Classification and visualization of longitudinal patterns of medication dose: An application to interferon‐beta‐1a and amitriptyline in patients with multiple sclerosis

Abstract

Purpose Describing patterns of use, including changes in dose and interruptions is challenging. Group‐based trajectory modelling (GBTM) can be used to identify individuals with similar dose patterns. We provide an intuitive graphical representation of dose patterns in groups identified using GBTM. We illustrate our approach using two drugs with different combinations of available dosages. Methods We drew data on patients with MS followed from 1977 to 2014 in Montréal using two sub‐cohorts of subjects. A sub‐cohort of patients taking interferon‐beta‐1a and another of patients taking amitriptyline were identified from the initial cohort. We use GBTM to identify groups of patients with homogeneous dose patterns for each of the two drugs. We compared the graphical representation obtained from the fitted values of GBTM with our proposed approach, which consisted of using step functions whose values corresponded to the mode. Differences in characteristics across groups were identified using chi‐squares and analysis of variance, both weighted by the posterior probability of group membership. Results Seven patterns of dose were identified for interferon‐beta‐1a and five for amitriptyline. The graphical representations of the patterns of dose from GBTM included values outside of the prescribed doses and did not capture changes in dose as clearly as the proposed representation using step functions. Conclusion Our proposed approach which is based on the mode at each visit in each pattern provides an intuitive and realistic representation of dose patterns in groups identified with GBTM.

Physical and Mental Health–Related Quality of Life Trajectories Among People With Multiple Sclerosis

“…Our finding of distinct clusters of people with MS who report similar physical or mental HRQoL trajectories is consistent with previous studies that found distinct trajectories among people with MS for outcomes such as cost of illness, work productivity, disability progression, and healthcare use before nursing home entry. [37][38][39][40][41][42][43][44] Collectively, this highlights the importance of accounting for heterogeneity when evaluating MS outcomes.…”

Assessing the performance of group‐based trajectory modeling method to discover different patterns of medication adherence

“…Diverse applications of GBTM, highlighting its significance in healthcare, are evident in the literature. 9,[13][14][15][16] Alternatively, we may use k-means for longitudinal data (KML). 17 KML has the ability to recover latent subgroups by using metrics such as the Euclidean distance to assign individuals to trajectory groups.…”

“…It is useful in summarizing developmental patterns of a time‐varying phenomena, 12 achieved by clustering similar profiles into homogeneous subgroups. Diverse applications of GBTM, highlighting its significance in healthcare, are evident in the literature 9,13–16 . Alternatively, we may use k‐means for longitudinal data (KML) 17 .…”