Discussion
This overview is the first to summarise both access and acceptance-related barriers to childhood vaccination from across a diverse range of separate systematic reviews. It is evident that the issues around childhood vaccination are multilayered and complex, with all the included reviews reporting multiple factors that require consideration by policy makers, public health services or immunisation providers. However, each review was also selective in its scope, as evidenced by the low degree of overlap between the reviews. This highlights the value of this overarching summary. Individual reviews reported on a limited range of barriers relevant to a specific population, setting or vaccine of interest. For example, most reviews and included studies were from HIC settings, with only one review from an LIC setting. Less than half of the included reviews reported barriers from all six of the overarching categories identified. Furthermore, only one review reported an empirical association with vaccine uptake—a critical gap in the literature.
The purpose of identifying and categorising these barriers is to bring attention to the diverse range of barriers that need to be considered when attempting to diagnose the reasons for under-vaccination in a particular population. There are many models of health behaviour that focus primarily on the internal decision-making process that leads to intention and then behaviour, such as the Health Belief Model or Theory of Planned Behaviour.45 46 These models incorporate psychological variables related to behaviour, but they do not address the myriad external factors that also impact behaviour in the real world. Nevertheless, they have been used to inform the development of instruments intended to measure vaccine hesitancy and predict vaccine uptake. Notable examples include the Parent Attitudes toward Childhood Vaccination scale,47 48 or the Vaccine Confidence Scale.49 These instruments measure the internal decision-making factors highlighted in the Health Belief Model and Theory of Planned Behaviour, such as perceived benefits of the vaccine and perceived risk of the disease. However, like the models themselves they are somewhat limited in their ability to predict actual vaccine uptake because they do not consider external factors related to access. As this overview demonstrates, there are many reasons a parent or individual might have trouble accessing vaccines, from time constraints to transportation difficulties to poor quality health facilities. By developing a comprehensive instrument that can measure both acceptance and access barriers, we can begin to target interventions effectively.
The comprehensive barrier list identified in this overview will be used in the development of such an instrument—the National Health and Medical Research Centre-funded VBAT. The VBAT will be able to diagnose both access and acceptance-related reasons for under-vaccination for children under 5 years and will be validated for use in Australia and New Zealand, with possible expansion to other settings in the future. A version of the VBAT will also be developed separately for use by Indigenous and Maori parents, led by Indigenous researchers and community advisors. In this overview and in the VBAT itself, we have categorised factors that can inhibit vaccine uptake as barriers, but most barriers could also be framed as facilitators. For example, a poor clinical encounter with a judgemental provider is a barrier to vaccination, while a positive clinical encounter with an empathetic provider can drive uptake. Users of the VBAT can be flexible in how they decide to frame the key issues they identify with this tool. Our decision to choose the barrier frame is purposeful because it is typically how health policy and public health interventions are targeted—to address barriers.
Although this overview captures and organises all potential barriers to childhood vaccination described in the literature, it is only one step in the development and validation of the VBAT instrument. We also wanted to ensure that the VBAT included theoretical barriers that may not have been described in existing literature. Therefore, in addition to thematically grouping the barrier descriptions into categories in this overview, we also mapped the barrier descriptions against two theoretical models or frameworks: the COM-B (Capability, Opportunity, Motivation, Behaviour) model50 and Theoretical Domains Framework (TDF).51 These models incorporate both psychological and practical factors impacting behaviour, though they are not specific to vaccination. This additional theoretical mapping process, described elsewhere,52 allowed us to identify potential gaps where barriers might exist, despite not appearing in the literature. This process also means that the thematic groupings of barriers that we applied in this overview to facilitate comparison and summarisation do not represent the factor groupings of the eventual VBAT instrument. The final VBAT will be theoretically informed, with conceptually distinct categories. Following this comprehensive barrier identification and mapping process, we generated survey questions related to every barrier. We tested and refined these questions, along with a range of response options, through cognitive interviews with a purposive sample of potential survey participants in Australia and New Zealand. Finally, we will undertake a three-part psychometric evaluation process to reduce the items and assess dimensionality, internal consistency, construct validity, and test–retest and predictive validity.
Despite our rigorous methodological approach, this overview has several limitations. As with any overview of reviews, the evidence available for our analysis depended on the original review’s focus and review author’s interpretations of primary studies. On occasion, particularly for qualitative data, it was not clear whether the data came from the primary studies or were subject to author interpretation. We acknowledge that due to the subjective nature of thematic coding, some barriers and categories in our conceptual framework may have been grouped differently by other authors. The additional mapping to theoretical models, described above, helps address this potential bias. Finally, we acknowledge that quantitative frequency counts (ie, the number of reviews contributing data on each barrier) are not traditionally compatible with qualitative coding. However, we believe these frequency counts support our aim to describe the range of review evidence on this topic and highlight the most frequently reported barriers in the existing literature. The frequency of reporting of different barriers should not be assumed to be reflective of the importance or prevalence of this barrier. In primary quantitative survey studies in particular, participants may only be able to select from a predefined list of barriers. Qualitative data are less constrained by such predefined limits, and therefore a strength of this review is its inclusion of reviews of mixed-methods studies. Finally, 67% of the included reviews were from HIC only, with only six reviews from LMIC settings. While this means more individual barrier descriptions were identified from HIC settings, every barrier category included data from LMIC settings as well.