Liaison-consultation psychiatry (LCP) represents a specialized domain of psychiatric practice that integrates mental health care into general medical settings (1). Psychiatrists working in this field provide expert clinical evaluations, diagnostic clarification, and treatment recommendations for patients with confirmed or suspected psychiatric conditions. These consultations, typically requested by other healthcare professionals, most commonly occur in inpatient medical wards or emergency settings. Within this framework, LCP holds a strategic position in general hospitals, frequently encountering patients admitted to medical units for stabilization following self-harm or suicide attempts (SA). This context offers a unique opportunity to identify and characterize individuals at high risk for suicidal behaviors, facilitating access to services, integrated care, and the implementation of early interventions that could interrupt the cycle of recurrent attempts (2). A core challenge in both research and clinical practice lies in accurately differentiating suicidal ideation (SI), SA, and non-suicidal self-injury (NSSI). This distinction carries critical implications for risk assessment, treatment planning, and legal considerations in hospital settings, where every decision can profoundly impact patient outcomes. Undeniably, suicide and suicidal behaviors represent major public health concerns with profound clinical, social, and economic implications that make suicide prevention a priority at the global level (3). The World Health Organization estimates over 700,000 annual suicide deaths worldwide, positioning suicide among the leading causes of death in young people and highlighting the urgency of prevention strategies across healthcare settings (4). The Global Burden of Disease study reported approximately 746,400 deaths and 33.5 million disability-adjusted life years due to self-harm—predominantly suicide—in 2021 (5). Suicide ranks as the third leading cause of death among individuals aged 15–29 years (6). In Italy, national surveillance data indicate thousands of annual suicide deaths, with persistent regional heterogeneity—such as elevated rates in northern regions—and temporal fluctuations in external-cause mortality (7, 8). Notably, non-fatal suicidal behaviors, particularly SA, far outnumber suicide completions, possibly serving as critical endpoints for secondary prevention, offering a window for timely psychiatric intervention (9). Risk factors for SA are multifactorial, spanning individual, clinical, sociodemographic, social, and environmental domains. At the individual level, psychiatric disorders—particularly major depressive disorder (MDD) and bipolar disorder (BD), schizophrenia-spectrum disorders, and substance use disorders—represent the strongest predictors (10, 11), alongside a history of prior attempts (12, 13) and NSSI (14). As for the latter, some evidence suggests NSSI may acutely reduce SI in the short term, functioning as a maladaptive coping mechanism for suicidal states, though it ultimately escalates long-term risk through habituation to self-injury (15). Additional clinical correlates of suicidality include comorbid anxiety disorders, SI, high impulsivity, aggression, psychological pain, and childhood adversity, which often interact synergistically. On the other hand, sociodemographic factors encompass younger age, female sex, unemployment, single/divorced/separated status, and lower educational attainment (16, 17). Social and environmental contributors include recent stressors, socioeconomic deprivation, social isolation, and exposure to suicide in family or community settings (17). Despite the crucial role of specific nosographic entities in defining higher suicide risk, diagnostic challenges in LCP must be acknowledged. Among these, the adequate and timely identification of mood disorders deserves specific attention, particularly concerning the elevated risk of misdiagnosis between MDD and BD. To note, BD carries an elevated suicidal burden, with risk amplified by younger age at onset, depressive predominant polarity, comorbid anxiety/substance use/cluster B personality disorders, and family history of suicide (10, 18). Patients labelled with recurrent treatment-resistant unipolar depression may represent unrecognized bipolar spectrum cases, especially when clinical information concerning illness course, mixed features, or family history are missing. Personality disorders also increase suicidal risk, particularly cluster B disorders, which are characterized by emotional dysregulation (19). Nevertheless, any personality disorder diagnosis heightens risk, compounded by substance use, insecure attachment, and childhood trauma, creating a complex interplay that demands holistic evaluation (20). In time-constrained LCP settings, identifying personality disorders remains challenging but essential for risk stratification and tailored discharge planning.

Despite the clinical salience of suicidal behaviors in general hospitals, data collected in Italian LCP settings concerning possible risk factors for suicidality are scant. This retrospective study aims to delineate sociodemographic, psychosocial, and clinical, correlates of SA among patients assessed by an LCP service in a tertiary general hospital, integrating the evaluation of self-injury history, diagnostic features, and psychosocial stressors in a real-world setting. Given the crucial role that LCP settings play in early detection and intervention, results from the present study may help enlightening a comprehensive risk profile to bolster community prevention efforts and reduce future attempts.

In this LCP sample, SA accounted for more than one third of all psychiatric referrals from medical wards, confirming suicidality as a highly prevalent and clinically relevant condition in the general hospital setting. This is consistent with previous findings, also confirming high suicidal risk in LCP setting (29) and suggesting that suicidality should routinely be investigated in the general hospital, not depending from the reason of admission. Beyond prevalence, our findings contribute to a more refined understanding of individual and clinical profiles of patients evaluated after a SA, identifying a limited set of robust clinical and psychosocial correlates that retain significance in the multivariable analysis. The strongest associations with a current SA were observed for previous NSSI and previous SA, both remaining highly predictive after adjustment for psychiatric diagnoses and recent stressors, in line with previous findings (12–14). This pattern supports the conceptualization of suicidal behavior as a longitudinal and dynamic process rather than an isolated event, in which repeated exposure to self-injury progressively lowers the threshold for subsequent suicidal acts. In this framework, NSSI emerges not merely as a comorbid behavior, but as a marker of enduring vulnerability, possibly related to impaired emotion regulation, behavioral disinhibition, and habituation to self-inflicted harm (30). The magnitude of this association in a medically hospitalized population underscores the clinical relevance of systematically assessing lifetime NSSI during liaison-consultation evaluations, even when the referral reason is not primarily psychiatric. Current depressive disorders and personality disorders were also independently associated with SA, as expected on the basis of previous findings in LCP settings (10, 19, 20). While the association between depressive features and suicidality is well established (31), our findings suggest that diagnostic categories retain predictive value even when proximal affective symptoms are excluded from the regression model. This supports a dimensional interpretation in which depressive disorders reflect broader vulnerability traits, including hopelessness, cognitive rigidity, and reduced stress tolerance. It should also be noted that more than a half of SA patients were assessed as euthymic during the psychiatric consultation. This finding support a dynamic framework for the acute suicide crisis, challenging the traditional opinion that suicidality arises from persistent depressive states. Indeed, some individuals may exhibit a decline in depressive symptoms in the weeks following a SA (32). The transient reduction in subjective distress following a SA may be the resultant of complex and potentially intertwined phenomena, e.g., emotional exhaustion, temporary relief after externalizing unbearable psychological pain, or the perception that the crisis has been momentarily resolved. However, this apparent “calm after the storm” may be misleading, since suicidal behavior should be conceptualized as a dynamic and longitudinal process rather than a discrete event, in which acute post-attempt calm may coexist with persistent vulnerability and elevated risk of recurrence. In this perspective, the association with personality disorders highlights the contribution of enduring psychopathological features—affective instability, impulsivity, and interpersonal dysregulation, as specifically seen in cluster B personality disorders (33) and people with high neuroticism (34)—that may amplify the impact of acute stressors and medical illness. In LCP settings, where diagnostic assessment is often constrained by time pressure and overall clinical complexity, the identification of personality characteristics may be particularly challenging but appears clinically meaningful for suicide risk stratification. An additional, non-mutually exclusive interpretative hypothesis is that part of the observed association between depressive and personality disorders and SA may reflect the presence of bipolar spectrum conditions that are not fully recognized at the time of assessment. In real-world clinical practice, both in acute hospital settings and in community mental health services, diagnostic formulations may legitimately prioritize symptom management and functional stabilization, while longitudinal features such as mood periodicity, polarity shifts, and lifetime course may receive less systematic attention. Within this framework, recurrent or treatment-resistant MDD diagnosis may, in some cases, represent provisional or pragmatic formulations applied to patients with underlying mood instability. Although this hypothesis cannot be directly tested within the present study, it underscores the importance of a longitudinal, course-based diagnostic approach in LCP, particularly in patients presenting after suicidal behavior. This should be empowered by the systematic evaluation of specific “red flags” that should be considered among risk factors for a bipolar diathesis, e.g., familiar history of BD and early onset (35, 36), which also turned out to be significant in our analysis. The presence of stressful life events in the six months preceding hospitalization emerged as an additional independent correlate of SA. This finding aligns with stress-diathesis models of suicidality, in which environmental stressors interact with pre-existing vulnerability to precipitate suicidal behavior. Importantly, stressful events retained significance after controlling for psychiatric diagnoses and prior suicidal behaviors, supporting their role as proximal triggers rather than epiphenomena. From a clinical standpoint, this reinforces the importance of integrating psychosocial assessment into LCP practice, rather than limiting evaluations to symptom-based screening or diagnostic labels (37). Moreover, the importance of life adversities in the suicidal crisis suggests that a thorough screening of these events should be performed with special interest to specific vulnerable groups, that are both prone to suicidality and often experience highly stressful events (38). Age also showed a significant association with suicidality, with younger individuals displaying a higher likelihood of SA. This result is consistent with epidemiological data indicating higher rates of SA and non-fatal suicidal behavior among younger patients (39, 40) and suggests that age-related factors may be particularly relevant in hospital-based populations. Several developmental and contextual determinants may contribute to this pattern, including greater impulsivity, emotional reactivity, and exposure to acute interpersonal stressors during adolescence and young adulthood. Moreover, this life stage is characterized by identity instability and ongoing psychosocial transitions, which may amplify vulnerability to suicidal crises (41, 42). Contemporary stressors, such as social media-related pressures, peer comparison, and cybervictimization, have also been linked to increased suicide risk in younger populations. Conversely, older adults may be underrepresented in hospital-based liaison-consultation samples because suicidal behavior in later life is more likely to result in medically lethal outcomes, thereby reducing the probability of subsequent psychiatric evaluation (17). Within this framework, the observed age effect likely reflects both developmental vulnerability and selection mechanisms intrinsic to general hospital settings. These findings support the need for age-sensitive preventive strategies, with particular attention to early identification and intervention in younger patients presenting with self-injurious behaviors. Although female sex was associated with SA at the bivariate level, it did not retain significance in the multivariable model, indicating that sex differences may be largely mediated by clinical and psychosocial variables rather than representing an independent risk factor in this context. Beyond variables that remained significant in the multivariable model, several additional factors showed significant differences between SA and non-SA patients at the bivariate level. This pattern suggests that these features may primarily operate as indicators of overall clinical severity and of more intensive prior contact with specialist services, rather than as specific and independent risk markers for SA in the acute hospital setting. Foreign nationality also emerged as significantly associated with SA at the bivariate level. This finding warrants clinical consideration, as migration-related factors may meaningfully shape vulnerability to suicidal behavior (43). Indeed, individuals with a migrant background often face cumulative psychosocial stressors, including language barriers, limited access to healthcare, cultural incongruence in help-seeking behaviors, and experiences of marginalization (44). Acculturative stress—defined as the psychological burden associated with adapting to a new sociocultural environment—has been consistently linked to increased depressive symptoms, psychological distress, and suicidal ideation, particularly in recently migrated or socially isolated individuals (45). Reduced mental health literacy and communication difficulties may hinder early detection of affective symptoms, leading to delayed referral and presentation at more advanced stages of crisis in general hospital contexts (46). From a clinical perspective, this may underscore the importance of culturally sensitive assessment in LCP, with systematic attention to language proficiency, migration history, social support, and acculturative difficulties. Integrating these dimensions into routine risk assessment may improve the understanding of suicidal crises and facilitate more tailored discharge planning and continuity of care. In line with previous work showing that the apparent impact of sociodemographic and treatment-related variables often attenuates after controlling for psychiatric disorders and prior suicidal behaviors, our findings support the notion that the core vulnerability is carried by lifetime suicidal trajectories and underlying psychopathology, while other bivariate correlates may function as proxies of such vulnerability or markers of clinical selection into care. From a clinical perspective, the higher intensity and complexity of pharmacological regimens and psychosocial interventions in the SA group likely mirror histories of recurrent crises and persistent risk, rather than implying a direct causal role of specific medications or therapeutic settings in precipitating attempts. Overall, these results support a model of suicidality in which LCP serves not only as an interface service, but as a critical clinical setting for identifying enduring vulnerability markers that may otherwise remain unrecognized in medically ill patients. Indeed, the post-attempt phase represents a clinically critical window, during which reduced overt distress and a state of apparent “calm” may often obscure persistent drivers of suicidality.

Suicidality represents a frequent and clinically complex presentation in liaison-consultation psychiatry. In this general hospital sample, suicide attempts were most strongly associated with prior self-injurious behaviors, depressive and personality disorders, recent stressful life events, and younger age, rather than with acute symptom severity alone. These findings highlight the importance of a longitudinal and multidimensional approach to suicide risk assessment during psychiatric consultations in medical wards, with systematic attention to lifetime suicidal behaviors and NSSI. In this context, integrating course-based diagnostic perspectives may be particularly relevant to avoid oversimplified formulations centered exclusively on current symptom states, especially in patients presenting after suicidal behavior. Liaison-consultation psychiatry offers a unique opportunity to intercept high-risk individuals at a critical point of contact with healthcare services, particularly those who are not engaged in ongoing psychiatric care. Early identification of patients with established vulnerability profiles should be prioritized for targeted preventive strategies and structured follow-up pathways. Future longitudinal studies are warranted to determine whether interventions initiated during medical hospitalization can effectively modify long-term trajectories of suicide risk and reduce recurrence.