Discussion

We show two opposing roles for disgust on vaccine uptake, a direct positive effect combined with an indirect negative effect through Vaccine Attitudes, which in-turn influences the manifest behaviour in the absence of other predictors. Additionally, we found that those who perceived themselves as more susceptible to infection were more likely to be vaccinated, while no indirect effect of Perceived Susceptibility through Vaccine Attitudes was observed. Nonetheless, these effects were small and attenuated by economic, logistic and psychological determinants shown in prior analyses to be associated with vaccine uptake [Reference Luz, Johnson and Brown23]. In particular, the psychological constructs of perceived susceptibility to flu and perceived benefits of flu vaccine more than doubled the odds of flu vaccine uptake while the more generic assessment of Perceived Susceptibility had only a minor effect size. Moreover, exceeding the effect sizes of perceived susceptibility to flu and perceived benefits of flu vaccine, factual (economic and logistic) and perceived barriers to getting the flu vaccine were the strongest predictors of flu vaccine uptake. Our inclusion of these variables in the models considered only their independent effect on the outcome, future studies should explore alternative conceptual models. Furthermore, our conceptual model assumed that a person's general attitudes towards vaccines led to their behaviour during the 2016 flu vaccine season yet it is also plausible that one's experience with the vaccine might later inform their attitudes. However, a recent analysis of antivaccination attitudes in 24 countries suggests an interesting model for understanding the roots of such beliefs [Reference Hornsey and Fielding34] which, in the case of vaccines, are ‘conspiratorial beliefs, reactance and disgust/fear towards blood and needles’ [Reference Hornsey, Harris and Fielding21].

We found that the six items intended to measure attitudes towards vaccines adequately converged into one factor. Our means of measuring Vaccine Attitudes expressed positions towards vaccines’ efficacy and safety with those having more negative attitudes more frequently agreeing that ‘People get sick from vaccines’ and that ‘Vaccines weaken the immune system’. We also found that those with heightened Disgust Sensitivity or Germ Aversion tended to hold more negative attitudes towards vaccines. Prior research has found that adults scoring higher on disgust sensitivity were more likely to adhere to the misconception that vaccines cause autism and were more skeptical of the effectiveness and safety of vaccines [Reference Clay19]. In another study, greater disgust sensitivity was associated with antivaccination beliefs, support for organic foods and opposition to genetically modified foods [Reference Clifford and Wendell20]. A 2018 study of 24 countries found that those with antivaccination attitudes scored higher on conspiratorial theory, reactance and disgust towards blood and needles [Reference Hornsey, Harris and Fielding21]. These findings suggest that attitudes towards vaccination may be less subject to rational thinking and more driven on an emotional basis and, in particular, disgust. Accordingly, our mediation analyses’ results show that Disgust Sensitivity and Germ Aversion had a negative indirect effect on vaccine uptake which was mediated by Vaccine Attitudes. That is, heightened Disgust Sensitivity, when mediated by Vaccine Attitudes, leads to a decreased odds of vaccine uptake. This finding may help explain the contradictory findings observed among health care workers, a high-risk population with knowledge of disease transmission mechanisms and access to vaccines, who seldom achieve coverage targets without additional incentives [Reference FitzSimons35] or resort to litigation when faced with mandated vaccination [Reference Opel, Sonne and Mello13]. A qualitative review among health care workers in Europe suggested an emotional basis for decision making highlighting health care worker's concern with vaccine side effects as well as distrusts of pharmaceutical companies [Reference Karafillakis36].

A recent ecological study evaluated the association between a population's cultural and social norms assessed via voting in the US 2012 presidential election with uptake of routine adolescent vaccines at the state level [Reference Bernstein37]. Those authors found that states classified as ‘red’ (i.e. with Republican affiliation) had significantly lower vaccine coverage of human papillomavirus (HPV), tetanus-containing (Tdap) and meningococcal (MCV4) vaccines. Political ideology and attitudes towards a diverse set of policies and social norms have been shown to correlate with disgust sensitivity in different cultures [Reference Barrett, Lewis and Haviland-Jones14]. Though most studies have been conducted in countries where English is the primary language, a recent Dutch study corroborated the association of disgust with political ideology showing how heightened disgust sensitivity correlated with the socially conservative as well as with negative attitudes towards immigrants, gays and lesbians, and greater nativism and isolationism [Reference Brenner and Inbar38]. As such, disgust is involved in the avoidance of physical threats (pathogens) as well as perceived ones and vaccine attitudes may result from disgust's role in an individual's political, social and moral beliefs. In a study of prejudice against immigrants and foreigners, authors found the link between disgust sensitivity and prejudice was not due to concerns about disease acquisition but by ideological orientation [Reference Hodson and Costello39].

Notwithstanding, we also found that Disgust Sensitivity and Germ Aversion had a direct positive effect on vaccine uptake, i.e. those with heightened disgust sensitivity were more likely to get vaccinated. Indeed, as suggested in the introduction, disgust may have two opposing associations with vaccine uptake. Since vaccines prevent infection, those with heightened disgust sensitivity might vaccinate more; this likely reflects a rational decision-making process (our direct positive effect). Our results also show that those who perceive themselves more susceptible to infectious diseases, that is, who scored higher in Perceived Susceptibility, were significantly more likely to report following through on the expected manifest behaviour, that is, getting vaccinated. A 2017 systematic review that included 470 studies reported similar findings: low perceived risk of acquiring the disease was a significant barrier to vaccine uptake [Reference Schmid5]. Interestingly, we found no indirect effect (be it positive or negative) of Perceived Susceptibility on vaccine uptake that could be mediated by Vaccine Attitudes.

We also found that the effects of Disgust Sensitivity, Germ Aversion and Perceived Susceptibility were no longer statistically significant when we included in the models other covariates such as workplace availability of the vaccine, belonging to a priority group and perceived barriers of flu vaccine [Reference Luz, Johnson and Brown23]. These results suggest that although heightened disgust sensitivity may correlate with negative attitudes towards vaccines, as suggested by us and others [Reference Clay19, Reference Clifford and Wendell20], its impact on vaccine uptake is minimal. However, it is important to note that our inclusion of these variables as covariates in the regression models imply a specific causal structure where these covariates are interpreted as ‘confounders’ and the regression coefficient the ‘adjusted effect’ of the treatment variable. It may well be that other theoretical models could consider the interplay of these factors in different manners, thus highlighting a possible effect that was obscured in this analysis. Indeed, this is what we see in the unadjusted effects of Disgust Sensitivity and Germ Aversion: had we only considered their total effect, we would have concluded that there was none. However, when we separated their effects into a direct effect on the outcome and an indirect effect mediated by Vaccine Attitudes, we observed the two opposing effects.

Moreover, the lack of effect of Disgust Sensitivity on reported flu vaccine uptake in the model that considered other covariates may be due to our assessment of the manifest behaviour as specific to the influenza vaccine, that is, disgust may impact vaccine uptake differently depending on the infection/vaccine [Reference Wheelock40]. However, a 2017 systematic review of 145 European studies on vaccine perceptions and concerns for various vaccines (seasonal and pandemic influenza, HPV, MMR, others) found that safety, in particular, mistrust, was the largest deterrent to vaccine uptake in all countries, for all vaccines [Reference Karafillakis and Larson41].

Some of the limitations of the present study have been discussed previously [Reference Luz, Johnson and Brown23] and include the small, non-probabilistic nature of the participant selection, the MTurk workforce [Reference Chandler and Shapiro25]. Moreover, as with any survey, the data are self-reported and not verified, however, the MTurk population has been shown to be at least as honest and reliable as other survey respondents [Reference Chandler and Shapiro25]. As for external validity, our subjects were younger and with higher income and education that the general US population. In the present, though we used two validated instruments to measure Disgust Sensitivity, Germ Aversion and Perceived Susceptibility, we highlight that others are available such as the Three-Domain Disgust Scale [Reference Tybur, Lieberman and Griskevicius42] and that future studies could address the roles the different domains of disgust might have. Finally, we highlight that our six-item Vaccine Attitudes measure, though adequately consistent, was not validated and suggest, for future studies, the currently available measures of vaccine hesitancy particularly the 5C scale (Confidence, Constraints, Complacency, Calculation and Collective) given its comprehensiveness in considering all psychological antecedents of vaccination, both individual and social [Reference Betsch43].

As suggested in multiple studies, a natural and necessary next step is the development of means of communicating with individuals to promote acceptance, uptake and adherence to vaccine recommendations. Our findings suggest disgust might drive negative attitudes towards vaccines though to improve flu vaccine uptake it would be more effective to address economic, logistic and psychological barriers. A 2017 review that integrates diverse findings from psychology, sociology, behavioural economics, public health and medicine suggests a similar path, that which acknowledges and includes psychological theories to promote vaccine uptake [Reference Brewer44]. Prior trials unfortunately show how tricky message framing for vaccine uptake can be. A 2014 randomised online trial tested four message framing interventions, none of which increased intention to vaccinate [Reference Nyhan18]. Two 2015 studies found that correcting myths about vaccine's side effects (i.e. flu vaccine causes the flu [Reference Nyhan and Reifler10] and measles, mumps and rubella (MMR) vaccine causes autism [Reference Horne12]) had no effect on vaccine intention or attitude, respectively. A 2017 study on the effect of self-affirmation exercises found it to be ineffective or even detrimental when correcting information regarding vaccine safety was not jointly provided [Reference Reavis11]. To this effect, recent work by Hornsey et al. [Reference Hornsey, Harris and Fielding21, Reference Hornsey and Fielding34] shed new light into the roots of antivaccination attitudes while also proposing a new model for promoting vaccine acceptance that seems promising.