, beliefs, and affective states associated with SI with EMDR would reduce the intensity of those states. We further theorized that beliefs would shift in a “bottom up” manner once the memories were desensitized, enabling the integration of more adaptive information from other experiences. For a deeper exploration of this hypothesis, refer to Winkler et al. (1). This perspective is also consistent with recent calls to view suicidality as a manifestation of adaptive processes gone awry, best studied through a neuroscience approach like the Research Domain Criteria (RDoC) framework, rather than imprecise “verbal theories” of suicide (49).

The primary objective of this study was to investigate whether web-based, synchronous clinician-delivered EMDR, primarily targeting experiences, beliefs or affective states associated with suicidal thinking, would aid in reducing SI compared to treatment as usual (TAU). Secondary objectives included examining the impact of web-based EMDR treatment on symptoms of depression, anxiety, posttraumatic stress, emotional dysregulation, and dissociation, as well as treatment safety and attrition.

This pilot study is a real-world non-blinded randomized controlled trial that aimed to evaluate the effects of web-based clinician-delivered synchronous EMDR for adult outpatients with SI. Those enrolled in the study were randomly assigned to either the EMDR treatment or TAU group using the randomization module of REDCap (Research Electronic Data Capture) (52).

Approval by the Health Research Ethics Board at the University of Alberta was obtained prior to commencing research activities (protocol ID number: Pro00090989). This study was registered on ClinicalTrials.gov (ID number: NCT04181047). For detailed information regarding the study protocol, rationale, and clinical safety procedures used, refer to Winkler et al. (1).

This study is unique and represents a significant contribution to the literature at several points. First, the study was originally conceived based on the AIP model and the hypothesis that EMDR could directly target the “roots” of suicidality. SI was therefore an inclusion criterion and therapists specifically targeted experiences associated with SI rather than a specific diagnosis. Second, it included a complex, real-world psychiatric population. At the time this study was initiated, many would have considered it too risky to focus on such distressing material. Studies of this nature are infrequently conducted given the perceived risks, including adverse events, dropouts, and the potential for negative results due to participant heterogeneity. Third, this study is among the first to report on remote delivery of EMDR in a real-world context, which was infrequent before the pandemic.

Study results were favorable, although methodological factors may have impacted outcomes. Both the EMDR and TAU care groups exhibited a decrease in self-reported SI and PTSD-like symptoms from baseline to four month follow up. The EMDR group experienced an additional significant decrease in depression and anxiety symptoms over the same time frame, a change that was not observed in the TAU group. Although the small sample size limits data interpretation, there appears to be a downward trend in DERS and DES-II scores in the EMDR group as well. Far from worsening symptoms, EMDR treatment focused on SI resulted in ongoing improvement across multiple symptom categories. Larger studies are needed to assess the reliability of this observation.

Conversely, symptoms in the TAU group appeared to trend downwards from baseline to midpoint, and then worsen thereafter ( Figure 2 ). This U-shaped curve, with rebounding symptom scores, may be related to the difference in treatment received by the TAU and EMDR groups in this study. Treatment as usual likely included temporary supportive and stabilization strategies, lessening SI without resolving underlying factors. As mentioned previously, EMDR desensitizes implicit and explicit memories of distressing experiences. In this context, when focused on experiences associated with SI, there may be potential to address repetitive suicidal states, thereby leading to more enduring improvements compared to TAU. Given the distress involved in focusing on such material, the lack of worsening SI in the EMDR group is notable. However, this rationale remains theoretical and speculative at present, awaiting further research to confirm this theory.

We hypothesize that multiple factors may have contributed to an underestimation of the effect of EMDR treatment on SI relative to TAU. Compared to similar studies, the enrolled participants appear to have greater trauma burden and psychiatric complexity. This may have translated into a more extensive memory network contributing to suicidal states. For example, a pilot study by Proudlock and Perris limited inclusion to mental health crises that were clearly linked to a trauma; participants at risk of dissociation were also excluded (14). Fereidouni et al. included a homogeneous population of depressed inpatients with SI but excluded those with previous suicide attempts. In addition, 80% reported single traumas, mostly in adulthood (12). Our study participants, however, reported a high incidence of early onset SI, chronic SI, multiple suicide attempts, BPD or BPD traits and multiple other psychiatric diagnoses. A significant minority reported lifelong existential themes related to their SI, such as “I’ve always felt like I should never have been born”, likely representing a more entrenched belief system. In this context, we believe more than 12 EMDR sessions may be necessary to address enough pathogenic memories to observe a larger difference between EMDR and TAU groups for SI. In addition, there were some participants in the EMDR group who could not utilize EMDR online, as they, for example, worried it would contaminate their “safe space” at home, or they needed the safety of an in-person therapist, diluting results. Nevertheless, the results from our intention to treat analysis indicate promise and warrant further investigation to explore whether different presentations of SI (e.g., trauma-based, depression-based, existential themes) or patient characteristics warrant specific EMDR strategies. Such strategies might also offer a pathway to resolve SI and suicide related beliefs that persist even after successful treatment of MDD, for example (64).

Research interest in using EMDR for complex populations at risk for suicide has grown in the last decade, with promising results. For example, EMDR has been used to treat PTSD in increasingly complex populations, often with high prevalence of SI, demonstrating good safety and efficacy (16, 65). There are also multiple positive studies for treating MDD, a condition frequently present in those with SI (19–21). Additionally, clinicians have developed specific strategies for addressing suicidal ideas, urges, and states in the context of BPD, MDD, or suicidal intrusions (11, 13, 30, 66), which remain to be fully studied, but are increasingly used in clinical practice. This exploration may stem from observed limitations of cognitive strategies, possibly because distress accompanying such conditions can overwhelm cognitive and emotion regulatory capacities, especially in populations with a history of early life stress or significant trauma. This emerging trend could also be attributed to the increasing recognition that EMDR is a transdiagnostic treatment, capable of reducing a range of symptoms associated with multiple diagnoses (12, 67–69). Our study contributes to this body of literature, supporting the potential for transdiagnostic treatments targeting specific symptom clusters (70, 71). This aligns with the recent establishment of the NIMH Research Domain Criteria (RDoC), promoting research that focuses on behavioral and neural systems within mental disorders, transcending traditional diagnostic categories (72).

This study also adds to the emerging literature suggesting that TFPs can be safely delivered in the context of SI, including through remote delivery. Despite the presence of SI at baseline, therapists directly focused on distressing content ordinarily prompting or closely related to the SI while providing treatment in an online milieu. There was no significant overall indication of symptom exacerbation, contrary to earlier speculation, and the EMDR group experienced numerically fewer serious adverse effects and instances of excessive healthcare utilization, such as emergency room visits and hospitalizations, compared to those receiving TAU. Our findings align with studies on PTSD that included participants with SI, showing a general pattern of reduced PTSD, depressive symptoms, and SI (for recent reviews, see 16,24). Our dropout rate was, notably, even lower than in other studies. The twice-weekly schedule may have influenced this, as intensive scheduling (e.g., frequency greater than weekly) has been linked with lower dropout rates (73).

The perception of treatment safety can have real world implications, influencing treatment strategies, such as limiting access to TFPs and overemphasizing stress avoidance. Multiple authors have noted clinicians’ reluctance to use TFPs due to concerns about possible clinical deterioration, including symptom exacerbation and worsening or treatment-emergent SI (23–25). It is noteworthy that symptom exacerbation is reported in 10 to 46% of participants in PTSD studies, although this is apparently unrelated to treatment response and dropout (23, 25, 74–76). Even so, the prevailing practice often involves restricting TFP until SI resolves, with a focus on “stabilization”. This stabilization typically centers on achieving a sense of safety, providing psychoeducation, and teaching skills. This is an area of ongoing debate, with some arguing that this approach leads to unnecessary delays in definitive treatment - for days to years (77–80).

There is evidence suggesting that excessive avoidance of distress may hinder the resolution of trauma-related symptoms, which we posit includes SI. Both emotional engagement and the mounting of a stress response during therapy sessions are linked to positive therapy outcomes (81, 82). Successful TFP has demonstrated improvement in autonomic sympathetic dysregulation (83, 84). According to memory reconsolidation theory, for a long-term memory trace to be changed, the distressing memory must be brought into awareness along with prediction error, resulting in new contextual learning and extinction (35, 85). In simpler terms, changing a traumatic memory requires accessing it while introducing something new or unexpected. This new experience can then be integrated during reconsolidation as new learning. Examples might include experiencing a distressing state or memory while feeling safe with a therapist, or envisioning urges play out imaginally to a place of triumph, resolution, or repair. The Fluid Vulnerability Theory of suicide aligns with this concept, proposing that the suicidal belief system is “potentially amenable to change during periods of activation, [and] activation is critical to treatment progress and success” (51). If true, this has important clinical and academic implications, including for suicide research.

Despite the promising results observed in this study, caution is warranted, particularly for those with severe dissociative symptoms who are at heightened risk for severe suicidal behavior (53). This specific cohort, which comprised one-third of referrals, was intentionally excluded from our study. Engaging in EMDR with that high-risk group presents challenges for many reasons, including extreme difficulty in accessing and tolerating emotion, dissociation hindering emotional learning, severe attachment difficulties impeding therapeutic progress, and scarcity of adaptive information from experiences related to safety and self-efficacy due to extensive early and repeated wounding (11, 28, 86, 87). There have also been reported cases of clinical deterioration when applying standard EMDR to severe dissociative disorders, such as Dissociative Identity Disorder (88). This has led to the development of specific, titrated EMDR protocols that incorporate stabilization techniques (89–92). We echo the work of others, underscoring the necessity for further research to precisely identify indicators of readiness and therapy suitability, address risk factors for poor outcomes, optimize preparation, individualize treatment, and offer choices for individuals (78).

This study provides preliminary evidence supporting the safe and effective use of web-based clinician-delivered EMDR as a transdiagnostic treatment aimed at reducing SI and symptoms of depression, anxiety, and posttraumatic stress. To our knowledge, it is also the first study to investigate the effectiveness and safety of web-based EMDR in a real-world, highly complex psychiatric population. While results are encouraging, addressing methodological limitations is needed before considering wider implementation. Further research is also needed to better understand who is or is not suitable for this intervention, including who would benefit from in-person versus web-based treatment. Finally, there is a need for clarity on optimal preparation, treatment duration and intensity.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

This study involved human participants and was approved by the Health Research Ethics Board, University of Alberta. This study was conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

LB: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Resources, Supervision, Validation, Writing – original draft, Writing – review & editing. SY: Data curation, Formal analysis, Project administration, Writing – original draft, Writing – review & editing. SP: Conceptualization, Methodology, Writing – review & editing. AA: Conceptualization, Methodology, Writing – review & editing. KO: Conceptualization, Methodology, Writing – review & editing, Resources. SB: Writing – review & editing. AG: Conceptualization, Formal analysis, Methodology, Supervision, Writing – review & editing. OW: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Resources, Supervision, Validation, Writing – original draft, Writing – review & editing.