The study was a cross-sectional online survey administered through the Singapore Population Health Studies Online Panel, approved by the National University of Singapore Institutional Review Board (NUS-IRB; reference number: H-18‐011). This panel served as an online research platform that facilitated public health research on a diverse range of topics through monthly surveys related to public health. When the survey was conducted, the panel comprised 2058 active participants who were demographically representative of the general population of Singapore.

Singapore is a city-state nation located in Southeast Asia with a population of over 5.7 million. Following the first COVID-19 case detected in January 2020, Singapore increased social distancing measures, with an eventual 2-month national partial lockdown (termed a circuit breaker) in April 2020 [15]. The COVID-19 vaccines were first introduced in Singapore in December 2020 [16]. Subsequently, a COVID-19 vaccine booster program was introduced in Singapore in September 2021 for older adults, residents of aged care facilities, and individuals with weakened immune systems, which was then gradually expanded to all other age groups [17]. In January 2022, it was announced that all individuals aged 18 and older would need to receive a vaccine booster no later than 270 days after their primary vaccination series to maintain their fully vaccinated status [18]. The survey was administered in June 2022, during the period when this policy was still in force.

As the pandemic situation stabilized, Singapore announced the easing of COVID-19 measures and the transition to an endemic phase of COVID-19 in February 2023 [19]. The nation has since been living with the endemic new normal; however, routine COVID-19 vaccination was still regarded as an important key preventive measure against severe COVID-19 illness and hospitalization [19,20].

The survey was designed with 2 main parts. The first part presented participants with two sets of DCEs, each with seven choice tasks, to elicit (1) the public’s level of acceptance of and WTP for COVID-19 vaccine boosters and (2) the nonpharmaceutical intervention measures they would be inclined to trade off if they did not want to receive and/or pay for the vaccine boosters. This manuscript focuses on the discussion of the first DCE, as the second DCE is of a different study design, with the results published elsewhere [21].

The second part of the survey consisted of 41 questions that collected sociodemographic information and assessed the level of risk acceptance and tolerance in the health domain using the Health Risk Attitude Scale (HRAS-13). The HRAS-13 is a psychometric instrument measuring risk attitudes in the health domain [22], which has demonstrated strong reliability and has been used in various population surveys and WTP studies [23,24].

The DCE data were analyzed using a mixed logit model [31] in the following 3 steps.

Singapore Population Health Studies Online Panel members had previously provided informed consent as part of a master ethical approval to participate in the panel (NUS-IRB reference number: H-18‐011). The members received notification by email and/or SMS text messages about this survey on June 20, 2022, to access the survey via the provided link and were given 10 days to complete the survey on the REDCap platform. At the start of the online survey, participants were presented with a participant information sheet. Participation in the survey was voluntary and anonymous, and participants only proceeded to the survey questions after acknowledging that they had read and received a copy of the participant information sheet. Ethical approval and a waiver of documented informed consent were obtained for this study from the National University of Singapore Institutional Review Board (reference number: NUS-IRB-2020‐82). To ensure anonymity, no personal identifiers were linked to the survey responses. Participants were reimbursed S$15 (S$1=US $0.77 as of July 10, 2026) upon survey completion.

A total of 1651 survey responses were received from the 2058 active participants on the online research platform, corresponding to a response rate of 80.2%. Of the 1651 survey respondents, 84 were excluded; 83 respondents failed the dominance test, and 1 respondent indicated an age less than 18 years old. Among the 1567 included respondents (Table 1), the majority were female (n=914, 58.3%), of Chinese ethnicity (n=1337, 85.3%), married (n=999, 63.8%), had no chronic disease (n=1238, 79%), had attained postsecondary or tertiary education (n=1211, 77.3%), lived in 4- to 5-room public housing (n=1005, 64.1%), had children (n=917, 58.5%), and worked full-time (n=877, 56.0%). Most of them reported that they had received at least 3 doses of COVID-19 vaccine (n=1443, 92.1%) and had not contracted COVID-19 (n=940, 60%). Slightly less than half of the included respondents (n=700, 44.7%) were aged 40 to 60 years. The HRAS-13 score had a mean of 41.89 (SD 7.90); a higher score indicates higher health-risk acceptance.

Table 2 shows the appearance and selection frequencies for the different levels of the attributes used in the DCE. In about one-fifth of the choice tasks, respondents chose not to accept the vaccine booster given their options made in the first stage of the choice task.

Table 3 summarizes the utility, or preference weight, from the mixed logit model. There was no evidence of left-and-right bias in choice selection (left=0.02, 95% CI –0.05 to 0.09; P=.48). Overall, the level of acceptance of the vaccine booster was high (coefficient for the none option=–6.58, 95% CI –6.91 to –6.24; P<.001), with a predicted uptake probability of 99.9% (95% CI 99.8% to 99.9%; P<.001). However, the random preference heterogeneity at the individual level for vaccine booster acceptance or hesitancy was large (SD for the none option=5.63, 95% CI 5.24 to 6.02; P<.001), indicating heterogeneous preferences among the respondents with respect to their decisions regarding accepting the vaccine booster in the real-world decision-making contexts.

The pattern of the estimated utilities for the attribute levels of risks of COVID-19 and vaccine booster characteristics was consistent with the expected utility theory [35]. Higher disutilities were exhibited for higher levels of risks of COVID-19, with increasing numbers of daily new cases, hospitalizations, and deaths. There were stronger preferences for a vaccine booster that offered stronger protection against COVID-19 in terms of its effectiveness and the duration of protection from COVID-19 variants. The relative utility increased with booster effectiveness (coefficient=0.05, 95% CI 0.04 to 0.05; P<.001) and booster effectiveness duration (coefficient=0.16, 95% CI 0.14 to 0.18; P<.001).

The WTP was generated using the coefficient of booster price as the base. The average WTP value for every percentage-point increase in booster effectiveness was S$1.43, and S$4.95 for every month increase in the duration of booster effectiveness. Respondents were willing to pay an additional S$55.23 for a vaccine booster under the hypothetical worst-case scenario of COVID-19 as presented in the DCE.

Results shown in Table 4 indicate that under the scenarios with the highest numbers of daily new cases and daily deaths, individuals’ utility for booster effectiveness duration further increased (coefficient=0.06, P<.005, respectively). However, people became more price sensitive in the scenario with the highest number of daily deaths (coefficient=−0.01, 95% CI −0.01 to −0.00; P=.04).

Of the 1567 included respondents, 1469 provided complete demographic information and were included in this analysis. Ninety-eight respondents were excluded due to “Prefer not to say” responses. Demographic factors associated with vaccine booster acceptance were female, non-Chinese ethnicity, aged younger than 40 years, having no children, high SES (defined as living in private housing), having received 2 shots of COVID-19 vaccine or less, having a chronic disease, and on chronic disease medication. On the other hand, vaccine booster hesitancy was noticed among those aged older than 60 years, with a lower education level, low SES (defined as living in 1- to 3-room public housing), who had previously contracted COVID-19, and with higher health-risk acceptance (see Figure 2). However, there were no strong associations observed between demographics and characteristics of vaccine booster. Results are provided in Multimedia Appendix 1.

This study examined the associations between sociodemographic characteristics, individual health-risk profiles, and public preferences for COVID-19 vaccine boosters, including out-of-pocket cost, among the Singapore population using a DCE. Overall, our study showed that booster preferences varied across sociodemographic and health-risk subgroups. Individuals with lower educational attainment (coefficient=2.22, 95% CI 1.83-2.63), older adults (coefficient=0.59, 95% CI 0.34-0.85), and those living in smaller public housing types (coefficient=0.34, 95% CI 0.01-0.69) showed lower acceptance of vaccine boosters, underscoring persistent socioeconomic gradients in booster uptake. Health-risk profiles also played an important role; we found that individuals with higher health-risk acceptance were less likely to accept booster vaccination (coefficient=2.37, 95% CI 1.77-2.97). These findings provide important insights into the key drivers of booster vaccination decisions and highlight critical subgroups that may require tailored communication and policy interventions to sustain booster uptake.

Respondents were willing to pay more to gain the benefit, that is, an additional S$4.95 for every month increase in the duration of vaccine booster effectiveness. These findings imply that the risks of COVID-19 may influence people’s WTP for vaccines. Respondents were willing to pay a much higher amount (ie, an additional S$55.23) for vaccine boosters to avoid the hypothetical worst-case scenario of COVID-19. The estimated average WTP value for every unit percent increase in booster effectiveness was S$1.43, which was much lower than that reported in the Australian study for COVID-19 vaccine (ie, Aus $23.92) [36]. This finding was not unexpected, as people are less willing to pay for the vaccine out of pocket considering the recurring costs of repeated vaccine boosters required. Additionally, Singapore has provided and continues to provide the COVID-19 vaccines and the booster doses free of charge.

In the multivariable analysis, we found that the health-risk profile and several sociodemographic characteristics, including age, gender, ethnicity, education level, SES, having children, chronic conditions, COVID-19 vaccination status, and previous history of COVID-19 infection, were associated with vaccine booster acceptance or hesitancy. Our findings identified that individuals with higher health-risk acceptance were less likely to accept a vaccine booster and preferred a vaccine booster with lower effectiveness. Moreover, our findings of greater booster acceptance among those who did not complete their primary vaccination series, had a chronic disease, and were taking medication for a chronic disease also provide evidence that individuals with higher risks of COVID-19 infection had a higher utility for vaccine boosters.

We found that individuals who reported that they had contracted COVID-19 were more hesitant to accept the vaccine booster. A study showed that natural immunity provides a comparable level of protection against new COVID-19 infection as vaccine-induced immunity [37]. Those with a previous history of COVID-19 infection might feel confident in the protection by infection-induced immunity and/or not be too concerned about the effects of an infection. As such, they might deem an additional booster dose as redundant [38]. Interestingly, we found that individuals with no children were more likely to accept boosters, suggesting that individuals with and without young dependents may perceive the risks and benefits of vaccine boosters differently for adults and children. As of April 2023, the Singapore Ministry of Health reported that more than two-thirds of children aged 5 to 11 years had not completed the vaccine doses required for minimum protection against COVID-19 [39]. Despite having confidence and trust in the COVID-19 vaccine for adults, parents, who are often the decision makers of child vaccination, may have concerns about the safety of the COVID-19 vaccine for their children and may reject booster doses for themselves when they are hesitant to vaccinate their children [40,41].

Our findings corroborate previous literature showing greater booster hesitancy among those with lower educational attainment [41-43] and greater booster acceptance among those with high SES groups [42,44] and those with a chronic disease [45]. On the other hand, we found some contrary results compared to previous studies. Our findings identified that there was greater booster acceptance among females, while the opposite was previously found in a meta-analysis [42]. A previous study conducted in Singapore found no gender differences in COVID-19 vaccine hesitancy [46]. However, greater vaccine acceptance was observed among females in a population-based study conducted in Hong Kong [47].

Although most studies have reported lower COVID-19 booster hesitancy among older adults [41,42,48], this study found that booster hesitancy was higher among older adults, whereas younger individuals exhibited greater acceptance of COVID-19 booster doses in the multivariable-adjusted model. Our findings are consistent with a recent systematic review and a study from Malaysia that found higher levels of booster hesitancy among older adults, suggesting that age-related differences in booster acceptance may vary across populations and contexts [43,49]. Furthermore, recent national data from Singapore indicated that, as of June 2024, booster vaccination rates among older adults aged 60 years and older remained low, lending further support to our findings [50]. Greater effort should be put forth on educational campaigns and in targeting public health messaging to individuals of older age, lower educational attainment, and lower SES to improve vaccine booster uptake among these subpopulations. This study also found that individuals of non-Chinese ethnicity were more likely to accept vaccine boosters, suggesting that the efforts made by the government and Muslim religious authorities to reduce vaccination barriers among minority groups have been effective in addressing earlier hesitancy [46].

The predicted booster uptake probability observed in this study was high, nearly 100%. Respondents may have been influenced by the prevailing policy environment, specifically the booster vaccination recommendations and the requirement to maintain fully vaccinated status, both of which were introduced in early 2022 and remained in effect at the time of data collection. This may partly reflect the broader policy environment, as well as social desirability bias in responses, whereby respondents reported choices that aligned with perceived public health expectations rather than reflecting their actual future behavior. It may also be attributable to the characteristics of the study sample, which was highly vaccinated: 92% of respondents self-reported having received three or more vaccine doses at the time of the survey. This suggests that respondents were already predisposed toward vaccine acceptance, and the near-universal predicted uptake may therefore reflect a pre-existing provaccination orientation and prior vaccination behavior, rather than the effects of the DCE attributes alone.

As with all DCE studies, our estimates reflect stated rather than revealed preferences. A positive attitude toward a vaccine booster does not necessarily always translate to actual uptake in reality [45]. The near-universal predicted uptake observed in this study contrasts sharply with actual booster uptake in Singapore, where about half among seniors aged 60 years and older had received an additional booster dose by November 2023 [51], and fewer than 20% of the local population had received a COVID-19 booster dose within the preceding year by May 2024 [52]. This discrepancy may also reflect the influential role of policy mandates and perceived disease risk in shaping vaccination behavior. During the initial booster rollout, uptake was high, with approximately 84.5% of eligible individuals having received a first booster dose by January 2022 [53], likely reflecting the combined effects of vaccination-differentiated measures and heightened concern about COVID-19. When these measures were withdrawn and the perceived threat of COVID-19 declined, uptake of subsequent annual boosters fell markedly.

On the other hand, the discrepancy between stated preferences and actual booster uptake may partly reflect factors beyond vaccine attributes and sociodemographic characteristics. Previous studies have highlighted the influence of psychological determinants, including safety, vaccine confidence or side effects, and trust, which may affect whether favorable intentions ultimately translate into vaccination behavior [54,55].

This study has several limitations. First, as this was a cross-sectional study, it captured a snapshot of public sentiment at a single point in time and could not account for potential changes in attitudes over time. As most countries have now transitioned to the endemic phase of COVID-19, public attitudes toward vaccine boosters are likely to evolve. Given that the study was conducted before the formal transition to the endemic phase, the findings may not be fully generalizable to the current endemic context. However, this timing allowed the study to capture public preferences during a critical phase when booster policy was being adapted from pandemic response to longer-term management under evolving risk perceptions and policy signals. Future research employing longitudinal designs would be valuable for monitoring shifts in public perceptions and acceptance of boosters in response to changing epidemiological conditions or policy cycles.

One other limitation is that DCEs may be subject to hypothetical bias. Nevertheless, we sought to reduce hypothetical bias by emphasizing the application to real-life choices. Future studies can assess the concordance between stated and revealed preferences within individuals [56], and expand on cheap-talk scripts to further reduce hypothetical bias [57]. Another limitation is that the worst-case scenarios included in the DCE (57,000 daily new cases, 10% of cases hospitalized, and 0.10% of cases died) exceeded the reported daily case burden observed in Singapore and may have been perceived as less realistic by some respondents. However, these levels were designed to represent a severe outbreak scenario and to capture preferences across a broad range of epidemiological conditions. Nevertheless, respondents who failed the dominant choice task were excluded from the analysis, providing some reassurance that participants remained engaged with and were able to meaningfully evaluate the hypothetical scenarios.

Finally, although the participant pool was demographically representative of the Singaporean population, the study was limited to those who were able to read and understand English. About 20% of the resident population aged 15 years and older are illiterate in English, with over half of them being older adults aged 55 years and older [58]. Future research could include other languages to capture individuals not included in the current study and improve the generalizability of the findings. This could be especially important to further inform comprehensive policy decision-making aimed at more targeted public health messaging. Despite these limitations, the study offers valuable insights into the determinants of booster acceptance and provides evidence to inform vaccination planning as COVID-19 management continues to evolve.

Vaccine boosters are a vital tool to maintain societal resilience against COVID-19. Our findings demonstrated that individuals generally preferred boosters offering longer-lasting protection, particularly in the context of heightened COVID-19 risks of infection and death. These findings underscore the dynamic nature of public attitudes, which may shift in response to evolving policy measures, vaccination recommendations, and epidemiological trends. In addition, this study identified the factors associated with booster acceptance and hesitancy, providing valuable insights that can shape targeted public health strategies and messaging to promote the uptake of vaccine boosters. These findings may extend beyond COVID-19, offering guidance for the development of vaccination policies for future pandemics and epidemics. The estimates of the WTP values in this study may also guide pricing decisions for COVID-19 vaccination in the long term. As preferences for preventive health interventions are highly context dependent, the results should be interpreted in the context of the period in which the study was conducted and may not fully represent the current setting or true endemic-phase perceptions.