Determinants of Early Non-Persistence to Medication Therapy for Chronic Diseases
Date
2025-03-04
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
ORCID
0000-0003-2298-9000
Type
Thesis
Degree Level
Masters
Abstract
Abstract
Background:
Medication non-persistence poses significant challenges in chronic disease management, particularly for conditions such as hypertension, diabetes, and dyslipidemia. This study aimed to explore predictors of non-persistence using two approaches: a subjective framework guided by the WHO (World Health Organization) theoretical framework or a factor analysis-based data-driven approach. Additionally, we sought to identify upstream influences of negative beliefs and side-effect expectations.
Methods:
A cross-sectional survey of 3,029 eligible respondents was conducted using a self-reported questionnaire capturing various predictors of medication adherence. Survey items were organized into domains using both a subjective-driven approach and a data-driven factor analysis approach. Logistic regression and structural equation modelling (SEM) were employed to evaluate associations between predictors and medication non-persistence.
Results:
Using factor analysis to organize variables did not improve prediction performance for non-persistence compared to a subjective approach guided by the WHO framework. We identified key upstream influences affecting non-persistence, particularly focusing on negative beliefs and side effect expectations. Poor healthcare provider support and low perceived knowledge were strongly associated with negative beliefs, though neither directly predicted non-persistence in the multivariable model. The factors associated with side effect expectations revealed complex patterns. Expecting side effects showed a protective effect against non-persistence when incident side effects occurred; however, it was also associated with negative beliefs in the SEM analysis.
Discussion:
The similar performance of prediction models created with subjective organization of survey items versus factor analysis affirm the validity of the WHO framework to guide appropriate groupings of risk factors. Factor analysis, despite its objectivity in grouping correlated items, resulted in domains that were less theoretically coherent compared to the subjective framework without any significant improvement in prediction performance.
Poor health care provider (HCP) support and low knowledge were associated with negative beliefs, suggesting a possible pathway leading to non-persistence. However, definitions of knowledge and HCP support lack standardization in the research field, so generalizations should not be made without scrutiny. Also, our study found apparent contradictions where side effect expectations was associated with negative beliefs but also high HCP support. As a result, we believe the reason for both positive and negative associations with persistence is due to a lack of specificity in questions asking patients whether they expect side effects. Further study is needed to ensure these questions are able to discriminate between underlying patient motivations, with some expectations stemming from preparedness and others from pessimism.1,2
Description
Keywords
Non-Persistence, Chronic disease, Medication Adherence, Side-effect expectations, Belief, Structural equation modeling, Factor analysis, knowledge, side effects, Hypertension, Diabetes, Dyslipidemia
Citation
Degree
Master of Science (M.Sc.)
Department
Pharmacy and Nutrition
Program
Pharmacy