Understanding antidepressant change patterns in the UK Biobank

Background Antidepressants are the most frequently prescribed medications in psychiatry. Medical records of thousands of individuals provide a valuable opportunity to explore prescribing patterns and identify factors that influence treatment outcomes. Aims To investigate antidepressant change patterns in depression and non-depression indications and assess clinical and genetic factors associated with outcomes of antidepressant treatment. Methods Using primary care records from the UK Biobank, we examined outcomes including number of antidepressant changes and discontinuation due to either side-effects or inadequate response. Genetic analyses including heritability estimation, genetic correlation and polygenic score association were performed. Results A total of 82 633 individuals were prescribed at least 1 antidepressant. Of these, 28 332 individuals with at least 1 primary care depression diagnosis were classified as the depression group, and 24 543 individuals without evidence of depression were classified as the non-depression group. Citalopram and fluoxetine were the most prescribed antidepressants for depression, whereas amitriptyline dominated prescriptions for non-depression indications. Individuals with depression were more likely to stay on antidepressants longer than those without depression and to follow preferred antidepressants that changed over time. Antidepressant changes and discontinuation were associated with a range of psychiatric and somatic conditions, including recurrent depression (early discontinuation: odds ratio = 1.96; late discontinuation: odds ratio = 2.63) and anxiety (early discontinuation: odds ratio = 1.37; late discontinuation: odds ratio = 1.99) in the depression group, and pain-related conditions in the non-depression group. Genetic analyses identified two novel variants associated with early discontinuation of selective serotonin reuptake inhibitors. Notable genetic overlap was shown between these outcomes and multiple psychiatric and physical traits, including the number of antidepressant changes with anxiety and depression ( r g = 0.81–0.83), and polygenic scores for depression and attention-deficit hyperactivity disorder showed significant predictive value with respect to treatment outcomes. Conclusions These findings characterise antidepressant change patterns in primary care records and highlight the potential value of integrating clinical and genetic data to better understand factors associated with treatment outcomes.