Explorative scoping review and bibliometrics analysis of big data applications for medication adherence

Objectives: As part of a wider project on approaches to medication adherence in chronic diseases, this exploratory study set out to map the methods reported in relevant quantitative research, including any involving ‘big data’, as well as to summarize evidence of intervention efficacy. The aim was to assess whether big data approaches have a role in improving knowledge about patterns of medication adherence.
Methods: Using an adapted version of Arksey and O’Malley’s framework for scoping reviews, the Scopus database was interrogated to identify, chart and summarize studies on medication adherence. Bibliometric analysis was then undertaken to map the evolution of this literature over time, and to chart the concepts represented in this knowledge domain.
Results: 533 articles were retrieved from the Scopus academic database, of which 61 met the inclusion criteria. 13 studies (21%) were Randomized Controlled Trials, 12 were retrospective cohort studies and 5 were prospective cohort analyses. The most adopted statistical methods were regression (multivariate and univariate), used in 51% of the studies. The Morisky adherence scale (36%) was the most widely adopted measurement tool and cardiovascular disease/hypertension was the most investigated condition (38%). No studies using advanced data mining techniques to study medication adherence in chronic conditions were found. Bibliometric analysis of the medication adherence literature showed an average of 6.7 citations per article. The most prolific countries were the USA with 225 citations and China with 40 citations. Analysis of key-words in article titles and abstracts showed patients’ beliefs and preferences as a key theme and a worthwhile area of investigation.
Conclusions: The use of big data techniques to understand medication adherence is still under-researched. A new framework for classifying methods, measurement tools and key variables in medication adherence research is proposed, along with recommendations for new studies to better understand adherence patterns in big data and how to translate these into actionable interventions.