Non-suicidal self-injury (NSSI) is defined as behavior of intentional harm to one’s body tissue in the absence of a suicidal purpose (1), although several studies have identified frequent coexistence between NSSI and suicidal behavior (2, 3). It is a transdiagnostic phenomenon; in fact, it is associated with a wide variety of psychiatric disorders in both outpatient and inpatient samples (4–7). Research showed a NSSI prevalence in nonclinical adolescent populations of 20% (8). This tends to rise in clinical populations to over 50% (5, 9). A further increase in NSSI among adolescents has been recorded during the Covid-19 pandemic (10, 11). The NSSI methods can be multiple (12), such as “cutting,” “scratching,” “head banging,” “burning,” and “hitting” (13, 14).

Nock and Pristein (2004) proposed a model that conceptualizes the different functions of NSSI along two axes: the positive/negative reinforcement axis and the intrapersonal/interpersonal axis (15). Subsequent research has shown that the one most reported is the intrapersonal function (16) in both inpatients (17) and outpatients (18). Unlike interpersonal function, intrapersonal function also positively predicts the severity of NSSI in terms of method versatility and NSSI frequency (19). The severity of the clinical picture has also been associated with more injured body sites or an injury in a body site other than the arms (20). Non suicidal self-injurers often reported greater psychosocial impairment (21, 22) and alexithymia, which is the difficulty in identifying and describing emotions (7, 23, 24). Moreover, Andover and Morris (2014) showed an association between NSSI and a low ability to regulate emotions adaptively (25). As a consequence, NSSI can be used as a dysfunctional coping mode that allows avoidance or reduction of negative emotions (15, 26).

The literature has also investigated the social and family context associated with NSSI, identifying dysfunctional family dynamics (27, 28), particularly in terms of the emotional dimension. A history of emotional abuse (29), a lack of parental emotional validation, a prevalence of negative affect and control (30), more rigid punitive methods, and an insecure parent-child attachment relationship (31) have been found. In addition, Halstead and colleagues (2014) reported poor family cohesion and communication issues in these families (32), whereas other authors found high family conflict (33, 34), low parental monitoring (34), and poor relationships with caregivers (33). A study conducted by Wang and colleagues (2022) showed a mediating effect of depressive symptoms on the relationship between poor family functioning and NSSI (35). Specifically, it has been suggested that the onset of depressive symptoms within a dysfunctional family context might be linked to the low development of the child’s ability to regulate emotions (36).

Emergency Department (ED) presentation for self-harm has increased in the last years, and this trend has accelerated, most of all in young females, since the Covid-19 outbreak (37). Nevertheless, a systematic review and meta-analysis conducted by Witt and colleagues (2023) showed that just one-in-five adults and young people who were referred for inpatient treatment were admitted following hospital presentation for self-harm, whereas just over half of young people referred for outpatient treatment after hospital presentation for self-harm attended at least one treatment session (38). In addition, Bridge and colleagues (2019) noted that only about one in four cases resulted in an outpatient follow-up appointment being scheduled before the patient leaves the emergency room (39), despite non-suicidal self-injury — especially if it occurs repeatedly over time — is thought to be a precursor of suicidal behavior (40). Moreover, the time after hospital discharge is considered to be the peak risk period for both non-suicidal self-harm and suicide (38). The choice regarding the level of care (hospitalization versus outpatient visit), in turn, depends on a number of factors that relate not only to suicidal risk (i.e., previous history of suicidal behaviors, depressive disorders, and high-lethality self-harm), but also to the severity of the individual’s psychosocial functioning, the ability of the family environment to adhere to and support treatment choices, and the need for ongoing monitoring of the patient (41). Family’s level of distress and the parental capacity to contain the child also play a role in the decision-making on the child’s psychiatric hospitalization (42).

Few studies have compared NSSI outpatients with NSSI inpatients. We are aware of only one study that investigated the course of suicidal and non-suicidal self-injury between the two clinical samples, identifying in the hospitalized sample a lower age of onset of such self-injurious phenomena and a significantly higher presence of suicidal ideation and suicidal planning (43). This study considered the current level of care as an inclusion criterion, such that some outpatients might have had previous hospitalizations. Further research, focusing exclusively on groups of self-harm inpatients, has identified an increased suicide risk among those with a previous hospitalization for self-harm (44) and a tendency for repeated self-harm over time, which, in turn, is influenced by traits of emotional dysregulation and internalizing symptoms (45). Moreover, other studies found that both externalizing (e.g., disruptive behavior disorder) and internalizing (e.g., anxiety and depressive disorder) problems were associated with inpatient treatment in adolescents (46, 47).

The current study aimed to further expand knowledge regarding features that characterize young self-harmers who receive different levels of care (i.e., inpatient admission vs. outpatient visits), with particular attention on psychopathological, family, and non-suicidal self-harm-related characteristics, as well as suicidality. Specifically, the objectives of the study were as follows: 1) identifying any difference between young NSSI inpatients and young NSSI outpatients in the non-suicidal self-harm-related characteristics (i.e., NSSI methods, NSSI frequency, NSSI method versatility, NSSI function, and number of injured body sites), suicidality (i.e., lifetime suicidal ideation and lifetime suicide attempts), and psychopathological features. We have considered variables that have previously been associated with NSSI, such as alexithymia, emotion dysregulation, psychosocial functioning, externalizing problems, internalizing problems, and total problems. According to the previous findings (41, 43, 48), we supposed that NSSI inpatients reported more lifetime suicidality, greater NSSI severity, and impaired psychosocial functioning than the outpatient group. Moreover, based on studies that focused on hospitalization-related psychopathological factors in psychiatric populations (46, 47, 49), we hypothesized that the inpatient group showed a greater severity of externalizing, internalizing, and total problems than the outpatient group, as well as more severe emotion dysregulation; 2) pinpointing any difference between young NSSI inpatients and young NSSI outpatients in the family interactive-relational variables and investigating any changes in family interaction dynamics in the inpatient and the outpatient groups. Plener and colleagues (2016) highlighted that inpatient treatment for youngs with NSSI was indicated when the environment is detrimental for the recovery in an outpatient setting (41), whereas other studies found that hospitalization in a Child Neuropsychiatric Unit was associated with intrafamily issues (50–52). Despite these findings, we are not aware of studies that focused on specific interactive family dynamics associated with hospitalization in NSSI cases; thus, we did not formulate any hypothesis and we decided to proceed in an explorative way.

The main purpose of the current study was to investigate psychopathological, family, and self-harm- related features that distinguish young NSSI inpatients from young non-suicidal self-injurers who received outpatient treatment. According to Glenn and colleagues (2017) (43), we found that inpatient self-harmers reported higher rates of lifetime suicidal ideation and lifetime suicidal attempts than the outpatient group. Differently from Glenn and colleagues (2017) (43), we did not include outpatients with a previous history of psychiatric hospitalization; thus, this highlighted that the two groups differ from the intensity of level of care received in their lifetime for mental health issues. Our findings were consistent with the fact that hospitalization is often the treatment of choice for individuals with high suicidal risk (69). Contrary to our expectations, the inpatient group did not show more severe NSSI in terms of method versatility and NSSI frequency, even though the inpatient group reported higher rates of scratching, which was associated in the previous studies with a more severe risk profile (70) and with interpersonal conflict management and regulation of negative emotions (71). Consistently, we found that the inpatient group showed a more severe emotion dysregulation profile, higher alexithymic traits, and worse psychosocial functioning than the outpatient group. In addition, clinical levels of externalizing and internalizing problems were more associated with the inpatient treatment. Overall, these findings supported a more severe global clinical picture of self-harmers who were required to have a more intensive level of care. Although the association between the level of care and certain variables (such as lifetime suicidal ideation, externalizing problems, and psychosocial functioning) was lessened when a drug regimen was taken into account, the degree of underlying psychopathology — rather than the severity of the NSSI characteristics — seems to be what distinguishes the inpatient sample from the outpatient sample. This suggests that NSSI should not be viewed as a symptom in and of itself, but rather as an expression of a larger psychopathological picture that requires thorough investigation.

The evaluation of interactive family dynamics through the LTP procedure highlighted that the inpatient group showed better scores in the variables ‘structure’ and ‘interactive mistakes during activities’ than the outpatient group. In fact, hospitalization at a Child Neuropsychiatric Unit requires that all family members define new roles and responsibilities together, adjust their daily routines and adapt themselves to the strict rules and routines imposed by the admission (72–74). As a result, our findings may be understood in view of the fact that the inpatient group’s families have become accustomed to changing their activities as part of an adaptation process in response to external circumstances (e.g., the hospital admission) and strictly adhering to explicit hospital rules. Moreover, these findings emphasize the role of hospitalization as a structural framework within which the young patient and family can get care and containment at a time when the family setting alone cannot provide a safe and conflict-free environment for the child’s growth.

In addition, unlike the outpatient group’s families, the inpatient group’s families showed a decrease in the overall interactive-relational performance during the course of the LTP four-part scenario, reaching the global lowest score in the fourth part, in which the play situation is mostly managed by the co-parenting system. This decreasing trend has also been found in some of the specific LTP interactive-relational variables, such as ‘role implication’, ‘parental scaffolding’, ‘conflicts’, ‘adolescent’s involvement’, and ‘adolescent’s self-regulation’. A two-way process emerged: both the parents’ ability to monitor and stimulate appropriately the child and the child’s ability to self-regulate and appropriately involve in family interactions collapsed into the fourth part of the family play situation, when the co-parenting system was left to manage the family interactions and the adolescent simply had to stay nearby. At this stage, conflicting and interfering parental interactions peaked.

A further contribution of our research consists of showing that the adolescent’s self-dysregulation is not a fixed dimension but changes across the LTP four-part scenario, according to different triadic configurations. It reached its maximum peak when the adolescent should act as a simple observer within a family context characterized by lowered parental scaffolding and high conflicts. These findings were consistent with the biopsychosocial model of mental health (75), which emphasizes the role of the family and social environment for the child’s development (28). Previous studies highlighted the role of the co-parenting system in the child’s mental health issues (76, 77). In fact, it has been demonstrated that families of hospitalized children were often characterized by high interparental conflicts (50, 51) and low parental capacity to contain the child (42), which, in turn, were associated with a low child’s ability to regulate own emotions (78, 79). To the best of our knowledge, the current study is the first to report findings on the changes in interactive family dynamics using an observational semi-standardized tool in hospitalized adolescents with self-harming, while also disentangling family, psychopathological, and NSSI-related features from an adolescent NSSI outpatient sample. Overall, our results highlighted the relevance of evaluating the co-parental system during the hospitalization process of young self-harmers using multi-informant and multi-method instruments, given that the quality of interparental interactions may have a role in the child’s self-regulation.

The current study presents some limitations. First, because of the current study’s small sample size and the sample enrollment occurring only at two Child Neuropsychiatry Units in northern Italy, it is necessary to be cautious in generalizing the results. Particularly, we were able to obtain data on interactive family dynamics just for a small subgroup of self-harmers due to the presence of family limitations (e.g., problematic family situations). Future research involving multisite sampling across different areas of Italy are needed. Moreover, all patients were females; thus, it did not allow us to extend the findings to the male population of self-harmers. Further studies should enroll larger samples, including male self-harmers, to conduct subgroup analysis according to gender and increase the external validity of study results. Given the cross-sectional nature of this study, it poses a challenge to determine whether the findings pertaining to clinical data and family functioning also serve as a predictor for the type of psychiatric treatment that follows. Hence, prospective longitudinal studies are needed for further research in this field. Although similar inclusion and exclusion criteria between groups have been chosen (with the exception of the variable ‘previous psychiatric hospitalizations’), it is possible that other sources of selection bias acted during the sample enrollment, given the several factors that influence the choice of the outpatient level of care vs. hospitalization. We only controlled the effect of gender, age, and psychiatric diagnosis on results by using pair matching without considering other potential confounding variables such as psychiatric comorbidity, physical illness, psychiatric familiarity, intrafamily issues, and traumatic events, which should be included in future research. The absence of assessment of the influence of other family factors (such as the number of family members, family income, and family beliefs) on interactive family dynamics is an additional limitation of the study. Finally, we did not consider other environmental factors such as negative peer relationships or school-related problems, which could contribute to mental health issues. The multi-informant and multi-method approach used in this study allows to overcome limitations connected to the exclusive use of self-report instruments, which may be influenced by individual biases such as social desirability or failure to understand questions. In conclusion, disentangling throughout multi-informant and multimethod evaluation the psychopathological, family, and NSSI-related characteristics of young patients with self-harm who required different levels of care (hospitalization vs. outpatient visits) is a promising research field with high clinical relevance for the advancement of the diagnostic and therapeutic process that characterizes daily clinical practice.

The current study highlighted that NSSI inpatients showed a more severe global clinical picture, confirming the need for a more severe level of care for this group. In fact, hospitalization represents a structural framework that works as a source of containment and support for young patients and their families that show greater difficulties for sustaining the child’s development at a certain time. The absence of group difference in NSSI-related characteristics suggested that NSSI should be considered an expression of a larger psychopathological picture, in which contextual factors such as family dynamics could play a role. Therefore, the need for a multi-informant and multi-method clinical assessment, which includes the investigation of environmental factors (e.g., family system and co-parenting system), is highlighted. Future research should further investigate the characteristics of family relationships of NSSI inpatients, using longitudinal studies that allow us to understand the changes of family dynamics over time.