Supplementary file 1_Network analysis of mental health knowledge and stigma among high school students in Sichuan, China.docx

BackgroundLack of mental health knowledge (MHK) and stigma toward mental illness pose significant barriers to help-seeking behaviors among Chinese high school students, amid intense academic pressures and cultural influences. Traditional aggregate scoring methods overlook dynamic interconnections between specific knowledge items and stigma attitudes. This study applies network analysis to model MHK and stigma as interconnected systems, identifying central nodes, bridges, and potential intervention targets in a large adolescent sample.
MethodsA cross-sectional survey was conducted among 12,537 high school students in Sichuan Province, China, between October 2024 and January 2025.The MHK was assessed using the 20-item Mental Health Knowledge Questionnaire (MHKQ), and the stigma via the 12-item Perceived Devaluation and Discrimination Scale (PDD). Networks were estimated with the IsingFit algorithm in R (v4.3.2), incorporating partial correlations. Centrality (strength), bridge expected influence, and stability were computed. The NodeIdentifyR algorithm (NIRA) was used to simulated aggravating and alleviating interventions on network sum scores. Gender invariance was tested using the NetworkComparisonTest package.
ResultsThe network revealed two communities (MHK and stigma) with dense intra-cluster connections and key bridges. MHKQ11 (“optimistic attitude, good relationships, and healthy habits help maintain mental health”) showed the highest centrality (strength: 12.06), serving as a MHK hub. MHKQ10 (“short-term medication suffices for severe mental illnesses like schizophrenia without long-term adherence”) bridged to stigma items (e.g., Stigma5: "most employers will not hire a person who has been hospitalized for mental illness"; bridge expected influence: 0.957). Network stability was robust (CS > 0.672). Aggravating simulations were associated with the highest sum scores for MHKQ10, MHKQ13, and MHKQ14. Alleviating interventions showed the greatest potential for score reduction via MHKQ11, MHKQ8, and MHKQ16. Gender networks showed invariance (global strength difference: 0.37, p = 0.693).
ConclusionsThis network analysis highlights MHKQ10 and MHKQ11 as pivotal targets for stigma reduction, with misconceptions about treatment adherence linking knowledge deficits to devaluation perceptions. Gender-invariant structures suggest universal applicability for school-based interventions, aligning with China’s mental health initiatives to enhance literacy and promote equity.

研究背景：在中国高中生群体中，受学业重压与文化环境影响，心理健康知识（Mental Health Knowledge, MHK）匮乏与对精神疾病的病耻感，是阻碍其求助行为的重要障碍。传统的总分计分法往往忽视了特定知识条目与病耻感态度间的动态关联。本研究采用网络分析方法，将心理健康知识与病耻感建模为相互关联的系统，并在大型青少年样本中识别出核心节点、桥接节点与潜在干预靶点。
研究方法：本研究于2024年10月至2025年1月间，在中国四川省对12537名高中生开展了横断面调查。采用20条目心理健康知识问卷（Mental Health Knowledge Questionnaire, MHKQ）评估心理健康知识水平，使用12条目感知贬值与歧视量表（Perceived Devaluation and Discrimination Scale, PDD）评估病耻感程度。基于R语言（版本4.3.2）中的IsingFit算法构建网络模型，纳入偏相关分析；随后计算节点中心性（强度）、桥接预期影响值与网络稳定性。采用NodeIdentifyR算法（NIRA）对网络总分模拟干预效果，分别模拟加重与缓解型干预。使用NetworkComparisonTest包检验网络的性别不变性。
研究结果：网络分析显示存在两个集群（心理健康知识与病耻感），集群内部连接紧密且存在关键桥接节点。条目MHKQ11（"乐观的心态、良好的人际关系与健康的生活习惯有助于维持心理健康"）具有最高的中心性（强度值：12.06），为心理健康知识集群的核心节点。条目MHKQ10（"精神分裂症等重型精神疾病仅需短期服药即可，无需长期遵医嘱"）为连接心理健康知识与病耻感的桥接节点，对应病耻感条目Stigma5（"多数雇主不会聘用因精神疾病住院治疗的人员"），其桥接预期影响值为0.957。网络稳定性良好（CS值>0.672）。加重型干预模拟结果显示，对MHKQ10、MHKQ13与MHKQ14进行干预会导致网络总分最高程度的升高。缓解型干预模拟结果则显示，针对MHKQ11、MHKQ8与MHKQ16开展干预可最大程度降低网络总分。性别分组网络不存在显著差异（全局强度差异值：0.37，p=0.693），即具有性别不变性。
研究结论：本网络分析明确了MHKQ10与MHKQ11为降低病耻感的关键干预靶点，而关于服药依从性的认知误区，会将心理健康知识匮乏与病耻感知相连接。网络结构具有性别不变性，提示本研究的干预方案可推广至全国中小学心理健康教育场景，契合中国当前提升全民心理健康素养、促进公平的精神卫生工作规划。