Abstract
BACKGROUND: Drug prescription networks (DPNs) model the temporal dynamics of medication co-prescription within a population. Understanding these networks can provide insights into polypharmacy and prescribing behaviors. OBJECTIVE: This study assesses the structural characteristics of temporal DPNs derived from daily co-prescriptions of antidepressants, anxiolytics, and other therapeutic drug classes. By analyzing these networks using eigenvector centrality, we identify influential medications and prescribing patterns. METHODS: We utilized the IADB.nl database, including prescriptions from 128 Dutch pharmacies (2018-2022). A cohort of patients prescribed antidepressants/anxiolytics was extracted. Medications were classified using the Anatomical Therapeutic Chemical (ATC) system into 24 therapeutic classes. Time-varying DPNs were constructed as undirected graphs using symmetric daily dose-adjusted co-prescriptions. Eigenvector centrality ( ce ) quantified relative nodal importance. Weekly-aggregated data included number of dispensing ( nc ) and eigenvector centrality, which were decomposed using a singular-spectrum approach. RESULTS: Antidepressants ( ce : 0.09, nc : 28,993) and anxiolytics ( ce : 0.05, nc : 14,061) had high eigenvector centrality, demonstrating frequent co-prescription. Other ATC groups with high centrality included those for the alimentary tract and metabolism (A01-A16), blood and blood-forming organs (B01-B06), cardiovascular system (C01-C10), respiratory system (R01-R07), and analgesics (N02). DISCUSSION: DPNs revealed key co-prescription patterns. High-centrality medications highlight potential targets for drug monitoring, such as identifying co-prescription trends that may warrant evaluation for safety, appropriateness, or policy oversight. This approach aids in identifying influential medications and refining prescribing oversight.