Development and validation of supporting ability scale for caregivers of adolescents with non-suicidal self-injury

Abstract

Background

Non-suicidal self-injury (NSSI) is a significant mental health concern among adolescents, characterized by deliberate self-harm without suicidal intent. Caregivers play a crucial role in supporting adolescents with NSSI, yet there is a lack of standardized instruments to assess their supporting ability. The aim of this study was to develop and validate a scale to evaluate the supporting ability of caregivers for adolescents with NSSI.

Objective

The goal was to develop and validate a scale that assesses the supporting ability of caregivers of adolescents with NSSI, incorporating dimensions of knowledge, attitudes, practice, health, and interpersonal relationships and resource utilization.

Methods

The scale development process involved item generation through a comprehensive literature review, scale refinement via a two-round Delphi expert consultation, and validation through exploratory factor analysis (EFA) and confirmatory factor analysis (CFA). The initial item pool was generated based on the "Knowledge-Attitude-Practice Theory" and Farran's caregiver skill model framework. A total of 15 experts participated in the Delphi consultation, and 431 caregivers of adolescents with NSSI were recruited for scale validation. The final scale was subjected to item analysis, EFA, and CFA to assess its psychometric properties.

Result

The Delphi consultation resulted in a refined scale with 37 items across five dimensions. Item analysis and factor analysis led to the scale with 32 items across 5 dimensions. The EFA revealed a five-factor structure with a cumulative variance contribution rate of 75.38% and factor loadings ranging from 0.771 to 0.858. The CFA confirmed the factor structure with satisfactory fit indices. Validity analysis demonstrated that the scale had satisfactory convergent validity (AVE: 0.545-0.637; CR: 0.828-0.924) and discriminant validity. The scale demonstrated high internal consistency, with a Cronbach’s α coefficient of 0.935 for the overall scale and coefficients ranging from 0.822 to 0.924 for individual dimensions. The split-half reliability analysis also showed a Spearman-Brown coefficient of 0.781.

Conclusion

The scale provides a preliminary evaluation framework to identify caregiver supporting ability for NSSI adolescents. However, we did not assess the scale's external validity or its predictive relationship with adolescent outcomes. Therefore, the scale's clinical utility and effectiveness in evaluating interventions remain to be established through further research. Future research should explore the scale's test-retest reliability, responsiveness to change, and its utility in predicting adolescent outcomes.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40359-026-04962-z.

Introduction

Non-suicidal self-injury (NSSI) is a significant and growing public health concern among adolescents, defined as the deliberate, repetitive, and socially unacceptable infliction of injury to one’s own body tissue without suicidal intent [1]. Though not a standalone diagnosis in DSM-5, NSSI was included in DSM-5-TR (2022) as an “Other Condition that may be a focus of clinical attention“ [2]. NSSI behaviors can take various forms, such as cutting, burning, or hitting oneself. The aggregate prevalence of NSSI among a non-clinical sample of adolescents was 22.0% over a lifetime [3]. NSSI is an even stronger predictor of future suicidal behavior than is a history of suicide attempts [4]. Adolescents who engage in NSSI often face a complex interplay of emotional, psychological, and social challenges [5]. They often experience intense emotions such as depression, anxiety, and distress, which can drive them to engage in self-injurious behaviors as a maladaptive coping mechanism [6]. Moreover, these adolescents often struggle with interpersonal relationships, academic performance, and overall functioning in daily life [7].

The presence of NSSI can also have a significant impact on their families and caregivers, who may feel overwhelmed, helpless, and uncertain about how to provide appropriate support [8]. Caregivers, including parents, guardians, and other primary caregivers, play a crucial role in the lives of adolescents with NSSI, serving as the first line of support and significantly influencing the recovery and well-being of these young individuals [9]. However, providing effective support to adolescents with NSSI requires a range of skills and abilities. Caregivers need to be able to recognize the signs of NSSI, understand the underlying emotional and psychological issues, and provide a supportive and non-judgmental environment [10]. They also need to be equipped with strategies to help manage crises, promote healthy coping mechanisms, and facilitate access to professional help when necessary [10, 11].

Despite the importance of caregivers in the support system of adolescents with NSSI, there is a lack of validated instruments to assess their supporting ability. Existing tools often focus on attitudes of health professionals towards self-injuring clients [12, 13], but few specifically address the unique challenges and competencies required to support adolescents engaging in NSSI. For instance, “Attitudes Scale” developed by Jennifer J Muehlenkamp et al. [12]. primarily assesses health professionals’ attitudes toward self-injuring clients, yet it fails to address the specific supportive capabilities required by caregivers of adolescents with NSSI. In addition, the “Suicidal Caring Ability Scale,” which was developed by researchers [14] based on adaptations of previous qualitative and quantitative studies [15, 16], focuses on suicidal populations but lacks targeted assessment tools for the specific competencies of NSSI caregivers. This gap in assessment tools limits our understanding of caregivers’ strengths and areas for improvement, as well as their overall effectiveness in supporting these vulnerable young individuals. To address this gap, we developed the Supporting Ability Scale for Caregivers of Adolescents with NSSI (SAS-CAN) grounded in the Knowledge-Attitude-Practice (KAP) model and Farran’s caregiver skill model. The KAP model illustrates that health behaviors are a continuous process evolved from knowledge acquisition to belief formation, and finally to behavior generation [17]. Farran’s model further expands this by suggesting that caregiver skill is not merely a set of tasks but involves complex interactions between the caregiver’s competence, their health status, and environmental resources [18]. Guided by an integration of the KAP model and Farran’s caregiver model, we define “supporting ability” in this study as a reflective, multidimensional construct (Fig. 1). It encompasses the cognitive knowledge, affective attitudes, behavioral practices, personal health resources, and social resource utilization required by caregivers to effectively respond to the needs of adolescents with NSSI. Unlike caregiver self-efficacy, which is a subjective belief that serves as a psychological antecedent to action [19], supporting ability represents the actual functional capability—an outcome of that belief combined with objective resources. Similarly, while caregiver competence focuses on the perceived adequacy of skills [20], supporting ability extends this to include the sustainability dimension, specifically the caregiver’s own health. Furthermore, we distinguish supporting ability from caregiver burden; whereas burden is a negative emotional outcome resulting from the caregiving process, supporting ability acts as a mediating resource that potentially mitigates such strain. Grounded in the KAP Theory and Farran’s caregiver skill model, the scale conceptualizes supporting ability as a multidimensional construct. It integrates cognitive knowledge, affective attitudes, and behavioral practice with essential personal health and relational interpersonal relationship and resource utilization dimensions. This scale aims to comprehensively assess the various dimensions of caregivers’ supporting ability. By developing a reliable and valid instrument, we can better identify caregivers’ needs, tailor interventions to enhance their supporting ability, and reduce caregivers’ burden and psychological distress, decrease NSSI frequency and improve mental health of adolescents ultimately.

Fig. 1.

Conceptual model of supporting ability for caregivers of adolescents with non-suicidal self-injury

Methods

The scale was developed by several key steps, including item generation by literature review, scale refinement by Delphi expert consultation and preliminary survey and formal scale testing by cross-sectional study.

Item generation

A comprehensive literature review was conducted to generate an item pool. According to the Farran caregiver support theory [18], the assessment of NSSI supporting ability, knowledge, attitude and behavior cannot comprehensively evaluate the ability of caregivers, and the personal quality and supportive relationship of caregivers should also belong to the category of care ability. Specifically, the knowledge dimension reflects caregivers’ cognitive understanding of NSSI, including its definition, risk factors, motives and interventions. The attitudes dimension pertains to caregivers’ affective beliefs and willingness to support adolescents with NSSI. The Practice dimension involves the behavioral strategies and actions caregivers employ when responding to NSSI situations. The health dimension addresses caregivers’ own physical and mental well-being. The interpersonal relationships and resource utilization dimension encompasses caregivers’ relational networks and their ability to seek and use external resources. By carefully reviewing literature, guided by the “K-A-P"Theory and caregiver skill model framework proposed by Farran [18], the literature review included searching keywords related to caregivers (e.g., “primary caregivers”, “parents”, “guardians”, “caregiver burden”, “caregiver stress”, “coping” and “family support”) and specific NSSI behaviors (e.g., “cutting,” “burning,” “scratching,” “banging,” “self-harm”, “self-injury,” “self-poisoning,” “overdose,” and “self-injurious behavior”). Furthermore, to ensure comprehensiveness, we examined NSSI subtypes and behaviors mentioned in established scales, such as the Inventory of Statements About Self-Injury (ISAS) [21] and Functional Assessment of Self-Mutilation (FASM) [22]. These search terms and scale-derived content were used to search databases such as Embase, CINAHL, Cochrane Library, PubMed, APA PsycArticles, APA PsycInfo, Psychology and Behavioral Sciences Collection, Web of Science, Scopus, CNKI (China National Knowledge Infrastructure), VIP Database, Wanfang Database, Chinese Biomedical Literature Database, and China Medical Academic Conference Full-text Database to collect relevant content related to the research topic. The study group conducted collaborative meetings to compile and refine ideas from relevant literature. Ultimately, a preliminary questionnaire was formed, encompassing five dimensions (Knowledge, Attitudes, Practice, Health, and Interpersonal Relationships and Resource Utilization) with a total of 42 items. Each item is rated on a 5-point Likert scale ranging from 1 (Strongly Disagree) to 5 (Strongly Agree), with higher scores indicating greater supporting ability.

Scale refinement

The selection and revision of scale items applied a two-round Delphi expert consultation method. The expert inclusion criteria were as follows: (1) Bachelor’s degree or above; (2) Senior professional title; (3) More than 10 years of experience in mental health, public health, or nursing practice in Psychiatric department. (4) All experts participated voluntarily in this study. A self-made expert consultation form includes four parts. (1) Guidance: Brief explains of the purpose, methods, significance of the study, and requirements for filling out the questionnaire; (2) Information of General Experts; (3) Consultation form for assessing items pool, including the expert’s degree of agreement with each item and the opinion modification column. A Likert 5-point scoring method is used for item importance rating, with 5 = very important, 4 = important, 3 = moderately important, 2 = unimportant, 1 = very unimportant. An “Add/Delete Items” option and a column for “Modification Suggestions” are provided. (4) Survey Form on the Expert’s Familiarity with the Study and Judgment Basis. The expert consultation item screening criteria are as follows: average importance rating > 3.50, coefficient of variation < 0.25, and satisfaction rate > 20%. After an objective evaluation by the research team, the questionnaire items were appropriately selected and modified based on expert opinions and item screening criteria, resulting in the initial draft of the questionnaire including 5 dimensions and 37 items.

Preliminary survey

A purposive sampling method was employed to conduct a preliminary survey on caregivers of 30 adolescents engaging in NSSI who were hospitalized in the psychiatric and psychological department of a tertiary hospital in Guangdong Province in March 2023. The aim was to further examine the clarity and understandability of each item. Caregivers were included if their child was diagnosed with NSSI according to DSM-5 for the first time [1]. The inclusion criteria are outlined below: Biological parents or grandparents whose child or grandchild was aged 12 to 19 years [23]; ② Long-term primary caregivers (defined as those who have been taking care of the adolescent continuously for more than 6 months); ③ Not currently in the acute stage of a psychiatric illness. Exclusion criteria: People who experienced major stress events such as bereavement and divorce within 3 months. A total of 30 questionnaires were distributed, and 30 valid questionnaires were collected, resulting in a response rate of 100%. Based on the preliminary survey results The language expression of 5 items was modified, and the number of items was not deleted or added.

Formal scale testing

The inclusion and exclusion criteria remained the same as in the preliminary survey. Using convenient sampling, caregivers of NSSI adolescents hospitalized in a Grade A tertiary general hospital in Guangdong Province were selected as the participants of the survey from March to May 2023 for project analysis, exploratory factor analysis, and reliability testing. Then, from June to August 2023, the same subjects were selected for confirmatory factor analysis. The study has been approved by the hospital’s ethics committee [No. KY2023-336-02], and all procedures were conducted in accordance with the principles outlined in the Helsinki Declaration. All participants voluntarily joined the study, signing an informed consent form. In this study, “supporting ability” was conceptualized as a reflective multidimensional construct. We posited that the latent trait of supporting ability manifests through its indicators; therefore, variations in the level of supporting ability are expected to be reflected in variations across the observed items. This theoretical assumption supports the use of internal consistency reliability (Cronbach’s α) and Confirmatory Factor Analysis (CFA) to evaluate the scale’s psychometric properties, as these methods are predicated on the reflective model logic where items are expected to correlate highly with each other.

To evaluate construct validity using exploratory factor analysis (EFA), a sample of five times the minimum number of items is required and a CFA requires a minimum sample size of 200 participants [24]. The preliminary SAS-CAN consisted of 37 items; therefore, 185 participants were needed for the EFA and 200 for the CFA. Considering a 85% valid response rate, a total of 453 caregivers were recruited. After excluding questionnaires that had duplicate or inappropriate answers, questionnaires from 431 participants were included in the analysis.

Statistical analysis

IBM SPSS statistical software version 26.0 and R version 4.4.2 were used for data analysis. The characteristics of the participants were analyzed using descriptive statistics, including frequency, percentage, mean, and standard deviation. Differences between the CFA and EFA groups were analyzed using the chi-square test for categorical variables and the independent samples t-test for continuous variables. For Item-analysis, The discriminatory power and discrimination of each item were examined using the critical-ratio method, the correlation-coefficient method, and Cronbach’s alpha coefficient [25]. For EFA, principal axis factoring with oblimin rotation was used to extract common factors [26]. The Kaiser-Meyer Olkin (KMO) test and Bartlett’s test for sphericity were conducted to determine whether the collected data were suitable for factor analysis. For CFA, the maximum likelihood estimation method was used. The goodness of fit of the model was evaluated using the following indices: the chi-square to degrees of freedom ratio (CMIN/df), Goodness of Fit Index (GFI), Standardized Root Mean Square Residual (SRMR), Root Mean Square Error of Approximation (RMSEA), Comparative Fit Index (CFI) and No correlated error terms or model modifications (NNFI) were applied during the CFA.

Results

Delphi survey results

A total of 15 experts were included in the study, with an age range of 37 to 54 years (mean age: 43.00 ± 6.26). Among them, 5 held a doctoral degree, 3 had a master’s degree, and 7 had a bachelor’s degree. The experts had a work experience ranging from 7 to 32 years (mean: 18.62 ± 8.03). Regarding professional titles, 5 held intermediate titles, 7 held associate senior titles, and 3 held senior titles. Two rounds of expert consultations were conducted, with positive coefficients of 87.50% and 92.86%, authoritative coefficients of 0.729 and 0.746, and Kendall’s concordance coefficients of 0.223 and 0.224 (all P < 0.001). The moderate consensus is due to the interdisciplinary backgrounds of the experts, while the statistically significant values, combined with high positive and authoritative coefficients, confirm acceptable content validity.

In the first round, the importance scores of the items ranged from 3.714 to 4.929, with a coefficient of variation between 0.0542 and 0.418. The full score percentage ranged from 28.57% to 92.86%. Ten experts provided feedback, leading to the modification or merging of 23 items. One new item was added: “I am willing to take responsibility for helping and caring for children with NSSI.” Ultimately, 5 dimensions and 37 items were retained.

In the second round, the importance scores of the items ranged from 4.615 to 5.000, with a coefficient of variation between 0 and 0.141. The full score percentage ranged from 61.54% to 100.00%. Six experts provided modification suggestions, resulting in the modification of 6 items. No new items were added or deleted. The final version retained 5 dimensions and 37 items ( See Table 1 for the full version of the scale).

Table 1.

Supporting ability scale for caregivers of adolescents with non-suicidal self-injury

NO	Items
K1	Nonsuicidal Self-Injury (NSSI) refers to an individual’s direct and intentional act of harming their own body without suicidal intent, which is not socially or culturally accepted.
K2	NSSI manifests in various forms, including common behaviors such as cutting, burning, scratching, abrasions, and biting, which interfere with wound healing.
K3	The number of adolescents with NSSI is increasing, and they account for a higher proportion among those with mental and psychological disorders.
K4	Most NSSI occurs between the ages of 12 and 16, gradually decreasing after the age of 18, and is more common in females than males.
K5	Children NSSI to alleviate negative emotions, punish themselves, express pain to others, attract attention or gain concern from others, etc.
K6	Risk factors for NSSI among adolescents include personal factors (such as introversion, low self-esteem, mental illness, etc.), family factors (such as childhood abuse, excessive control, absence of family members, etc.), and academic factors (school bullying, academic pressure, imitation by classmates), etc.
K7	Signs of NSSI include: unexplained cuts, burns, or bruises on arms, legs, and abdomen; searching for razors, sharp blades, or other items that could be used for NSSI; increased time spent alone; wearing clothing that doesn’t match the weather, etc.
K8	If there is a risk to life (such as suicidal thoughts in a child), they must be taken to a psychiatric department for evaluation. If the child refuses, police assistance can be sought.
K9	If there is no risk to life, you can take the child to see a psychologist or mental health worker.
K10	With proper intervention, NSSI tends to decrease or even disappear with age. Conversely, it can develop into lifelong behavior or serious consequences such as suicide.
A1	I believe that NSSI, accompanied by physical and mental pain, is an issue that needs to be addressed rather than a shameful secret.
A2	I believe that primary caregivers play an important role in helping adolescents with NSSI.
A3	I am willing to take responsibility for helping and caring for children with NSSI.
A4	I am willing to learn relevant knowledge and skills to care for children with NSSI.
A5	I believe I can effectively identify NSSI behavior in children.
A6	I believe I can face children’s NSSI behavior head-on and handle it calmly.
A7	I am very patient with children who repeatedly NSSI and believe they will get better because treatment takes time and requires comprehensive intervention.
A8	I believe I can help the child to reduce or even stop self-harming.
P1	When a child engages in NSSI, express concern and understanding for them, and view and analyze the behavior from their perspective.
P2	When a child engages in NSSI, I will patiently guide them to confide in me to understand the underlying motives and alleviate the situation, avoiding suicidal impulses.
P3	When a child does not want to confide, I will choose to respect them, accompany them quietly, calmly express my concern, and tell them that I will try to talk again.
P4	When a child engages in NSSI, remove tools used for NSSI and use distraction techniques to control the urge to NSSI.
P5	Limit access to NSSI peers, internet and social media when necessary.
P6	Provide children with trusted human resources to turn to when they feel unsafe, including family members, friends, and mental health services.
P7	Discuss with children the decision to seek medical and mental health assistance to alleviate their fear of seeking further help from these services.
P8	I am able to help children properly treat wounds.
H1	I have abundant physical strength and energy, or although I have minor health issues, I can find suitable ways to manage physical fatigue.
H2	I have not been diagnosed with a mental illness, or I have been diagnosed with a mental illness but my current mental state is stable and I am able to take care of my child.
H3	I occasionally feel psychological pain, but I can find suitable ways to relieve it.
H4	I have strong adaptability and can calmly cope with changes in my life and work caused by a child’s NSSI behavior.
R1	I often have heart-to-heart talks with my child and trust each other.
R2	I have harmonious relationships with other people besides my child (family members, teachers, classmates).
R3	I can obtain reasonable advice, help, and psychological support for coping with NSSI from family members and friends.
R4	Medical institutions can provide me with reasonable advice and help for coping with NSSI.
R5	My child’s school can provide me with reasonable advice and help for coping with NSSI.
R6	Books and online information recommended by mental health professionals can provide me with reasonable advice and help for coping with NSSI.
R7	I can obtain support or help from social groups recommended by professionals.

The final version excluded items K2, K3, K10, A8 and P1, with 32 items remaining in the end

Characteristics of participants

A total of 431 participants were included in the formal scale testing (220 for the EFA and 211 for the CFA). There were no significant differences between the EFA and CFA groups in terms of gender, age, education, marital status, role of caregiver, and place of residence (Table 2).

Table 2.

Demographic characteristics of participants (n = 431)

Variables	Categories	Total (n = 431) n(%) or M ± SD	EFA group (n = 220) n(%) or M ± SD	CFA group (n = 211) n(%) or M ± SD	χ2/t	P
Gender	Male	92(21.35)	46(20.91)	46(21.80)	0.051	0.821
	Female	339(78.65)	174(79.09)	165(78.20)
Age (year)		44.00 ± 5.88	43.68 ± 5.41	44.33 ± 6.33	-1.135	0.257
Education	Primary school or below	48(11.14)	23(10.45)	25(11.85)	6.455	0.168
	Junior high school	122(28.31)	65(29.55)	57(27.01)
	Senior high school / vocational school / Technical school	83(19.26)	46(20.91)	37(17.54)
	Junior college	72(16.71)	42(19.09)	30(14.22)
	Bachelor’s degree or above	106(24.59)	44(20.00)	62(29.38)
Marital status	Married	375(87.01)	199(90.45)	176(83.41)	7.293	0.063
	Divorced	42(9.74)	14(6.36)	28(13.27)
	Widowed	4(0.93)	3(1.36)	1(0.47)
	Remarried	10(2.32)	4(1.82)	6(2.84)
Role of caregiver	Father	88(20.42)	46(20.91)	42(19.91)	1.139	0.566
	Mother	339(78.65)	173(78.64)	166(78.67)
	Others	4(0.93)	1(0.45)	3(1.42)
Place of Residence	Rural area	79(18.33)	39(17.73)	40(18.96)	0.109	0.741
	Urban area	352(81.67)	181(82.27)	171(81.04)

Item analysis

Critical ratio method

The results of this study using the critical ratio method indicate that there is a statistically significant difference in item scores between the high-scoring group and the low-scoring group (all P < 0.05). This suggests that all 37 items demonstrate good discriminatory power (Supplementary file 1).

Correlation coefficient method

In the Pearson correlation analysis between each item and the total questionnaire, the correlation coefficients for the initial 37 items ranged from 0.148 to 0.672 (all P < 0.05). While most items demonstrated acceptable correlations with the total score, the lower bound (0.148) fell below the conventional threshold of 0.30, indicating weak discrimination for Item A8. This item was subsequently subjected to further analysis and removal (Supplementary file 2). In addition, average inter-item correlation analysis was conducted to assess the scale’s internal consistency reliability. The average correlation coefficient of the 37 items was 0.321, falling within the ideal range of 0.15 to 0.50. However, this analysis identified high inter-item correlations: K3 correlated highly with the other 9 items in the Knowledge dimension (r = 0.703–0.726); K2 with K1, K3, K5 and K9 (r = 0.703–0.734); K10 with K3, K8 and K9 (r = 0.709–0.726); K5 also with K6 and K8 (r = 0.728, 0.702), and P1 with P2 (r = 0.703). Accordingly, an expert panel (3 psychiatrists, 2 clinical psychologists, 2 adolescent mental health specialists, 1 senior clinical nurse) was convened to discuss these items. The panel reached a consensus: K3 (NSSI epidemiological trends) was background knowledge irrelevant to caregiving ability, with its core information covered by other items; K2 (specific NSSI behaviors) was redundant for caregivers of adolescents with NSSI, as caregivers intuitively grasp such behavioral details given their children’s self-injury, and the item added few value to assessing caregiving ability; K10 assessed NSSI prognostic knowledge (the risk of chronic lifelong NSSI or subsequent suicide), which overlapped substantially with K8 (emergency responses to suicide risk). Caregivers recognizing the severe adverse outcomes of NSSI (K10) typically understand the need for urgent intervention (K8), so K10 was deleted to ensure item independence; P1 and P2 both focused on empathic understanding and communication, with the panel recommending only P2 be retained. Though K5 correlated highly with K6 and K8, the panel confirmed it measured the key cognitive dimension of NSSI motivation and was thus retained to ensure the scale’s content comprehensiveness. Thus, K2, K3, K10 and P1 were removed, with 33 items remaining for further analysis.

Internal consistency coefficient method

The internal consistency of the 33 items was assessed using Cronbach’s α coefficient, resulting in a coefficient of 0.937. when the item A8 was deleted, the reliability coefficient showed a noticeable increase. The corrected item-total correlation (CITC) value for this item was 0.107, which is less than 0.3. The Cronbach’s α for the remaining 32 items was 0.940, indicating high internal consistency. (Supplementary file 3). The possible psychological reasons for the poor performance of this item may include caregivers’ unrealistic expectations of their own abilities, perceived lack of control over the adolescent’s NSSI behavior, or ambiguity in their responsibilities in helping the adolescent stop self-harming. After the deletion of this item, caregiver self-efficacy is partially captured by other items such as “I believe I can effectively identify NSSI behavior in children” and “I believe I can face children’s NSSI behavior head-on and handle it calmly”. The final scale retained 32 items.

Exploratory factor analysis

The initial EFA was conducted on the 32 items, resulting in a Kaiser-Meyer-Olkin (KMO) value of 0.931 and a significant Bartlett’s sphericity test value of 5983.638 (P < 0.001). This indicated suitability for factor analysis. The cumulative variance contribution rate of the extracted common factors was 75.38%. The loading values of each item on the respective factors ranged from 0.771 to 0.858, as shown in Table 3. The EFA-derived five-factor structure is both theory and empirically based. Guided by the KAP Theory and Farran’s caregiver skill model, we proposed five dimensions, and EFA grouped the items into these theoretically consistent factors, each representing a distinct psychological construct: Factor 1 (Knowledge) reflects the cognitive component (caregivers’ understanding of NSSI definition, risk factors, warning signs, and intervention methods); Factor 4 (Attitudes) the affective component (caregivers’ beliefs and willingness to support NSSI adolescents); Factor 2 (Practice) the behavioral component (specific supportive strategies and actions for NSSI); Factor 5 (Health) the self-regulatory component (caregivers’ physical and mental health, and stress-coping ability); Factor 3 (Interpersonal Relationships and Resource Utilization) the relational component (caregivers’ interpersonal bonds and external resource utilization capacity). Consistent with the theoretical framework, EFA results suggest a conceptual link between caregivers’ knowledge, attitudes and supportive behaviors, with no directional or causal relationships confirmed due to the cross-sectional design.

Table 3.

Rotated factor pattern matrix of EFA (n =220)

Items	Factor Loading Coefficient					Communalities (Common Factor Variance)
	Factor 1	Factor 2	Factor 3	Factor 4	Factor 5
K1	0.176	0.845	0.149	0.097	0.140	0.796
K4	0.087	0.832	0.165	0.103	0.136	0.756
K5	0.082	0.836	0.135	0.199	0.057	0.767
K6	0.156	0.822	0.171	0.136	0.139	0.767
K7	0.107	0.831	0.128	0.108	0.080	0.737
K8	0.138	0.836	0.115	0.174	0.160	0.786
K9	0.118	0.824	0.148	0.132	0.174	0.763
A1	0.141	0.105	0.805	0.175	0.105	0.721
A2	0.148	0.129	0.829	0.166	0.073	0.758
A3	0.146	0.146	0.825	0.106	0.030	0.736
A4	0.065	0.183	0.828	0.119	0.075	0.743
A5	0.135	0.144	0.808	0.213	0.084	0.745
A6	0.139	0.153	0.783	0.199	0.123	0.711
A7	0.120	0.135	0.842	0.172	0.068	0.776
P2	0.084	0.118	0.264	0.793	0.092	0.728
P3	0.037	0.140	0.199	0.842	0.057	0.772
P4	0.160	0.113	0.211	0.797	0.096	0.727
P5	0.128	0.167	0.149	0.820	0.134	0.756
P6	0.123	0.091	0.079	0.853	0.142	0.777
P7	0.118	0.140	0.132	0.833	0.077	0.750
P8	0.168	0.171	0.139	0.771	0.129	0.687
H1	0.264	0.163	0.080	0.140	0.783	0.735
H2	0.092	0.222	0.070	0.246	0.810	0.780
H3	0.140	0.139	0.168	0.117	0.814	0.744
H4	0.109	0.206	0.119	0.105	0.833	0.773
R1	0.840	0.125	0.118	0.108	0.074	0.752
R2	0.838	0.054	0.176	0.160	0.076	0.769
R3	0.858	0.119	0.114	0.092	0.096	0.781
R4	0.836	0.095	0.163	0.122	0.118	0.763
R5	0.827	0.169	0.119	0.131	0.064	0.748
R6	0.852	0.136	0.102	0.069	0.079	0.766
R7	0.822	0.124	0.082	0.110	0.200	0.749
Eigenvalue (Before Rotation)	11.791	3.765	3.483	3.047	2.034	-
Variance Explained (%) (Before Rotation)	36.85%	11.77%	10.88%	9.52%	6.36%	-
Cumulative Variance Explained (%) (Before Rotation)	36.85%	48.62%	59.50%	69.02%	75.38%	-
Eigenvalue (After Rotation)	5.380	5.368	5.214	5.194	2.964	-
Variance Explained (%) (After Rotation)	16.81%	16.78%	16.29%	16.23%	9.26%	-
Cumulative Variance Explained (%) (After Rotation)	16.81%	33.59%	49.88%	66.11%	75.38%	-
KMO Value	0.931
Bartlett's Sphericity Value	5983.638
df	496
P	< 0.001

Confirmatory factor analysis

A CFA was performed on the five factors and 32 items of the scale. All items had a standardized factor loading of 0.40 or higher (see Table 4 for detailed estimates). To further validate the hierarchical structure of “supporting ability,” we compared three competing models: a five-factor model (the original first-order model), a one-factor model (all dimensions loaded on a single factor), and a second-order model (a higher-order factor “Supporting Ability” driving the five first-order dimensions). The fit indices are presented in Table 5. The one-factor model showed poor fit, failing to meet acceptable criteria. The five-factor model demonstrated acceptable fit as reported. The second-order model yielded comparable fit indices (CFI = 0.921, RMSEA = 0.050 ~ 0.064) and smaller AIC (130.632) and BIC (361.910) values than the five-factor model (AIC = 147.920, BIC = 409.365), indicating greater parsimony. Thus, the second-order model supports the theoretical premise that “Supporting Ability” is a higher-order construct reflected by its five dimensions.

Table 4.

Results of the confirmatory factor analysis (n =211)

Item	Standardized estimates	SE	z-value	p
K1	0.751	-	-	-
K4	0.767	0.091	11.439	< 0.001
K5	0.833	0.085	12.568	< 0.001
K6	0.837	0.087	12.620	< 0.001
K7	0.836	0.086	12.603	< 0.001
K8	0.783	0.086	11.706	< 0.001
K9	0.773	0.091	11.536	< 0.001
A1	0.746	-	-	-
A2	0.683	0.090	9.818	< 0.001
A3	0.796	0.079	11.577	< 0.001
A4	0.800	0.080	11.642	< 0.001
A5	0.744	0.088	10.765	< 0.001
A6	0.772	0.090	11.212	< 0.001
A7	0.701	0.087	10.089	< 0.001
P2	0.816	-	-	< 0.001
P3	0.820	0.074	13.636	< 0.001
P4	0.768	0.080	12.481	< 0.001
P5	0.621	0.090	9.513	< 0.001
P6	0.733	0.075	11.729	< 0.001
P7	0.764	0.080	12.381	< 0.001
P8	0.768	0.082	12.482	< 0.001
H1	0.726	-	-	-
H2	0.595	0.113	8.036	< 0.001
H3	0.770	0.105	10.312	< 0.001
H4	0.852	0.111	11.109	< 0.001
R1	0.743	-	-	-
R2	0.693	0.090	9.902	< 0.001
R3	0.752	0.094	10.797	< 0.001
R4	0.703	0.091	10.054	< 0.001
R5	0.677	0.095	9.657	< 0.001
R6	0.790	0.093	11.381	< 0.001
R7	0.801	0.098	11.548	< 0.001

The dash ‘-’ indicates that the item is a reference category

Table 5.

Goodness of fit indices for the model (n =211)

Fitness index	Criteria	Model 1 (five-factor)	Model 2(One-factor)	Model 3(Second-order)
X2	(> 0.05)	5000.449	2500.435	777.367
CMIN/df	≤ 3	1.649	5.389	1.694
GFI	> 0.80	0.811	0.446	0.816
SRMR	≤ 0.08	0.053	0.134	0.061
RMSEA (90% CI)	≤ 0.08	0.049 ~ 0.062	0.129 ~ 0.151	0.050 ~ 0.064
CFI	≥ 0.90	0.924	0.493	0.921
NNFI	> 0.90	0.918	0.458	0.914
AIC	-	147.920	104.299	130.632
BIC	-	409.365	318.818	361.910

CMIN/df Chi-square minimum / degrees of freedom, GFI Goodness of fit index, SRMR Standardized root mean-square residual, RMSEA Root mean square error of approximation, CFI Comparative fit index, NNFI Non-Normed Fit Index, AIC Akaike Information Criterion, BIC Bayesian Information Criterion

Reliability analysis

The overall Cronbach’s α coefficient for the questionnaire was 0.935, indicating high internal consistency. The Cronbach’s α coefficients for each dimension ranged from 0.822 to 0.924, suggesting good reliability within each dimension (Supplementary file 4).The McDonald’s omega coefficient for this scale was 0.941, exceeding the threshold of 0.9, thereby indicating high reliability quality of the research data. In the split - half analysis of the questionnaire, the Cronbach’s α coefficient of the first half of the questionnaire is 0.904, and that of the second half is 0.900. The Spearman - Brown coefficient is 0.781, which higher than 0.70, indicates a high level of reliability of the data.

Validity analysis

Convergent validity

As shown in Table 6, Average Variance Extracted (AVE) values ranged from 0.545 to 0.637 (all > 0.5) and Composite Reliability (CR) values from 0.828 to 0.924 (all > 0.75), and the Maximum Shared Variance (MSV) was examined with AVE values exceeding corresponding MSV values for all factors, indicating strong convergent validity.

Table 6.

Convergent and discriminant validity analysis (n =211)

Factor	AVE	CR	MSV	Knowledge	Attitude	Behavior	Health	Interpersonal Relationship and Resource Utilization
Knowledge	0.637	0.924	0.186	0.798
Attitude	0.562	0.900	0.347	0.307	0.750
Practice	0.575	0.904	0.347	0.373	0.523	0.758
Health	0.550	0.828	0.365	0.336	0.305	0.464	0.742
Interpersonal Relationship and Resource Utilization	0.545	0.893	0.365	0.396	0.358	0.410	0.510	0.738

CR Composite Reliability, AVE Average Variance Extracted; The blue number is the square root of AVE

Discriminant validity

We performed the Fornell-Larcker criterion test. The square roots of AVE (diagonal values) were compared with inter-construct correlations. The results (Table 6) confirm that the square root of AVE for each construct is greater than its correlation with other constructs, supporting discriminant validity.

Discussion

World Health Organization and the United Nations Children’s Fund (UNICEF), has been developed to support the implementation of the WHO Guidelines on mental health promotive and preventive interventions for adolescents. The Toolkit includes a core set of evidence-informed strategies, including the provision of support to parents and other caregivers to promote and protect adolescent mental health [27]. The key to supporting parents and other caregivers is to understand their care giving capacity and the challenges they face. The results of this study provide initial evidence for the reliability and validity of the SAS-CAN. The confirmatory factor analysis supported the hypothesized five-factor structure, aligning with the integrated theoretical framework derived from KAP theory and Farran’s caregiver skill model. However, the study’s cross-sectional nature and current scope of validation necessitate careful interpretation of these findings.

Caregivers’ response styles had a strong influence on the willingness of NSSI adolescents to seek help and clinical outcomes [28]. Parental response style to adolescent self-harm was related to their mental health [9]. As surveyed in a study, parental psychiatric treatment peaked during the three-month period after offspring self-harm [29]. And greater caregivers’ burnout was predictive of youth’s greater depressive and anxiety symptoms [30]. Thus, mental health of caregiver plays a vital role in supporting adolescents of NSSI. While traditional literature often conceptualizes health as an antecedent or moderator of supporting outcomes [9], we argue that in the context of chronic NSSI supporting, health is an integral component of ability itself. This dimension reflects the caregiver’s self-regulatory capacity, linking it to established psychological concepts such as emotional regulation, resilience, and caregiver burnout. For instance, caregivers with robust physical and mental health demonstrate better emotional regulation abilities and higher resilience, which directly sustain the supporting process. Conversely, parental burnout and poor mental health compromise the ability to provide effective support [30]. Therefore, we treat health not just as a precondition, but as a core functional aspect of the ‘supporting ability’ construct, as caregivers must sustain the physical and emotional demands of supporting to execute their supportive role effectively. Therefore, Caregiver health–encompassing physical, mental, and emotional well-being–is a fundamental personal factor that significantly influences a caregiver’s capacity to provide sustained, effective support to adolescents with NSSI [31].

In addition, the caregiver’s interpersonal relationships and ability to utilize resources are critical to the health of adolescents. Caregivers’ interpersonal relationships, including relationships with patients and people around them, particularly with healthcare providers, can significantly influence adolescent health outcomes [32]. Our previous research also indicated that caregivers lacked effective coping strategies and hoped to receive more external support in dealing with their child’s NSSI behaviors [8]. Therefore, in addition to knowledge, attitudes, and behaviors, we also included the caregiver’s health and interpersonal relationships and ability to utilize resources as factors in the assessment of care giving capacity.

Regarding the validity of the instrument, study findings demonstrated adequate content validity for the questionnaire. Thus, the items composing the instrument accurately and clearly reflect what the instrument is intended to measure and the domain it targets. The EFA indicated good structural validity of the questionnaire. The CFA demonstrated that the model fit indices met acceptable standards, confirming the stability of the questionnaire’s factor structure. Reliability and validity outcomes further supported the scientific rigor and robustness of the questionnaire. The EFA and CFA results supported the scale’s factorial validity. The second-order model yielded comparable fit indices to the five-factor model and smaller AIC/BIC values, indicating that the five dimensions share a common higher-order construct. This supports our theoretical framework, which conceptualizes “supporting ability” as a comprehensive construct encompassing knowledge, attitudes, practice, health, and interpersonal relationships. The scale also demonstrated good internal consistency. Furthermore, the SAS-CAN distinguishes itself within the broader landscape of caregiver assessment instruments. While established tools like the Zarit Burden Interview [33] primarily focus on the negative impacts of caregiving, the SAS-CAN adopts a strength-based perspective by evaluating the multidimensional “supporting ability.” Unlike instruments targeting health professionals (e.g., the Attitudes Scale [12]) or specific suicidal populations (e.g., the Suicidal Caring Ability Scale [14]), the SAS-CAN is specifically tailored to the unique needs of NSSI caregivers. By integrating dimensions of health and resource utilization, it extends beyond traditional Knowledge-Attitude-Practice models to capture the sustainability of care.

A key conceptual boundary of the SAS-CAN is its primary focus on the distal and intermediate contexts of supporting ability. While the Practice dimension includes items on immediate action protocols (e.g., removing tools, wound care), the scale does not fully capture the dynamic, dialectical process of real-time crisis responses—such as navigating an adolescent’s specific trauma triggers and acute relational dynamics. Notably, it is precisely this immediate response style that adolescents directly experience as supportive and that significantly influences their clinical outcomes and willingness to seek help [9, 28]. Thus, the SAS-CAN evaluates foundational readiness for support rather than moment-to-moment interactive competence.

Thus, the scale’s high level of reliability and validity is particularly important given the lack of standardized instruments available to assess caregivers’ ability to support adolescents with NSSI. The scale has the potential to be a useful tool for researchers and practitioners in mental health, enabling them to better understand the factors that may influence caregivers’ responses to NSSI in adolescents.

This study has several limitations. First, convenience sampling from a single tertiary hospital introduces notable selection bias: (1) The sample was predominantly female (78.65%, mostly mothers), underrepresenting fathers, grandparents, and other caregivers—potentially limiting applicability to non-maternal groups due to gender-related differences in supporting styles and NSSI knowledge; (2) 81.67% of participants were urban residents, excluding community-dwelling/outpatient adolescents with NSSI; urban caregivers’ greater access to mental health resources and higher health literacy may inflate scores and alter the factor structure; (3) Only adolescents with newly diagnosed NSSI were included, excluding those with chronic/recurrent NSSI (caregiver supporting abilities may differ here). Additionally, Chinese cultural norms (e.g., strong emphasis on parental responsibility, mental health stigma) may have influenced responses to items like “I am willing to care for self-harming children,” which may not translate to non-Chinese contexts with distinct supporting roles and stigma. Furthermore, certain items reflect specific legal and clinical contexts in China. For instance, Item K8 regarding “seeking police assistance” is grounded in the Mental Health Law of the People’s Republic of China, which authorizes guardians to seek police help when an individual poses a serious risk to themselves and refuses voluntary admission. While this is a standard safety protocol in China, it may raise ethical or legal ambiguities in other jurisdictions where police involvement in mental health crises is discouraged or differently regulated. Thus, cross-cultural validation of the scale’s factor structure and item relevance—including the adaptation of crisis intervention items to fit local legal frameworks—is required before non-Chinese use.

Second, the study did not assess criterion validity. The scale was not validated against established instruments (e.g., Zarit Burden Interview [33]) or clinical outcomes, so it remains unclear whether scores correlate with caregiver burden or adolescent recovery. Current use should be limited to descriptive assessment rather than predictive purposes.

Third, the health dimension relies solely on self-reported physical and mental health, without objective clinical screening or validation against standardized health scales. Third, regarding the H2 item used in the survey was a compound statement: “I have not been diagnosed with a mental illness, or I have been diagnosed with a mental illness but my current mental state is stable and I am able to take care of my child.” The sentence structure is relatively complex. Future versions of the scale should simplify this item to a direct functional description (e.g., “I feel mentally well enough to support my child”) to improve readability without losing content validity. And Future research should also incorporate objective health indicators to examine how actual health status influences supporting ability.

Fourth, while the Practice dimension includes items on immediate response actions (e.g., removing tools, wound care), the scale primarily assesses general behavioral strategies. It may not fully capture the dynamic, dialectical process of an immediate supportive response, which involves navigating an adolescent’s specific temperament, trauma triggers, and acute relational dynamics during a crisis. Future research could focus on developing specific modules to assess these nuanced, real-time interaction skills.

Finally, Item K4 relies on the current epidemiological consensus. As epidemiological trends may vary by region and period. Future revisions of the scale should consider regularly updating the knowledge items to align them with the latest evidence. It is recommended that Item H2 be revised to “I feel mentally well enough to support my child,” in future version as the original item was problematic due to its binary nature, potential for stigma, and the conflation of diagnostic status with functional capacity.

The scale provides a preliminary evaluation framework to identify caregiver supporting ability for NSSI adolescents. However, given the limitations of single-center convenience sampling and the lack of external validation, the scale’s clinical utility and effectiveness in evaluating interventions remain to be established. Future research should prioritize multi-center validation involving diverse demographic and clinical subgroups to test the scale’s generalizability and measurement invariance. Additionally, future studies should explore the scale’s test-retest reliability, responsiveness to change, and its utility in predicting adolescent outcomes.

Authors’ contributions

Conception and Design: Hui-Ping Li, Chun-Yan Sun; Writing the Manuscript: Chun-Yan Sun, Hui-Ping Li; Analysis and Interpretation: Chun-Yan Sun; Obtaining Funding: Hui-Ping Li, Chun-Yan Sun; Besides the authors mentioned above, all the other authors contributed to the article.

Funding

This study was supported by grants from Medical Scientific Research Foundation of Guangdong Province of China (No.A2024283, A2024012), the Party Building Research Project of Guangdong Provincial People’s Hospital (No.GDPHDJ202509) and Nursing Innovation Development Research Project of the Guangdong Nursing Association and the Lingnan Nightingale Nursing Research Institute (No.GDSHLXHYJYY202616).

Data availability

Data is provided within the manuscript or supplementary information files.

Declarations

Ethics approval and consent to participate

Research protocol approval was authorized by the Ethics Committees of Guangdong Provincial People's Hospital (No.KY2023-336-02), and all procedures were conducted in accordance with the principles outlined in the Helsinki Declaration. All participants voluntarily joined the study, signing an informed consent form.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.