Comparison of algorithm-based versus single-item phenotyping measures of depression and anxiety disorders in the GLAD Study cohort

Background: Research to understand the complex aetiology of depressive and anxiety disorders often requires large sample sizes, but this comes at a cost. Large-scale studies are typically unable to utilise "gold standard" phenotyping methods, instead relying on remote, self-report measures to ascertain phenotypes. Aims: To assess the comparability of two commonly used phenotyping methods for depression and anxiety disorders. Method: Participants from the Genetic Links to Anxiety and Depression (GLAD) Study (N = 37,419) completed an online questionnaire including detailed symptom reports. They received a lifetime algorithm-based diagnosis based on DSM-5 criteria for major depressive disorder (MDD), generalised anxiety disorder (GAD), specific phobia, social anxiety disorder, panic disorder, and agoraphobia. Any anxiety disorder included participants with at least one anxiety disorder. Participants also responded to single-item questions asking whether they had ever been diagnosed with these disorders by health professionals. Results: Agreement for algorithm-based and single-item diagnoses was high for MDD and any anxiety disorder but low for the individual anxiety disorders. For GAD, many participants with a single-item diagnosis did not receive an algorithm-based diagnosis. In contrast, algorithm-based diagnoses of the other anxiety disorders were more common than the single-item diagnoses. Conclusions: The two phenotyping methods were comparable for MDD and any anxiety disorder cases. However, frequencies of specific anxiety disorders varied depending on the method. Single-item diagnoses classified most participants as having GAD whereas algorithm-based diagnoses were more evenly distributed across the anxiety disorders. Future investigations of specific anxiety disorders should use algorithm-based or other robust phenotyping methods.