Lessons Learned from Analysis of GP Free-text Data (A Case Study on Depression)

Depression is a major cause of disability, affecting about 4.4% of the global population. Most treatment is handled in primary care. During GP encounters the patient is assessed on their mood and relevant symptoms (e.g., sleep or appetite), their change compared to previous encounters, and response to antidepressants (including adverse reaction) with potential adjustment (dosage, switching to a different antidepressant). Encounters are recorded into hospital systems using structured Read codes (representing conditions or tests) and free-text data (allowing finer detail). Here we describe our experience in analysing GP data within a Secure Data Environment (SDE) in the context of depression as part of the AMBER project including statistics from a sample of records of patients with depression, the outcomes of discussions with clinical experts, and we highlight the opportunities and issues around GP data. While our work focuses on depression, our findings may be relevant to other psychiatric conditions or general recording of illness and treatment in primary care.