Category: Research Methods and Statistics

PS13- #B60 - Creating a Multivariate Risk Prediction Algorithm of Depression Using Individual Patient Data to Identify Risk Factors of Depression Onset

Saturday, Nov 18
2:45 PM – 3:45 PM
Location: Indigo Ballroom CDGH, Level 2, Indigo Level

Keywords: Adult Depression | Prevention | Risk / Vulnerability Factors

Even though psychological interventions for the prevention of depression have shown to be effective, only a minority (< 25%) of those targeted with preventative interventions would actually develop a Major Depressive Disorder (MDD). One strategy to increase the effects of prevention programs is to target individuals at ultra high risk. Through the combination of a broad range of different risk factors, multivariate risk prediction algorithms allow the prediction of disorder onset on an individual level with a high prediction accuracy. Such risk-prediction algorithms are often used in somatic-medicine, but less applied in mental health settings. One general problem is that very large sample sizes are needed to develop such multivariate algorithms, and single trials are usually underpowered. Using individual patient-data meta-analysis allows pooling of several studies to gain sufficient statistical power. The aim of the study is to develop and test a multivariate risk factor algorithm, to identify individuals at ultra high risk of developing a depression among those participating in preventive interventions. An individual patient data meta-analysis was conducted on randomized controlled trials that examined a psychological indicated preventive intervention compared to a control group on the onset of major depressive disorder in adults. Individual-level predicted probabilities will be created, receiver operating characteristic (ROC) curves generated, and to evaluate prediction accuracy area under the curve (AUC) values will be calculated. Predicted probabilities will discretize into 10 risk deciles and cross-classified with observed cases to visualize the concentration of risk associated with high composite predicted probabilities. 12 randomized controlled trials were identified including N=2372 adults. Analyses of risk factors of depression in the non-treatment group are currently in progress and will be presented during the conference.By identifying risk factors and developing a multivariate risk prediction algorithm of depression a first step is taken to increase the effectiveness of prevention programs by targeting actual at risk individuals.

Kiona Weisel

Research associate
Friedrich-Alexander University Erlangen-Nürnberg, chair of clinical psychology and psychotherapy

Jo Annika Reins

Leuphana Universität Erlangen-Nürnberg

Claudia Buntrock

Friedrich-Alexander Universität Erlangen-Nürnberg

Matthias Berking

Head of chair of clinical psychology and psychotherapy
Friedrich-Alexander University Erlangen-Nürnberg, chair of clinical psychology and psychotherapy

Johannes Zimmermann

Psychologische Hochschule Berlin

Pim Cuijpers

VU University Amsterdam

David D. Ebert

Lead e-Mental Health Unit
Friedrich-Alexander University Erlangen-Nürnberg, chair of clinical psychology and psychotherapy