The key aims of this study were to identify and describe a subgroup of university students who show high levels of psychiatric symptoms but do not have suicidal thoughts and to compare them with both peers who have similar symptoms and have suicidal thoughts and healthy students. The study successfully found a distinct subgroup—called the invisible group—characterized by high depressive symptoms, poor sleep quality, and increased impulsivity, even though they do not have suicidal thoughts. Although these individuals did not express thoughts of death or self-harm, they showed significant psychological distress and behavioral risks.

Importantly, their profiles differed not only from those of healthy students with low symptom burden, but also from those of the visible group—students with similarly high symptomatology who did endorse suicidal ideation. These differences were especially noticeable in sleep quality, interpersonal distress, and cannabis use. The existence of this hidden subgroup highlights a significant limitation of traditional screening methods, which often rely solely on detecting suicidal ideation as a risk indicator [25,26]. These findings challenge the implicit assumption that the absence of suicidal ideation reflects low risk, suggesting instead that individuals may experience substantial psychological vulnerability even without endorsing suicidal thoughts. The findings suggest that a notable portion of high-risk individuals may go unnoticed when suicidal ideation is absent or unreported, emphasizing the need for more comprehensive, symptom-based approaches to suicide risk assessment. Importantly, the identification of this group should not be interpreted as indicating direct or imminent suicide risk. Rather, the “invisible” group appears to reflect a state of significant psychological distress and behavioral vulnerability that may remain undetected when screening relies exclusively on suicidal ideation. This distinction is crucial, as not all individuals with elevated psychiatric symptoms will develop suicidal behavior, but some may nonetheless represent a population requiring clinical attention and early intervention [27].

The invisible group showed sleep and affective impairments that were intermediate between the visible and healthy groups. Specifically, their mean PSQI score was substantially higher than that of the healthy group, indicating clinically poor sleep, but slightly lower than that of the visible group. This is consistent with the existing literature showing that sleep disturbances are a transdiagnostic vulnerability factor for emotional dysregulation, mood instability, and suicide risk, even when suicidal ideation is not present [10,25]. Likewise, levels of thwarted belongingness and perceived burdensomeness were elevated in the invisible group, reflecting significant interpersonal strain. Although these constructs are classically studied in relation to suicidal ideation and behavior [24,28], our findings suggest that they are also relevant for understanding psychological suffering in individuals who do not express suicidal thoughts.

A further mechanism that may help explain the clinical vulnerability of the “invisible” group concerns the effects of sleep deprivation on frontolimbic connectivity and behavioral regulation. Experimental and neuroimaging studies consistently show that sleep loss is associated with functional disconnection between the amygdala and prefrontal control regions, particularly the medial and dorsolateral prefrontal cortex. Under sleep deprivation, the amygdala exhibits exaggerated reactivity to emotional stimuli, while top-down inhibitory control from the prefrontal cortex is markedly reduced [29,30]. This neurobiological pattern mirrors that observed in individuals characterized by affective instability, impulsivity, and impaired decision-making.

Beyond emotional dysregulation, sleep deprivation has been shown to alter higher-order cognitive and moral processes. Laboratory studies indicate that insufficient sleep increases risk-taking, reduces sensitivity to negative consequences, and alters ethical and moral decision-making, leading to more self-interested, norm-violating, or dishonest behavior [31,32]. These effects appear to be mediated by impaired prefrontal functioning and reduced integration of affective signals into decision-making.

Within this framework, individuals in the invisible group, who report poor sleep quality yet deny suicidal ideation, may still experience heightened emotional reactivity alongside weakened cognitive and behavioral control. This condition may not present as explicit suicidal thoughts but rather as behavioral dysregulation, substance use, and interpersonal difficulties, thereby increasing vulnerability to impulsive or unplanned self-harm under acute stress. From a clinical perspective, this supports the view that sleep disturbances may act not only as correlates of suicidal ideation but also as transdiagnostic facilitators of risk by disrupting the neural systems underlying emotional regulation, moral judgment, and behavioral inhibition.

One of the most salient findings was the high prevalence of cannabis use in the invisible group. Over half of these individuals reported lifetime use of cannabis—substantially more than in the Visible group and more than double the rate observed in the Healthy group. Interestingly, the invisible group also exhibited the highest cannabis-to-alcohol use ratio, suggesting a relative preference for cannabis as a coping mechanism or form of self-regulation. This pattern may reflect attempts to manage dysphoric affect, anxiety, sleep difficulties, or social disconnection through substance use, particularly among individuals who are reluctant to disclose or confront suicidal ideation. Prior research has shown that cannabis use is often employed by young adults as a means of affect modulation and stress management, particularly in the context of psychiatric vulnerability [33,34,35].

However, the relationship between cannabis use, sleep disturbance, and mood symptoms is likely bidirectional. Cannabis may be used as a maladaptive coping strategy to manage sleep difficulties or emotional distress, while at the same time potentially exacerbating sleep disruption, affective instability, and cognitive dysregulation.

Indeed, this behavior may mask underlying risk by providing transient relief while maintaining emotional suppression and social disengagement.

The concept of the “invisible” group also aligns with emerging theoretical critiques of ideation-focused suicide models. While many frameworks, including the Interpersonal Theory of Suicide [36] and the Integrated Motivational–Volitional Model [37], recognize that suicidal ideation is not always a precursor to attempts, clinical practice often continues to rely on the presence of explicit suicidal thoughts as a gatekeeper for intervention. However, longitudinal studies and meta-analyses have shown that a significant proportion of suicide attempts occur in individuals who did not previously endorse ideation [7,38]. Our data underscore this concern: the invisible group exhibits traits commonly linked to suicide risk, such as impulsivity, poor sleep, substance use, and interpersonal distress, despite the absence of ideation. These individuals may represent a group at elevated latent risk who could transition rapidly to suicidal behavior under acute stress or crisis.

Clinically, this study highlights a critical blind spot in mental health screening practices. University campuses increasingly employ brief risk assessments and digital tools to identify students in crisis [39,40]. In practical terms, screening protocols based solely on suicidal ideation may fail to identify a subgroup of students who are already experiencing clinically significant distress but remain undetected within existing systems.

While these are valuable, they often prioritize detecting suicidal ideation as the primary indicator of need. Our findings suggest that such strategies may fail to capture a sizable segment of students who, while not endorsing suicidal thoughts, still experience serious psychological difficulties and engage in potentially harmful behaviors. The current findings advocate integrating broader symptom-based criteria, particularly those involving sleep, impulsivity, and social disconnection, into risk screening protocols. Doing so may help identify students in the early stages of distress who would otherwise remain invisible to the clinical system until their difficulties escalate.

An additional aspect that warrants consideration is the potential role of sex and gender differences in shaping the observed patterns. Although no significant differences in sex distribution were found across groups, it is well established that males and females may differ in the expression of depressive symptoms, sleep disturbances, impulsivity, and substance use behaviors [41,42]. For instance, females tend to report higher levels of internalizing symptoms such as depression and sleep difficulties, whereas males may exhibit higher levels of externalizing behaviors, including substance use and impulsivity.

These differences may influence not only the manifestation of psychological distress but also help-seeking behaviors and the likelihood of reporting suicidal ideation. In particular, some individuals—especially males—may be less likely to disclose suicidal thoughts despite experiencing significant distress, potentially contributing to the identification of “invisible” high-risk profiles [42,43].

From a clinical perspective, these findings highlight the importance of adopting gender-sensitive screening and intervention strategies, which take into account different pathways to distress and risk. Future research should further explore sex- and gender-specific mechanisms using stratified analyses or interaction models to better understand how vulnerability unfolds across different populations.

From a clinical and service perspective, these findings indicate that university mental health services could benefit from adding brief symptom-based screening tools into existing prevention strategies. In practice, this might involve using standardized measures of sleep quality (e.g., PSQI), depressive symptoms (e.g., CES-D), and impulsivity in routine digital or face-to-face screening procedures, even if students do not report suicidal thoughts. Students showing high symptom levels across multiple areas could then be provided with early, easily accessible interventions, such as psychoeducation, sleep-focused programs, or short-term psychological support. Furthermore, stepped-care models could be adjusted to treat this group as a priority for monitoring, rather than relying only on triage based on ideation. These approaches may enhance early detection and lower the risk of escalation for individuals who might otherwise go unnoticed.

Furthermore, the high rates of cannabis use among invisible participants raise important questions for prevention and health education. While cannabis is increasingly perceived as a benign or normalized substance among young adults, its use in the context of poor sleep, mood dysregulation, and social withdrawal may represent a maladaptive trajectory. These findings suggest the need for targeted interventions that address not only substance use per se, but the emotional and social contexts that make it appealing to at-risk individuals. Interventions targeting emotion regulation, peer connectedness, and non-substance-based coping strategies could be particularly beneficial in this subgroup.

This study is not without limitations. First, the cross-sectional design prevents any conclusions regarding directionality or causality among the observed relationships. It remains unclear whether poor sleep quality constitutes a risk factor for suicidal ideation or a consequence of it, or whether impulsive behavior precedes or results from substance use. Future longitudinal or prospective studies are needed to clarify the temporal sequence of these phenomena and to identify potential causal pathways linking them.

A further limitation concerns the assessment of suicidal ideation, which was based on a single dichotomous self-report item (“yes/no in the past three months”). While this approach is practical for large-scale epidemiological surveys and has been widely adopted in population-based research, it does not capture important clinical distinctions such as passive versus active ideation, intensity, persistence, or degree of impulse control. In adult populations, suicidal ideation represents a heterogeneous and clinically meaningful construct, and its presence is generally considered a psychiatric emergency regardless of severity. The use of a binary item, therefore, limits the ability to detect subthreshold, fluctuating, or ambivalent forms of ideation and may result in an underestimation of the true prevalence and complexity of suicidal thoughts. Future research should employ multi-item, validated instruments—such as the Beck Scale for Suicidal Ideation or the Columbia–Suicide Severity Rating Scale—to provide a more nuanced and clinically informative assessment of suicide risk.

The grouping strategy, based on a simultaneous median split of depressive symptoms (CES-D), sleep quality (PSQI), and impulsivity (UPPS-P), represents another methodological limitation. Although this approach allowed for the identification of a clearly defined subgroup characterized by a convergent and high psychiatric burden, it is inherently arbitrary and reduces the dimensional complexity of the data. In particular, requiring participants to exceed the median on all three measures may have excluded clinically relevant profiles, such as individuals with severe insomnia and depressive symptoms but only moderate impulsivity. This criterion increases group selectivity while reducing representativeness and may underestimate the heterogeneity of high-risk presentations.

This selection strategy may also limit external validity, as individuals with partially overlapping or discordant symptom profiles were excluded, potentially reducing the representativeness of the analyzed sample.

Importantly, the absence of data on suicidal behaviors (e.g., suicide attempts or non-suicidal self-injury) represents a key limitation. As a result, it is not possible to determine whether individuals in the “invisible” group are at increased risk of engaging in suicidal behavior. Therefore, the findings should be interpreted as reflecting psychological vulnerability rather than direct suicide risk. Future studies should incorporate detailed assessments of suicidal behaviors to better clarify the clinical significance of this subgroup within a suicide prevention framework.

Future studies should therefore favor dimensional or data-driven approaches, including multivariate regression models, clustering techniques, or latent class analyses, to better capture the continuity and complexity of psychiatric vulnerability.

The measurement of substance use also poses interpretative challenges. The dichotomous “ever used” classification does not distinguish between occasional, habitual, or problematic use, nor does it capture motivational aspects such as self-medication versus social or recreational use.

In addition, the dataset did not include information on other substances or psychotropic medications (e.g., benzodiazepines), which may also influence sleep, mood, and behavioral regulation.

Consequently, associations between substance use and suicidal ideation should be interpreted as preliminary. Future studies should incorporate more detailed quantitative and qualitative assessments to better characterize substance use patterns and their clinical significance.

From a statistical perspective, the reliance on bivariate analyses (ANOVA and chi-square tests) limits the ability to account for potential confounders. More sophisticated multivariate models—such as logistic or linear regression analyses—would allow for the inclusion of relevant covariates (e.g., age, sex, socioeconomic status, psychiatric history), thereby clarifying the independent contribution of each psychological factor.

In particular, relevant confounding variables such as psychiatric history, psychotropic medication use, socioeconomic factors, and recent life stressors were not available in the dataset and could not be included in the analyses, potentially influencing the observed associations.

In addition, the absence of sensitivity analyses and limited information regarding missing data management may affect the robustness and generalizability of the findings. Future research should adopt strategies such as multiple imputation and sensitivity testing to minimize potential bias.

Interpretatively, the observed associations suggest a possible interaction between depressive symptoms, sleep disturbances, and impulsive traits in conferring vulnerability to suicidal ideation. However, the modest effect sizes indicate that these factors alone are unlikely to fully explain suicide risk. Other dimensions—such as emotion regulation capacities, social connectedness, and coping strategies—may act as mediators or moderators. Future research should therefore investigate integrated, multifactorial, and transdiagnostic models combining psychological, biological, and contextual variables.

Alternative analytical approaches, such as decision tree models or machine learning techniques, may offer additional insights into complex interactions among risk factors and could enhance the identification of high-risk subgroups in future studies.

Finally, the generalizability of the present findings is limited by the use of a single university-based sample, which may not be representative of the broader population of young adults or individuals outside academic settings. Replication in multicenter, cross-cultural, and age-diverse cohorts will be essential to validate these preliminary observations and to inform more inclusive and generalizable prevention strategies.

In conclusion, this study provides a valuable initial insight into subtle psychological vulnerabilities that may underlie suicidal ideation among ostensibly healthy young adults. However, its findings should be viewed as exploratory rather than definitive. The main takeaway is that depressive symptoms, poor sleep, and impulsivity represent potential indicators of latent risk rather than deterministic predictors. Clarifying causal mechanisms, testing mediation and moderation effects, and adopting robust multivariate models in longitudinal frameworks represent key directions for future research aimed at developing reliable screening and early prevention strategies.

Despite these limitations, this study offers a novel contribution by empirically identifying a subgroup of high-symptom individuals who remain undetected by ideation-focused screening methods. The findings support a reconceptualization of suicide prevention approaches that extend beyond self-reported suicidal thoughts, emphasizing instead broader symptom-informed frameworks capable of capturing risk in its early, less visible forms.