(128) Network Analysis of Chronic Medication Use in Older Adults Living with Dementia Prior to and Following Diagnosis: A Web-based Visualisation Tool

Background: Inappropriate medication use is of concern for those living with dementia, and concurrent medication use has been associated with adverse outcomes. Limited attempts have been made to map medication use prior to and following dementia diagnosis for all chronic medications, which may identify any potentially inappropriate prescribing changes post-dementia diagnosis. With vast amounts of data, new tools are needed to adequately communicate research outputs with various stakeholders, including clinicians, carers, and patients.

Objectives: To produce an interactive web-based visualisation tool that shows changes in prevalence of chronic medication and patterns of concurrent medication use in the 15 years prior to and following dementia diagnosis.

Methods: Using Clinical Practice Research Datalink GOLD and Hospital Episode Statistics, 184,806 individuals were identified with a dementia diagnosis at age ≥ 65 years between 1998 and March 2020. Chronic medications for common comorbidities were included and labelled according to their BNF subsection. Treatment lengths were estimated such that there was no gap larger than 28 days between prescriptions within a treatment episode. Concurrent medication between BNF subsections was identified by finding overlapping treatment episodes of at least 14 days. Stratifying by sex, age group (5-year age bands) when first diagnosed with dementia, and time before or since dementia diagnosis, matrices were generated showing the proportion of patients co-medicated with two BNF subsections. D3.js was used to build a web-based visualisation tool that produced interactive force-directed networks and adjacency matrices for users to explore the results. Users can explore trends in the prevalence of each BNF subsection by sex and age group in the 15 years prior to and 5 years following diagnosis.

Results: The web-based interactive visualisation enables users to explore trends in both prevalence and concurrent use of chronic medications across 20+ years. Concurrent use of cardiovascular drugs was highly prevalent both before and after diagnosis, and more prevalent in males. Psychotropic medication use was more prevalent in females across the time periods as well as concurrent use of different types of psychotropic medications. Results were similar across age bands.

Conclusions: Our interactive tool demonstrates how researchers can use web-based visualisation techniques to improve communication of results to the public when using vast amounts of data. The tool visualises 108 different networks—each with 5,995 pairwise combinations of BNF subsections—and changes in prevalence over time of 110 different drug classes. The tool may help identify potential areas of common inappropriate prescribing to those living with dementia.