Understanding the public’s decision-making about seasonal flu vaccination during a pandemic: Application of the precaution adoption process model

Factors influencing flu vaccine uptake

Previous research highlights that vaccine acceptance is a complex decision-making process influenced by a wide range of psychological, health, and socio-demographic factors (Butter et al., 2022; Goffe et al., 2024). Seven psychological and behavioural factors have been most frequently and consistently identified as being significantly associated with flu vaccine intention and uptake (Bish et al., 2011; Lorenc et al., 2017; Maltezou et al., 2010; Schmid et al., 2017). These include: perceived susceptibility to disease (not perceiving oneself as vulnerable to the disease) (Han et al., 2016), perceived severity of the disease and perceived benefits of vaccination, attitude towards flu vaccination, vaccine-related side effects (Nagata et al., 2013), social norms (i.e. the perceived pressure or expectation that significant others would want them to get the vaccine) (Quinn et al., 2017a), anticipating feelings of regret (i.e. the impact of missing a vaccination opportunity) (Gallagher and Povey, 2006), and previous flu vaccinations (Gidengil et al., 2012; Sherman et al., 2021a). These factors represent key constructs in theories of behaviour change such as the Theory of Planned Behaviour (TPB) (Ajzen, 1991) and Health Belief Model (HBM) (Rosenstock, 1974; Rosenstock et al., 1988). Contextual factors include physical barriers such as access, vaccine availability, and previous vaccination experiences (Boey et al., 2018; Gallant et al., 2023). Importantly, socio-demographic factors such as age (Abbas et al., 2018), gender (Mo and Lau, 2015a), educational level (Takayama et al., 2012) and ethnicity (Bish et al., 2011; Quinn et al., 2017b) have also been associated with flu vaccine intention and uptake.

Theoretical framework

The WHO Scientific Advisory Group for Emergencies (SAGE) Working Group on Vaccine Hesitancy defines vaccine hesitancy as a continuum ranging from complete refusal of all vaccines, delay or refusal of some vaccines but acceptance of others, to acceptance of all vaccines and defines it as a delay in acceptance or refusal of vaccination despite availability of vaccination services (Larson et al., 2014; MacDonald, 2015). Thus, people may go through various stages of decision-making before having a vaccine which can be best captured using a stage of change-based framework. The Precaution Adoption Process Model (PAPM) explains how individuals make decisions to adopt new precautions or cease risky behaviours, a process that involves conscious, deliberate steps (stages) rather than automatic habits, like for example, adopting physical activity habits (Weinstein et al., 2020). Stages are theoretical constructs, with each representing an ‘ideal’ prototype, that not all individuals align with perfectly. Unlike theories that view behaviour as a binary decision to act or not act, stage theories, like the PAPM, typically assume that people move through a series of different stages before taking action and that predictions about behaviour require different factors to be in place for each stage. However, the progression is not always straightforward and can vary, with individuals spending different amounts of time in each stage, and some may decide not to act at all. This variability reflects the complex nature of decision-making and behaviour change (Weinstein et al., 2020). The PAPM postulates that adopting preventive health behaviour involves six distinct sequential stages of decision-making: (1) being unaware of the health behaviour; (2) being unengaged in the decision; (3) being undecided about whether to act or not; 4() deciding not to act; (5) deciding to act (intending) and (6) acting (having the vaccination) (Weinstein et al., 2008; Weinstein and Sandman, 1992). The PAPM’s detailed breakdown – ranging from being unaware to taking action or deciding not to – offers a nuanced understanding of flu vaccination adoption. This stage-based approach enables a more precise identification of barriers and facilitators at each step. However, the PAPM complements other models, as it does not provide a predetermined set of variables for distinguishing between stages. Instead, individuals’ progress through the PAPM stages is based on their attitudes and beliefs (Elliott et al., 2007). To enhance this understanding, constructs from other models, such as perceived benefits, control, susceptibility, and disease severity from the HBM and the extended TPB, can be integrated to better understand how individuals transition between stages. Previously, this model has been applied to human papillomavirus (HPV) vaccination decision-making (Shapiro et al., 2018; Waller et al., 2020) and more recently to COVID-19 vaccination (Meyer et al., 2023), but not, to our knowledge, to flu vaccination. An understanding of factors associated with each stage of flu vaccine decision-making can inform the design of more tailored interventions at different stages to increase flu vaccination rates ahead of future flu seasons and pandemics. This targeted approach can enhance the effectiveness of public health campaigns aimed at promoting flu vaccination. Importantly, the PAPM considers the dynamic nature of behaviour change by acknowledging that individuals may progress through different stages over time. This sequential perspective is crucial for understanding the evolving nature of flu vaccine decision-making, especially in response to changing public health contexts.

Ethics

Ethics approval was obtained from Newcastle University Ethics Committee (Reference: 13754/2020). Informed consent was obtained from all participants involved in the study.

Design

This was an exploratory cross-sectional study about seasonal flu vaccination, conducted as part of a larger study examining factors influencing vaccine decision making for both the COVID-19 booster vaccine (Meyer et al., 2023) and seasonal flu vaccine in the second year of the Covid-19 pandemic. We administered an online, cross-sectional survey to people living in England approximately 4 weeks after the COVID-19 booster vaccine programme commenced and the seasonal flu vaccine programme was underway (October 11–20, 2021) (Department of Health & Social Care, 2021; Public Health England, 2021).

Participants

Participant inclusion criteria aligned with age-based vaccine eligibility criteria at the time we conducted the study: participants needed to live in England, be 50 years of age or older, and have previously completed the primary course of the COVID-19 vaccine (i.e. received two doses). Quotas were used to ensure a nationally (English) representative sample in terms of gender and region. Participants were recruited using Qualtrics panel recruitment. Participants who completed the survey were incentivised through Qualtrics automatic rewards-based programme as reimbursement for their time.

Materials

The online questionnaire was adapted from one previously used to investigate seasonal flu vaccine intention (Antonopoulou et al., 2022) and was informed by the PAPM (Weinstein and Sandman, 1992), the extended Theory of Planned Behaviour (Ajzen, 1991) and the Health Belief Model (Rosenstock, 1974). Additional questions asked about past flu vaccine experience and personal health and socio-demographic characteristics. All questions are available in Supplemental File 2.

PAPM stage for seasonal flu vaccination

Participants were asked ‘Which of the following best describes your thoughts about having a seasonal flu vaccine this year?’ and were asked to select one of six options that related to Stages 2–6 of the PAPM (see Supplemental File 2). Options pertaining to Stages 1 (unaware) and 7 (maintenance) were omitted because the seasonal flu vaccination programme was well established and therefore, it was highly unlikely that participants would be unaware of the COVID-19 and the flu vaccination programmes. Similarly, Stage 7 (maintenance) was not included as although the behaviour can be repeated annually, it is not a behaviour that needs to be maintained. Omitting Stage 7 streamlined our survey to concentrate on stages characterised by more active decision-making. Overall, our targeted approach was intended to ensure that our research captured the more pronounced elements of flu vaccine decision-making without unnecessary redundancy on stages where awareness is reasonably assumed (Stage 1) or maintenance is inherently straightforward (Stage 7).

Previous vaccine experience

Questions focussed on individuals’ previous flu vaccination behaviour (2 items) and if applicable, their overall experience of receiving their previous flu vaccine (4 items). All items were measured on a 5-point Likert scale, with higher scores representing higher frequency or more positive experiences. A mean subscale score was computed for previous flu vaccine experience as the scale was found to have good internal consistency (Cronbach’s alpha = 0.77).

Beliefs and attitudes about seasonal flu vaccination

Questions informed by the extended TPB pertained to vaccine attitudes (2 items), vaccine subjective norms (4 items), vaccine perceived control (1 item), and anticipated regret (3 items). Questions informed by the HBM represented perceived severity (1 item), perceived susceptibility (2 items), perceived benefits (5 items), and perceived vaccine safety (2 items). Additional questions asked about knowledge of vaccine safety and effectiveness (3 items), trust in Government (1 item; adapted from (Meyer, 1988)), and fears of having a vaccine (4 items; (Myers and Goodwin, 2011)). Survey questions were derived from published studies that also examined the psychological determinants of vaccine intention (Hamilton et al., 2021; Myers and Goodwin, 2011; Sherman et al., 2021b; Ziarnowski et al., 2009). All items (except ‘fears of having a vaccine’) were measured on a 5-point Likert scale, with higher scores representing higher levels of agreement/more anticipated regret. All constructs aside from ‘perceived susceptibility’ had good internal consistency (Cronbach’s alpha 0.69–0.96) and were represented as a single mean score.

Personal health characteristics and socio-demographic characteristics

Participants were asked to indicate their age, gender, region, ethnicity, education status, employment status, keyworker status, household income, general health, and whether or not they were asked to shield from COVID-19 during the pandemic.

Patient and public involvement

The survey was reviewed twice by our Unit’s dedicated Patient and Public Involvement (PPI) team (N = 6, five aged 50 and over, consistent with target population) to check the relevance and understandability of survey items and to ensure the online survey was easy to use.

Data analysis

STATA (version 16) was used to carry out all analyses. Univariate associations between PAPM stage and all potential variables were first examined using univariate, multinomial logistic regression models and significant variables (p < 0.10) were considered for inclusion in the multivariate model. Multicollinearity was checked by inspecting variation inflation factor (VIF) values; all were less than 3.0 indicating minimal evidence of multicollinearity. Pairwise correlations were computed between all continuous variables; the variable ‘vaccine attitude’ was dropped because it was highly correlated with ‘perceived safety’ (r = 0.719) and ‘subjective norms’ (r = 701). All other correlations were less than 0.7 (see Supplemental File 3). Two variables were excluded from multivariate modelling due to high numbers of missing data, including overall experiences with past flu vaccination (n = 1460; 27% missing data) and fear of vaccine side effects lasting longer than 2 days (N = 1563; 22% missing data). A multivariate, multinomial logistic regression model was fitted to the data to represent factors associated with each stage of the PAPM model; participants in PAPM Stage 6 (vaccinated) served as the reference group and was compared to participants in Stage 2 (unengaged), Stage 3 (undecided), Stage 4 (decided not to vaccinate), and Stage 5 (decided to vaccinate). Backward stepwise regression was used to simplify model selection by identifying the most relevant predictors in a data-driven way, which aligns well with the exploratory nature of our analysis. A conservative p-value of 0.10 was used for factor inclusion; however, statistical significance was inferred by p-values <0.05. The relative risk ratios (RRR) and 95% confidence intervals are reported; RRRs < 1 indicate a negative association with PAPM stage and RRRs > 1 indicate a positive association. Goodness of fit is represented by the pseudo r² value.

Comparison with COVID-19 vaccination

When examining factors associated with COVID-19 booster vaccine decision-making also using the PAPM model for another study focussing on COVID-19 booster vaccine with the same cohort of participants (Meyer et al., 2023) knowledge about vaccine safety and effectiveness and perceived susceptibility (i.e. perception that the immune system was strong enough to protect against COVID-19) were correlated with being ‘unengaged’ or ‘undecided’ about having a booster vaccine. When compared to the present study examining decision-making about the seasonal flu vaccine, it is clear that perceived susceptibility is an important factor for both vaccines; however, knowledge about vaccine safety and effectiveness appears less important once engaged in the decision-making process.

In contrast to the flu findings, only people who had not yet engaged in the decision-making process reported weaker subjective norms for the COVID-19 booster vaccine (Meyer et al., 2023), whereas weaker subjective norms were reported across all PAPM stages for the flu vaccine except for those who had decided to have the flu vaccine. Thus, creating societal expectations around vaccination appears warranted for individuals at all PAPM stages in relation to flu vaccination but only for engaging people in the decision-making process in relation to COVID-19 booster vaccination. A potential explanation for these differences may be that while the urgency and heightened risk perception surrounding COVID-19 led to rapid vaccine uptake in some populations, flu vaccination might involve different decision-making processes due to the lower perceived severity and urgency. This could help explain why individuals move through the stages of the PAPM differently for these two vaccines. Another possible explanation is that these differences reflect the distinct communication strategies used during the two vaccination campaigns, as well as the differing public health priorities during the recent pandemic, when health concerns about COVID-19 were more prominently prioritised. Interestingly, socio-demographic factors appeared to account for more variance in vaccine decision-making for COVID-19 relative to seasonal flu. Where being unengaged in the decision to have a COVID-19 vaccine was associated with being employed, not having a formal educational qualification, and having a household income less than £30,000; and being undecided was associated with being an ethnicity other than White (Meyer et al., 2023); younger age was the only significant corelate of being ‘unengaged’ or ‘deciding to vaccinate’ in relation to flu vaccination.

Policy implications

As the decline in antibodies against the influenza viruses over time necessitates annual revaccination, particularly for older adults and other at-risk groups, the need for successful public health campaigns every year is evident. The current findings can be useful at a strategic level for targeting groups identified as being at different levels of decision-making by highlighting which constructs are useful or which ones are under-utilised in the promotion of flu vaccine uptake. Previous studies conducted on people’s preferences for vaccination suggest that a more nuanced approach to the selection of persuasive campaign elements is the most effective strategy (Ling et al., 2019; Roller-Wirnsberger et al., 2021; Su et al., 2021). Our findings indicate that targeting vaccine knowledge may be most crucial among those not engaged in the decision-making process, whereas providing reassurance regarding flu vaccine safety and highlighting the benefits of vaccination may be most crucial among people who are undecided or who have decided not to vaccinate. Addressing social norms, anticipated regret, and perceived susceptibility could promote vaccine uptake among all people who are at earlier stages of decision-making, although could be most useful if targeting people who are unengaged or who have decided not to vaccinate. Importantly, however, flu vaccine frequency is negatively correlated with all earlier stages of decision-making indicating the need to focus attention on people who have never or rarely had the flu vaccine previously.

The application of the PAPM in this study provides a dynamic, stage-based understanding of vaccination decision-making, which contrasts with single-point theoretical models. By examining decision-making as a process with distinct stages, even if the boundaries between the stages aren’t always clear, this approach enables the identification of common barriers faced by people in the same stage and can help improve intervention effectiveness by designing targeted interventions. Future research could build on this methodological approach by applying the PAPM to other health behaviours or vaccination campaigns, allowing for a more comprehensive understanding of how decision-making processes evolve over time and in response to different public health communications.

Strengths and limitations

Some limitations of this study are important to consider when interpreting these findings. Firstly, this was an online cross-sectional self-report study and as such, while it facilitates timely data collection, this method may introduce biases in the data in terms of sampling representativeness (requiring internet access, being registered with the survey platform) and it reflects vaccination attitudes and behaviour at the specific time point of data collection. Therefore, future research with longitudinal designs would be crucial to validate and expand upon our findings. Longitudinal studies would allow for a deeper understanding of how temporal factors influence behaviour over time and can help confirm the robustness of the stage-based models such as the PAPM. Secondly, our sample was a predominantly white British sample (92% vs ca. 80% in the population of England based on the latest census (ONS 2011 Census aggregate data., 2021) and therefore more research is needed to better examine the ethnic disparities in flu vaccine uptake. Thirdly, it should be noted that the current data reflects the decision-making processes of a population during a pandemic situation and thus, the profile of PAPM stages may have been unusually influenced by this. However, as we still in the aftermath of the pandemic and the flu with likely co-circulate with COVID and other infectious diseases, key lessons can still be learned that can promote vaccine acceptance for flu in future years. The strength of this study lies in the well-validated theoretical constructs on which the survey questionnaire was developed, and the large dataset of responses collected, which allowed for advanced statistical analyses, from a wide sample across all geographical regions in England.