Malaria is a significant public health challenge, especially in poor-resourced areas and has dire consequences for individuals, especially children and pregnant women, in endemic countries. In 2021, of the 619,000 malaria case fatalities that occurred globally, 77 percent involved children aged below five years [1]. In Ghana, malaria is an endemic disease and puts the entire population at risk [2]. It remains a leading cause of morbidity and mortality in children, accounting for more than a quarter of all visits to Ghanaian hospitals and more than half of all outpatient visits in the country [3, 4]. Consequently, malaria remains a significant public health concern despite the availability of effective treatments and prevention strategies [5].
Several malaria control initiatives have been implemented in sub-Saharan Africa (SSA). In Ghana, interventions such as targeted indoor residual spraying (IRS), long-lasting treated Nets (LLINs), intermittent preventive treatment for pregnant women (IPTp), malaria monitoring, and malaria chemotherapy have been explored [6]. Vaccines are among the most effective and cost-effective treatments for reducing childhood mortality across low- and middle-income countries, including those caused by malaria [7]. The RTS, S/AS01 (Mosquirix) vaccine, primarily targeting children in SSA, was endorsed by the World Health Organisation (WHO) in 2021 after Phase III clinical trials demonstrated its efficacy in reducing malaria in children [8]. This vaccine has been a significant milestone in the fight against malaria, with trials beginning in 2009 [9]. To ensure the efficacy and safety of RTS,S malaria vaccines, several clinical trials have been conducted by the WHO and other agencies [10–12].
The Malaria Vaccine Implementation Programme (MVIP) in Ghana was first introduced through the Ministry of Health and the Ghana Health Service in some districts in the Brong Ahafo Region (now the Bono, Bono East and Ahafo Regions), Central, Volta (now Oti and Volta) and Upper East Regions [13]. The MVIP recommends administering four doses of the RTS, S/AS01E vaccine to children between the ages of 6 and 24 months old [14]. To successfully ensure the enrolment and uptake of the malaria vaccine, the Ministry of Health and Ghana Health Service linked the implementation of the RTS, S vaccine to the Expanded Programme on Immunisation (EPI), which has been enacted since 1976. When it comes to the EPI, Ghana has a solid record [15], but the country faces significant difficulties with continuity. Especially after the first year, the risk of default increases. According to Mekuria, Hailu, Bedimo and Tefera [16], vaccination default is a situation whereby a child skips a minimum of one dose from a recommended vaccination schedule. A high default rate could jeopardise the expansion of the MVIP and to create efficient implementation strategies, it is crucial to comprehend the dynamics and causes of such defaults.
While achieving high vaccine coverage is pivotal for disease prevention and control, such as with malaria, understanding the determinants of vaccine coverage is essential for crafting effective public health strategies. Previous research [17–21] have highlighted factors like education, occupation, parental perceptions, and vaccine-related febrile reactions as pivotal in influencing the uptake of the RTS, S vaccine. Similarly, Yeboah, Owusu-Marfo, and Agyeman [22] underscored the importance of health education and trust in vaccine efficacy. However, a gap remains in the literature regarding the predictive power of different classification models for vaccine defaulting. Our study aims to fill this void by employing and evaluating three classification models—binary logistic regression, decision tree, and random forest—to ascertain which most accurately predicts vaccine defaulting and its influencing factors. This methodological comparison, focused squarely on the prediction of vaccine defaulting, aims to build upon the foundational work of previous studies by offering a novel analytical perspective on the factors affecting RTS, S vaccine uptake. By doing so, we not only address the limitations identified in prior research but also expand the discussion to include an evaluation of predictive models, thus enriching the understanding of vaccine coverage dynamics.
The analytical approach adopted by this study not only addresses urgent issues in a timely manner but also helps to pinpoint more critical factors in comparison to others. Furthermore, considering that the RTS, S vaccine is integrated within the EPI schedule, we postulate that the factors that lead to defaulting for the RTS, S vaccine may parallel those of other multiple-dose vaccines within the same schedule, thus enhancing the applicability of our findings across various immunisation programs.