Preliminary Psychometric Evaluation of the Inventory of Statements About Self-Injury (ISAS) in a Greek Adolescent Inpatient Psychiatric Sample

Abstract

Non-suicidal self-injury (NSSI) is a major concern in adolescent mental health, yet the psychometric properties of the Inventory of Statements About Self-Injury (ISAS) have not previously been examined in a Greek adolescent inpatient sample. This preliminary study evaluated the internal consistency, factorial structure, and construct validity evidence of the ISAS in 95 Greek adolescents receiving inpatient psychiatric care (mean age = 14.69 years, SD = 1.30; 86.3% female). Data were obtained retrospectively from clinical records. Psychometric evaluation included Cronbach’s alpha coefficients, exploratory factor analysis (EFA), confirmatory factor analysis (CFA), and correlations with the Youth Self-Report (YSR). The ISAS showed satisfactory internal consistency at the higher-order factor level, with alpha coefficients of 0.82 for the interpersonal factor and 0.85 for the intrapersonal factor, although Autonomy and Interpersonal Boundaries showed weak reliability. EFA supported a broad two-factor intrapersonal–interpersonal structure. CFA findings were exploratory and mixed. The original CFA models showed inadequate fit, whereas exploratory modified and post hoc sensitivity models showed improved but non-definitive fit. Convergent validity evidence was modest and was supported by associations between the original broad ISAS intrapersonal score and YSR self-harm behavior, suicidal ideation, internalizing-related dimensions, and Total Problems. Discriminant-pattern evidence was limited. Overall, the findings provide preliminary support for the clinical usefulness of the ISAS as an adjunctive assessment tool in this population, but they do not constitute definitive validation. Further prospective validation in larger, more diverse, and independent samples is needed.

1. Introduction

Non-suicidal self-injury (NSSI) is a major concern in adolescent mental health worldwide. It refers to the intentional destruction of body tissue without suicidal intent and through methods that are not socially sanctioned, such as cutting, scratching, burning, or hitting [1,2]. Although distinct from suicidal behavior, NSSI is closely associated with emotional distress, psychiatric comorbidity, and substantially elevated risk for subsequent suicidal ideation and suicide attempts [3,4,5,6,7,8]. For this reason, NSSI is increasingly recognized as a clinically significant phenomenon in child and adolescent psychiatry.

NSSI is relatively common during adolescence, with lifetime prevalence estimates in community samples generally ranging from approximately 17% to 23%, although individual studies have reported wider variation depending on population and methodology [1,3,9,10]. Rates are markedly higher in psychiatric inpatient populations, where approximately half to two-thirds of adolescents may report a history of NSSI [11,12]. Onset typically occurs in early to mid-adolescence, most often between 12 and 16 years of age [1,11]. NSSI is usually repetitive rather than episodic, and its most common functions include emotional regulation, relief from distress, and the management of anger or other overwhelming affective states [1,9,11,13,14,15,16,17,18,19].

A growing body of research has identified multiple risk domains associated with adolescent NSSI, including depression, anxiety, emotional dysregulation, adverse childhood experiences, bullying, trauma exposure, family dysfunction, and problematic peer and parent–child relationships [1,2,3,4,5,6,7,20,21,22]. From a clinical perspective, understanding not only the presence of NSSI but also the psychological functions it serves is essential for accurate assessment, case formulation, and intervention planning [3].

Several self-report instruments have been developed to assess NSSI [23,24,25,26,27,28,29]. Among them, the Functional Assessment of Self-Mutilation [30] has been widely used; however, it does not cover the full range of self-injury functions described in the literature, particularly those related to sensory seeking, anti-suicide motives, and interpersonal boundaries [2]. The Inventory of Statements About Self-Injury (ISAS) [26,31] was developed to provide a broader assessment of both the characteristics and functions of NSSI. In addition to recording descriptive features of self-injurious behavior, the ISAS assesses 13 distinct functions and has been widely used in studies examining the intrapersonal and interpersonal motives underlying NSSI [26,31].

Previous studies from different countries and clinical contexts have generally supported the reliability and validity of the ISAS [32,33,34,35,36,37,38,39,40,41,42,43,44,45,46]. Most validation studies have identified a broad distinction between intrapersonal and interpersonal functions, although some variability has been reported in the placement of specific subscales across samples and cultural settings. Overall, the available literature suggests that the ISAS is a clinically useful instrument, while also indicating that its factor structure may vary across age groups and populations [32,33,34,35,36,37,38,39,40,41,42,43,44,45,46].

Despite the growing importance of NSSI assessment, no previous study has systematically examined the psychometric properties of the ISAS in a Greek adolescent inpatient psychiatric sample. This represents an important gap, as the lack of Greek psychometric evidence limits standardized assessment in child and adolescent psychiatric settings and reduces comparability with the international literature. The present study therefore aimed to provide a preliminary psychometric evaluation of the Greek ISAS in an inpatient sample of adolescents receiving psychiatric care. Specifically, the study examined internal consistency, exploratory factorial structure, CFA-based model fit for illustrative and sensitivity purposes, and construct validity evidence using Youth Self-Report (YSR) [47,48] as an external criterion measure. Given the small sample size and the use of both EFA and CFA in the same dataset, all factor-analytic findings were interpreted as exploratory and requiring independent replication.

2. Materials and Methods

2.1. Participants

Clinical records were screened for the study period from 20 September 2020 to 21 April 2025 in the inpatient unit of the Department of Child and Adolescent Psychiatry, School of Medicine, National and Kapodistrian University of Athens, “Aghia Sophia” Children’s Hospital, Athens, Greece. During this period, 146 clinical records were screened for eligibility. Of these, 95 adolescents were included in the final analytic sample and 51 were excluded because the required questionnaire data were unavailable or because routine questionnaire completion had not been feasible. All included participants had received a psychiatric diagnosis at the time of assessment according to ICD-10 criteria. Sociodemographic variables extracted from the clinical records included age, sex, family income, parental educational level, and family status.

The sample included 82 females (86.3%) and 13 males (13.7%). Thus, although the final analytic sample allowed preliminary psychometric analyses of the ISAS functional scores, the sample was small for factor-analytic procedures and was markedly imbalanced by sex. The mean age of the participants was 14.69 years (SD = 1.30; range = 11–17). The most common primary diagnoses were depressive disorders (44.5%), autism spectrum disorder (15.2%), and anxiety disorders (14.6%).

2.2. Procedure

This was a retrospective file-based study using questionnaire and clinical data recorded as part of routine inpatient assessment. Because the study relied on pre-existing clinical records, the data reflect NSSI that was disclosed by adolescents and documented during routine inpatient assessment. Undisclosed self-injury, self-injury not captured by the questionnaires, or clinically relevant information not recorded in the file could not be identified retrospectively. Therefore, the descriptive findings should be interpreted as clinical-record-based estimates rather than as prevalence estimates of all NSSI occurring in this inpatient population. Participants were not recruited prospectively specifically for the purposes of the present study. Psychiatric diagnoses had been assigned by a team of child psychiatrists working in the unit using ICD-10 diagnostic codes. For the purposes of the present analyses, data extracted from the clinical records included questionnaire responses and relevant demographic and clinical variables.

Records were excluded when the required questionnaire data were not available or when routine questionnaire completion had not been feasible because of moderate to profound intellectual disability, age younger than 11 years, reduced level of consciousness, insufficient knowledge of the Greek language, or inability to physically complete the questionnaires.

2.3. Measures

2.3.1. Inventory of Statements About Self-Injury

The Inventory of Statements About Self-Injury (ISAS) [26] is a publicly available self-report questionnaire developed to assess both the descriptive characteristics and the perceived functions of non-suicidal self-injury (NSSI). The ISAS consists of two sections.

The first section assesses descriptive characteristics of self-injurious behavior, including type, frequency, predominant method, age at onset, pain during the act, whether the behavior occurs in the presence of others, the time interval between urge and act, and the desire to stop self-injuring. The listed behaviors include cutting, severe scratching, biting, banging or hitting oneself, burning, interfering with wound healing, carving, rubbing the skin against a rough surface, pinching, sticking oneself with needles, pulling hair, and swallowing dangerous substances, with an additional open category for other methods.

The second section comprises 39 items assessing 13 potential functions of self-injury and is completed by respondents who report a history of self-injurious behavior. Responses are rated on a three-point scale ranging from 0 (“not relevant”) to 2 (“very relevant”). Each function is represented by three items, yielding subscale scores ranging from 0 to 6. The assessed functions are Affect Regulation, Interpersonal Boundaries, Self-Punishment, Self-Care, Anti-Dissociation/Feeling Generation, Anti-Suicide, Sensation Seeking, Peer Bonding, Interpersonal Influence, Toughness, Marking Distress, Revenge, and Autonomy. Each item begins with the stem “When I self-harm, I am …”. Example items include “calming myself down” (Affect Regulation), “creating a boundary between myself and others” (Interpersonal Boundaries), “punishing myself” (Self-Punishment), “giving myself a way to care for myself (by attending to the wound)” (Self-Care), “causing pain so I will stop feeling numb” (Anti-Dissociation/Feeling Generation), “avoiding the impulse to attempt suicide” (Anti-Suicide), “doing something to generate excitement or exhilaration” (Sensation Seeking), “bonding with peers” (Peer Bonding), “letting others know the extent of my emotional pain” (Interpersonal Influence), “seeing if I can stand the pain” (Toughness), “creating a physical sign that I feel awful” (Marking Distress), “getting back at someone” (Revenge), and “ensuring that I am self-sufficient” (Autonomy).

A Greek-language clinical version of the ISAS, which had already been in routine clinical use in the inpatient unit for several years, was used in the present study. This version had undergone forward translation and back-translation procedures as part of its earlier linguistic and cultural adaptation for clinical use in the unit, although, to our knowledge, no previous peer-reviewed psychometric validation study or formal publication of the adaptation process has been published. Accordingly, the present study represents the first peer-reviewed psychometric evaluation of this Greek version in an adolescent inpatient psychiatric sample.

For the purposes of the present analyses, the 13 function scores derived from the second section of the ISAS were treated as the main observed variables in the psychometric evaluation. Selected descriptive characteristics from the first section were also examined for descriptive purposes, including predominant method, age at onset, recent self-injury, and desire to stop self-injuring. Previous international studies have supported the reliability and validity of the instrument across different cultural and clinical contexts [32,33,34,35,36,37,38,39,40,41,42,43,44,45,46].

2.3.2. Youth Self-Report

The Youth Self-Report (YSR) [47] is a self-report questionnaire completed by adolescents and consists of two sections. The first section collects demographic information and provides a broad qualitative overview of the adolescent’s functioning, difficulties, and strengths. The second section contains 112 items assessing emotional and behavioral problems. Responses are rated on a three-point scale from 0 (“not true”) to 2 (“very true or often true”). Item scores are combined to generate eight empirically derived syndrome scales, namely Anxious/Depressed, Withdrawn/Depressed, Somatic Complaints, Social Problems, Thought Problems, Attention Problems, Rule-Breaking Behavior, and Aggressive Behavior, as well as three broad summary scales: Internalizing Problems, Externalizing Problems, and Total Problems. In addition, the YSR Other Problems score was included in the correlation table for completeness.

In the present study, the YSR was used to assess convergent validity. Analyses focused on the Internalizing, Externalizing, and Total Problems summary scales, relevant syndrome scales, and the individual items assessing self-harm behavior and suicidal ideation. The instrument has been standardized in Greek and has demonstrated satisfactory psychometric properties in Greek adolescent samples [48].

No additional NSSI-specific or suicidality-specific validation scale, such as the Suicidal Ideation Questionnaire–Junior [49], was available in the retrospective clinical records. Therefore, the YSR self-harm and suicidal ideation items were used as the closest available external clinical indicators. These items were considered proxy indicators for convergent validity rather than comprehensive external validation measures of NSSI or suicidal ideation.

2.4. Statistical Analysis

Missing data were handled using Multiple Imputation by Chained Equations (MICE), implemented according to the Fully Conditional Specification approach [50]. Prior to imputation, the extent and pattern of missingness were examined descriptively for all study variables. Multiple imputation was performed under the assumption that data were missing at random. The imputation model included all ISAS functional scores, YSR variables used in the convergent validity analyses, and the demographic and clinical variables extracted from the records. Estimates from the imputed datasets were pooled according to Rubin’s rules. Multiple imputation was selected as a pragmatic approach for handling missingness across the set of study variables, given the distributional characteristics of the data and the analytic framework used.

Prior to the final analyses, descriptive variables were checked for implausible values. Implausible values for age at NSSI onset were treated as data-quality errors and recoded as missing before descriptive statistics were calculated. Specifically, values indicating onset at 1 year of age were considered implausible for self-reported NSSI onset and were not included in the final age-at-onset estimate.

Ninety-five percent confidence intervals (95% CIs) were calculated for the primary descriptive and psychometric estimates. For continuous variables, 95% CIs were calculated for means; for categorical variables, 95% CIs were calculated for proportions. For internal consistency estimates, bootstrap 95% CIs were reported. For correlations, 95% CIs were calculated using Fisher’s z transformation and pooled across imputed datasets where applicable. For factor-analytic results, 95% CIs were reported for standardized loadings where available. For CFA model fit, RMSEA was reported with its 95% CI, and the remaining fit indices were interpreted descriptively in conjunction with the full pattern of model results.

The internal consistency of the ISAS subscales and broader factors was assessed using Cronbach’s alpha coefficients. Confidence intervals for reliability estimates were calculated using bootstrap procedures with 500 repetitions.

To investigate factorial validity, an exploratory factor analysis (EFA) was first conducted on the ISAS functional variables. Factor extraction was based on principal axis factoring, and an oblique rotation was used because correlations between factors were expected. The number of factors to retain was determined based on Horn’s parallel analysis, the Minimum Average Partial (MAP) test, and visual inspection of the scree plot.

Subsequently, confirmatory factor analysis (CFA) was performed to examine the extent to which alternative factor structures suggested by the EFA and by previous validation studies [32,33,34,35,36,37,38,39,40,41,42,43,44,45,46] were consistent with the observed data. Model evaluation was based on standard fit indices, including the comparative fit index (CFI), Tucker–Lewis index (TLI), root mean square error of approximation (RMSEA), standardized root mean square residual (SRMR), and chi-square statistics.

The unifactorial and two-factor models were compared using the overall pattern of fit indices, including CFI, TLI, RMSEA, SRMR, χ², and χ²/df. Modification indices were inspected to identify localized areas of model strain. Because of the small sample size, post hoc modifications were restricted to a small number of theoretically interpretable residual covariances between functionally related ISAS indicators and were interpreted as exploratory sensitivity analyses. Cross-loadings were not added. In addition, an exploratory modified three-factor 13-function sensitivity model was examined to evaluate whether a more differentiated data-informed structure improved model fit. This model was not specified a priori and was therefore interpreted only descriptively and cautiously. The model referred to as “bifactorial” in the review process is described here as a two-factor model, because it included correlated interpersonal and intrapersonal factors rather than a general factor plus specific factors.

Given the relatively small sample size, additional sample-size and power considerations were reported for the factor-analytic and correlation analyses. For the EFA, no single universally accepted post hoc power index is available for a multidimensional exploratory factor solution; therefore, sample adequacy was evaluated using the participant-to-indicator ratio and an approximate sensitivity analysis for correlations underlying the factor model. With N = 95, a two-sided test at α = 0.05 has approximately 80% power to detect correlations of |r| ≈ 0.28. Thus, the study was better powered to detect moderate factor loadings and moderate validity correlations than small loadings, weak cross-loadings, or small convergent validity associations.

For the CFA, the sample size was considered limited, particularly because the model was examined in the same dataset used for EFA and because sample-size requirements for covariance structure modeling are strongly influenced by model complexity, degrees of freedom, and expected model misfit [51]. RMSEA-based sensitivity/power calculations were used to contextualize the limited sample size [51]. Using N = 95, df = 41, α = 0.05, a close-fit null hypothesis of RMSEA = 0.05, and alternative RMSEA values of 0.08 and 0.10, estimated power was approximately 0.36 and 0.73, respectively. These results indicate limited power to distinguish close model fit from modest model misfit and reinforce the need to interpret the CFA findings as exploratory and illustrative rather than confirmatory.

Measurement invariance by gender and age was considered but not tested because the available subgroups were too small and uneven for reliable multi-group CFA. In particular, the male subgroup included only 13 participants, and age-based subdivision would have produced small groups within a narrow age range. Accordingly, no claims are made regarding measurement equivalence of the ISAS across gender or age groups in the present sample.

To examine construct validity more completely, correlations were calculated between the ISAS total functional score based on all 13 original functions, the original broad ISAS intrapersonal and interpersonal function scores, YSR syndrome scales, YSR summary scales, and the YSR items assessing self-harm behavior and suicidal ideation. Correlations were examined primarily using Pearson’s coefficients, with Kendall’s tau used as a supplementary robustness check given the distributional characteristics of the data. Stronger positive correlations with theoretically related variables, particularly YSR self-harm behavior, suicidal ideation, and internalizing-related dimensions, were interpreted as preliminary evidence of convergent validity. Weaker correlations with less directly related YSR dimensions, particularly rule-breaking behavior, aggressive behavior, and externalizing problems, were interpreted as preliminary discriminant-pattern evidence. Because no independent NSSI-specific or suicidality-specific instrument was available, formal divergent validity could not be tested. All statistical findings, particularly the factor-analytic results, were interpreted with caution given the sample size, the clinical heterogeneity of the sample, and the use of both EFA and CFA within the same dataset.

3. Results

3.1. Descriptive Findings

Descriptive statistics for selected sample characteristics and the main ISAS variables are presented in

Table 1. Descriptive statistics for the sample and main ISAS/NSSI variables with 95% confidence intervals.

Variable	Estimate	95% CI	Range/Denominator
Number of participants	n = 95	—	—
Female participants	82 (86.3%)	78.0–91.8%	n = 95
Male participants	13 (13.7%)	8.2–22.0%	n = 95
Age, years	14.69 (1.30)	14.43–14.95	11–17
Family income: high	13.0%	7.7–21.2%	n = 95
Family income: medium	67.0%	57.0–75.6%	n = 95
Family income: low	20.0%	13.2–29.1%	n = 95
Parental educational level: high	41.3%	31.9–51.4%	n = 95
Parental educational level: medium	40.9%	31.6–51.0%	n = 95
Parental educational level: low	17.8%	11.4–26.7%	n = 95
Married/cohabiting parents	42.7%	33.2–52.7%	n = 95
Other family arrangements	57.3%	47.3–66.8%	n = 95
Depressive disorders	44.5%	34.9–54.5%	n = 95
Autism spectrum disorder	15.2%	9.3–23.8%	n = 95
Anxiety disorders	14.6%	8.9–23.1%	n = 95
Age at NSSI onset, years	11.56 (2.07)	11.12–12.00	5–15; n = 87
Total ISAS functional score	23.93 (11.53)	21.61–26.25	0–78
Interpersonal functions	8.53 (6.82)	7.16–9.90	0–48
Intrapersonal functions	15.40 (6.81)	14.03–16.77	0–30
“Releasing emotional pressure that has built up inside of me”	1.51 (0.73)	1.36–1.66	0–2
“Punishing myself”	1.33 (0.77)	1.17–1.49	0–2
Predominant method: carving	42.1%	32.7–52.2%	n = 95
Predominant method: cutting	17.9%	11.5–26.8%	n = 95
Self-injury during previous month	71.9%	62.1–80.0%	n = 95
Wish to stop self-injuring	74.2%	64.6–81.9%	n = 95

Table note. Values are means with standard deviations or n (%). CIs for continuous variables are 95% confidence intervals for the mean. CIs for categorical variables are Wilson 95% confidence intervals for proportions. Implausible values indicating NSSI onset at 1 year of age were treated as missing before calculating the age-at-onset estimate. Because the data were obtained retrospectively from clinical records, these estimates describe disclosed and documented NSSI-related information rather than true population prevalence. The ISAS total functional score refers to the sum of the 13 original ISAS function scores. The interpersonal and intrapersonal function scores also follow the original broad ISAS scoring structure and should not be interpreted as scores from the post hoc 11-function CFA sensitivity model.

. The mean age of the participants was 14.69 years (SD = 1.30), and the sample was predominantly female (86.3%). Family income was high in 13.0% of cases, medium in 67.0%, and low in 20.0%. Parental educational level was high in 41.3% of cases, medium in 40.9%, and low in 17.8%. Family status was classified as married/cohabiting parents in 42.7% of cases and other family arrangements in 57.3%.

All primary psychometric analyses were conducted on the final analytic sample of N = 95 after multiple imputation. The EFA and original CFA models used the 13 ISAS functional subscale scores, corresponding to a participant-to-indicator ratio of approximately 7.3:1. The post hoc exploratory CFA sensitivity model included 11 indicators after excluding Autonomy and Interpersonal Boundaries, corresponding to a participant-to-indicator ratio of approximately 8.6:1.

After recoding implausible values of 1 year as missing, the mean age at onset of non-suicidal self-injury was 11.56 years (SD = 2.07; range = 5–15; n = 87; 95% CI = 11.12–12.00). The mean total ISAS functional score, based on all 13 original functions, was 23.93 (SD = 11.53; 95% CI = 21.61–26.25). Regarding the two original broad ISAS functional dimensions, the mean score for interpersonal functions was 8.53 (SD = 6.82; 95% CI = 7.16–9.90), whereas the mean score for intrapersonal functions was 15.40 (SD = 6.81; 95% CI = 14.03–16.77).

At the item level, the most frequently endorsed statement was “releasing emotional pressure that has built up inside of me”, with a mean score of 1.51 (SD = 0.73). This was followed by the item “punishing myself”, with a mean score of 1.33 (SD = 0.77).

Regarding documented self-injury characteristics, the most commonly reported predominant methods were carving (42.1%; 95% CI = 32.7–52.2%) and cutting (17.9%; 95% CI = 11.5–26.8%). In addition, 71.9% of participants reported at least one episode of self-injury during the previous month, and 74.2% reported a wish to stop self-injuring. These estimates should be interpreted as based on disclosed and recorded clinical information and may underestimate NSSI that was not reported or documented.

3.2. Exploratory Factor Analysis

Exploratory factor analysis supported a two-factor solution as the most parsimonious and interpretable structure. Horn’s parallel analysis suggested a three-factor solution when based on Pearson correlations but a two-factor solution when based on Kendall correlations. Given the non-normal distribution of the data, greater interpretive weight was given to the Kendall-based solution. In addition, the Minimum Average Partial (MAP) test supported a two-factor solution for both Pearson and Kendall correlations, and inspection of the scree plot also favored a two-factor structure.

The first factor primarily reflected interpersonal functions and included Peer Bonding, Toughness, Interpersonal Influence, Interpersonal Boundaries, Revenge, Marking Distress, and Autonomy. The second factor primarily reflected intrapersonal functions and included Affect Regulation, Anti-Dissociation/Feeling Generation, Anti-Suicide, Self-Punishment, and Sensation Seeking. The complete standardized loading matrix, including 95% confidence intervals, is presented in

Table 2. Exploratory factor analysis loadings for ISAS functional subscales with 95% confidence intervals.

ISAS Function	Interpersonal Loading	95% CI	Intrapersonal Loading	95% CI
Affect Regulation	−0.01	−0.20 to 0.24	0.72	0.39 to 0.86
Interpersonal Boundaries	0.60	0.09 to 0.83	0.28	0.03 to 0.65
Self-Punishment	0.18	−0.01 to 0.37	0.59	0.36 to 0.73
Self-Care	0.35	0.07 to 0.56	0.38	0.19 to 0.63
Anti-Dissociation/Feeling Generation	0.12	−0.10 to 0.34	0.74	0.50 to 0.90
Anti-Suicide	0.10	−0.16 to 0.39	0.57	0.23 to 0.81
Sensation Seeking	0.27	−0.14 to 0.66	0.33	0.02 to 0.67
Peer Bonding	0.64	0.27 to 0.84	−0.15	−0.35 to 0.27
Interpersonal Influence	0.60	0.17 to 1.03	0.04	−0.16 to 0.46
Toughness	0.73	0.34 to 0.87	0.17	−0.01 to 0.58
Marking Distress	0.44	0.09 to 0.63	0.37	0.16 to 0.63
Revenge	0.36	0.00 to 0.62	0.10	−0.16 to 0.44
Autonomy	0.42	0.06 to 0.72	0.26	−0.02 to 0.64

Table note. Loadings are standardized. Confidence intervals are bootstrap 95% confidence intervals. Loadings with confidence intervals crossing zero should be interpreted cautiously. The two-factor solution was retained only as a preliminary exploratory structure. Several cross-loadings and broad confidence intervals indicate that the structure should not be interpreted as definitively established.

. Overall, the loading pattern provided preliminary support for a broad intrapersonal–interpersonal distinction. However, the structure was not clean for all functions. Self-Care showed a less distinct pattern, with similar loadings on the interpersonal and intrapersonal factors (0.35 and 0.38, respectively), and was therefore interpreted cautiously. Autonomy and Interpersonal Boundaries loaded more strongly on the interpersonal factor, but both showed weak internal consistency and psychometric instability in subsequent analyses. These results support an exploratory interpretation rather than definitive factor-structure confirmation.

3.3. Internal Consistency

The ISAS demonstrated satisfactory to very good internal consistency at the higher-order factor level, whereas internal consistency at the level of individual subscales was more variable. Cronbach’s alpha coefficients for the individual subscales and the two broader factors are presented in

Table 3. Cronbach’s alpha coefficients for ISAS subscales and higher-order factors with 95% confidence intervals.

Subscale/Factor	Cronbach’s Alpha	95% CI
Affect Regulation	0.72	0.58–0.81
Interpersonal Boundaries	0.42	0.29–0.58
Self-Punishment	0.72	0.60–0.82
Self-Care	0.71	0.56–0.81
Anti-Dissociation/Feeling Generation	0.70	0.57–0.80
Anti-Suicide	0.63	0.48–0.76
Sensation Seeking	0.54	0.36–0.70
Peer Bonding	0.88	0.75–0.95
Interpersonal Influence	0.57	0.25–0.75
Toughness	0.58	0.32–0.72
Marking Distress	0.74	0.62–0.83
Revenge	0.81	0.66–0.90
Autonomy	0.33	0.03–0.60
Interpersonal factor	0.82	0.76–0.87
Intrapersonal factor	0.85	0.78–0.89

Table note. Confidence intervals are bootstrap 95% confidence intervals. The broader interpersonal and intrapersonal reliability estimates were calculated using the original broad 13-function ISAS scoring structure. Autonomy and Interpersonal Boundaries showed the weakest reliability. They were not removed from the ISAS instrument but were excluded from a post hoc exploratory CFA sensitivity model.

. Reliability estimates ranged from α = 0.33 for Autonomy to α = 0.88 for Peer Bonding. Lower internal consistency was observed for Autonomy and Interpersonal Boundaries, whereas the broader factors showed strong reliability, with α = 0.82 for the interpersonal factor and α = 0.85 for the intrapersonal factor.

3.4. Confirmatory Factor Analysis

Confirmatory factor analysis was used to examine alternative factor structures for illustrative and exploratory purposes. Because the same dataset was used for both EFA and CFA, and because the sample size was small for CFA, these analyses should not be interpreted as independent confirmation of the ISAS factor structure.

As shown in

Table 4. Comparison of confirmatory factor analysis models.

Model	χ2	df	χ2/df	CFI	TLI	RMSEA	95% CI for RMSEA	SRMR
Unifactorial 13-function model	228.52	65	3.52	0.53	0.44	0.16	0.13–0.19	0.13
Original theoretical two-factor 13-function model	178.89	64	2.80	0.67	0.60	0.14	0.11–0.17	0.12
Exploratory modified three-factor 13-function model	75.31	57	1.32	0.95	0.93	0.06	0.00–0.09	0.08
Post hoc 11-function two-factor sensitivity model	83.37	43	1.94	0.85	0.79	0.11	0.07–0.15	0.10
MI-informed 11-function two-factor sensitivity model	59.88	41	1.46	0.89	0.85	0.07	0.03–0.11	0.08

Table note. CFI = comparative fit index; TLI = Tucker–Lewis index; RMSEA = root mean square error of approximation; CI = confidence interval; SRMR = standardized root mean square residual; MI = modification index. Fit indices are reported descriptively and were averaged/pool-summarized across imputed datasets. The 11-function models were post hoc exploratory sensitivity models and should not be interpreted as validated shortened versions of the ISAS. The exploratory modified three-factor 13-function model was data-informed, was not specified a priori, and allowed residual covariances between Peer Bonding and Interpersonal Influence, Peer Bonding and Marking Distress, Affect Regulation and Sensation Seeking, Anti-Suicide and Revenge, and Toughness and Peer Bonding. The MI-informed 11-function two-factor sensitivity model allowed residual covariances between Peer Bonding and Marking Distress and between Interpersonal Influence and Peer Bonding. These residual covariances were added because they were suggested by modification indices and were theoretically interpretable as localized overlap among functionally related self-injury functions. Model evaluation was based on the overall pattern of fit indices rather than any single statistic.

, the unifactorial model showed poor fit, indicating that the ISAS functional subscales were not adequately represented by a single general function factor. The original theoretical two-factor model showed inadequate fit, despite some improvement relative to the unifactorial model. A modified three-factor 13-function model and an 11-function two-factor sensitivity model excluding Autonomy and Interpersonal Boundaries showed improved but still non-definitive fit. The modified three-factor model was specified as an exploratory, data-informed sensitivity model based on the exploratory results, theoretical interpretability, and localized areas of strain in the original models, and included three correlated exploratory domains. The first domain included Self-Care, Anti-Dissociation/Feeling Generation, Anti-Suicide, Affect Regulation, and Self-Punishment, reflecting primarily intrapersonal and self-directed affect-regulation functions. The second domain included Toughness, Autonomy, Peer Bonding, Sensation Seeking, and Interpersonal Boundaries, reflecting a mixed identity-, peer-, boundary-, and sensation-related domain. The third domain included Interpersonal Influence, Revenge, and Marking Distress, reflecting primarily interpersonal influence and distress-signaling functions. This modified three-factor model also allowed residual covariances between Peer Bonding and Interpersonal Influence, Peer Bonding and Marking Distress, Affect Regulation and Sensation Seeking, Anti-Suicide and Revenge, and Toughness and Peer Bonding. Because this specification was exploratory, data-informed, examined in the same dataset, and not independently cross-validated, it was interpreted only as a sensitivity model rather than as evidence for a stable three-factor structure. The exclusion of Autonomy and Interpersonal Boundaries in the 11-function sensitivity model was post hoc and was based on their weak internal consistency and unstable model behavior; therefore, the 11-function model should be interpreted only as an exploratory sensitivity analysis, not as a validated shortened version of the ISAS.

Modification indices for the 11-function two-factor sensitivity model primarily indicated localized residual associations among conceptually related interpersonal functions. The largest theoretically interpretable modification indices involved residual covariances between Peer Bonding and Marking Distress and between Interpersonal Influence and Peer Bonding, as shown in

Table 5. Modification-index-informed residual covariances tested in the two-factor sensitivity model.

Modification Suggested	Average MI	Expected Parameter Change	Decision
Peer Bonding ~~ Marking Distress	43.76	Positive residual association	Retained
Interpersonal Influence ~~ Peer Bonding	28.26	Positive residual association	Retained
Interpersonal Influence ~~ Toughness	21.99	Positive residual association	Not retained
Interpersonal Influence ~~ Marking Distress	12.97	Positive residual association	Not retained
Toughness ~~ Peer Bonding	11.38	Positive residual association	Not retained

Table note. MI = modification index. Only theoretically interpretable residual covariances were retained in the MI-informed 11-function two-factor sensitivity model. No cross-loadings were added. Additional modification-index suggestions were not retained in order to limit overfitting in a small sample.

. These residual covariances were therefore added to an MI-informed two-factor sensitivity model.

The MI-informed two-factor sensitivity model showed improved fit relative to the unifactorial model and the unmodified two-factor models: χ² = 59.88, df = 41, χ²/df = 1.46, CFI = 0.89, TLI = 0.85, RMSEA = 0.07, 95% CI = 0.03–0.11, and SRMR = 0.08. However, because this model was data-informed, post hoc, and tested in the same small dataset, it should be interpreted as provisional and exploratory rather than confirmatory. No CFA model provided definitive evidence of factorial validity.

In the MI-informed sensitivity model, the intrapersonal factor included Affect Regulation, Self-Punishment, Anti-Dissociation/Feeling Generation, Anti-Suicide, Sensation Seeking, and Self-Care. The interpersonal factor included Interpersonal Influence, Toughness, Peer Bonding, Revenge, and Marking Distress. Autonomy and Interpersonal Boundaries were excluded only from this exploratory sensitivity model. As shown in Table 2, Table 3 and

Table 6. Evidence considered for excluding Autonomy and Interpersonal Boundaries from the post hoc CFA sensitivity model.

ISAS Function	Interpersonal EFA Loading	Intrapersonal EFA Loading	Cronbach’s Alpha	Decision
Interpersonal Boundaries	0.60	0.28	0.42	Excluded only from post hoc CFA sensitivity model
Autonomy	0.42	0.26	0.33	Excluded only from post hoc CFA sensitivity model

Table note. Autonomy and Interpersonal Boundaries were not removed from the ISAS instrument and should continue to be reported descriptively. Their exclusion was limited to the post hoc exploratory CFA sensitivity model. Interpersonal Boundaries showed a moderate primary loading but weak internal consistency and some cross-loading. Autonomy showed weak internal consistency and weaker factor definition. The exclusion decision was based on combined psychometric evidence and should be replicated before any change to the instrument is recommended.

, their exclusion was not based solely on EFA loadings but on the combined evidence of weak reliability, unstable confirmatory performance, and limited interpretability in this sample.

3.5. Convergent and Discriminant-Pattern Validity

The ISAS demonstrated modest preliminary evidence of convergent validity through positive associations with theoretically relevant YSR indicators. As shown in

Table 7. Correlations between ISAS scores and YSR variables with 95% confidence intervals.

YSR Variable	ISAS Total Functional Score r	95% CI	p	ISAS Intrapersonal r	95% CI	p	ISAS Interpersonal r	95% CI	p
YSR Item 18: self-harm behavior	0.29	0.10 to 0.47	0.004	0.39	0.21 to 0.55	<0.001	0.10	−0.10 to 0.30	0.339
YSR Item 91: suicidal ideation	0.29	0.10 to 0.47	0.004	0.40	0.22 to 0.56	<0.001	0.09	−0.11 to 0.29	0.364
Anxious/Depressed	0.24	0.04 to 0.42	0.018	0.36	0.17 to 0.52	<0.001	0.05	−0.15 to 0.25	0.618
Withdrawn/Depressed	0.32	0.13 to 0.49	0.001	0.32	0.13 to 0.49	0.001	0.23	0.02 to 0.41	0.028
Somatic Complaints	0.20	0.00 to 0.39	0.048	0.16	−0.04 to 0.35	0.121	0.18	−0.02 to 0.37	0.076
Social Problems	0.31	0.12 to 0.48	0.002	0.32	0.13 to 0.49	0.001	0.20	0.00 to 0.39	0.049
Thought Problems	0.17	−0.03 to 0.36	0.091	0.23	0.03 to 0.41	0.026	0.07	−0.14 to 0.27	0.519
Attention Problems	0.26	0.07 to 0.44	0.010	0.31	0.12 to 0.48	0.002	0.14	−0.07 to 0.33	0.187
Rule-Breaking Behavior	0.15	−0.05 to 0.34	0.146	0.20	0.00 to 0.39	0.053	0.06	−0.15 to 0.25	0.593
Aggressive Behavior	0.15	−0.06 to 0.34	0.161	0.20	0.00 to 0.38	0.054	0.05	−0.16 to 0.25	0.647
Other Problems	0.32	0.13 to 0.49	0.001	0.34	0.14 to 0.50	<0.001	0.21	0.00 to 0.39	0.045
Internalizing Problems	0.27	0.07 to 0.45	0.008	0.32	0.12 to 0.49	0.002	0.14	−0.07 to 0.33	0.182
Externalizing Problems	0.17	−0.03 to 0.36	0.102	0.23	0.03 to 0.41	0.027	0.06	−0.14 to 0.26	0.572
Total Problems	0.31	0.11 to 0.48	0.002	0.38	0.19 to 0.54	<0.001	0.15	−0.06 to 0.34	0.160

Table note. YSR = Youth Self-Report; ISAS = Inventory of Statements About Self-Injury; CI = confidence interval; p = p-value. Correlations are pooled Pearson’s r values across imputed datasets. Confidence intervals are pooled 95% confidence intervals based on Fisher’s z transformation. YSR Items 18 and 91 were used as the closest available clinical indicators of self-harm behavior and suicidal ideation, respectively. Because no independent NSSI-specific or suicidality-specific validation measure was available, these findings should be interpreted as convergent and discriminant-pattern evidence rather than formal divergent validity. ISAS total, intrapersonal, and interpersonal scores were calculated using the original 13-function ISAS scoring structure; they do not refer to the post hoc 11-function CFA sensitivity model.

, among the a priori YSR criterion items, the original broad ISAS intrapersonal score showed positive associations with YSR self-harm behavior (Item 18; r = 0.39, 95% CI = 0.21–0.55, p < 0.001) and suicidal ideation (Item 91; r = 0.40, 95% CI = 0.22–0.56, p < 0.001). The ISAS total functional score based on all 13 original functions was also positively associated with YSR self-harm behavior (r = 0.29, 95% CI = 0.10–0.47, p = 0.004) and suicidal ideation (r = 0.29, 95% CI = 0.10–0.47, p = 0.004).

Correlations with broader YSR scales were generally small to moderate. The original 13-function ISAS intrapersonal score was positively associated with internalizing-related YSR dimensions, including Anxious/Depressed, Withdrawn/Depressed, and Internalizing Problems, and also showed a moderate association with YSR Total Problems (r = 0.38, 95% CI = 0.19–0.54, p < 0.001). Associations involving the original ISAS interpersonal score were generally weaker than those involving the intrapersonal score. Small positive associations were observed between the interpersonal score and Withdrawn/Depressed, Social Problems, and Other Problems, whereas associations with self-harm behavior, suicidal ideation, Internalizing Problems, Externalizing Problems, Rule-Breaking Behavior, Aggressive Behavior, and Total Problems were weak and non-significant. Overall, correlations with Rule-Breaking Behavior, Aggressive Behavior, and Externalizing Problems were smaller than those observed for self-harm behavior, suicidal ideation, internalizing-related scales, and Total Problems. However, because no independent NSSI-specific or suicidality-specific measure was available, these findings should be interpreted as preliminary convergent and discriminant-pattern evidence rather than as definitive evidence of divergent validity.

4. Discussion

The present study provides a preliminary psychometric evaluation of the Inventory of Statements About Self-Injury (ISAS) in a Greek inpatient psychiatric sample of adolescents. This is a clinically relevant population in which NSSI is common and closely related to emotional distress, psychiatric comorbidity, and suicide-related risk. Overall, the findings provide initial support for the potential usefulness of the ISAS as an adjunctive clinical assessment tool in this population. However, the results should not be interpreted as definitive validation. The sample was small and predominantly female, EFA and CFA were conducted in the same dataset, and several CFA models showed inadequate or only partially improved fit. Accordingly, the factor-analytic findings are best understood as exploratory and hypothesis-generating.

4.1. Psychometric Interpretation of the Findings

One of the main exploratory findings was a broad two-factor structure reflecting intrapersonal and interpersonal functions of self-injury. This pattern is broadly consistent with the original conceptualization of the ISAS and with previous studies identifying intrapersonal and interpersonal domains of NSSI functions [26,33,34,35,38,43,44,45]. In the present sample, intrapersonal functions were more strongly endorsed than interpersonal functions, suggesting that self-injury was often experienced as a means of regulating internal distress. However, because the EFA and CFA were conducted in the same dataset and because the CFA models did not provide definitive fit, this structure should be interpreted as provisional and requiring replication in an independent Greek adolescent sample.

The CFA findings further support a cautious interpretation. The original theoretical models did not show satisfactory fit. Although the modified three-factor model and the MI-informed 11-function sensitivity model showed improved fit, both were exploratory, data-informed, examined in the same small dataset, and not independently cross-validated. Therefore, neither model should be interpreted as confirming a revised Greek ISAS structure, because model respecification within the same dataset may capitalize on sample-specific variance and requires independent replication [52]. Rather, these models identify provisional patterns that may guide future research.

The exclusion of Autonomy and Interpersonal Boundaries from the post hoc CFA sensitivity model requires careful interpretation. These functions were not removed from the ISAS instrument, and their exclusion should not be understood as evidence that they are clinically irrelevant. The EFA did not indicate that these functions were entirely unrelated to the broader interpersonal domain. Interpersonal Boundaries showed a moderate loading on the interpersonal factor, and Autonomy also loaded more strongly on the interpersonal than on the intrapersonal factor. However, both subscales showed weak internal consistency, particularly Autonomy, and their inclusion worsened the stability and interpretability of the CFA solution. This may reflect developmental factors, difficulties in interpreting autonomy- and boundary-related motives during adolescence, or the clinical severity and heterogeneity of the inpatient sample [53,54]. These interpretations remain tentative and require replication.

Self-Care also requires cautious interpretation. Although it was retained in the exploratory sensitivity model, its loading pattern was not fully distinct and suggested some overlap between intrapersonal and interpersonal domains. This is consistent with previous evidence that the placement of Self-Care may vary across samples [26,32,41]. In the present study, its interpretation should therefore remain provisional rather than definitive.

The validity findings provide modest preliminary support for the construct validity of the ISAS. The original broad ISAS intrapersonal score was positively associated with YSR self-harm behavior and suicidal ideation, which is theoretically consistent with the role of intrapersonal NSSI functions in emotional relief, self-punishment, anti-dissociation, and management of suicide-related distress [33,46,55,56]. Correlations with broader YSR syndrome and broadband scales were generally small to moderate, with the strongest broader YSR association observed between the original broad ISAS intrapersonal score and YSR Total Problems. Given that the study had limited power to detect small correlations, non-significant or weak associations should be interpreted cautiously. Therefore, the present findings should not be overstated as strong convergent validity evidence.

The findings also provide some preliminary evidence of construct discrimination, although not formal divergent validity. The original broad ISAS intrapersonal score showed stronger associations with YSR self-harm behavior, suicidal ideation, internalizing-related dimensions, and Total Problems than with rule-breaking behavior, aggressive behavior, or externalizing problems. In contrast, the original broad ISAS interpersonal score showed weaker associations than the intrapersonal score, with only small associations observed for some broader YSR dimensions. However, because all external validity indicators were derived from the YSR, and because no independent NSSI-specific or suicidality-specific measure was available, the findings should not be interpreted as definitive evidence of divergent validity.

4.2. Clinical Implications

These findings have several clinical implications, provided they are interpreted cautiously. First, the ISAS may be useful as an adjunctive tool in routine assessment in adolescent inpatient psychiatric settings. In such settings, clinicians often need to assess not only whether self-injury has occurred, but also why the adolescent uses self-injury and what psychological or interpersonal function it serves. The ISAS may help structure this assessment by distinguishing between original broad intrapersonal motives, such as affect regulation, self-punishment, anti-dissociation, anti-suicide functions, and marking distress, and original broad interpersonal motives, such as interpersonal influence, peer bonding, revenge, toughness, autonomy, interpersonal-boundary functions, self-care, and sensation seeking [3,45,55,57,58].

Second, the predominance of intrapersonal functions in the present sample suggests that, for many hospitalized adolescents, NSSI may primarily serve as a strategy for regulating intense internal distress. Clinically, high intrapersonal ISAS scores may indicate the need for interventions targeting emotion regulation, distress tolerance, self-criticism, dissociative experiences, and alternative coping strategies. They may also indicate the need for careful assessment of suicide-related risk, particularly because the original broad ISAS intrapersonal score was associated with YSR self-harm behavior and suicidal ideation. Although NSSI and suicidal behavior are conceptually distinct, the presence of strong intrapersonal motives, especially anti-suicide and self-punishment functions, may identify adolescents requiring closer monitoring and more detailed safety planning.

Third, interpersonal functions, although less strongly endorsed overall, remain clinically important. Elevated interpersonal ISAS scores may suggest that self-injury is embedded in difficulties with communication, attachment, family conflict, peer relationships, or boundary formation. In these cases, treatment planning may need to include family-based work, interpersonal problem solving, communication skills, and interventions addressing relational stressors.

Finally, the ISAS should not be used as a stand-alone diagnostic or risk-assessment instrument. Instead, ISAS results should be interpreted alongside clinical interview findings, psychiatric diagnosis, suicide-risk assessment, family information, and longitudinal observation during hospitalization. Given the preliminary nature of the present findings, ISAS scores should guide clinical hypotheses rather than determine clinical decisions on their own.

4.3. Limitations and Future Research

The findings should be interpreted in light of several limitations. First, the sample was relatively small for factor-analytic procedures, particularly CFA, and was drawn from a single psychiatric inpatient setting. Although the final analytic sample included N = 95 adolescents, the sample was markedly imbalanced by sex, with 82 females and 13 males. According to commonly cited sample-size guidance, including Comrey and Lee’s criteria [59], this sample size is small for factor analysis and substantially below what would be considered good for stable factor-analytic estimation. The sample size allowed only a preliminary examination of the ISAS functional structure and was better powered to detect moderate than small factor loadings or validity correlations. With N = 95, the study had approximately 80% power to detect correlations of about |r| = 0.28, indicating limited power for small convergent validity effects. RMSEA-based CFA sensitivity analyses also indicated limited power to distinguish close model fit from modest misfit, particularly for RMSEA = 0.08. Therefore, the CFA findings should be interpreted as exploratory and not as definitive confirmation of the retained sensitivity model.

Second, both EFA and CFA were conducted within the same dataset. This violates the ideal principle of cross-validation and increases the risk of capitalization on chance, particularly when post hoc model modifications are introduced [52]. The CFA results are therefore best interpreted as illustrative and hypothesis-generating. Independent replication in a larger sample is necessary before any firm conclusions can be drawn about the factor structure of the Greek ISAS.

Third, measurement invariance across gender and age could not be tested because the male subgroup was very small and age-based subdivision would have produced small and uneven groups. Consequently, the present findings should not be interpreted as evidence that the ISAS operates equivalently across gender or age groups. The predominance of female participants and the single-site inpatient design also limit generalizability to male adolescents, outpatient adolescents, community samples, and non-clinical populations.

Fourth, the study relied on questionnaire data recorded in clinical files as part of routine inpatient assessment rather than data collected prospectively under standardized research conditions. This design may have led to under-ascertainment of NSSI. The clinical records could only capture self-injury that adolescents disclosed and that was documented during assessment; unreported, minimized, forgotten, or clinically undocumented NSSI episodes could not be recovered retrospectively. This limitation is particularly important in an inpatient psychiatric sample, where acute distress, shame, fear of consequences, dissociation, depressive symptoms, limited insight, or the immediate clinical context may influence how adolescents recall and report self-injury.

The use of routine clinical data has both strengths and limitations. On the one hand, it reflects real-world inpatient assessment and therefore provides clinically relevant information from a high-risk population. On the other hand, routine clinical data are not collected under standardized research conditions. Questionnaire completion may have varied according to clinical workload, patient distress, staff availability, and the immediate priorities of inpatient care. As a result, the findings should be interpreted as reflecting disclosed and documented clinical information rather than systematically ascertained research data.

Fifth, selection bias may have been introduced because only records with available questionnaire data were included. Adolescents who were unable or unwilling to complete questionnaires, or for whom routine questionnaire administration was not feasible, may have differed systematically from those included in the final sample. In addition, questionnaire administration occurred as part of routine clinical care rather than a standardized research protocol, and no formal inter-rater reliability assessment was available for chart extraction. These factors may have introduced variability in administration conditions and clinical documentation.

Sixth, the retrospective clinical records did not include an independent NSSI-specific or suicidality-specific validation measure. Therefore, formal divergent validity and more specific convergent validity with established NSSI or suicidality measures could not be evaluated.

Finally, several ISAS functions showed weaker psychometric performance. Autonomy and Interpersonal Boundaries showed low internal consistency, and Self-Care showed a less distinct loading pattern. These findings suggest that some functions may be less stable in younger or clinically severe adolescent samples, but the developmental interpretation remains speculative and requires further study.

These limitations point to several directions for further study. Larger and more diverse samples, including outpatient and community-based adolescent populations, are needed to examine the psychometric performance of the Greek ISAS more robustly. Prospective studies are particularly needed. Such studies should administer the ISAS under standardized research conditions, include structured clinical assessment of NSSI, record reasons for missing or incomplete responses, and, where ethically and clinically appropriate, use repeated assessments during hospitalization and follow-up. Future research should also include independent NSSI- and suicidality-specific validation measures, such as the Suicidal Ideation Questionnaire–Junior [49], the Deliberate Self-Harm Inventory [60], the Self-Injurious Thoughts and Behaviors Interview [61], or the Non-Suicidal Self-Injury Assessment Tool [28]. Including such measures would allow stronger tests of convergent validity, divergent validity, temporal stability, and the clinical predictive value of ISAS function scores. Finally, larger samples are needed to test measurement invariance across gender and age and to examine whether low-performing functions such as Autonomy, Interpersonal Boundaries, and Self-Care require developmental adaptation or rewording for younger respondents.

5. Conclusions

To our knowledge, this is the first study to examine the psychometric properties of the Inventory of Statements About Self-Injury (ISAS) in a Greek adolescent inpatient psychiatric sample. The findings suggest that the ISAS may be a promising adjunctive instrument for assessing NSSI functions in adolescents receiving psychiatric care. The scale showed satisfactory internal consistency at the broader factor level and modest preliminary evidence of convergent validity with clinically relevant YSR indicators, particularly self-harm behavior and suicidal ideation. The exploratory factor-analytic findings were broadly consistent with an intrapersonal–interpersonal distinction, but the CFA findings were not definitive.

Autonomy and Interpersonal Boundaries showed weak internal consistency and were excluded only from a post hoc exploratory CFA sensitivity model, although their EFA loadings suggested that they may still be related to the broader interpersonal domain. Their exclusion should therefore be interpreted as a psychometric decision specific to this exploratory sensitivity analysis rather than as evidence that these functions are clinically unimportant or should be removed from the ISAS. Because the sample was small, predominantly female, and drawn from a single inpatient setting, measurement invariance by gender and age could not be examined and generalizability is limited. In addition, because the data were obtained retrospectively from clinical records, the descriptive NSSI estimates may underestimate undisclosed or undocumented self-injury.

Clinically, the ISAS may help inpatient teams identify the functions maintaining NSSI and use this information to inform individualized formulation, treatment planning, safety planning, and discharge preparation. However, ISAS scores should not be used as stand-alone diagnostic or risk-assessment indicators. Further validation in larger, more diverse, prospective, and independent samples is needed.