心理科学 ›› 2025, Vol. 48 ›› Issue (3): 732-742.DOI: 10.16719/j.cnki.1671-6981.20250321

• 临床与咨询 • 上一篇    下一篇

焦虑与抑郁的共病:基于交叉滞后路径模型和网络分析的纵向研究*

张森森1, 丁凤琴**1, 李宁1,2, 靖治1,3   

  1. 1宁夏大学教师教育学院,银川,750021;
    2宁夏银川市宁安精神卫生医院,银川,750021;
    3鄂尔多斯应用技术学院心理健康咨询中心,鄂尔多斯,017000
  • 出版日期:2025-05-20 发布日期:2025-05-30
  • 通讯作者: **丁凤琴,E-mail: dingfqin@nxu.edu.cn
  • 基金资助:
    *本研究得到国家自然科学基金(32360206)的资助

Comorbidity of Anxiety and Depression in a Clinical Sample: A Longitudinal Study Using Causal-Lagged Path Model and Cross-Lagged Panel Network Analysis

Zhang Sensen1, Ding Fengqin1, Li Ning1,2, Jing Zhi1,3   

  1. 1Institute of Teacher Education, Ningxia University, Yinchuan, 750000;
    2Ningan Mental Health Hospital, Yinchuan, 750000;
    3Mental Health Counseling Center of Erdos College of Applied Technology, Ordos, 014300
  • Online:2025-05-20 Published:2025-05-30

摘要: 采用交叉滞后路径模型和网络分析方法分析399名被诊断为焦虑症和抑郁症的患者在敏感期(T1)、巩固期(T2)和维持期(T3)的追踪数据。结果发现:(1)焦虑和抑郁的水平在T2和T3相较T1显著降低,且其在T2和T3未发生显著改变,而具体症状发生显著变化,这凸显了整体诊断之外对症状评估的必要。(2)核心症状和预期影响节点是临床干预的靶点:恐慌和疲劳是T1、T2和T3共有的核心症状,空虚感是T2和T3的核心症状;T1的预期影响节点手足颤抖、绝望和无用感能预防T2的症状,T2的躯体疼痛、睡眠障碍可以预防T3的症状。研究发现为不同发展期调整干预策略提供了潜在靶点,为预防健康人群潜在症状进一步发展为病症提供了参考。

关键词: 焦虑, 抑郁, 交叉滞后路径模型, 交叉滞后网络分析模型, 纵向研究

Abstract: Anxiety and depression represent a substantial burden to modern society and severely affect the quality of life of individuals. This study aimed to investigate the dynamic nature of anxiety and depression using a longitudinal approach, and to analyze causal relationships and interaction mechanisms between symptoms at different stages of treatment. By combining causal-lagged path model (CLPM) and cross-lagged panel network analysis (CLPN), we sought to gain a comprehensive understanding of the causal relationships and evolution of symptoms, identifying key predictive symptoms, and suggest possible intervention strategies for different treatment stages. CLPM and CLPN may offer distinct perspectives for investigating the relationship between anxiety and depression. The CLPM adopts a syndrome-oriented approach and explores longitudinal relationships among two or more latent variables. Conversely, the CLPN integrates the strengths of latent variable modeling and network theory, assuming that relationships occur at the symptom level over time. It captures the longitudinal evolution of specific symptoms and their interactions through directed symptom networks, identifying symptoms that play predictive or influential roles in understanding cross-diagnostic processes.
Therefore, we recruited 399 outpatient patients (Mean age = 39.6 years; female = 374) diagnosed with severe anxiety and depression from a psychiatric hospital in Western China. Clinical interviews were conducted by two clinicians, which provided objective assessments of the patients’ conditions. CLPM and CLPN were employed to analyze three sets of data from the Self-rating Anxiety Scale (SAS) and the Self-rating Depression Scale (SDS), which were completed by the patients during the acute period (T1), continuation period (T2), and maintenance period (T3). The aim was to draw syndrome-oriented and symptom-oriented inferences regarding the mechanisms of complicated interactions and dynamic evolutionary processes of the disorder.
Results showed that there were high temporal correlations between anxiety and depression at all three time points, emphasizing their strong association. Panic and fatigue emerged as core symptoms across all periods, with emptiness identified as a shared core symptom at T2 and T3. Additionally, when comparing T2 and T3 to T1, anxiety and depression levels significantly decreased (p < .05). Notably, specific symptom relationships highlighted the importance of certain nodes. Depressive mood (i.e., depressed, sad, and blue), tachycardia, and fatigue acted as bridging symptoms, suggesting their role in activating opposing symptom clusters. This emphasizes the need to consider multiple dimensions of symptoms during interventions to disrupt the pathway of comorbidities. Moreover, the global strength of the three network structures did not differ significantly (p > .05), but T2 showed the highest one. It may indicate variations in treatment effects at different stages, with patients’ sensitivity to specific symptoms changing throughout the treatment process. Local strength analysis revealed specific changes in symptom sensitivity, emphasizing the importance of adjusting coping strategies for different symptoms during treatment. Additionally, the 95% confidence intervals of the bootstrapped edge weights of the network were relatively narrow, and there was no overlap with strongest edges, indicating the accuracy of the estimated network edges at each time point. The centrality stability coefficient (CS) estimated through bootstrapped subset procedures reveals that the CS of nodal strength, in-expected influence (iEI) and out-expected influence (oEI) were all greater than .25 at T1, T2, and T3. Specifically, in the T1→ T2 network and in the T2→ T3 network, the CS coefficients of iEI and oEI were also all greater than .25. Moreover, differences in nodal strength centrality indicated significant variations among several symptoms, suggesting stable and generalizable findings.
In conclusion, the present study sheds light on the nuanced interplay between anxiety and depression in outpatients. These findings have significant implications for the understanding and prevention of anxiety and depression, offering clinical recommendations and potential intervention targets for adjusting treatment strategies at different stages of treatment to mitigate symptom development. Thus, it is recommended that network analysis be intergrated into current diagnostic, treatment, and follow-up procedures to promote individualized interventions and improve patient recovery.

Key words: anxiety, depression, causal-lagged path model, cross-lagged panel network analysis, longitudinal study