Dialectical behavior therapy combined with parental support in adolescents with non-suicidal self-injury: a randomized controlled trial

Abstract

Background

Non-suicidal self-injury (NSSI) is highly prevalent among adolescents. Although dialectical behavior therapy for adolescents (DBT-A) is effective, its standard format can be lengthy (often > 19 weeks) and may insufficiently address deeper family dynamics, limiting accessibility and impact. This study evaluated the efficacy of a more time-contained DBT protocol, augmented with a structured parental support (PS) program, in reducing NSSI frequency and suicidal ideation (primary outcomes), as well as emotion dysregulation and depressive symptoms (secondary outcomes) in adolescents.

Methods

This study included 100 adolescents with depressive disorders who exhibited NSSI behavior and randomly assigned them to an intervention group (DBT + PS, n = 50) or a control group (DBT only, n = 50). The control group received 13 weeks of DBT (6 weeks face-to-face, 7 weeks online). The intervention group received the same DBT plus a 10-week parallel PS program. Outcomes were assessed using the Self-rating Idea of Suicide Scale, Adolescent Non-Suicidal Self-injury Assessment Questionnaire, Difficulties in Emotion Regulation Scale, Patient Health Questionnaire-9, and Montgomery–Asberg Depression Rating Scale at baseline, 6-week (mid-intervention), and 13-week (post-intervention). Data were analyzed using repeated-measures ANOVA.

Results

This study was completed by 91 participants (intervention: n = 44; control: n = 47). Significant group-by-time interactions were observed for all outcomes (p < 0.05). Compared to the Control group, the Intervention group exhibited significantly greater reductions in NSSI frequency at 6 weeks (p = 0.027) and in all measured outcomes at 13 weeks, including suicidal ideation, NSSI frequency and function, emotion dysregulation, and depressive symptoms (all p < 0.01).

Conclusions

Supplementing DBT with a structured PS program is more effective than DBT alone in improving clinical outcomes for adolescents with NSSI. Hospitals with relevant resources should consider implementing such combined interventions to improve patient recovery.

Clinical trial registry

Offline and online dialectical behavioral therapy combined with offline parental support in adolescents with non-suicidal self-injury: a randomized controlled trial (https://www.chictr.org.cn/) (ChiCTR2500107404).

Supplementary Information

The online version contains supplementary material available at 10.1186/s13034-026-01036-1.

Background

Non-suicidal self-injury (NSSI) is a common and severe mental health issue characterized by direct, deliberate, and recurrent self-harm without suicidal intent [1]. Its prevalence among adolescents is significantly high [2], with a reported rate of 25.2% among Chinese adolescents, frequently coexisting with psychological disorders such as depression and anxiety [3, 4].

Various intervention strategies for teenage NSSI are currently available, including cognitive and emotional interventions; however, their efficacy differs significantly [5, 6]. Dialectical behavior therapy (DBT), the most widely recognized intervention [5], is based on biosocial theory and dialectics, highlighting the balance between ‘acceptance’ and ‘change’ to enhance the psychological condition from a dialectical perspective [7]. It is significantly effective in diminishing suicidal behavior and NSSI as an evidence-based psychotherapy and is extensively utilized in treating patients with NSSI [8–10]. DBT for adolescents (DBT-A), adapted from DBT, is intended for adolescents diagnosed with borderline personality disorder. DBT-A, which targets internalizing symptoms [11], generally includes a combination of individual and family sessions designed to help adolescents make changes [12].However, DBT-A has some drawbacks. A systematic review by Fang et al. [13] demonstrated that DBT-A primarily focuses on emotional regulation and crisis management, frequently neglecting deeper family dynamics such as ineffective communication patterns. Moreover, as noted by Witt et al. [12], the extended duration of DBT-A—with most interventions lasting for 19 weeks or longer [14]—places significant time and financial burdens on families, limiting its accessibility.

The mental health of adolescents is closely associated with family dynamics [15]. Young people experiencing suicidal thoughts and behaviors, especially those engaging in NSSI, often encounter challenges, including ineffective parental protection, excessive parental control, and feelings of being misunderstood, unsupported, and unencouraged within their family relationships [16, 17]. Family therapy (FT) emphasizes the utilization of family resources and the relationships, roles, and communication patterns among family members to assist the family system in achieving a balance between acceptance and change [18]. Previous studies have demonstrated that the integration of FT can diminish self-injurious behaviors in adolescent patients [19], and it is recommended that families be involved at the earliest opportunity. Therefore, the parental support (PS) program focuses on transforming existing family interaction patterns and improving parents’ emotional management skills. This approach complements the principles of DBT, collaboratively guiding parents on how to optimally support patients.

The integrated model that combines online therapy with face-to-face treatment is increasingly utilized in chronic disease management. It has demonstrated favorable outcomes in managing conditions including depression, diabetes, and adolescent NSSI. This approach diminishes the stigma patients face regarding their disease, facilitates healthcare access, boosts treatment efficacy, and addresses disparities in healthcare delivery [20–22]. In a hybrid model combining online and face-to-face therapy, a robust therapeutic alliance is initially established through in-person sessions to gain the patient’s trust, enhancing the implementation of complex therapeutic techniques. Subsequently, online platforms improve treatment accessibility, decrease dropout rates, and provide continuity of care after hospital discharge [23].

Therefore, building on the principles of DBT-A but aiming to address its practical limitations (e.g., duration, cost, and depth of family involvement), we developed a more time-contained DBT protocol integrated with a structured PS program. This randomized controlled trial aimed to investigate whether this DBT + PS combination is more efficacious than DBT alone in diminishing NSSI behaviors, suicidal ideation, emotion dysregulation, and depressive symptoms in adolescents. We also sought to investigate the preliminary feasibility of delivering this combined intervention through a hybrid online-face-to-face model.

Methods

Study design and participants

This study employed G-Power 3.1.2 for sample size calculation. The sample size calculation established the effect size at 0.25, and the test efficacy (1-β) at 0.80, with two sample groups, three measurements, and a significance level (α) as a two-tailed test at 0.05. The requisite sample size for each group was calculated to be 44 cases, and the dropout of potential samples was also considered. Therefore, 100 participants were determined to be enrolled in the study.

We recruited 100 adolescent patients with NSSI, admitted to a tertiary psychiatric hospital in Henan Province, China, between March 2024 and December 2024. A statistician uninvolved in participant’s recruitment or assessment generated a randomization sequence using a computer-generated random number table, and participants were randomly assigned to a control group (50 individuals) or an intervention group (50 individuals). Three individuals in the control group were excluded due to disease recurrence, insufficient time to complete remaining training sessions, and inability to establish contact. Six participants from the intervention group were excluded owing to unavailability, unwillingness to continue training, time constraints, perceived ineffectiveness, or reluctance to have their faces recorded by the camera. Consequently, 44 participants in the intervention group received intervention and completed study. In the control group, 47 participants completed the study. Figure 1 depicts the flow diagram for this study.

Fig. 1.

Flowchart of the study

The inclusion criteria were as follows: (1) Patients who met the Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-V) diagnostic criteria for depression and had ≥ 1 NSSI during hospitalization; (2) patients aged 12 to 18 years old; (3) educated patients beyond primary school level who are capable of communicating normally and expressing feelings adequately; (4) patients proficient in using the Mental Rehabilitation Intelligent Cloud Platform system; (5) and patients whose guardian is the patient’s mother or father and possesses an education level of primary school or above and can use the cloud platform.

The exclusion criteria are as follows: (1) patients with Montgomery and Asberg Depression Rating Scale (MADRS) ≥ 35 points and assessed as severe risk; (2) patients or guardians with psychotic symptoms or those diagnosed with organic mental disorder or mental retardation; (3) patients with severe conditions, including neurological diseases, or severe substance abuse, and their guardians; (4) patients or guardians with formal education in psychology; (5) patients are currently participating in other group or individual therapy.

Introduction to the cloud platform utilized in this research

This platform was developed by a certain information technology company in China and is structured around a video conferencing framework. The study participants utilized this platform to engage in online therapy following their discharge from the hospital. To ensure patient treatment compliance and attendance rates, access to the platform was restricted until a manager inputted the patient’s national identity card number and mobile phone number into the system. Before participating in online treatment sessions, patients were required to book appointments through the platform, with successful bookings validated by SMS notification. If a patient fails to attend a session punctually, the staff will call to remind them to attend promptly, thus facilitating patient attendance. The online treatment sessions were administered in a group setting. To ensure the delivery of the online therapy according to the plan, the platform’s backend records patients’ login times in real time. Moreover, if the platform is unable to identify the patient’s head within the video frame (patient-side network disconnection or the patient genuinely not displaying their face on the screen), the screen will turn black. The therapist will notify the patient if their screen turns black, to ensure the delivery of the online therapy according to the plan. To ensure patients can operate the platform correctly, the research team has developed a user manual for the platform to assist patients in its proper use, thereby preventing any interruptions to online treatment.

Research team

The research team comprises one team leader, twelve rehabilitation therapists, six psychiatrists, two administrators, six psychiatric nurses, and six psychotherapists. All rehabilitation therapists hold DBT operational certification. All psychotherapists hold Chinese qualifications in psychotherapy and family therapy. The team leader supervises comprehensive coordination. Psychiatrists are responsible for subject screening, disease diagnosis, pharmacological treatment, and scale assessments. Psychiatric nurses perform patient risk evaluations and administer self-report questionnaires. Lead rehabilitation therapists facilitate DBT group skills training. Assistant rehabilitation therapists oversee classroom dynamics and motivate silent participants to engage actively. Administrators collect patient demographics, organize online therapy schedules, and manage human resources.

Control group

Patients were stratified into six groups, each comprising 8–10 participants. Patients engaged in face-to-face DBT from weeks 1 to 6, followed by online DBT from weeks 7 to 13. Under the supervision of the lead rehabilitation therapist, patients participated in DBT group skills training through role-playing, group discussions, assignment presentations, and several other methods. Each treatment session lasts between 50 and 60 min, with the final five minutes dedicated to participants sharing their learning experiences and providing feedback. During this concluding period, the therapist assesses participants’ immediate comprehension of the taught skills and their level of engagement through real-time feedback, whilst identifying any practical difficulties. Such feedback is subsequently utilised to make immediate adjustments to subsequent treatment segments, thereby enhancing participant adherence of the intervention.

This study developed a 13-week program (Table 1) based on the DBT group skills training manual for Chinese adolescents compiled by Wang Chun [24]. The manual comprises a therapist’s guide, a patient workbook, and a weekly scheduled homework booklet. It includes scripted dialogue examples, scenario case studies, and specific training content, organized into four modules: mindfulness, distress tolerance, emotion regulation, and interpersonal effectiveness.

Table 1.

Arrangement of the content of DBT group skills therapy for 13 weeks

Weeks	Content of training	Forms of treatment
1st	Introduction to Group Skills Training and Positive Thinking Skills Training (Wisdom Mind)	Group lectures
2nd	Chain analysis of problem behavior	Group lectures, individual sharing
3rd	Positive thinking: what it is and how to do it	Group lectures, simulated demonstration
4th	Pain tolerance: TIP skills	Group lectures, role-playing, group discussions
5th	Pain tolerance: distraction and self-soothing	Group lectures, role-playing, group discussions
6th	Pain tolerance: total acceptance	Group lectures, simulated demonstration
7th	Emotion regulation: recognizing emotions	Group lectures, Homework (Emotional Diary) Presentation
8th	Emotion regulation: contrary actions	Group lectures, Personal sharing
9th	Emotional regulation: positive thoughts about the emotions of the moment and accumulating positive emotions	Group lectures, simulated demonstration
10th	Emotion regulation: building self-control, pre-response, and PLEASE skills	Group lectures, simulated demonstration
11th	Interpersonal Effectiveness: goals and distractors	Group lectures, group discussions
12th	Interpersonal Effectiveness: As You Like It	Group lectures, role-playing
13th	Interpersonal effectiveness: maintaining relationships and self-esteem	Group lectures, Personal sharing

Intervention group

The intervention group commenced simultaneously with the control group. In addition to the treatment administered to the control group, the intervention group additionally provided PS to the patients’ guardians. PS is conducted simultaneously with DBT.

Based on the findings from the preliminary survey of parental preferences regarding the program, we developed the PS program, which incorporates family involvement during the patient’s hospitalization. This was subsequently enhanced through expert consultation employing the Delphi approach to establish the final version. The PS material comprises a 10-week course. Initially, 25 experts were invited (15 clinicians and 10 family guidance educators), and 22 experts ultimately completed all correspondence rounds (88.00% response rate), with a Kendall W = 0.68 (p < 0.01) and an expert authority coefficient Cr = 0.82. The training program was titled “Courses for Guardians,” comprising weekly in-person sessions lasting 50–60 min each. Table 2 presents specific topics and methods.

Table 2.

Arrangements for “Courses for Guardians” classes

Weeks	Thematic	Objectives	Main elements and methodology	Forms of treatment
1st	Building relationships of trust	Understanding the patient’s family situation and the purpose of self-injury	Building trusting relationships and ensuring that everyone is open to expressing feelings and experiences	Questionnaires, interviews, observations
2nd	Knowledge transfer	Explaining the psychological and physiological mechanisms of NSSI	Teaching about NSSI to help the patient’s carers understand the causes and effects of self-injury	Group lectures, discussions
3rd	Family communication training	Enhancing effective communication in the family	Instructing the patient’s carers in using communication techniques such as active listening and non-violent communication	Mock demonstrations, group lectures
4th	Emotional support training	Promoting emotional connections in the family	Using emotional expression and empathy training	Role-playing and simulation exercises
5th	Conflict resolution training	Avoiding being emotional	Learning compromise and co-operation techniques, such as the stop yelling trilogy	Group lectures, simulation exercises
6th	Establishment of family rules	Providing a harmonious and safe family environment	Jointly establishing family rules and ensuring consistent enforcement	Discussion, written contributions
7th	Family interaction	Learning how to display love	Learning the five languages of love, such as physical touch and exchanging gifts	Hands-on
8th	Crisis intervention	Providing timely support and interventions to overcome difficulties together	Developing a crisis response plan, such as background to the crisis, response plan, and how to call for help	Written material, simulation exercises
9th	Ongoing family support	Ensuring a continuous support system for families	Verbal-behavioral transmission, such as couples in the same frequency, classmates, empathy, and gay orientation	Group lectures, simulation exercises
10th	Maintenance and consolidation	Consolidation of treatment effects and prevention of recurrence	Developing a long-term plan with regular follow-up visits	Group presentations, questionnaires

Measures

The patient’s baseline information

A self-administered questionnaire was utilized to collect baseline information regarding the patients. The questionnaire included sex, age, education level, disease duration, and frequency of hospitalizations. Baseline data on guardians were obtained using a self-administered general guardianship questionnaire. The questionnaire encompassed sex, age, education level, occupation, and marital status.

The patient’s suicidal ideation

Suicidal ideation was assessed using the Self-rating Idea of Suicide Scale (SIOSS) [25]. This scale comprises 26 entries in four dimensions: despair, optimism, sleep, and masking. A total score of ≥ 12 on the first three signifies suicidal ideation, whereas a masking factor of ≥ 4 is considered unreliable. Each item was scored as ‘1’ for ‘yes’ with a ‘0’ for ‘no’. The total score of the scale is the sum of the despair factor, optimism factor, and sleep factor. Higher scores signify greater suicidal ideation. This scale demonstrates strong reliability and validity in China [26]. The test-retest reliability coefficient is 0.922, and its Cronbach’s α coefficient is 0.762. The five factors collectively explain 41.78% of the total variance.

The patient’s NSSI behavior and functioning

NSSI behaviors and their functions were assessed using the Adolescent Non-Suicidal Self-injury Assessment Questionnaire (ANSAQ) [27, 28]. The questionnaire comprised two parts: a behavioral questionnaire (featuring 12 entries covering 12 common forms of self-harm, including cutting, scratching, burning, hitting, deliberately preventing wound healing, and hair pulling) and a functional questionnaire (comprising 19 entries covering 19 motivations, including expressing anger, seeking revenge, and garnering attention). The questionnaire utilized a 5-point Likert scale, where scores 1–5 represented ‘none, occasional, sometimes, often, and always,’ indicating that higher scores reflect greater severity of self-injurious behavior. The Cronbach’s alpha coefficient of the behavioral questionnaire was 0.92, the split-half reliability was 0.85, the retest reliability was 0.84, and the cumulative variance contribution rate was 64.91%. The Cronbach alpha coefficient for the overall score of the function questionnaire was 0.90, split-half reliability was 0.78, retest reliability was 0.80, and the cumulative variance contribution rate was 53.87%. The questionnaire demonstrates strong reliability and validity and can be utilized as an effective tool to assess the non-suicidal self-injurious behaviors and their functions among adolescents in China.

The patient’s emotional regulation ability

Emotion regulation ability was assessed using the Difficulties in Emotion Regulation Scale (DERS) [29, 30]. This scale comprises 36 items categorized into six factors: difficulty accepting emotional reactions, challenges in goal orientation, difficulty controlling impulses, difficulty perceiving emotions, challenges using regulation strategies, and difficulty understanding emotions. Items 2, 6, 8, 10, 17, 20, 24, 34, and 36 require reverse scoring. The Cronbach’s α internal consistency coefficient for the total scale is 0.96. Confirmatory factor analysis demonstrates an ideal model fit: χ²/df = 1.05, CFI = 0.99, TLI = 0.99, RMSEA = 0.015. Exploratory factor analysis identified six factors, with each factor loading exceeding 0.62. The scale was scored on a 5-point Likert scale, with 1 being ‘rarely’ and 5 being ‘almost always.’ A higher score indicates greater difficulty in emotion regulation, and the lower the level of emotion regulation ability.

The severity of the patient’s depression

The severity of depressive symptoms was assessed using the Patient Health Questionnaire-9 (PHQ-9) [31, 32] and the Montgomery and Asberg Depression Rating Scale (MADRS) [3, 33]. The PHQ-9 is a self-report questionnaire, and it has nine entries designed for adolescent depression screening, assessing feelings in the past two weeks. Each entry is rated on a 4-point scale from 0 to 3. The total score is the sum of the scores from the 9 entries, ranging from 0 to 27, with elevated scores signifying more severe depressive symptoms. The single-factor model of this scale revealed a good fit (CFI = 0.953, TLI = 0.937, RMSEA = 0.077) and exhibited equivalent measurement properties among Chinese adolescents. The Cronbach’s α coefficient for the total scale was 0.85.

The MADRS is a scale of 10 items for others to evaluate, and it comprises 10 items. It employs a 7-point rating scale, ranging from 0 to 6 points, where higher scores signify more severe depression. The Cronbach’s α coefficient is 0.82, the test-retest reliability coefficient is 0.73, and the correlation coefficient is 0.8. It is appropriate for evaluating depressive symptoms in adolescents, accurately indicating variations in depressive symptoms and treatment efficacy.

Data collection

The team members informed the patients and guardians about the study’s objective, schedule, and material. Researchers collected patients’ self-reported measures (SIOSS, ANSAQ, DERS, and PHQ-9) and clinician-assessed MADRS scores at three time points: pre-intervention (T0), mid-intervention (week 6, T1), and post-intervention (week 13, T2). The patient self-assessment scale was established to complete the questionnaire within 48 h. The online platform was programmed to prevent the submission of questionnaires with any blank items. Therefore, any questionnaire with one or more missing items could not be submitted and was effectively excluded from the dataset.

Data analysis

The Statistical Package for the Social Sciences software (version 27) was utilized for data analysis. The count data are expressed as relative numbers. Comparisons between the two groups were conducted utilizing the χ² test. Measurement data conformed to normal distribution and are expressed as “x̄ ± s.” The paired t-test was employed to compare the two groups. Measurement information of patients’ scale scores at multiple time points in groups before and after the intervention was analyzed using repeated measures of analysis of variance (ANOVA). In case of a significant interaction or main effect, post-hoc pairwise comparisons with Bonferroni correction were performed to analyze the differences between specific time points (pre- versus mid-, mid- versus post-, and pre- versus post-intervention) within each group. For these within-group pairwise comparisons between time points, a Bonferroni correction was applied to adjust for multiple comparisons. Therefore, the significance level was set at p < 0.05/3 (p < 0.016). A single session was recorded as an absence if the patient’s screen remained black (indicating disconnection or non-participation) for over 15 min. Participants were classified as dropouts and withdrawn from the study if they accumulated ≥ 3 such absences.

Results

Sample characteristics

This study was completed by 91 adolescent patients, comprising 44 in the intervention group and 47 in the control group, resulting in an overall dropout rate of 9.0%. No statistically significant differences were observed between the two groups at baseline regarding demographic characteristics (sex, age, and years of education) and clinical features (duration of illness, number of hospitalizations, and guardian information) (p > 0.05), signifying that the groups were comparable at baseline. Tables 3 and 4 present the demographic characteristics of the patients.

Table 3.

Background variables for adolescent patients in each group

Variables	Intervention group (n = 44)	Control group (n = 47)	χ2/Z/t	P
Sex			0.0201	0.886
Male	7 (15.9%)	8 (17.0%)
Female	37 (84.1%)	39 (83.0%)
Age	15.54 ± 1.40	15.08 ± 1.62	1.4403	0.153
Years of education	8.63 ± 1.29	8.27 ± 1.34	1.2973	0.198
Duration of illness (years)	1.44 ± 0.82	1.62 ± 0.85	− 1.0383	0.302
Number of hospitalizations			− 1.1642	0.244
1	21 (47.7%)	25 (53.2%)
2	14 (31.8%)	20 (42.5%)
≥ 3	9 (20.5%)	2 (4.3%)

¹χ²; ² Z; ³t

Table 4.

Background variables for guardians in each group

Variables	Intervention group (n = 44)	Control group (n = 47)	χ2/Z/t	P
Sex			0.2001	0.655
Male	6(13.6%)	8(17.0%)
Female	38(86.45%)	39(83.0%)
Age	44.90 ± 5.94	45.46 ± 6.44	− 0.4133	0.681
Educational attainment			− 0.2102	0.833
Junior high school	17(38.6%)	19(40.4%)
High school	14(31.8%)	15(31.9%)
Undergraduate	11(25.0%)	11(23.4%)
Master’s degree or above	2(4.6%)	2(4.3%)
Careers			2.2261	0.694
Farmer	13(29.5%)	17(36.2%)
Worker	7(16.0%)	4(8.5%)
Profession	6(13.6%)	6(12.8%)
Out of work	12(27.3%)	16(34.0%)
Others	6(13.6%)	4(8.5%)
Marital status			0.6061	0.895
Married	31(70.5%)	30(63.8%)
Remarried	6(13.6%)	7(15.0%)
Divorced	4(9.1%)	5(10.6%)
Widow	3(6.8%)	5(10.6%)

¹χ²; ² Z; ³t

Primary clinical outcome measures: suicidal ideation and NSSI behaviors in patients

The results of the repeated measures ANOVA indicated an interaction effect between the intervention group and the control group in suicidal ideation (SIOSS) and non-suicidal self-injury Behavioral and Functional Questionnaires (ANSAQ) (p < 0.01). For the ANSAQ Behavioral Questionnaire, post-hoc pairwise comparisons revealed that both groups exhibited significant reductions from pre- to mid-intervention (both p < 0.001), from pre- to post-intervention (both p < 0.001), and mid- to post-intervention (Intervention group p = 0.002, Control group p = 0.047). Notably, independent samples t-tests at each time point revealed no significant difference between the groups at pre-intervention (p = 0.307); however, the intervention group exhibited significantly lower scores than the control group at mid-intervention (p = 0.027) and post-intervention (p = 0.001). This suggests that the superior effect of the combined intervention in reducing NSSI frequency was fully evident by mid-intervention (Table 5).

Table 5.

Scale scores, repeated measures analysis of variance, and independent samples t-test results

Scale	Group	Score (x̄ ± s)			Repeated measures ANOVA			Two-sample t-test
		Pre-intervention	Mid-intervention	Post-intervention	F time	F groups	F interact	Pre-intervention	Mid-intervention	Post-intervention
SIOSS	Intervention group	12.50 ± 2.41	9.04 ± 2.25	4.65 ± 1.46	257.078 p < 0.001	8.067 p = 0.006	5.773 p = 0.004	0.229 p = 0.819	0.452 p = 0.653	− 5.870 p < 0.001
	Control group	12.61 ± 2.45	9.25 ± 2.18	6.61 ± 1.70
ANSAQ Behavioral Questionnaires	Intervention group	22.79 ± 3.25	10.29 ± 5.11	6.75 ± 5.11	246.52 p < 0.001	8.714 p = 0.004	5.856 p = 0.003	1.028 p = 0.307	− 2.243 p = 0.027	− 3.281 p = 0.001
	Control group	22.02 ± 3.88	12.38 ± 3.69	10.42 ± 5.54
ANSAQ Functional Questionnaires	Intervention group	30.34 ± 10.91	14.65 ± 5.17	7.61 ± 1.10	178.662 p < 0.001	9.306 p = 0.003	1.25 p = 0.289	− 0.276 p = 0.783	− 2.272 p = 0.026	− 4.921 p < 0.001
	Control group	30.91 ± 8.86	18.29 ± 9.37	11.36 ± 4.93
DERS	Intervention group	94.40 ± 12.52	82.54 ± 7.69	76.45 ± 6.03	72.581 p < 0.001	1.711 p = 0.194	3.978 p = 0.022	0.251 p = 0.802	− 0.259 p = 0.796	− 3.510 p < 0.001
	Control group	93.78 ± 11.10	83.00 ± 8.94	81.25 ± 6.94
PHQ-9	Intervention group	15.59 ± 3.24	12.93 ± 2.49	9.02 ± 2.81	58.25 p < 0.001	20.854 p < 0.001	10.857 p < 0.001	− 0.128 p = 0.899	− 1.559 p = 0.123	− 6.719 p < 0.001
	Control group	15.68 ± 3.46	13.76 ± 2.60	12.91 ± 2.70
MADRS	Intervention group	29.38 ± 4.29	15.00 ± 2.96	11.29 ± 4.52	446.719 p < 0.001	4.607 p = 0.035	7.893 p < 0.001	1.345 P = 0.182	− 2.269 p = 0.026	− 3.774 p < 0.001
	Control group	28.08 ± 4.89	16.63 ± 3.83	14.40 ± 3.27

Secondary clinical outcome measures: patients’ emotional regulation capacity and depressive symptoms

Repeated assessments ANOVA revealed significant group-by-time interaction effects for emotional regulation difficulties (DERS) and depressive symptoms (PHQ-9 and MADRS scores) (all p < 0.05; see Table 5).

Emotion Regulation (DERS): Post-hoc pairwise comparisons indicated significant improvements in emotion regulation from pre- to mid-intervention and pre- to post-intervention in both groups (both p < 0.001). However, the course of change differed. No significant between-group difference was observed at the mid-intervention point (p = 0.796), indicating that the addition of PS had not yet conferred a notable benefit in emotion regulation at that stage. The post-intervention assessment revealed that the intervention group exhibited significantly lower DERS scores than the control group (p 0.001), highlighting the time needed for the combined intervention to fully impact this domain.

Depressive Symptoms (PHQ-9 and MADRS): For self-reported depressive symptoms (PHQ-9), post-hoc analyses revealed significant within-group reductions from pre- to mid-intervention (Intervention group p < 0.001, Control group p = 0.003), from pre- to post-intervention (both p < 0.001), but not from mid- to post-intervention (Intervention group p < 0.001, Control group p = 0.124). No significant difference was observed between the groups at mid-intervention (p = 0.123). However, by the end of the study, the intervention group reported a significantly greater reduction in depressive symptoms than the control group (p < 0.001). For clinician-rated depressive symptoms (MADRS), significant within-group improvements were observed from pre- to mid-intervention (both p < 0.001), from pre- to post-intervention (both p < 0.001), and from mid- to post-intervention (Intervention group p < 0.001, Control group p = 0.003). A significant between-group difference emerged earlier for this measure. The intervention group exhibited significantly lower MADRS scores than the control group at mid-intervention (p = 0.026) and post-intervention (p < 0.001), indicating that clinicians detected the benefits of the combined intervention on depressive symptomatology sooner than was self-reported by the adolescents (Table 5).

Discussion

This randomized controlled trial evaluated the efficacy of DBT combined with PS versus DBT alone for adolescents with NSSI. The combined intervention exhibited enhanced improvements in suicidal ideation, NSSI behaviors, emotional regulation, and depressive symptoms, which is consistent with the findings of reviews by Fang et al. [13] and Witt et al. [12] on the efficacy of the DBT-A intervention. The dose-effect relationship of PS component is a significant contributor; however, the varying time of impacts across outcomes necessitates additional consideration alongside other potential explanatory factors.

The earlier emergence of significant between-group differences in NSSI frequency (ANSAQ-Behavioral) at mid-intervention, compared to emotional regulation (DERS), which demonstrated significant differences only post-intervention, may reflect the distinct characteristics and temporal responsiveness of the targeted skills. The PS program’s early focus on crisis management, communication, and establishing family safety plans [8, 34] could have offered more immediate external scaffolding and environmental changes to mitigate self-injury episodes, even before adolescents had fully internalized and automated their emotion regulation skills through DBT. This aligns with models indicating that behavioral containment frequently precedes deeper cognitive-emotional transformation in therapeutic processes [35]. Conversely, improvements in emotion regulation, a more complex internal ability, might necessitate longer practice and consolidation of individual (DBT) and familial (PS) skills before differential effects manifest [36].

The significantly greater reduction in clinician-rated depressive symptoms (MADRS) in the combined intervention group at mid- and post-intervention, compared with the self-reported measure (PHQ-9), which demonstrated between-group differences only at post-intervention, is notable. Clinicians, attuned to observable changes in family interactions and adolescent engagement during sessions, might have identified improvements in socio-functional dimensions of depression (interpersonal engagement and affective expression) before any shifts in the adolescents’ subjective mood [17]. The PHQ-9, which emphasizes anhedonia and negative affect, may only detect changes after prolonged practice of DBT skills such as mindfulness and positive emotion accumulation, reinforced by a more supportive family environment [37].

The synergistic advantages of integrating PS with DBT are certainly beyond mere inclusion of parent training. The “Courses for Guardians” specifically addressed parental emotional response, communication habits, and family interaction rules. For instance, when parents learned conflict resolution techniques such as the “three-step approach to stop shouting,” this potentially diminished a common catalyst for adolescent emotional dysregulation and self-harm [38]. This intervention offered adolescents models of emotion regulation and lowered overall emotional arousal within the family system. Parental supportive responses are a known buffer for adolescent emotion dysregulation [36]. The transition from focusing solely on “individual skills” to modifying the “systemic environment” may be a crucial mechanism for the enhanced efficacy of the combined program [39].

The PS component likely contributed to reducing suicidal ideation and NSSI functions by establishing a familial safety net through enhanced parental crisis recognition and response, moving beyond panic and blame toward providing assistance [34, 35]. Moreover, by enhancing familial support and parent-child bonds through techniques such as teaching expressions of affection and active listening, PS may have mitigated feelings of isolation and worthlessness associated with depression [17, 40], thereby disrupting the depression-self-harm cycle.

Strengths and limitations

A notable strength of this study is the hybrid intervention delivery model, which integrated initial face-to-face sessions with continuing online therapy. This model probably facilitated the low dropout rate and high feasibility. The face-to-face phase was essential for establishing therapeutic rapport, conveying complex skill instructions, and fostering trust, whereas the online phase enhanced accessibility, ensured continuity of care post-discharge, and potentially reduced stigma [20–22, 41]. This methodology facilitates the establishment of effective “hospital-home” management systems for adolescent NSSI and offers a clear, actionable protocol for clinicians, consistent with recommendations for utilizing online components to sustain engagement [41, 42].

This study also has several limitations. The primary caregivers of the participants were primarily mothers, with lower paternal involvement, indicating societal division of labor within the Chinese cultural context [43]. The sample was derived from a single center and did not account for cultural and economic variations across regions, potentially affecting the generalizability of findings. The exclusion of patients with major depressive disorder and other severe comorbidities might have impacted the generalizability of results. Online therapy relies on patients’ active participation; some individuals were unable to fully complete the intervention due to limitations in their local internet access, thereby affecting treatment efficacy. This study evaluated outcomes solely on weeks 6 and 13, precluding assessment of the intervention’s long-term efficacy. The measurement tools employed were restricted to questionnaires, lacking objective physiological indicators.

Future studies should further investigate training programs tailored to the paternal role, including developing courses that address fathers’ interests and communication characteristics. Conducting multicenter, large-scale studies integrating process-based indicators, including family functioning and parent-child interactions, mechanism variables and long-term follow-up, and adding physiological indicators, will enable more precise evaluation of intervention outcomes and validate their effectiveness across diverse populations.

Conclusions

Combining DBT with a structured PS program was more effective than DBT alone for adolescents with NSSI. The combined intervention resulted in significant decreases in suicidal ideation, NSSI behaviors, and depressive symptoms, and enhanced emotional regulation. These findings support the implementation of such combined group training programs in clinical settings with appropriate resources to enhance recovery for adolescents with NSSI.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

NSSI

Non-suicidal self-injury

DBT

Dialectical behavior therapy

PS

Parental support

FT

Family therapy

ANSAQ

Adolescent Non-Suicidal Self-Injury Assessment Questionnaire

DERS

Difficulties in Emotion Regulation Scale

PHQ-9

Patient Health Questionnaire-9

MADRS

Montgomery and Asberg Depression Rating Scale

ANOVA

Analysis of Variance

Author contributions

Yanping Zhang was responsible for designing the research plan and drafting the initial version of the paper; Yue Zhou was responsible for revising the paper; Bingyi Du and Jixuan Hou were responsible for creating the tables; Running Hou and Binyuan Zhang were responsible for implementing the research plan; Shouxun He was responsible for data analysis; Chuansheng Wang and Fang Yan were responsible for the feasibility analysis, quality control, and data review of the paper.

Funding

This study was supported by the Henan Province science and technology research and development plan joint fund (industry) project (235101610004), the Henan Provincial Key Scientific and Technological Research Project (242102310286), the Open topic of Collaborative Innovation Center for Prevention and Treatment of mental illness in China Henan Province (XTkf12), the open project of Henan Province psychology in the second Affiliated Hospital of Xinxiang Medical University (XYEFYJSSJ-2023-01), and the Henan Provincial Medical Science and Technology Research Programme (RKX202502028).

Data availability

All data used in this study are publicly available as an appendix.

Declarations

Ethics approval and consent to participate

This study passed the ethical approval of the Ethics Committee of the Second Affiliated Hospital of Xinxiang Medical University (XYEFYLL-(Research)-2022-48), and the study followed the principle of “informed consent” to ensure that participants were informed of the purpose of the study. All patients participating in the study received written informed consent. In addition, participants were informed that their ability to opt out of the study was protected and guaranteed confidentiality, and that their information would not be shared elsewhere. The Intelligent Cloud Platform System for Psychiatric Rehabilitation used in the study has been applied for a software copyright in China (No. 14906983).

Competing interests

The authors declare no competing interests.