Related Experiment Video
Updated: Jun 7, 2025
Network Analysis of Foramen Ovale Electrode Recordings in Drug-resistant Temporal Lobe Epilepsy Patients
Published on: December 18, 2016
Prediction of electroconvulsive therapy outcome: A network analysis approach
Tessa F Blanken1, Rob Kok2, Jasmien Obbels3
1Department of Psychology, University of Amsterdam, Amsterdam, Netherlands.
Objective:
While electroconvulsive therapy (ECT) for the treatment of major depressive disorder is effective, individual response is variable and difficult to predict. These difficulties may in part result from heterogeneity at the symptom level. We aim to predict remission using baseline depression symptoms, taking the associations among symptoms into account, by using a network analysis approach.
Method:
We combined individual patient data from two randomized controlled trials (total N = 161) and estimated a Mixed Graphical Model to estimate which baseline depression symptoms (corresponding to HRSD-17 items) uniquely predicted remission (defined as either HRSD≤7 or MADRS<10). We included study as moderator to evaluate study heterogeneity. For symptoms directly predictive of remission we computed odds ratios.
Results:
Three baseline symptoms were uniquely predictive of remission: suicidality negatively predicted remission (OR = 0.75; bootstrapped confidence interval (bCI) = 0.44-1.00) whereas retardation (OR = 1.21; bCI = 1.00-2.02) and hypochondriasis (OR = 1.31; bCI = 1.00-2.25) positively predicted remission. The estimated effects did not differ across trials as no moderation effects were found.
Conclusion:
By using a network analysis approach this study identified that the presence of suicidal ideation predicts an overall worse treatment outcome. Psychomotor retardation and hypochondriasis, on the other hand, seem to be associated with a better outcome.
