Predictive Modelling of ART Non-Adherence: Interactions of Patient, Health System, and Socio-Cultural Determinants in Rural and Urban Settings of Ghana’s Western Region
DOI:
https://doi.org/10.64261/ijaarai.v1n3.003Keywords:
ART adherence, predictive modelling, HIV, stigma, rural–urban disparitiesAbstract
Background: Achieving sustained adherence to antiretroviral therapy (ART) is critical for viral suppression and HIV epidemic control. In Ghana’s Western Region, rural–urban disparities in healthcare access, socio-cultural norms, and economic opportunities may differentially influence adherence.
Objective: To apply predictive modelling to identify patient, health system, and socio-cultural determinants of ART non-adherence in rural and urban settings of Ghana’s Western Region.
Methods: A cross-sectional study was conducted between March and August 2023 among 620 people living with HIV (320 rural, 300 urban) who had been on ART for at least six months. Data were collected through structured interviews and verified with pharmacy refill records. Independent variables included socio-demographics, travel time to clinic, stigma, disclosure status, appointment attendance, and provider communication. Binary logistic regression identified predictors of non-adherence (adherence <95%), with separate rural and urban analyses. Model performance was assessed using the area under the receiver operating characteristic curve (AUC).
Results: Overall non-adherence prevalence was 19.1%, higher in rural settings (23.7%) than urban settings (14.0%). Significant predictors of non-adherence included stigma (AOR 2.54, p < 0.001), missed clinic appointments (AOR 2.87, p < 0.001), travel time >60 minutes (AOR 2.26, p = 0.001), low education (AOR 1.95, p = 0.006), poor provider communication (AOR 1.74, p = 0.023), younger age (AOR 1.78, p = 0.015), and non-disclosure of HIV status (AOR 1.69, p = 0.031). The final model demonstrated good discrimination (AUC = 0.82).
Conclusion: ART non-adherence in Ghana’s Western Region is driven by a combination of structural and behavioural factors, with distinct rural–urban dynamics. Integrating predictive modelling into programme monitoring could enable early identification of high-risk patients, while geographically tailored interventions—such as decentralized ART delivery in rural areas, stigma reduction campaigns, and enhanced patient–provider communication—may improve adherence outcomes.
Keywords: ART adherence, predictive modelling, HIV, stigma, rural–urban disparities
