Narrative Drawing Intervention for Adolescents Following Earthquake Exposure in Rural Western China: A Quasi-Experimental Study

Abstract

Background: Adolescents are particularly vulnerable to psychological distress following natural disasters, especially in low-resource settings. This study examined the short-term psychosocial outcomes associated with Narrative Drawing Intervention (NDI), a structured, trauma-informed, school-based group counselling program integrating expressive drawing and guided narrative reflection, among students affected by an earthquake in rural western China. Methods: Using a quasi-experimental design, 30 trained educators facilitated eight NDI group sessions for 150 students. Of the 120 students who completed the intervention, a randomly selected subset completed standardized psychological assessments. The final analyzed sample included 64 participants (44 intervention; 20 control). Results: The intervention group demonstrated significant reductions in anxiety (p = 0.011, d = 0.40) and PTSD symptoms (p = 0.008, d = 0.42), with a reduction in stress approaching statistical significance (p = 0.063, d = 0.29). In contrast, the control group showed significant increases in anxiety, stress, and PTSD symptoms over the same period. Depressive symptoms did not significantly change in either group. Descriptive drawing comparisons indicated increased visual elaboration and more centralized figure placement following the intervention. Conclusions: Within the context of a quasi-experimental and exploratory design, the findings provide preliminary support for the feasibility of NDI and suggest potential short-term psychosocial benefits in post-disaster school settings. While baseline group differences and the lack of randomization suggest the need for further investigation, the results provide a foundation for future randomized and longitudinal studies that further examine causal pathways and the sustainability of observed effects.

1. Introduction

A recent magnitude 6.8 earthquake in a rural region of China resulted in casualties, infrastructural damage, and widespread displacement. The disruption of daily life, loss of loved ones, and uncertainty about the future are likely to exacerbate psychological strain. This psychological impact on vulnerable populations, including children and adolescents, is of significant concern. Broader research on seismic events provides insights into potential mental health challenges faced by adolescents, such as post-traumatic stress disorder (PTSD), depression, and anxiety, with some studies highlighting gender differences and the effects of recurrent disasters.

1.1. Effects of Earthquakes on Adolescents: A Global Perspective

Natural disasters, particularly earthquakes, pose profound and multifaceted challenges to human well-being, not only causing immediate physical harm and socio-economic disruption but also contributing to deep and lasting psychological trauma. The psychological effects of such events may persist for years and, in some cases, produce long-term impacts on survivors. Adolescents, positioned at a critical juncture of biopsychosocial development, are among the most vulnerable groups following exposure to such catastrophic events. They may face substantial risks of developing a range of mental health disorders, including PTSD, depression, anxiety, and complicated grief (Şam et al. 2025; Tanaka et al. 2016; Yılancıoğlu and Özbaran 2023).

The impacts of earthquakes on adolescent mental health have been studied worldwide, with particular focus on the Chinese context due to its history of significant seismic activity. The 2008 Wenchuan earthquake stands as a pivotal event that generated extensive research on child and adolescent trauma. Following this devastating event, Zhang et al. (2012) conducted a pioneering longitudinal study of 548 adolescents and found PTSD prevalence rates of 9.7% six months after the event. This prevalence notably declined to approximately 1.3% and 1.6% at 12 and 18 months, respectively. This trend revealed both the remarkable resilience of many young people and the persistent risk for others over time. Key predictors of persistent PTSD symptoms included female gender, prior trauma exposure, and comorbid depression (Zhang et al. 2012). The finding that depression often co-occurs with PTSD suggests a shared etiology and highlights the need for interventions that address complex symptomatology.

Another longitudinal study conducted after the Wenchuan earthquake found that adolescents may experience chronic, delayed, or relapsing trauma symptoms (Fan et al. 2015). Cheng et al. (2018) further demonstrated that PTSD symptoms strongly predicted depressive symptoms following the Wenchuan earthquake, illustrating the intertwined and often sequential nature of trauma-related disorders.

Adolescents may be particularly vulnerable to traumatic responses following disaster exposure due to still-developing emotional regulation capacities (Şam et al. 2025). Brain regions responsible for emotional regulation, such as the prefrontal cortex, are still developing, while the amygdala—responsible for threat detection and fear responses—is highly active. This neurobiological imbalance can lead to intense emotional reactions and difficulty modulating negative emotions (Schwarz and Perry 1994). Emotional dysregulation may negatively impact family dynamics, social connectedness, and education (Pate et al. 2017). Trauma may also hinder identity formation, leading to long-term effects on autonomy and peer relationships (Waterman 2020).

Gender differences consistently emerge in earthquake-related mental health outcomes. Female adolescents are more greatly affected by PTSD and depression compared to male adolescents (Fan et al. 2015; Wang et al. 2025; Zhang et al. 2012). This disparity may reflect biological influences on stress response systems as well as socialization processes that favor internalization of stress among females (Hodes and Epperson 2019). Such findings underscore the importance of gender-sensitive approaches in post-disaster mental health interventions.

Moreover, cumulative trauma is particularly relevant in regions prone to recurrent disasters. Ni et al. (2022) found a dose–response relationship between disaster exposure and psychological symptoms among adolescents. Xu et al. (2018) reported PTSD rates peaking at 15.9% among adolescents exposed to successive earthquakes, underscoring compounding effects. Anniversary reactions may further reactivate trauma symptoms years later (Leaune et al. 2025).

1.2. Art-Based Interventions in Post-Disaster Psychology

Expressive arts-based interventions have gained empirical support in post-disaster trauma recovery, particularly for children and adolescents for whom verbal processing may be limited (Malchiodi 2020). A systematic review and meta-analysis of 15 studies found robust reductions in PTSD symptoms following creative arts interventions (Morison et al. 2022).

Art therapy has shown effectiveness across cultural contexts. Jumilia and Kanathasan (2025) reported significant reductions in trauma symptoms among children in Indonesia. In a randomized inpatient study, Lyshak-Stelzer et al. (2007) found that adolescents receiving trauma-focused art therapy demonstrated greater reductions in PTSD symptoms compared with treatment-as-usual. Versitano et al. (2025) found enduring effects of art therapy up to six months post-treatment.

Art therapy may also foster social development and emotional regulation (Vogel et al. 2025; van der Kolk 2014). These findings provide theoretical grounding for Narrative Drawing Intervention (NDI), which in the present study is examined as a structured, trauma-informed group counselling model embedded within a school context.

1.3. Narrative Drawing Intervention (NDI): Theory and Mechanism

NDI integrates narrative therapy and art therapy into a manualized framework designed to facilitate trauma processing. The intervention emphasizes three pathways—seeing the self, accepting the self, and empowering the self—and three dialogues: spoken narratives, the drawing itself, and body language.

NDI has been applied in clinical contexts by multidisciplinary helping professionals, including psychologists, social workers, and counselors. Early descriptive accounts indicate that practitioners have found the structured integration of drawing tasks and facilitated narrative dialogue useful in supporting therapeutic engagement and emotional exploration (Wong 2015). However, systematic empirical evaluation of its effectiveness—particularly within school-based, post-disaster settings—remains limited. The present study therefore examines NDI specifically as a structured psychosocial group counselling intervention delivered in an educational context, focusing on short-term emotional and trauma-related outcomes rather than broad clinical or preventive claims.

The training framework includes a “train-the-trainers” model to ensure scalability and cultural adaptability.

1.4. NDI in the Chinese Cultural and Educational Context

Art therapy research in China is developing but remains limited (Cui and Wang 2022). Rural areas often lack trained mental health professionals (Edwards et al. 2023). Stigma remains a barrier to mental health care (Zhang et al. 2025). In such contexts, school-based, non-stigmatizing interventions are critical.

NDI’s adaptability allows application in group counselling settings. Training teachers as seed facilitators may contribute to sustainable mental health capacity building (Ho et al. 2014).

1.5. Identification of Research Gap and Theoretical Positioning

Despite growing global evidence supporting expressive arts interventions in post-disaster contexts, rigorous evaluation of Narrative Drawing Intervention (NDI) among mainland Chinese adolescents remains limited. Furthermore, while NDI has been described in therapeutic and empowerment-oriented terms, its theoretical positioning and hypothesized mechanism of change have not been systematically articulated within a school-based psychosocial research framework.

To clarify the conceptual grounding of the present study and delineate interpretive boundaries, the theoretical foundation and mechanism underlying NDI are outlined below. This framework guides the evaluation of short-term psychosocial outcomes within the current quasi-experimental design.

2. Conceptual Framework and Theoretical Positioning of Narrative Drawing Intervention

Adolescence is a developmental period marked by rapid neurobiological maturation, identity consolidation, and increasing autobiographical narrative integration. Exposure to natural disasters during this sensitive developmental phase may disrupt emotional regulation capacities and fragment the coherence of self-narrative. Such disruptions have been associated with heightened vulnerability to post-traumatic stress symptoms and internalizing distress. Within this context, structured psychosocial interventions that support emotional articulation and narrative integration may be developmentally appropriate and culturally adaptable forms of post-disaster support.

2.1. Conceptual Positioning of NDI

Narrative Drawing Intervention (NDI) was originally developed within a Chinese clinical context as a structured expressive therapeutic model. In the present study, however, NDI is examined specifically as a school-based, trauma-informed psychosocial group counselling intervention, rather than as a formal clinical psychotherapy or a comprehensive public health prevention framework.

Within this framing, NDI operates through structured peer engagement under facilitator guidance. While the model retains its therapeutic origins, its implementation in the current study emphasizes group-based psychosocial support delivered in educational settings.

The intervention is grounded in the premise that individuals are primary agents in understanding and resolving their own difficulties. Rather than positioning the facilitator as an interpretive authority, NDI emphasizes accompaniment, containment, and structured scaffolding of peer dialogue within a psychologically safe group environment. This epistemological stance aligns with empowerment-oriented and narrative-based approaches that conceptualize participants as experts in their lived experience.

Thus, this study evaluates NDI as a moderate-intensity, school-embedded group counselling model targeting short-term psychosocial outcomes.

2.2. Developmental Pathways: Seeing, Accepting, and Empowering the Self

At the core of NDI is a three-stage developmental pathway through which participants are guided:

• Seeing oneself—facilitating self-recognition and awareness of emotional experience;
• Accepting oneself—supporting acknowledgment of distress without avoidance or self-rejection;
• Empowering oneself—encouraging the mobilization of strengths and adaptive agency.

This progression reflects an increasing movement from emotional recognition to integration and adaptive engagement. In disaster-affected adolescents, whose traumatic experiences may remain fragmented or unarticulated, such structured progression aims to support the reconstruction of a coherent and agentic self-narrative.

2.3. Mechanism of Change: From Implicit Distress to Narrative Transformation

NDI integrates expressive drawing and guided narrative articulation within a trauma-informed group environment. The hypothesized mechanism of change unfolds through three interrelated processes:

(1)

Making implicit experience explicit

Traumatic experiences may be stored in sensory, affective, or non-verbal forms. Structured drawing provides a symbolic medium through which diffuse or pre-verbal distress can become consciously represented. This process may facilitate the transition from diffuse emotional experience to articulated awareness.

(2)

Externalization

Through visual representation and narrative sharing, distressing experiences are symbolically externalized. Externalization reduces psychological fusion with traumatic material, allowing participants to view their experiences as events that have occurred rather than as defining features of the self. This distancing process is consistent with narrative therapeutic principles.

(3)

Transformation through re-authoring within a peer context

Within the safety of the structured group environment, participants are invited to reflect upon their drawings and narratives in the presence of peers. Through facilitated dialogue, reciprocal witnessing, and shared meaning-making, individuals may revise or expand the meanings attributed to their experiences. Observing peers articulate distress, strengths, and adaptive reinterpretations may contribute to normalization of emotional responses and vicarious learning. This reflective process supports the transformation of a trauma-saturated narrative into a more integrated and strength-oriented account.

Importantly, this transformation is conceptualized as relational and process-based rather than interpretively imposed. The facilitator’s role is to scaffold peer interaction and maintain emotional safety rather than assign symbolic meaning. The present study does not directly measure these proposed mechanisms; however, the processes described above offer a thoughtful conceptual framework to guide interpretation of the findings. These hypotheses provide a valuable theoretical lens for understanding potential pathways underlying the observed changes. Future research can build on this foundation by explicitly examining these mechanisms to further clarify and strengthen the evidence base. These mechanisms remain theoretical propositions and were not empirically tested within the present design.

2.4. Mobilization of Psychosocial Resources: The “Three Treasures”

A central component of NDI is the identification and activation of three categories of psychosocial resources:

• Internal strengths (e.g., courage, perseverance, personal qualities).
• External resources (e.g., family, peers, teachers, community support).
• Hope (future-oriented aspiration and perceived possibility).

These “three treasures” function as organizing anchors within the narrative reconstruction process. Rather than focusing exclusively on symptom reduction, the intervention encourages participants to situate distress within a broader landscape of resilience and relational support.

In structured group settings, witnessing peers identify internal strengths and relational supports may further enhance normalization and social cohesion. The articulation of hope within a shared context may also strengthen collective resilience by situating individual recovery within a broader relational framework.

2.5. Integration Within a Trauma-Informed School Context

NDI operates within a trauma-informed framework emphasizing safety, voluntary participation, emotional containment, and collaborative reflection. When implemented in school settings following disaster exposure, the intervention functions as a structured psychosocial support mechanism rather than as individualized psychotherapy.

Within this model, the expected proximal outcomes are:

• Enhanced emotional articulation.
• Increased narrative coherence.
• Strengthened perceived agency.
• Short-term reduction in self-reported distress symptoms.

Importantly, these proximal outcomes may differentially affect symptom domains characterized by heightened emotional arousal and stress reactivity. Anxiety-, stress-, and trauma-related symptoms are closely linked to dysregulated autonomic and affective activation, processes that may be more immediately responsive to structured expressive and regulatory interventions. In contrast, depressive symptoms in adolescence often involve more persistent cognitive patterns such as hopelessness, negative self-appraisal, and motivational withdrawal, which may require longer-term or cognitively intensive approaches to demonstrate measurable change. Therefore, the present study places particular emphasis on anxiety-, stress-, and PTSD-related outcomes while examining depressive symptoms as a secondary indicator of broader mood shifts.

The present study evaluates these proximal psychosocial indicators within a quasi-experimental design, with particular attention to short-term changes in anxiety-, stress-, and trauma-related symptoms. The investigation does not assess long-term developmental transformation, preventive efficacy, or macro-level policy impact. Accordingly, interpretations are limited to short-term psychosocial outcomes within the specific post-disaster educational context examined.

2.6. Scope of Theoretical Claims

Although NDI has been described in broader terms as empowering or preventive, the current investigation limits its claims to short-term symptom change and exploratory expressive indicators within a specific post-disaster educational context. Broader implications for identity development, community resilience, or policy integration are acknowledged as theoretical extensions that require longitudinal and methodologically rigorous investigation.

The present findings should therefore be interpreted as outcome-level observations that do not adjudicate among competing theoretical mechanisms.

3. Materials and Methods

3.1. Study Design

This quasi-experimental study examined the effectiveness of Narrative Drawing Intervention (NDI), implemented as a structured school-based group counselling intervention, in improving negative emotional states and trauma-related symptoms among adolescents affected by a magnitude 6.8 earthquake in rural western China in September 2022. A non-equivalent control group design was employed due to practical constraints within the school setting. Pre- and post-intervention assessments were conducted between September and November 2023. Students in the control group continued with standard school programming and did not receive any structured psychosocial or arts-based intervention during the study period. No additional services were introduced specifically for research purposes.

3.2. Participants and Setting

Participants were recruited from a rural middle school located in an ethnically diverse region of western China that had been affected by a recent earthquake.

A total of 150 students were recruited across participating classes. Of these, 120 students completed the intervention and provided post-intervention responses. Of the 120 students who completed the intervention, only a randomly selected subset completed standardized psychological assessments. Due to logistical and resource constraints within the rural school setting, it was not feasible to administer standardized questionnaires to all participants. Therefore, 45 students from the intervention group were selected using a simple random sampling procedure based on the full roster of intervention completers. Selection was conducted prior to data analysis and was independent of students’ symptom levels, session attendance patterns, or facilitator recommendations. All eligible participants had an equal probability of inclusion. One case was excluded due to incomplete data, resulting in 44 valid intervention participants for standardized outcome analysis. In the control group, 20 students completed both baseline and follow-up assessments and were included in the analysis.

Thus, the final analyzed sample comprised 64 participants (44 intervention; 20 control). To evaluate the representativeness of the assessed subgroup, available demographic characteristics (age and gender distribution) were descriptively compared between the assessed intervention participants (n = 44) and the broader intervention cohort (n = 120). The comparison indicated similar age ranges and gender proportions, suggesting demographic comparability of the groups. Although standardized psychological measures were administered only to the selected subgroup—precluding direct confirmation of baseline symptom equivalence—this analysis provides partial support for representativeness. Future studies incorporating comprehensive baseline assessments across all participants would further strengthen confidence in subgroup generalizability.

Demographic information, including age and gender, was collected. Baseline analyses indicated that female students reported higher levels of psychological distress across several measures.

3.3. Facilitator Training and Implementation Model

Thirty trained educators delivered eight NDI group sessions. Fifteen educators were middle school teachers from the participating school, and fifteen were recruited from a nearby local college. All facilitators received standardized training in the Narrative Drawing Intervention protocol prior to implementation. The training followed a train-the-trainers framework designed to support sustainability within low-resource educational contexts. Facilitators received guidance in trauma-informed group management, structured drawing activities, and facilitation of peer dialogue while maintaining psychological safety.

The intervention was delivered within regular school hours in designated classroom spaces. This embedded implementation approach was intended to enhance accessibility and reduce stigma associated with mental health services.

3.4. Intervention Procedures

Students in the intervention condition participated in eight structured small-group counselling sessions conducted from September to November 2023. Each session integrated guided expressive drawing activities with facilitated peer reflection and narrative dialogue.

Sessions followed a consistent structure:

• Warm-up activity to promote engagement and psychological safety.
• Thematic drawing task focused on emotional expression or experience processing.
• Structured peer sharing and facilitator-guided reflection.
• Closing activity emphasizing internal strengths, external supports, and future-oriented hope.

The intervention emphasized participant agency, emotional containment, and peer witnessing rather than facilitator-led interpretation. Facilitators were instructed to scaffold discussion and maintain supportive group norms rather than assign symbolic meanings to drawings.

3.5. Measures

3.5.1. Depression Anxiety Stress Scales (DASS-21)

The DASS-21 was used to assess levels of depression, anxiety, and stress. The instrument was developed by Lovibond and Lovibond (University of New South Wales, Sydney, NSW, Australia). The instrument consists of 21 self-report items rated on a four-point Likert scale. Subscale scores were calculated for each domain. The DASS-21 has been widely used in adolescent populations and demonstrates acceptable psychometric properties.

3.5.2. PTSD Checklist for DSM-5 (PCL-5)

The PCL-5 was developed by the National Center for PTSD (U.S. Department of Veterans Affairs, Washington, DC, USA) and administered to assess trauma-related symptoms consistent with DSM-5 criteria for PTSD. Participants rated the severity of symptoms experienced in the past month. Total scores were calculated according to established scoring guidelines. Although originally developed for adult populations, the PCL-5 has been used in adolescent samples in prior post-disaster research in China and internationally. Nonetheless, continued evaluation of its developmental sensitivity and linguistic appropriateness for younger adolescents remains important when interpreting findings.

3.5.3. Structured Self-Portrait Drawing Analysis

Participants were asked to complete a self-portrait drawing before and after the intervention. Drawings were examined descriptively for changes in size, placement, color use, line quality, facial expression, and contextual elements. No symbolic interpretation framework or projective analytic model was applied; observations were restricted to surface-level visual features. The drawing analysis was exploratory and intended to complement standardized self-report measures by capturing expressive aspects of self-representation. No formal projective scoring system was applied.

3.6. Data Analysis

Statistical analyses were conducted using IBM SPSS Statistics Version 31.0.2 (IBM Corp., Armonk, NY, USA). (N = 64).

Paired-samples t-tests were performed to examine within-group changes from pre- to post-intervention for the experimental and control groups separately. Independent-samples comparisons were used descriptively to examine baseline differences between groups.

Given the non-randomized design, unequal group sizes, and baseline differences, the analyses were tailored to best suit the structure of the data. Rather than conducting formal interaction analyses (e.g., repeated-measures ANOVA or mixed-effects modeling), which would have required stronger assumptions and greater statistical power, the study focused on descriptive comparisons of pre–post changes. The relatively small and non-equivalent control group guided this decision, ensuring that results were interpreted within a stable and methodologically appropriate framework. Accordingly, between-group differences in change over time were evaluated descriptively, consistent with the exploratory aims of the investigation and with an emphasis on careful and transparent interpretation.

Statistical significance was set at p < 0.05. Effect sizes were calculated where appropriate to provide additional context for interpreting observed changes.

In light of the practical realities of field-based data collection in a rural school setting, data entry focused on subscale and total scores to ensure feasibility and efficiency, and item-level responses were not archived. As a result, internal consistency coefficients (e.g., Cronbach’s α) could not be recalculated for the present sample. However, prior validation research in Chinese populations has demonstrated strong internal consistency for the DASS-21 (α = 0.83–0.92; Wang et al. 2016) and the PCL-5 (α = 0.86–0.94; Cheng et al. 2020), supporting the reliability of these instruments. While the absence of sample-specific reliability estimates modestly limits psychometric precision, future studies can build on this work by retaining item-level data to enable reliability verification and more advanced measurement modeling.

4. Results

Baseline comparisons indicated that the intervention group reported higher symptom levels on certain measures relative to the control group at pre-test (

Table 1. Pre- and Post-intervention Means, Standard Deviations, and Within-Group Comparisons.

Measure	Group	Pre M (SD)	Post M (SD)	t (df)	p	Cohen’s d
Anxiety	Intervention (n = 44)	6.68 (5.26)	4.84 (4.99)	−2.65 (43)	0.011	0.40
	Control (n = 20)	4.35 (3.25)	5.70 (3.88)	2.63 (19)	0.016	0.59
Stress	Intervention	7.11 (5.18)	5.80 (4.83)	−1.91 (43)	0.063	0.29
	Control	4.55 (3.23)	4.90 (4.31)	2.43 (19)	0.025	0.54
Depression	Intervention	6.95 (6.44)	6.05 (5.76)	−1.34 (43)	0.188	0.20
	Control	3.40 (3.94)	4.00 (4.63)	1.33 (19)	0.199	0.30
PTSD (PCL-5)	Intervention	25.77 (17.64)	19.75 (18.92)	−2.77 (43)	0.008	0.42
	Control	12.25 (8.92)	16.00 (17.11)	2.73 (19)	0.013	0.61

Note: Cohen’s d reflects within-group effect size based on pre–post difference scores.

). Such differences are consistent with the non-randomized, institutionally assigned design and limit the strength of between-group causal inferences. Independent-samples t-tests were conducted to examine baseline differences.

At baseline, the intervention group demonstrated significantly higher anxiety scores than the control group, t(62) = 2.32, p = 0.023, and significantly higher PTSD scores, t(62) = 3.66, p < 0.001. Baseline differences in stress and depression were not statistically significant.

Given the non-randomized design and baseline group differences, analyses focused primarily on within-group pre–post changes.

Given that the intervention group entered the study with significantly higher baseline anxiety and PTSD symptom levels, regression to the mean cannot be ruled out as a potential partial contributor to observed symptom reductions. Participants with elevated initial distress scores may show some natural decline over time independent of intervention exposure. Accordingly, symptom change patterns should be interpreted cautiously and within the exploratory scope of the present quasi-experimental design.

4.1. DASS-21 Outcomes

4.1.1. Anxiety

In the intervention group, anxiety scores decreased significantly from pre-test (M = 6.68, SD = 5.26) to post-test (M = 4.84, SD = 4.99), t(43) = −2.65, p = 0.011, d = 0.40, indicating a small-to-moderate effect.

In contrast, the control group showed a significant increase in anxiety from pre-test (M = 4.35, SD = 3.25) to post-test (M = 5.70, SD = 3.88), t(19) = 2.63, p = 0.016, d = 0.59.

These findings indicate reduced anxiety in the intervention group, whereas anxiety increased in the control group over the same period.

4.1.2. Stress

For the intervention group, stress scores showed a non-significant decrease from pre-test (M = 7.11, SD = 5.18) to post-test (M = 5.80, SD = 4.83), t(43) = −1.91, p = 0.063, d = 0.29.

The control group demonstrated a significant increase in stress from pre-test (M = 4.55, SD = 3.23) to post-test (M = 4.90, SD = 4.31), t(19) = 2.43, p = 0.025, d = 0.54.

4.1.3. Depression

Depression scores in the intervention group showed no significant change from pre-test (M = 6.95, SD = 6.44) to post-test (M = 6.05, SD = 5.76), t(43) = −1.34, p = 0.188, d = 0.20.

Similarly, the control group showed no significant change in depression scores from pre-test (M = 3.40, SD = 3.94) to post-test (M = 4.00, SD = 4.63), t(19) = 1.33, p = 0.199, d = 0.30.

These findings indicate that depressive symptoms did not demonstrate significant short-term change in either group.

4.2. PTSD Symptoms (PCL-5)

In the intervention group, PTSD scores decreased significantly from pre-test (M = 25.77, SD = 17.64) to post-test (M = 19.75, SD = 18.92), t(43) = −2.77, p = 0.008, d = 0.42.

In contrast, the control group showed a significant increase in PTSD symptoms from pre-test (M = 12.25, SD = 8.92) to post-test (M = 16.00, SD = 17.11), t(19) = 2.73, p = 0.013, d = 0.61.

It is possible that structured emotional expression within a psychologically contained group environment may have played a role in the observed reductions in self-reported anxiety and trauma-related symptoms.

Figure 1 presents mean pre- and post-intervention scores for both groups across outcome measures.

4.3. Gender Differences in Psychological Outcomes

Exploratory paired-samples analyses were conducted within the intervention group to examine gender-specific symptom changes. Among female participants (n = 28), anxiety decreased significantly from pre-test (M = 8.18, SD = 5.85) to post-test (M = 6.04, SD = 5.67), t(27) = −2.64, p = 0.013, d = 0.50. Reductions in stress, t(27) = −1.74, p = 0.093, d = 0.33; depression, t(27) = −1.54, p = 0.136, d = 0.29; and PTSD symptoms, t(27) = −0.86, p = 0.398, d = 0.16, did not reach statistical significance.

Among male participants (n = 16), significant reductions were observed in stress, t(15) = −2.21, p = 0.043, d = 0.55, and PTSD symptoms, t(15) = −2.62, p = 0.019, d = 0.66. Depression demonstrated a trend toward reduction, t(15) = −2.03, p = 0.061, d = 0.51, whereas anxiety did not significantly change, t(15) = −0.96, p = 0.352, d = 0.24.

These subgroup analyses were exploratory and not powered for formal interaction testing; therefore, interpretations regarding gender differences should be made cautiously.

Consistent with prior literature, female participants reported higher baseline distress across measures. Within-group analyses indicated that females demonstrated a statistically significant reduction in anxiety. Male participants, in contrast, demonstrated significant reductions in stress and PTSD symptoms, with anxiety remaining stable.

Descriptive patterns indicated that females maintained higher overall levels of depression, stress, and PTSD symptoms at both time points. However, both genders showed parallel directional trends across symptom domains.

Figure 2 illustrates pre- and post-intervention DASS-21 subscale scores and PTSD symptoms by gender in the analyzed sample (N = 64). Error bars represent ±1 standard deviation.

These findings suggest possible gender-related variations in symptom patterns within this sample; however, differential responsiveness to the intervention cannot be established without formal interaction testing.

4.4. Drawing Analysis

Descriptive comparison of pre- and post-intervention self-portraits revealed observable shifts in structural and thematic features among participants in the intervention group.

Post-intervention drawings were generally characterized by:

• Increased figure size and more central placement on the page.
• Greater use of color.
• Improved line continuity and structural completeness.
• More detailed facial features and contextual elements.

These observable changes indicate increased visual elaboration and more centralized compositional placement over the course of the intervention. Psychological interpretation of these visual features remains speculative and should be approached cautiously.

Representative examples are presented in Figure 3, Figure 4, Figure 5 and Figure 6 to illustrate qualitative trends.

The male participant’s pre-intervention drawing (Figure 3) depicted a relatively small figure positioned near the edge of the page, with limited color use and light line quality. In the corresponding post-intervention drawing (Figure 4), the figure was centrally placed, larger in size, and rendered with stronger lines and more vibrant color, indicating greater visual intensity and compositional prominence relative to the pre-intervention drawing.

Similarly, the female participant’s pre-intervention drawing (Figure 5) portrayed a kneeling figure positioned toward the lower margin of the page, accompanied by textual elements reflecting perceived academic criticism. In the post-intervention drawing (Figure 6), the figure appeared centrally positioned, larger, and more compositionally balanced, with an altered facial expression and increased visual detail.

Importantly, these examples are intended to illustrate qualitative patterns rather than to provide diagnostic interpretation. No standardized projective coding system or inter-rater reliability procedures were applied; therefore, interpretations remain exploratory. Drawing-based observations are presented to complement standardized self-report measures and provide a multi-method perspective on short-term psychosocial change.

The qualitative shifts appear descriptively parallel to quantitative symptom patterns; however, no statistical association or causal linkage was examined. The drawing analysis is intended to provide illustrative visual context rather than evidence of psychological mechanism or diagnostic change.

5. Discussion

5.1. Overall Intervention Effects

Consistent with the primary analyses, participants in the intervention group demonstrated significant reductions in anxiety and PTSD symptoms relative to baseline, along with a non-significant trend toward reduced stress. In contrast, the control group showed significant increases in anxiety, stress, and PTSD symptoms over the same period.

The moderate effect sizes observed across several domains suggest that even a time-limited intervention may be associated with practically meaningful changes. Importantly, symptom reductions were observed across multiple internalizing domains, supporting the conceptualization of expressive group interventions as transdiagnostic approaches targeting shared mechanisms such as emotional regulation, cognitive reframing, and peer-supported disclosure.

However, given the quasi-experimental design and institutional group assignment, causal interpretations remain limited. Baseline group differences and contextual factors should be considered when interpreting between-group comparisons. Nonetheless, the alignment of statistical significance, meaningful effect sizes, and domain-specific change patterns offers preliminary indications of possible intervention-related improvements. While baseline differences and the quasi-experimental structure warrant careful interpretation, these findings suggest possible intervention-related improvements that support further investigation into the intervention’s effectiveness.

5.2. Gender-Specific Symptom Patterns

Exploratory subgroup analyses suggested differentiated patterns of symptom response by gender within the intervention group. Female participants demonstrated significant reductions in anxiety symptoms with a moderate effect size, whereas changes in stress, depression, and PTSD did not reach statistical significance. In contrast, male participants exhibited significant reductions in stress and PTSD symptoms, with moderate-to-large effect sizes, and a trend toward improvement in depressive symptoms, while anxiety did not significantly change.

Although these analyses were not powered for formal interaction testing and should be interpreted cautiously, the observed pattern may reflect gender-related differences in emotional expression and stress processing during adolescence. Females may be more likely to report internalizing symptoms characterized by worry and cognitive rumination. It is possible that structured emotional articulation and peer-based reflective dialogue aligned with these reported symptom profiles; however, such alignment was not directly examined in the present study. Therefore, conclusions regarding gender-specific mechanisms remain speculative.

In contrast, the observed reductions in stress and PTSD symptoms among male participants offer exploratory insight into how this subgroup may experience and report symptom change. The program’s multimodal and activity-based components may be particularly well suited to supporting a range of emotional processing styles. While the current design does not permit firm conclusions about gender-specific pathways or mechanisms, these patterns provide valuable exploratory insights that can inform future research aimed at clarifying how different individuals engage with and benefit from the intervention. These interpretations remain tentative. The relatively small subgroup sizes limit statistical power, and baseline symptom differences may partially account for the observed effects. Future research employing adequately powered randomized designs is necessary to determine whether gender moderates treatment response or whether these patterns reflect sample-specific variability. Nevertheless, the observed subgroup patterns highlight potential directions to design interventions that are responsive to diverse emotional expression styles. These insights lay a strong foundation for future, adequately powered studies to further explore whether gender plays a meaningful role in shaping intervention outcomes.

5.3. Theoretical and Contextual Implications

The findings contribute to the growing literature supporting school- and residential-based mental health interventions for adolescents in rural or socioeconomically constrained environments. Rural youth may face compounded stressors, including parental migration, reduced access to psychological services, and limited mental health literacy. Structured group-based programming implemented within educational settings may serve as a scalable and contextually appropriate model for early intervention.

The intervention’s emphasis on emotional awareness, structured reflection, and peer normalization aligns with theoretical models of trauma-informed and resilience-based practice.

Structured emotional expression within a psychologically supportive group environment may have contributed to reductions in hyperarousal and stress reactivity. Although these mechanisms were not directly assessed in the present study, this interpretation may offer a direction for future research to further explore and clarify the processes underlying the observed improvements. Furthermore, the integration of expressive and narrative elements may be particularly relevant in collectivist cultural contexts, where shared experience and relational cohesion play central roles in meaning-making. While the present study did not directly assess cultural mediators, the group-based format may have enhanced engagement and perceived safety among participants.

5.4. Implications for Psychosocial Capacity Building in Belt and Road Regions

Beyond the immediate study setting, these findings may hold relevance for psychosocial capacity-building initiatives across Belt and Road (BRI) partner regions facing comparable structural and mental health service gaps. These reflections are offered as contextual considerations rather than as direct empirical implications derived from cross-national data in the present study. Many BRI countries are characterized by uneven access to school-based psychological support, limited numbers of trained professionals, and persistent stigma surrounding adolescent mental health. The present intervention model—brief, structured, and delivered within existing school systems—illustrates a potentially adaptable framework that does not depend on high-resource clinical infrastructure. Importantly, its feasibility within routine educational environments suggests alignment with broader regional efforts emphasizing human capital development, social resilience, and cross-sector collaboration.

However, transferability should be approached cautiously. Differences in cultural norms, educational governance structures, and workforce training capacity require context-sensitive adaptation rather than direct replication. Future cross-national comparative studies within BRI contexts could clarify which components are universally effective and which require localization. In this sense, the study contributes not as a prescriptive or policy-endorsing model, but as a context-specific empirical reference point for further investigation of scalable, school-based mental health strategies within emerging and transitional economies.

As China advances the professional development of art therapy, NDI, a culturally grounded expressive psychotherapy model developed by Chinese practitioners, has emerged as a distinctive modality within this evolving field. Integrating narrative traditions with structured drawing-based techniques, NDI has been preliminarily implemented in group counselling with earthquake-affected adolescents in remote regions such as Shimian, Sichuan, demonstrating contextual feasibility within post-disaster school settings.

Building on these initial applications, the structured features of NDI may warrant cautious exploration in other disaster-prone regions, including selected Belt and Road contexts such as Indonesia and Nepal, where adolescents face recurrent natural hazards. Any potential dissemination should prioritize culturally responsive adaptation through collaboration with local mental health professionals, educators, and community stakeholders to ensure alignment with indigenous symbolic systems and healing practices. A train-the-trainer framework may offer one sustainable approach to strengthening local capacity, particularly in low-resource educational environments.

Alignment with internationally recognized mental health and psychosocial support frameworks, such as the IASC MHPSS intervention pyramid, could facilitate conceptual integration within broader service systems. However, cross-national implementation would require rigorous empirical evaluation. Longitudinal comparative studies retaining item-level data for psychometric validation, as well as structured process evaluation, would be necessary to establish generalizability, fidelity, and contextual effectiveness. The development of linguistically adapted manuals and culturally informed supervision models may further support implementation integrity. These considerations are presented as future research directions rather than as established evidence of cross-contextual applicability.

5.5. Limitations and Future Directions

The findings of this study should be interpreted within several methodological and contextual considerations that also point toward important directions for continued investigation.

The quasi-experimental design, without random assignment or matching procedures, limits the strength of causal inference. Baseline differences between groups, particularly in anxiety and PTSD symptoms, further constrain interpretation of between-group effects. Although descriptive pre–post patterns provide preliminary support for intervention-related change, randomized controlled trials with larger and more balanced samples would allow for more definitive evaluation. Cluster randomization at the classroom or school level may be especially appropriate in educational settings and would strengthen internal validity while preserving ecological relevance.

Sample size and attrition patterns also warrant attention. The smaller control group and lower completion rates may have influenced variance estimates and reduced statistical power for between-group comparisons. Future studies should prioritize balanced recruitment and retention strategies to enhance statistical stability and interpretive precision.

Trauma exposure history beyond the focal earthquake event was not systematically quantified, and formal verification of DSM-5 Criterion A trauma was not conducted. While elevated PTSD symptom scores served as inclusion criteria, incorporating validated trauma history inventories in future research would clarify the role of prior exposure, cumulative adversity, and potential dose–response relationships.

With respect to measurement, although the DASS-21 and PCL-5 have demonstrated validity in Chinese populations, reliability indices were not recalculated for the present sample due to the absence of archived item-level data. While this limits the ability to confirm internal consistency within this specific cohort, it highlights an important opportunity for future research to strengthen psychometric precision through more comprehensive data archiving practices. In addition, the PCL-5 was originally developed for adult populations. Although it has been applied in adolescent disaster research, developmental differences in reading comprehension, symptom interpretation, and self-report insight may influence response patterns among younger participants. At the same time, these considerations provide valuable direction for refining assessment approaches in youth samples. Future studies may benefit from incorporating adolescent-adapted PTSD instruments or conducting measurement invariance testing to further ensure developmental appropriateness in school-based contexts.

The qualitative drawing analysis was exploratory in nature and did not employ standardized coding manuals, inter-rater reliability procedures, or blinded raters. While the findings offer illustrative insight into expressive processes, future research incorporating structured coding frameworks and independent reliability testing would enhance reproducibility and methodological rigor.

The absence of longitudinal follow-up limits conclusions regarding durability of effects. Although reductions in anxiety and stabilization of trauma-related symptoms were observed, longer-term assessments at six- and twelve-month intervals would clarify sustainability and inform whether booster sessions or ongoing supports are warranted. Such follow-up would also help contextualize the modest fluctuation in depressive symptoms observed across groups.

Mechanisms of change remain to be more precisely specified. Relational factors hypothesized to contribute to outcomes—such as group cohesion, peer validation, and perceived self-efficacy—were not directly measured. Incorporating validated measures of group process and mediational analyses would allow for more refined testing of theoretical pathways underlying expressive group interventions.

In addition, the absence of psychological data for non-assessed participants limits certainty regarding full-cohort representativeness. It remains possible that unmeasured differences in baseline symptom severity or engagement patterns existed between assessed and non-assessed students. Future investigations should prioritize full-sample assessment or implement stratified sampling and baseline equivalence testing to further strengthen internal validity. Moreover, dismantling studies examining the relative contribution of narrative expression, structured activities, and peer interaction may also inform intervention optimization.

Replication in adequately powered controlled trials will be necessary before conclusions regarding effectiveness can be drawn.

Taken together, these considerations frame the present findings as an important step in an evolving program of research. Continued methodological refinement, expanded sampling, and deeper investigation of mechanisms will contribute to a more comprehensive understanding of Narrative Drawing Intervention within school-based trauma-informed practice.

6. Conclusions

This study evaluated a structured, school-based psychological intervention implemented within a resource-constrained educational setting. This investigation should be characterized as a quasi-experimental, field-based study conducted under real-world implementation conditions rather than as a randomized or controlled test of intervention efficacy. The absence of random assignment and full experimental control limits causal inference. The findings suggest that participation in the program was associated with reductions in selected emotional and behavioral symptoms, with some variation across gender groups. While these results provide preliminary support for the feasibility of brief, school-embedded psychosocial interventions, they should be interpreted with appropriate caution.

Several methodological considerations limit the strength of causal inference and generalizability. The study design, contextual specificity, and duration of follow-up constrain conclusions regarding long-term effectiveness and broader applicability. In addition, unmeasured contextual or implementation-related factors may have influenced observed outcomes. As such, the findings should be understood as indicative rather than definitive evidence of intervention impact.

Although the results contribute practice-based evidence regarding the feasibility of implementing brief, school-embedded psychosocial interventions in naturalistic settings, they require replication in adequately powered, controlled trials before conclusions about effectiveness can be drawn.

Despite these limitations, the study contributes contextually grounded data to an area where empirical evidence remains comparatively limited. Future research employing randomized or multi-site designs, longer follow-up periods, and cross-regional comparisons will be essential to clarify causal mechanisms, sustainability, and the conditions under which similar interventions may be effectively adapted or scaled.