Vaccines remain an essential tool to contain the Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2) pandemic. Besides the Coronavirus Disease 2019 (COVID-19) vaccines that have been in use so far, new vaccines adapted to virus variants are currently being rolled out. In addition, more than 200 vaccine candidates are under development, including vaccines with new targets.1 Vaccines, however, can only be effective in keeping the immunity in the general population up if people get vaccinated.
More than two years into the pandemic, though, vaccination campaigns are facing the growing challenge of widespread vaccine fatigue.2 In most developed nations, only a minority of people has remained unvaccinated and a large majority has already received either one, two, three or more doses of a COVID-19 vaccine. For many, however, further booster vaccinations may soon be needed to restore waning immunity. To minimize the risk of upcoming infection waves, many countries have developed medium-term scenarios requiring either widespread or targeted booster vaccinations with updated vaccines. Yet, given high levels of vaccine fatigue, it is unclear to what extent these vaccination efforts can be effective in achieving sufficient vaccine acceptance to avoid putting health services under undue pressure.
This study aimed to provide evidence on how to reduce vaccine fatigue and increase the number of vaccinations. To do so, we designed two conjoint experiments portraying possible future scenarios and assessing their impact on vaccination intentions and related attitudes. Conjoint experiments3 are highly suitable to model the likely outcomes of alternative futures as they allow the researcher to manipulate multiple attributes of a scenario and measure the responses of participants considering all attributes jointly. To identify relevant attributes, we reviewed the literature on COVID-19 vaccine acceptance. In particular, properties of vaccines4–6, communication (e.g. campaign messages7–10, expert consensus11, celebrity endorsement12–14, costs/incentives14–17, and legal rules (e.g. vaccine passports17,18, vaccine mandates19,20) may matter. Results from earlier survey experiments on COVID-19 vaccine acceptance, however, have only limited applicability as they mainly focused on the initial vaccinations and circumstances no longer present in the current situation (e.g. vaccine envy). More evidence is therefore needed on how to overcome vaccine fatigue and keep levels of immunity up to date.
The conjoint experiments of our study were embedded in a cross-sectional online survey conducted in two European countries – Austria and Italy – simultaneously between July 19th and August 8th, 2022. Like many other developed nations, both countries had been experiencing stagnating vaccination rates. Both samples matched the target distributions regarding gender, age groups, regions, and education. Overall 6,357 respondents took part in the survey (Austria: n = 3,187; Italy: n = 3,170). Further details on the survey are provided in the Supplemental File 1.
The readiness to get vaccinated was measured on a scale from 0 to 10 and was higher in Italy (5.8 ± 2.6) than in Austria (5.3 ± 3.3), averaging across all scenarios in both conjoint experiments. Respondents in both countries reported high levels of pandemic fatigue (Supplemental File 2) and showed low to medium levels of trust in parliament and government (Supplemental File 3). 61% of respondents had already received three or more doses of a COVID-19 vaccine, 14% had received one or two doses, and 25% reported not being vaccinated. The triple vaccinated reported the highest likelihood to get vaccinated again across all scenarios with a mean of 6.6 (+/-2.5), those with one or two doses showed medium levels with a mean of 4.7 (± 2.6), whereas the unvaccinated were the least likely to declare vaccination readiness under any of the shown scenarios with a mean of 3.4 (± 3.0) (Supplemental File 4). The triple vaccinated differed from the other two groups exhibiting higher mean age, levels of education and trust in institutions (Supplemental File 3 and 4). Interestingly, pandemic fatigue was highest among those with one or two vaccinations (Supplemental File 2).
In experiment 1, we showed the respondents two alternate scenarios for a hypothetical vaccination campaign in fall. We first asked the respondents to assess in which scenario they would evaluate the vaccination campaign more favorable (binary choice). Then, respondents were asked to rate their likelihood of getting vaccinated for each scenario on a 0 to 10 scale (ratings). The manipulated attributes in this experiment (Fig. 1) included the severity of the circulating virus variant (‘Variant‘), the availability of new vaccines with inactivated virus components (‘Vacc‘) or Omicron-adaption (‘Omic‘), costs/incentives (‘Incen‘), and campaign messages (‘Motiv‘). For the full wording of the experimental treatments and outcome variables, see Supplemental File 6. The results are shown in Fig. 1.
Figure 1. Effects of scenarios for a hypothetical vaccination campaign (experiment 1)
>> Please include Fig. 1 here.
We found limited evidence that more severe virus variants would result in a higher likelihood of getting vaccinated. In both countries, most effects were insignificant. Yet, the triple vaccinated reported a higher likelihood of getting vaccinated (0.122, CI 0.007–0.236) when facing a more severe virus variant. In contrast, the effects were insignificant for those with one or two doses (0.101, CI -0.140–0.341) and the unvaccinated (-0.074, CI -0.266–0.118) in the escalation scenario. Once and twice vaccinated respondents at least evaluated the vaccination campaign more favorably in the light of a more severe variant, which could imply a heightened receptiveness for campaigns under such circumstances.
Our results support the notion that new vaccines are likely to play an important role in future campaigns. Most notably, we found consistent evidence for both countries that Omicron-adapted vaccines could contribute to a greater readiness to get vaccinated (Austria: 0.256, CI 0.125–0.387; Italy: 0.037, CI 0.007–0.223), in particular, among those with one or two doses (0.038, CI 0.012–0.432) and the triple vaccinated (0.279, CI 0.182–0.377). Also, vaccination campaigns were evaluated more positively when Omicron-adapted vaccines were available. In contrast, the evidence in favor of inactivated virus vaccines was somewhat mixed. New vaccines of this type seem most relevant to those vaccinated once or twice, reporting a higher likelihood of getting vaccinated (0.208, CI 0.000–0.416) that is only marginally significant, though. For the unvaccinated, we see no change in vaccination intentions under any scenario, although they seem to evaluate vaccination campaigns more favorably when more new vaccines of either kind are available.
In line with previous research, the results confirm that costs and incentives are likely to matter for vaccination decisions in future scenarios. We found evidence in both countries that even minor costs (20 Euro) could strongly reduce vaccine uptake (Austria: -0.505, CI -0.688 – -0.322; Italy: -0,647, CI -0,801 – -0.494). Increasing vaccine acceptance with positive incentives, such as vouchers or monetary rewards, is more challenging. Even with a fairly generous reward of 500 Euros, the effects remained conditional and country-specific in nature. In particular, those vaccinated already once and twice reacted most strongly to incentives and reported a higher propensity of getting vaccinated when offered positive incentives (Cash: 0.722, CI 0.429–1.014; Voucher: 0.670, CI 0.373–0.967). We also found that respondents from Austria were more susceptible to incentives (Cash: 0.307, CI 0.127–0.487; Voucher: 0.384, CI 0.199–0.570).
As the final component of experiment 1, we tested different motivational appeals as campaign messages. We mostly found no effects of motivational appeals on vaccination intentions, suggesting that most messages were similarly effective or ineffective. Among the unvaccinated and compared to the baseline message warning of the re-infection risk (‘Infect_Ego’), however, emphasizing people’s sense of community (‘Community’: 0.342, CI 0.019–0.666) may be able to increase vaccine uptake slightly. Otherwise, the motivational appeals mainly affected how positively respondents evaluated the vaccination campaign, with most messages performing about equally well.
In experiment 2, we investigated the role of the wider information environment. We showed each respondent two fictional media reports and asked them to assess based on which report they would trust the vaccine more (binary choice) and rate their likelihood to get vaccinated for each scenario on a 0 to 10 scale (ratings). The manipulated attributes of each media report included the consensus of experts (‘Cons‘), celebrity endorsement (‘Celeb‘), the prevalence of Long Covid (‘LongCo‘), and the legal rules, such as vaccine passports (‘GreenPa‘) and vaccine mandates (‘Mand‘). For the full wording of the experimental treatments and outcome variables, see Supplemental File 7. The results are shown in Fig. 2.
Figure 2. Effects of scenarios of media communication about vaccinations (experiment 2)
>> Please include Fig. 2 here.
Our results confirm that communicating expert consensus will likely increase COVID-19 vaccinations.11 Specifically, the triple vaccinated show a lower likelihood of getting vaccinated when scientists (-0.133, CI -0,263 – -0.002) or doctors (-0.161, CI -0.293 – -0.030) disagree in the false balance scenarios as compared to when there is a consensus. We also see that trust in the vaccine is significantly lower in both countries when people are facing expert dissensus. Also, a lack of expert consensus most negatively affected trust in the vaccine among those respondents who had three or more vaccinations.
Likewise, celebrity endorsement was particularly relevant for the triple vaccinated. They show an increased likelihood of getting vaccinated if a celebrity recommends getting vaccinated (0.169, CI 0.039–0.299), falls ill and regrets not having been vaccinated (0.134, CI 0.004–0.264) as compared to when a celebrity openly opposes vaccinations. Celebrity endorsement also contributed to significantly higher levels of trust in the vaccine in both countries among the majority. Only the unvaccinated showed slight signs of backlash, trusting the vaccine less when endorsed by a celebrity.
The prevalence of Long COVID and vaccine passports mattered only in a few instances. For example, in Austria, we see a heightened readiness to get vaccinated at a prevalence of 20% of Long COVID (0.213, CI 0.030–0.395). A high prevalence of Long COVID and the requirement of a vaccine passport also contributed to greater trust in the vaccine among the triple vaccinated. Finally, media reports about a vaccine mandate showed no visible effects on the likelihood of getting vaccinated but affected the trust in vaccines strongly negatively. The most substantial negative effect on trust in the vaccine was observed for a general vaccine mandate applying to the adult population with a high fine (“18 + years; 1,500€”), but also a more limited mandate for the elderly with a lower fine (“50 + years; 100€”) undermined trust in the vaccine.
Although conjoint experiments are an excellent tool to explore the possible impact of future scenarios, they also have limitations. Most notably, reading a hypothetical scenario description may differ in significant ways from experiencing how such a situation plays out in real life. It is therefore important to bear in mind that the exact size of effects is somewhat contingent on how real and tangible the conditions become for individuals. Most notably, the impact of new vaccines will most likely depend on whether they will be perceived to be effective against currently circulating virus variants.
Overall, the results show that new adapted vaccines promising enhanced protection, the cost-free provision of vaccines as well as building and communicating medical consensus are the mainstays for further COVID-19 vaccination campaigns to succeed. To some degree, positive incentives, celebrity endorsements and a better understanding of the risk of Long COVID may motivate citizens to get the vaccine. These effects were, however, more subgroup- or country-specific in nature. In addition, the support for vaccination campaigns hinges on well-designed campaign messages. As persuasive effects are overall only modest, vaccination campaigns should focus on mobilizing those with a high vaccination readiness. Stricter policy instruments, such as vaccine mandates, bear a risk of backlash and their risks and benefits must be weighed carefully. We hope that these results will inform policymakers in charge of conducting COVID-19 vaccination campaigns, health professionals responsible for informing about vaccination and experts speaking to the public.