Depression screening tools are used for many purposes in research and practice. Most depression screening tool accuracy studies, however, have been conducted in samples too small to precisely estimate accuracy or conduct subgroup analyses. Many selectively publish results using “optimal cutoff” scores based on data-driven analysis, instead of standard cutoffs. These practices have led to distorted accuracy estimates when published results are synthesized in meta-analyses. Furthermore, due to the small samples in existing primary studies, researchers dichotomize screening scores as positive or negative; thus, all patients above or below a cutoff are assigned the same probability of having depression. In reality, however, the probability of having depression varies with continuous screening scores and patient characteristics (e.g., sex, age, medical conditions).
The DEPRESsion Screening Data (DEPRESSD) Project was set up to gather and synthesize large datasets to evaluate evidence on the accuracy of depression screening tools using individual participant data meta-analyses (IPDMAs). DEPRESSD now includes over 300 investigators from 58 countries who contribute primary datasets on commonly used depression screening tools, including the PHQ-9 (105 studies; 46,277 participants), HADS (82 studies; 17,202 participants), and EPDS (60 studies; 15,841 participants). By using IPDMA with large datasets, DEPRESSD is working to address limitations in the existing evidence based on screening tools and how we approach screening. DEPRESSD investigators are also conducting methodological research on topics such as the performance of different depression reference standards, patterns and influence of selective cutoff reporting, and data-driven selection of cutoff thresholds for identifying possible cases in screening. This talk will cover the details of the IPDMAs conducted as well as their implications for use of depression screening tools.
Brief Biography
Dr. Thombs completed his PhD in clinical psychology in 2004 and began his career at McGill University in 2006. He has published over 260 peer-reviewed articles, including many in the world’s top biomedical journals, such as JAMA, BMJ, JAMA Internal Medicine, and CMAJ. Dr. Thombs focus on innovative methodological principles allows his team to examine important, but previously ignored or poorly addressed problems. He founded and directs the Scleroderma Patient-centered Intervention Network (SPIN), leads an international collaboration to conduct individual participant data meta-analyses on the accuracy of depression screening tools (DEPRESSD) and an active meta-research program. Additionally, he is involved in policy as Chair of the Canadian Task Force on Preventive Health Care. Dr. Thombs is a Fellow of the Canadian Academy of Health Sciences and a member of the College of New Scholars, Artists, and Scientists of the Royal Society of Canada.