摘要正念冥想对焦虑症状的干预效果结论不一，系统考察其干预效果及影响因素，有助于理解其作用机制，并为正念冥想干预焦虑提供实践指导。本研究遵循PRISMA-Protocol，运用元分析技术，以即时效果与追踪效果为结果变量，考察被试特征、干预特征、研究特征和结果分析4个方面(11个变量)的调节作用。通过数据库检索与筛选，最终纳入东西方国家实证研究共55项 (68个独立效应量，4595名被试)。结果发现：正念冥想对焦虑干预的整体即时效果量达到显著的中等到大效果(g = 0.60)，但追踪效果不显著。单因素回归模型分析表明，研究质量、国家地域、年龄、干预形式(团体练习 vs. 个体练习)、家庭练习与脱落率显著调节干预即时效果，其中国家地域变量对效果量影响程度最大；多元回归分析显示，对照组类型、国家地域、练习经验、数据分析方式(ITT vs. PP)对干预效果存在调节作用；此外，国家地域与其他变量存在交互作用。后续研究需在研究质量、成本效益、不同焦虑类型及其他调节变量和提供客观评价指标上作改进。
Abstract：In such an era which increasingly emphasizes the practice of bio-psycho-social model in the field of modern medicine, mindfulness meditation (MM) has enjoyed a growing population in healthcare and demonstrated significant effectiveness in alleviating anxiety for nonclinical and clinical populations based on mounting empirical literature. However, the result of effect size has remained inconsistent and the moderators have not explored comprehensively yet. Accordingly, we conducted meta-analysis by systematically reviewing the studies regarding mindfulness-based intervention for anxiety to determine the acute and follow-up efficacy, and examine whether the features of subjects (e.g. age, area), intervention (e.g. duration, home practice), study design (e.g. type of controlling), as well as data analysis (e.g. attrition) moderate the magnitude of effect (11variables).
The review was performed according to PRISMA-Protocol in a rigor method. The randomized controlled trial researches, which have been published, were selected by searching databases of VIP Journal Integration Platform, Wan fang, China National Knowledge Infrastructure, Scopus, Web of Science, Embase, the Cochrane Library, PubMed. Keywords encompassed mindfulness, meditation, MBSR, MBCT, anxi*, mood, intervention, therapy, program. Using random effect model, we pooled the effect size (Hedge’s g), meanwhile, we conducted publication bias evaluation, moderating effect analysis, interaction analysis in CMA 3.3.
We finally identified fifty-five RCTs from both East and West (k = 68 samples, N = 4595 participants), mindfulness meditation for anxiety are efficacious with a medium to large acute effect (g = 0.60), but is not stable at follow-up. Post-heterogeneity test turned out that the use of random effect model is reasonable. Univariate meta-regression models analysis has yielded that study quality, area, age, intervention form (Group vs. Individual), home practice, and attrition rate remarkably influence the acute effect size, while type of controlling, health condition, mindfulness practice experience, intervention duration, or statistical analysis methods (ITT vs. PP) don’t evidently moderate the acute efficacy for anxiety. Additionally, area affects the effect size most. Multiple meta-regression model suggested that type of controlling and area, as well as statistical analysis methods significantly moderate the effect size of intervention.
The results strongly support the efficacy of mindfulness-based practices for anxiety, but may not last. Besides, the variance of area influences the efficacy most among these moderators, and practitioners in the east may benefit more than those who are in the west. Likewise, moderators including study quality, age, mindfulness practice experience, intervention form, home practice and attrition also affect the effect size to a certain extent. Future research in methodologically sound studies are warranted to provide more clear and reproducible evidence, and explore more underlying moderators for the intervention effect size, such as participant satisfaction and so forth.