Methodological and Statistical Practices of using Symptom Networks to Evaluate Mental Health Interventions: A Systematic Review

Background: The network approach to psychopathology, which assesses associations between individual symptoms, has recently been applied to evaluate treatments for mental disorders. While various options for conducting a network analysis in intervention research exist, no structured guidelines have been established yet. To gain an overview and evaluate the potential of different analytic options, we conducted a systematic review on these studies. Methods: We systematically searched the literature with combining terms on network analysis, mental health problems, and intervention studies. Studies were included if they constructed a symptom network, analysed data that was collected before, during or after a treatment for mental disorders, and yielded information about the treatment effect. Information on sample characteristics, the intervention, the research design, the estimated networks, the statistical analyses, and open science practices were extracted. Results: Across the 56 included studies, network analyses varied widely. About half of the studies estimated cross-sectional networks without a treatment node, e.g., for different treatment groups separately or for a single group before, and after treatment. About 20% of the studies analysed cross-sectional networks including a treatment node and a third of the studies estimated longitudinal networks. Studies differed on how networks were estimated, which network parameters were calculated, and which statistical tests were applied. Conclusion: This review highlighted that many different analytic options exist when applying the network approach to intervention research. As the potential of the network approach depends on the applied methodologies, the analytic choices need to be further investigated and structured guidelines need to be developed.