A systematic review of parental vaccine hesitancy and refusal in childhood vaccination

Background Acceptance of vaccines is an important predictor of vaccine uptake. This has public health implications as those who are not vaccinated are at a higher risk of infection from vaccine preventable diseases. We aimed to systematically review studies of parental attitudes and beliefs in childhood vaccination, with a focus on the methods used to measure hesitancy and refusal. Methods We identied and reviewed primary research studies using quantitative methods and investigating vaccine attitudes and beliefs published between January 2012 and May 2018. Studies were included if they involved a quantitative survey of the attitudes and beliefs of parents about vaccinations recommended for children. We undertook a narrative synthesis of the results with a focus on evaluating variation in the use of behavioural theories, validated survey instruments, and localisation and adaptation of questions to suit local context. Results A total of 116 studies met the inclusion criteria; 99 used a cross sectional study design, 5 used a case control study design, 4 used a pre-post study design and 8 used mixed methods study designs. Sample sizes of included studies ranged from 49 to 12,259. Thirty-six countries were represented in the included studies. The most commonly used tool was the Parent Attitudes about Childhood Vaccines (PACV) Survey. Questions eliciting vaccination attitudes and beliefs varied widely. Conclusions There was heterogeneity in the types of measures used in studies investigating attitudes and beliefs about vaccination in parents. Broader and more consistent use of validated survey instruments for measuring parental attitudes and beliefs about childhood vaccination would help to better understand localised differences in the reasons for vaccine hesitancy and refusal.


Introduction
Childhood vaccination rates vary widely by country and region, and the reasons for these variations are likely to be context-speci c (1)(2)(3). While access to vaccination is a perennial challenge, acceptance also remains an issue of importance to uptake (4). Acceptance of vaccines is important to health because problems with acceptance can have a substantial impact on vaccination coverage and risk of outbreaks, particularly in high income settings (5). Not only are unvaccinated individuals at higher risk of infection and adverse health outcomes, but under-vaccinated populations are at higher risk of more severe outbreaks (6-8).
A range of questionnaires have been developed and tested for measuring vaccination attitudes and beliefs (9).

Inclusion criteria
Studies were included if they were quantitative primary studies investigating parental vaccine attitudes and/or beliefs; regardless of whether they considered one or a combination of vaccines or vaccine-preventable diseases; and published after January 2012.
Studies were excluded if they investigated vaccination barriers not associated with attitudes or beliefs (e.g. measuring access other than as a factor affecting convenience); adult and adolescent vaccination; or if they were not reported in English. We applied no geographical constraints.
All titles and abstracts or executive summaries found through the search strategy were screened independently by two authors (Adam Dunn and Amalie Dyda) to determine if they were relevant to the review. The full text of those articles that appeared to meet the inclusion criteria were retrieved and reviewed for relevance independently by the same two authors. The reference lists of all included items were searched to identify any additional items for inclusion.
Data extraction and synthesis Page 5/29 Data were extracted by one author (Amalie Dyda) and confirmed by a second author (Adam Dunn). A standard data extraction form developed by the authors was used. For each study, study design information extracted from the articles included the method of recruitment and the location and type of participants; the number of participants recruited (and completing the study, where appropriate); the vaccine or set of vaccines of relevance to the study; and details of the questions used to measure attitudes and belief about vaccination including any description of behavioural theories used to inform the questionnaire design, and whether the questions were taken directly or adapted from existing instruments.
Data extracted from each study were tabulated and grouped by study type and study characteristics including sample size, recruitment method, and location. We also extracted information about the use of behavioural theories and the use or adaptation of validated questionnaires. We summarised associations between vaccine hesitancy and intentions or status by country and questionnaire to determine whether there were any consistent differences by country after accounting for differences in questionnaire designs.

Results
The initial search strategy returned 41,570 titles and abstracts, of which 23,201 were removed as duplicates. Title and abstract screening identified 673 full text items for review.
Of these, 116 met the inclusion criteria ( Figure 1). A review of the included articles' reference lists did not identify any additional items for inclusion.

Summary of included studies
Of the included studies, 99 (85.3%) used a cross sectional study design (Table 1) Thirty-four countries were represented in the included studies ( Figure 2). The most common country in which studies were conducted was the United States (n=36), followed by Canada (n=9) and the United Kingdom (n=8). When aggregated by the number of participants, the United States included the largest number (40,155 participants), followed by Canada (7,200 participants), and the United Kingdom (3,273 participants).

Use of questionnaires and survey instruments
The questions asked of participants varied substantially across the set of included studies.
There was heterogeneity both in terms of the specific questions asked of participants as well as the provenance of those questions in theory or from standardised questionnaire sets.
The most commonly used standard questionnaire was the Parent Attitudes about Childhood Vaccines (PACV) Survey Tool, used in 4 studies with its full format with 15 questions (15)(16)(17)(18). In some studies, the PACV questions were adapted to match the local context or study population, such as in Malaysia (17)  Among the 16 (13.8%) studies in which parents were directly asked about whether they would have children vaccinated for all childhood vaccines, the percentages ranged from 75% in a study involving 200 parents in the United States (56) to 98% in a study involving 54 parents in Canada (31). For the 9 (7.8%) studies that asked about intentions in relation to influenza vaccination, the percentages ranged from 29% in a study involving 236 parents in Canada (57) to 92% in a before and after study at a clinic involving 5,284 and 5,755 different groups of parents in rural Kenya (58).

Associations between attitudes and beliefs with vaccination intentions or vaccination status
We identified 13 (11.2%) studies reporting on the association between attitudes and beliefs with vaccine intentions. All of these studies identified a significant association between attitudes and beliefs with intentions. In 8 of the 13 studies, associations were found for questions related to the severity of the disease (18, 21, 30, 32, 38, 59-61), and these studies were conducted in the United States (n=3), the Gambia, Israel, Germany, the United Kingdom and Sweden. In 7 of the 13 studies, associations were found for questions related to the susecptability of the disease (18, 21, 30, 32, 38, 45, 53), and these studies were conducted in the United States (n=2), the Gambia, Israel, the Netherlands, the United Kingdom and South Korea. In 4 of the 13 studies, associations were found for questions related to the efficacy of the vaccine (21, 43, 61, 62), and these studies were conducted in the United States, Canada, Italy and Sweden. In 3 of the 13 studies, associations were found for questions related to safety of the vaccines (18, 60, 62), and these studies were conducted in the United States (n=2) and Italy.
Among the 116 included studies, 57 (49.1%) reported on the association between attitudes and beliefs with vaccination status. All studies identified an association between attitudes and beliefs with vaccination status, with concern about safety and efficacy the most

Discussion
There was little consistency in the provenance of the questions used to measure attitudes and beliefs across studies. For theoretical frameworks, we found that the HBM was most commonly used to support the development of questionnaires, which was consistent with previous reviews (10). The HBM posits that perceptions of susceptibility, severity, benefit and barriers, cues to action and self-efficacy predict behaviour. This and other models place emphasis on risk appraisals as important predictors of vaccination. Use of the HBM is complicated by the fact that all related perceptions could apply to vaccination uptake as much as disease outcomes. Since these models look at individual psychological factors by design, they are weaker at measuring other factors like false contraindications, social influence, or access to services or vaccines. Further, many models fail to measure trust, yet trust in vaccination arises as a relevant phenomenon in both qualitative accounts of undervaccination and the influence of vaccine safety scares (90). Trust is often "ill-defined and a loosely measured concept" (91). Recent work on the moral foundations of behaviour suggests that measuring constructs such as contamination and liberty are also relevant (92,93). Further work is needed to incorporate moral foundations, other feelings and attitudes and beliefs, trust, and practical barriers into a single model of vaccination behaviour and test its robustness.
The geographical distribution of primary studies included in the review was generally consistent with a previous review on attitudes and beliefs regarding vaccination (10) (94). The potential for mainstream media, social media, and other sources of misinformation to introduce or exacerbate concerns in the community may outpace our collective ability to measure and report on attitudes and beliefs, which may hinder our ability to support the design of evidence-informed and localised interventions for debunking or mitigating the impact of misinformation. Hence, there is a need to identify attitudes and beliefs with methods that survey individuals but also with methods that rapidly collect data in real-time using novel methods.
There were two main limitations to the review approach and conduct. The first limitation was that the geographical distribution of the studies included in the review may be biased by the exclusion of studies not written in English. Second, we did not undertake a metaanalysis to compare results across studies because of the heterogeneity of the questions asked in each of the included surveys, which limits our ability to summarise associations between attitudes and intentions or status within or across countries.

Declarations
Ethics approval and consent to participate Not applicable.

Consent for publication
Not applicable.

Availability of data and materials
Not applicable.

Competing interests
The authors declare that they have no competing interests.      Among the set of 116 included studies, 34 countries were represented.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. App2.docx