A big data approach to understanding anxiety disorders

Student thesis: Doctoral ThesisDoctor of Philosophy


Background: Anxiety disorders are similarly disabling and prevalent as depressive disorders; however, research on anxiety disorders lags behind the depression literature. For example, twin studies have demonstrated that anxiety and depressive disorders are between 30-60% heritable. Yet the latest genome-wide association study to identify genes associated with depression included over 800,000 participants, whereas the largest on anxiety included around 200,000 individuals. Furthermore, much of the literature focuses on anxiety disorders as a general category. Less is known about the disorder-specific environmental and genetic risk factors of anxiety disorders.

Aims: The aims of this thesis were to 1) introduce a new, large-scale study to enable research on anxiety disorders, 2) compare the most common remote assessments of specific anxiety disorder diagnoses, and 3) explore differences in sociodemographic factors, traumatic experiences, clinical characteristics, and familial risk factors associated with anxiety comorbidity groups: single anxiety, anxiety-anxiety comorbidity, anxiety-MDD comorbidity, and single depression.

Methods: Participants in the United Kingdom were recruited to the Genetic Links to Anxiety and Depressive (GLAD) Study and COVID-19 Psychiatry and Neurological Genetics (COPING) study, which are part of the NIHR BioResource. The GLAD and COPING studies were conducted entirely online. For GLAD, individuals consented to medical record linkage and recontact, and completed a questionnaire. Participants also provided a saliva sample for genotyping. For COPING, members of the NIHR BioResource (COPING NBR), including physical health and general population cohorts, and the GLAD Study were emailed invitations which included a link to the online information sheet, consent form, and baseline and optional questionnaires. The GLAD and COPING questionnaires included self-reported diagnoses as well as symptom-based assessments of lifetime major depressive disorder (MDD) and the five primary anxiety disorders: generalised anxiety disorder (GAD), specific phobia, social anxiety disorder, panic disorder, and agoraphobia. Self-reported and symptom-based diagnoses were compared using sensitivity and specificity analyses. Logistic regression analyses were used to examine the association between sociodemographic factors, traumatic experiences, and clinical characteristics and anxiety comorbidity groups. Familial risk was assessed using two different indices, self-reported family history of mental health diagnoses and genetic factors (i.e., polygenic risk scores), and were compared for subsamples of participants with available data.

Results: The GLAD and COPING studies were successful at online recruiting, with 43,202 GLAD and 17,065 COPING NBR participants who had completed the questionnaires, respectively, as of May 2022. Participants had a mean age of 39 (ranging from 16-99 years) and majority were female (79%). Responses to the self-reported and symptom-based measures of MDD and any anxiety had moderate agreement. However, self-report measures of the individual anxiety disorders classified the majority of participants as having GAD, whereas the respective symptom-based measures had a greater distribution across the disorders. Results from the comorbidity group comparison replicated prior research, demonstrating that anxiety-anxiety comorbidity is more severe and complex than single anxiety and has a distinct clinical profile to anxiety-MDD comorbidity. Minimal differences in familial risk were found between the comorbidity groups, likely due to small sample sizes.

Discussion: The GLAD and COPING studies, part of the NIHR BioResource, have contributed a valuable resource of detailed data on anxiety disorders from participants who have consented to be recontacted for future research. Utilising data from this resource, we demonstrated that the two most common online assessment methods for diagnosing anxiety disorders had high disagreement in disorder categorisation of participants, which may impact the generalisability of findings and has implications for meta-analyses of studies or samples. We additionally found that comorbidity between anxiety disorders is similarly severe and complex as anxiety-depression comorbidity but different clinical features, and therefore merits further investigation. More research to better assess and understand anxiety disorders is needed, and the GLAD and COPING studies provide a powerful resource to facilitate this goal.
Date of Award1 Sept 2022
Original languageEnglish
Awarding Institution
  • King's College London
SupervisorThalia Eley (Supervisor) & Kimberley Goldsmith (Supervisor)

Cite this