The study was conducted at the Department of Suicide Research and Prevention at the Moscow Research and Clinical Center for Neuropsychiatry. The Center specializes in the treatment of patients with non-psychotic mental disorders. The study cohort is represented by patients with SI and NSSI aged >18 years and older, identified from a consecutive cohort of patients with non-psychotic mental disorders and suicidal ideation. Patients with primary psychotic disorders, current substance use disorders, cognitive deficits below the level of comprehension on self-report scales and interviewer questions were excluded from the study.

All patients are screened on admission for lifetime SI, suicide attempts (SA), and NSSI. The screening includes the first items in the relevant sections of the Self-Injurious Thoughts and Behaviors Interview (SITBI) – “Have you ever had thoughts of killing yourself?” “Have you ever actually made a plan to kill yourself?” and “Have you ever actually engaged in NSSI?” All patients are then seen by an experienced psychiatrist to confirm the diagnosis of a non-psychotic mental disorder according to ICD-10 criteria.

All eligible patients then completed the study questionnaires and were interviewed by the investigator to collect basic socio-demographic information and data on self-injurious thoughts and behaviors. The first patient was enrolled in January 2018 and the last in December 2019.

The Inventory of Statements about Self-Injury-II (ISAS-II) consists of 39 items with responses rated on a 3-point Likert scale from 0 (not relevant) to 3 (very relevant). The list of items begins with an opening statement: “When I self-harm, I am…” According to the authors of the original study, the ISAS-II has two higher-order functions (Interpersonal and Intrapersonal) and 13 lower-order facets. The Intrapersonal function in the original ISAS-II version consists of motives for NSSI such as “Affect regulation” (… releasing emotional pressure that has built up inside of me), “Anti-dissociation/feeling-generation” (… causing pain so I will stop feeling numb), “Anti-suicide” (… avoiding the impulse to attempt suicide), “Marking distress” (… creating a physical sign that I feel awful), “Self-punishment” (… punishing myself) and the Interpersonal function includes “Autonomy” (… demonstrating that I do not need to rely on others for help), “Interpersonal boundaries” (… demonstrating that I am separate from other people), “Interpersonal influence” (… letting others know the extent of my emotional pain), “Peer bonding” (… bonding with peers), “Revenge” (… getting back at someone), “Self-care” (… creating a physical injury that is easier to care for than my emotional distress), “Sensation seeking” (… doing something to generate excitement or exhilaration) and “Toughness” (… seeing if I can stand the pain). The Cronbach’s alpha coefficient of the original ISAS-II version indicated good internal consistency for the interpersonal (0.88) and intrapersonal (0.80) factors (23).

With the permission of the author of the instrument (prof. ED Klonsky), the Russian version of the ISAS-II (Appendix 1) was developed using the back-translation method, which is recommended as a first step in cross-cultural adaptation of instruments. The original version of the ISAS-II was translated from Russian by two native Russian psychiatrists, both of whom were fluent in English. The Russian version of the tool was then back-translated into English by another bilingual translator on the research team to confirm that the translation was consistent with the wording of the original scale. Finally, the Russian version was compared with the original version by an English-speaking consensus committee of clinicians. If there were discrepancies between the two versions, individual items underwent additional rounds of back-translation until they were deemed satisfactory. The developed version of the tool was tested on 16 patients with NSSI (8 females). In their feedback, none of the participants reported any difficulties in understanding the instructions for the tool or the meaning of the items. This version of the ISAS-II was later used in our study.

The Self-Injurious Thoughts and Behaviors Interview (SITBI) is a structured clinical diagnostic interview consisting of the following blocks: suicidal thoughts, suicide plans, suicide attempts, suicidal gestures, non-suicidal self-injurious thoughts and non-suicidal self-injurious behaviors (40). Each block contains up to 30 questions on prevalence, frequency, and contextual factors. The Russian version of the SITBI is mainly used for scientific purposes (17, 41), but in some clinical centers it is also used in routine clinical practice. For the purposes of this study, we used the SITBI questions on prevalence and age of onset of SI, SA, and NSSI, as well as items on methods of self-injury and questions assessing descriptive and contextual factors of NSSI.

The Personality Inventory for DSM-5 and ICD-11 Brief Form Plus-Modified (PID5BF + M) is part of the DSM-5 family of personality trait questionnaires (42, 43). The PID5BF + M was proposed by Bach et al. (44) and consists of 36 questions rated on a 4-point Likert scale from 0 (very untrue or often untrue) to 3 (very true or often true). The questionnaire assesses 6 domains, including 3 facets, each consisting of two questions: Negative Affectivity (emotional lability, anxiousness, and separation insecurity), Detachment (withdrawal, anhedonia, intimacy avoidance), Antagonism/Dissociality (manipulativeness, deceitfulness, grandiosity), Disinhibition (impulsivity, irresponsibility, distractibility), Anankastia (rigidity, perfectionism, and orderliness), and Psychoticism (unusual beliefs, perceptual dysregulation, eccentricity). The instrument showed good internal consistency (McDonald’s omega coefficients ranged from 0.83 to 0.90) and its factor structure was fully consistent with the results of the study of the original version of the questionnaire (45).

The Beck’s Depression Inventory (BDI) was developed to assess the severity of depressive symptoms (46). It consists of 21 items, each rated on a 4-point Likert scale from 0 to 3. The psychometric properties of the Russian version of the BDI have been previously evaluated, with Cronbach’s alpha coefficients of 0.86 (47).

The State–Trait Anxiety Inventory (STAI) consists of two parts assessing state (STAI-S) and trait (STAI-T) anxiety (48). Each part contains 20 items and is scored on a 4-point Likert scale ranging from 1 (not at all) to 4 (very much so). The questionnaire has previously been validated in Russian by Khanin (49), with Cronbach’s alpha of 0.89 and 0.85 for the state and trait anxiety modules, respectively (50).

Categorical variables are presented as frequencies (%) and continuous variables as arithmetic means (standard deviation).

Factor structure was assessed using exploratory factor analysis (EFA) with standard geomin (oblique) rotation and the weighted least square mean and variance adjusted (WLSMV) estimator. WLSMV is a robust estimator suitable for responses with four or fewer response categories and can be corrected for non-normality in the data set (51–53). Kaiser-Meyer-Olkin (KMO) and Bartlett’s sphericity tests were used to assess the suitability of the data for EFA.

The internal consistency of the questionnaire was assessed using McDonald’s omega scores (ω) (53). Average inter-item correlations of the NSSI functions were also calculated to assess item homogeneity (54).

Correlational analysis was used to assess the relationship between NSSI functions and sociodemographic and clinical variables. Correlations between inter/intrapersonal functions and binary variables (age, mental disorder diagnoses, lifetime and 12 months suicide attempts, and 12 months NSSI and NSSI medical treatment) were assessed using biserial correlations. Polyserial correlations were used to examine associations between NSSI functions and ordinal variables (SITBI items 153–162, 165–169) and Pearson’s correlations for continuous variables.

Calculations were performed using Mplus v7.0, Jamovi v2.3.17.0 and BlueSky Statistics v10.3.0.

Of the 3,644 patients screened, 655 (18.0%) were positive for lifetime SI and NSSI. Forty-one patients refused to participate in the study or completed the ISAS-II incorrectly. Thus, 614 patients were included in the final study calculation, of which 14 refused to undergo the SITBI but did not object to the use of their completed questionnaire data in the calculation. Data from these patients were therefore excluded from the analysis of the relationship between NSSI function and the SITBI variables.

Clinical and demographic variables are presented in Table 1. The majority of patients were assigned female at birth (N = 543; 88.4%) with a mean age of 24.86 (SD 7.86; range 18–72) years. Fifty-one patients (8.3%) had an alternative gender identity (those with a non-binary gender identity (e.g., gender fluid, agender, and bigender) and those receiving gender affirming care).

The most common diagnoses were affective disorders [bipolar disorder – 160 (26.1%); depressive disorder – 165 (26.9%)] and personality disorders [162 (26.4%)]. Fifty-five (9.0%) patients were diagnosed with more than one mental disorder. The lifetime prevalence of suicide attempts was 44.7%.

The mean age of onset of NSSI was 15.8 (6.4) years. Approximately 75% had engaged in NSSI in the past 12 months. The most common methods of NSSI were cutting or carving (N = 457; 75.7%), hitting (N = 371; 61.4%), biting (N = 322; 53.3%), and scratching (N = 319; 52.8%). The mean number of NSSI methods was 4.2 (2.0), and 77 (12%) of patients had previously sought non-psychiatric medical help for the consequences of NSSI.

The EFA yielded a two-factor solution that accounted for 38.4% of the total variance. Factor 1 had an eigenvalue of 10.1 and included Intrapersonal functions, and Factor 2 had an eigenvalue of 4.9 and included Interpersonal functions. The two factors had an intercorrelation of 0.22. As can be seen in Table 2, 34 of the 39 item loadings were consistent with the loadings reported by Klonsky and Glenn (23). Three items (11, 24, and 37) loaded on the Interpersonal rather than the Intrapersonal factor, and item 17 loaded on the Intrapersonal rather than the Interpersonal factor. Item 7 was cross-loaded on two factors, but both factor loadings were low (0.25). In order to maintain the integrity of functions such as “Sensation Seeking” and “Self-care,” it was decided to place items 7, 17 in the original factors. Thus, Intrapersonal functions included functions such as “Affect regulation,” “Self-punishment,” “Anti-suicide” and “Anti-dissociation,” and interpersonal functions – “Marking distress,” “Self-care,” “Interpersonal influence,” “Toughness,” “Sensation-seeking,” “Interpersonal boundaries,” “Autonomy,” “Revenge” and “Peer bonding.”

The McDonald’s omega coefficient for the ISAS-II was 0.85, indicating good internal consistency. The internal consistency coefficients for the functions of the NSSI are presented in Table 3. McDonald’s omega coefficients for Intrapersonal and Interpersonal functions were greater than 0.80. The internal consistency of most of the NSSI motives was acceptable (ω ≥ 0.70), except for the coefficients for “Self-Care,” “Sensation seeking,” “Interpersonal boundaries,” and “Peer bonding,” which ranged from 0.55 to 0.70, indicating less adequate reliability.

The average inter-item correlation coefficient for the ISAS-II was 1.51. As shown in Table 3, the average inter-item correlations for the Interpersonal and Intrapersonal functions and most of the subfunctions were within the acceptable range of 0.15–0.50.

As shown in Table 3, the most common functions of NSSI in non-psychotic patients with SI were “Affect regulation” (N = 590; 96.1%), “Self-punishment” (N = 535; 87.1%), “Marking distress” (N = 484; 78.8%), “Anti-suicide” (N = 466; 75.9%) and “Anti-dissociation” (N = 446; 72.6%). However, they rarely self-harmed for “Revenge” (N = 119; 19.4%) or “Peer bonding” (N = 73; 11.9%).

The results of the correlation analysis are presented in Table 4. A total of Intrapersonal self-injurious motives had significant positive weak correlations with and BDI score (r = 0.34) and negative significant weak correlations with age (r = −0.39). Significant weak positive correlations were found between Intrapersonal NSSI functions and psychoticism.

Intrapersonal NSSI functions were significantly positively correlated with 12 months history of NSSI (r = 0.30), number of NSSI methods (r = 0.41), automatic negative (r = 0.49) and positive reinforcement (r = 0.50), and likelihood of future NSSI (r = 0.35). Interpersonal functions had significant positive correlations with both negative (r = 0.40) and positive social reinforcement (r = 0.42).

As the presence of NSSI and SI was assessed in all patients admitted to the inpatient unit, we were able to determine the prevalence of combined NSSI+SI in the hospital population of patients with non-psychotic mental disorders, which was 18.0%. Basic socio-demographic and clinical characteristics of the sample are shown in Table 1. The predominance of women in the sample is consistent with data on their greater use of psychiatric care. Recent study has found that over 76% of patients in the Moscow clinic for patients with non-psychotic mental disorders are women (55). On the other hand, the predominance of young people, those assigned female at birth and a significant proportion of participants with alternative gender identities is consistent with findings from studies conducted in other countries (56–58).

The distribution of diagnoses in the sample reflects the peculiarities of the organization of psychiatric services in the Russian Federation. In Russia, clinics treat patients with so-called non-psychotic mental disorders separately from patients with primary psychotic disorders, patients with mood disorders with psychotic features and patients with addictive or organic psychotic disorders. At the same time, patients with schizotypal and borderline personality disorders, even if they have transient psychotic symptoms, are predominantly treated in clinics for patients with non-psychotic mental disorders.

The Russian version of the ISAS-II has good overall internal consistency (ω = 0.85), and the omega coefficients for the Intrapersonal (0.83) and Interpersonal (0.85) functions were either greater than 0.80 (Table 3). These data indicate a good internal consistency between these factors and are in line with the results of a study by Klonsky and Glenn (23) (Intrapersonal – 0.88 and Interpersonal – 0.89) and some other studies (61). Interestingly, the internal consistency of the Intrapersonal factor functions was higher (range 0.71–0.83) than that of some of the Interpersonal factor functions (range 0.53–0.78). At the same time, although most of the NSSI motives had acceptable internal consistency (ω ≥ 0.70), the coefficients for “Self-care,” “Sensation seeking,” “Interpersonal boundaries” and “Peer bonding” ranged between 0.55 and 0.70, indicating less adequate reliability. The correlation between the Intrapersonal and Interpersonal functions in our sample was low (0.22). The average inter-item correlations for the Interpersonal and Intrapersonal functions and most of the subfunctions were within the acceptable range of 0.15–0.50.

An analysis of the distribution of items by factor (Intrapersonal and Interpersonal) in the Russian version of the ISAS-II revealed a number of differences from the original version (Table 2). The main difference between the Russian version of the ISAS-II and the original version of the questionnaire was that all 3 items of the “Marking distress” subscale loaded more strongly on the Interpersonal factor. Previously, a study by Reinhardt et al. (62) found that “Marking distress” on a par with “Interpersonal boundaries,” “Sensation Seeking,” “Toughness” and “Autonomy” motives may have both Intrapersonal and Interpersonal components. In the same study, the “Marking distress” function had a salient loading on the Interpersonal factor in male adolescents. Our results are consistent with the study by Vigfusdottir et al. (63), conducted on a sample of Norwegian students, which also showed that the “Marking distress” function loaded more on the Interpersonal factor, in contrast to the original study by Klonsky and Olino (22). In the Korean study (28), − item 11 of the “Marking distress” function also loaded on interpersonal rather than intrapersonal functions.

Another feature of the Russian version of the ISAS-II was that the item 17 (“Creating a physical injury that is easier to care for than my emotional distress”), which in the original version belonged to the Interpersonal function “Self-care” (self-injuring to create a physical wound that one can care for more easily than one’s emotional distress) (23), had a greater load on the Intrapersonal factor. The authors originally expected the “Self-care” function to be theoretically related to the intrapersonal factor, but in the original study this function had a higher load on the Interpersonal (0.41) than on the Intrapersonal factor (0.33) (23). It is interesting to note that in the original study the loading on the Interpersonal factor was only slightly higher than on the Intrapersonal factor. In a more recent study by Klonsky et al. (64), item analysis was performed and showed that item 17 loaded on the Intrapersonal (0.50) rather than the Social factor (0.26), which is fully consistent with our results. In the Russian version of the ISAS-II, the other two items of the “Self-care” function had high loadings on the Interpersonal factor, but the factor loadings were low. The “Self-care” function was also clearly loaded on the Intrapersonal function in the studies by Kortge et al. (65), Vigfusdottir et al. (63), and Pérez et al. (25). In the last of the above-mentioned studies (25), the model in which the “Self-care” function was included in the Intrapersonal factor was found to be preferable to the original model in which it was identified as an Interpersonal function. It is also noteworthy that the other “Self-care” function items, although more heavily loaded on the Interpersonal factor, had some of the lowest loadings of all questionnaire items (0.4 for item 4 and 0.35 for item 30). Given the instability of the “self-care” subfunction (both the items within it and the subfunction as a whole) identified in several studies, we believe that developing a revised version of the ISAS based on the experience of the validations may be the best decision. A new version of the instrument should either clarify the wording of the questions or remove this function altogether.

Another ISAS-II item that showed low factor loadings on both factors was item 7 (“Doing something to generate excitement or exhilaration”), which refers to the “Sensation-seeking” function. Previously, the Korean study (28) found that, contrary to the original research, item 7 loaded more on the Intrapersonal than on the Interpersonal function.

In our study, patients practiced in NSSI for both Intrapersonal and Interpersonal motives (Table 3), which is consistent with most studies, but the prevalence differed from that presented in the meta-analysis by Taylor et al. (66). The three most common Intrapersonal NSSI functions in our sample were “Affect regulation” (96.1%), “Self-Punishment” (87.1%), and “Anti-Suicide” (75.9%), and the most common Interpersonal functions were “Marking distress” (78.8%), “Self-care” (68.9%), and “Interpersonal influence” (54.9%). In a meta-analysis by Taylor et al. (66), avoidance or escape from an unwanted internal state was also the most common Intrapersonal motive for NSSI, while communicating the level of distress and interpersonal influence were found to be the most common Intrapersonal motives. It is noteworthy that the “Self-care” motive for NSSI in the Russian version of the ISAS-II includes both Intrapersonal and Interpersonal items and cannot be completely reduced to just one of the two functions.

We conducted a correlational analysis of Interpersonal and Intrapersonal functions with sociodemographic, clinical and personality profile variables. We also examined correlations between ISAS-II functions and parameters from the SITBI Non-suicidal self-injury module (items 144–169). We found small positive correlations (0.3–0.5) between Intrapersonal NSSI functions and lower age, higher BDI score, NSSI episode in the past 12 months, higher number of NSSI methods, “mental state at the time” as a cause of NSSI, and self-reported high likelihood of engaging in NSSI in the future. Our findings in the Russian-speaking patients on the relationship between Intrapersonal functions and the less favorable course of NSSI in patients with non-psychotic mental disorders are consistent with the results of studies conducted in other linguistic and cultural samples (67, 68).

To the best of the authors’ knowledge, no correlational analysis has previously been conducted between DSM-5 personality trait domains and NSSI functions. At the same time, the relationship between NSSI behavior and higher levels of Negative affectivity, Detachment, Antagonism and Psychoticism has been previously reported by researchers (69–73). The Psychoticism domain of the PID-5 includes traits such as Unusual Beliefs & Experiences, Eccentricity, and Perceptual Dysregulation (74). According to Peng et al., patients with depression and NSSI have significantly higher levels of psychoticism than healthy controls and patients with depression alone (73). In addition, regression analysis showed that psychoticism was an independent risk factor for NSSI in depressed patients. These data are supported by the results of the study by Kang et al. (74). In our study, Intrapersonal NSSI functions had the highest correlations with psychoticism (0.32). In our opinion, this may be explained by the fact that psychoticism is associated with a well-known difficulty in dealing effectively with emotional conflict. For example, a study by Granieri et al. found that two immature defense mechanisms – autistic fantasy and isolation – predicted the PID-5 Psychoticism domain in participants (75). While attempts to cope with the unwanted effects of dissociation are among the most common Intrapersonal motives for NSSI, psychoticism has been repeatedly reported to be associated with a high risk for the development of dissociative phenomena (76).

In our study, NSSI Interpersonal functions had the highest correlations with the Antagonism domain (0.25). An association between NSSI and low agreeableness has also been reported in some previous studies (76–78). This variable indirectly reflects antagonism, which, according to the authors of the PID-5 (74), includes traits such as Manipulativeness, Deceitfulness and Grandiosity. The relationship found between Interpersonal functions and Antagonism is consistent with the suggestion by some authors that some NSSI behaviors directed at the environment may involve aggressive goals (e.g., “Revenge” motives).

The SITBI questions related to self-harm motives (items 153–156) had the highest correlations with NSSI functions. The results were as expected: Interpersonal functions correlated with “… to communicate with someone else or to get attention?” (0.42) and “… to get out of doing something or to get away from others?” (0.40), and Intrapersonal functions with statements such as “… a way to get rid of bad feelings?” (0.49) and “… to feel something, because you were feeling numb or empty?” (0.50). These data provide general support for the convergent validity of the SITBI, but it is worth noting that none of the correlations exceeded the level of 0.5.