Reply to the Comments on “The Efficacy and Durability of Mindfulness-based Cognitive Therapy in the Treatment of Anxiety and Depressive Disorders: A Systematic Review and Meta-analysis”

Dear Commentators,

Thank you for showing interest in our study, ¹ and we appreciate the first comment concerning diagnostic heterogeneity among the included studies; however, our study was designed to represent “clinically significant anxiety and depressive symptomatology” rather than disorder-specific labels. Second concern raised by the commentator regarding the tool prioritization, as we are confronted with few studies only regarding the selection of the tool while extracting the data, and we prefer to select the tool which have wider coverage of assessment of anxiety and depression and has gold standard psychometric properties such as Beck Depression Inventory (BDI), Beck Anxiety Inventory (BAI), and DASS (Depression, Anxiety and Stress Scale). Further, greater emphasis was given on self-reported discomfort assessed through questionnaires and inventories, rather than on clinician ratings. Furthermore, concerns related to the non-supplementation of extracted data from individual studies will be addressed, and the data will be presented in the forest plots as a corrigendum.

The commentator raised concerns regarding the computational process of “dppc2.” We also clarify that in our analysis, we initially applied a fixed correlation value of 0.50, consistent with previous practices of many studies. Although discrepancies about the selection of correlation value are present in the literature, as some studies opted 0.59 a fixed correlation,^2,3 and others have used fixed correlations between 0.70 and 0.90.^4,5 However, researchers have criticized this approach because it may introduce bias when correlation values are unknown. ⁶ Therefore, in the revised analysis, we imputed a range of plausible correlation values (r = 0.25, 0.50, 0.75), as recommended in many methodological guidelines.^7,8 As Borenstein et al. ⁷ demonstrate, varying “r” within this range yields variance intervals between 0.075 and 0.025. Therefore, we applied to impute study-level correlations, and details about the sensitivity analysis will be provided as a corrigendum. Furthermore, as all included studies provided mean and standard deviation values for the relevant conditions, no data conversions were required.

The revised analysis with different imputed r values reduced the heterogeneity; however, as per the third comment and the reviewer’s suggestion, we will additionally conduct subgroup analyses based on diagnostic threshold, illness stage, and potential risk of biases to explore possible sources of heterogeneity. The findings of sub-group analysis will also be reported as a corrigendum.

We appreciate the thoughtful comment regarding the definition of durability. As highlighted in the comment, equal deterioration across groups could misleadingly indicate durability despite a loss of treatment benefit. Although this approach has been used in many meta-analytical reviews. ³ To address this limitation, we will also include pre- to follow-up comparisons as a corrigendum.

Further, the components of treatment-as-usual, including pharmacotherapy and psychotherapy, will be detailed as a corrigendum.

We agree that interpreting self-reports and waitlist controls requires caution when applying risk of bias (RoB) 2.0. In psychotherapeutic RCTs, blinding participants is not feasible, and outcomes are often self-reported, typically resulting in “some concerns” in the measurement of the outcome domain rather than an automatic “high risk.” ⁹ Additionally, most studies implemented an eight-week structured mindfulness-based cognitive therapy (MBCT), minimizing deviations; hence, the majority were rated as low risk. Additionally, some researchers argue that subjective measures warrant “some concerns,” while others note that validated questionnaires mitigate bias, reflecting the persistent struggle between reliance on psychological tools and the demand for rigorous, reproducible risk of bias evaluations. ¹⁰ As our RoB 2.0 evaluations were heavily influenced as per the recommendation of Munder and Barth, ⁹ therefore most studies were rated as low risk of bias. Nevertheless, we will reanalyze and report the results as a corrigendum in compliance with RoB 2.0 guidelines.

We thank the commentators for acknowledging our effort in the conduct of this research.