Public preferences and decision-making for mpox vaccination in the African region: a multinational discrete choice experiment

Abstract

Introduction

After the WHO prequalified the first vaccine against mpox, we aimed to identify the influence of vaccine attributes on mpox vaccination preferences among the African adults.

Methods

A discrete choice experiment was conducted among 1832 African adults across six countries. Respondents answered eight questions, each requiring them to choose between two hypothetical vaccines, with variations in distance from home to vaccination facilities, cost, effectiveness, duration of the protective effect, supply and side effects. A mixed logit model was employed to estimate vaccination preferences. Willingness to pay and changes in probability were also estimated from the regression coefficients.

Results

The strongest vaccine attribute was the higher effectiveness of vaccines (≥90% vs <60%: b=1.196, 95% CI 1.089 to 1.303), then followed by a longer duration of protective effect (lifetime vs <6 months: b=1.053, 95% CI 0.920 to 1.186), a low risk of side effects (<30% vs ≥30%: b=0.495, 95% CI 0.427 to 0.562) and sufficient vaccine supply (sufficient vs limited: b=0.417, 95% CI 0.360 to 0.475). Although compared with a walking distance of 60 min, a walking distance of 45 min was significant (b=0.402, 95% CI 0.296 to 0.508), there was no significant difference for walking distance at 15 and 30 min. Scenario prediction analysis showed that higher vaccine effectiveness (≥90%: 53.55%; 80%–89.99%: 51.41%; 60%–79.99%: 25.85%), a longer duration of protective effect (lifetime: 48.28%; 12–36 months: 25.51%; 6–11 months: 14.39%), lower vaccine costs (US$0: 28.10%; US$20: 25.42%; US$100: 14.34%), a risk of side effects of less than 30% (24.24%) and sufficient vaccine supply (20.57%) all increased the probability of vaccine uptake. Populations living with children preferred vaccines with sufficient supply and lower cost, compared with those living without children.

Interpretation

In Africa, alongside providing more reliable mpox vaccines, offering sufficient vaccine free of charge, particularly to those living with children, would encourage higher vaccine uptake.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• The relative importance of vaccine attributes on vaccination decisions among African populations remains unknown, as does the variation in their impact across different subgroups, making it difficult to scientifically and quantitatively plan a comprehensive and systematic vaccination programme.

WHAT THIS STUDY ADDS

• Except for vaccine effectiveness and the duration of protective effects that could significantly influence vaccine preferences, other attributes including sufficient vaccine supply, a lower risk of side effects and a walking distance of 45 min from home to vaccination facilities could promote vaccination uptake.

• However, compared with a walking distance of 60 min, walking distances of 15 and 30 min did not promote vaccination uptake.

• Populations living with children prefer vaccines that have sufficient supply and lower cost, compared with those living without children.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Our discrete choice experiments suggested that in Africa, in addition to providing more reliable vaccines, offering free vaccination and ensuring a sufficient supply, particularly in populations living with children, would encourage mpox vaccine uptake.

Introduction

Mpox (formerly known as monkeypox) is a sporadic zoonotic disease caused by the mpox virus (MPXV). Prior to 2022, it was primarily confined to rural rainforest villages in western and central Africa.^{1 2} However, the mpox epidemic began to spread to other continents outside Africa in 2022.³ Due to its expanding spread and changing epidemiological characteristics, the WHO Director-General declared that the mpox epidemic constituted a public health emergency of international concern (PHEIC) on 23 July 2022.⁴ With the control measures such as enhanced surveillance, increased public awareness, improved health education and travel-related interventions (including warnings, port screening and quarantine for travellers or exposed individuals) implemented by governments, the mpox epidemic appears to have become less severe in most regions. Nevertheless, the African region has reported a significant increase in cases during the first half of 2024 compared with the total cases in 2023.⁵ More importantly, the emergence of a new offshoot of Clade I, known as Clade Ib, has complicated the global response to the outbreak. Its rapid spread, along with recent detections in Sweden and Thailand, highlights the growing threat of potential international transmission.⁶ The current outbreak in the Democratic Republic of the Congo is attributed to clade Ib, which is associated with more severe disease than clade II, particularly among children.^{7 8} Consequently, the WHO declared a PHEIC again on 14 August 2024. As of 6 October 2024, a total of 9939 laboratory-confirmed cases, including 55 deaths, have been reported to the WHO.⁵ Of these cases, more than 75% (7535/9939, 32 deaths) were from the African region.⁵

Notably, mpox vaccination remains the most effective method to prevent infection and severe outcomes. The WHO reported in its Mpox global strategic preparedness and response plan (August 2024) that several potential vaccine candidates are under consideration, including MVA-BN, LC16m8, ACAM2000, BNT166a and BNT166c.⁶ Following the re-emergence of the PHEIC, on 13 September 2024, the WHO approved an application for emergency authorisation for the use of MVA-BN. On the same date, the WHO and its partners established an access and allocation mechanism for mpox vaccines. However, the high cost of the vaccine and limited availability of vaccines pose significant barriers to coverage, particularly in low-income and low-middle-income countries.⁸ Vaccine supply is also a critical issue in the most affected regions of Africa, as most mpox vaccines are sourced from countries outside the African region.⁸ Although 620 000 doses of the MVA-BN vaccine and 3 million doses of the LC16 vaccine have been pledged to affected countries by various nations and organisations,⁹ the available vaccines remain insufficient to address the recent surge in mpox cases. Additionally, other factors including the enabling environment, social influences and motivation also affect vaccination uptake.¹⁰ These factors may interact and complicate the decision-making process regarding vaccination. Discrete choice experiments (DCEs) have been widely used to examine preferences for vaccination programmes, such as those for human papilloma virus, seasonal influenza and COVID-19.¹⁰ Importantly, no research has investigated the mpox vaccination preferences among African adults. It is challenging to scientifically and quantitatively plan a comprehensive and systematic vaccination programme based on qualitative research. One meta-analysis reported a low mpox vaccine uptake of 5% and the high vaccine hesitancy rate of 58.1% in 2023 within the African region. Therefore, it is essential to understand the factors influencing mpox vaccination and to identify strategies for improvement based on DCE.¹¹ Undoubtedly, those findings are vital for formulating effective mpox vaccine contribution plans and policies, especially given the scarcity of data regarding mpox in the African region.¹²

This study aimed to investigate the key attributes that may influence decision-making regarding mpox vaccination in the African region. In response to the risk posed by the new clade Ib strain, international organisations, including the WHO, the Vaccine Alliance and other partners, have issued tenders to help secure mpox vaccines for the countries most affected. Key factors that promote the improvement of vaccination programmes include vaccine supply, vaccine cost and vaccination schedules. Given the high disease burden of mpox and the ongoing vaccination promotion programmes, understanding how preferences for mpox vaccination vary with sociodemographic factors could provide valuable insights for future mpox vaccination programmes in the African region.

Materials and methods

This online multinational DCE was conducted from 1 October 2024 to 10 October 2024, following the WHO’s approval of an application for emergency authorisation of the MVA-BN vaccine. Participants aged 18 years and older were recruited through Dynata across six African countries: Uganda, Nigeria, Morocco, Egypt, Kenya and South Africa. Our aim was to include all African countries with reported mpox cases, based on WHO records as of 11 September 2024.⁵ Ultimately, we included the six aforementioned countries due to the limited reach of the Dynata platform within the African region. Anonymised Self-Administered Questionnaire, translated into the official languages of each country, was distributed using multistage stratified random sampling. Respondents were purposively selected based on age-specific strata (18–44 years old, 45–64 years old and ≥65 years old) from each gender until the required sample size was achieved in each country.¹³ Dynata, a professional international survey platform, provided the sampling database. In our study, Dynata randomly sent email invitations to participants who met the gender and age criteria in each country.

DCE experimental design

We conducted a rapid review to identify the primary attributes influencing mpox vaccine uptake. Articles were retrieved from PubMed, Embase, Scopus, Web of Science, WANFANG DATA, CNKI, SinoMed and EBSCO from their inception to 4 September 2024, with no language restrictions. The search strategy, developed to focus on factors affecting decision-making regarding mpox vaccine uptake, is presented in online supplemental Table S1.

After conducting a focus group discussion with three experts specialising in epidemiology of infectious diseases, health management and health big data to rank the 21 shortlisted attributes and identify those potentially significant, we ultimately identified six mpox vaccine-related attributes for the final experimental design. These attributes include the distance from home to vaccination facilities (walking time, minutes),^{14 15} vaccination cost, vaccine effectiveness,¹⁶ duration of the protective effect,¹⁷ vaccine supply and side effects of the vaccine (any adverse event, primarily local adverse events).¹⁸ The attributes and their corresponding levels are presented in table 1.

Table 1. Attributes and corresponding levels chosen for this experiment.

Attributes	Level 1	Level 2	Level 3	Level 4
Distance from home to vaccination facilities (walking time, minutes)	15 min	30 min	45 min	60 min
Vaccination cost	US$0	US$20	US$100	US$200
Effectiveness of vaccines	<60%	60%–79.99%	80%–89.99%	≥90%
Duration of vaccine protective effect	<6 months	6–11 months	12–36 months	Lifetime
Vaccine supply	Sufficient	Limited	–	–
Risk of side effects of the vaccine (any adverse event mainly including local adverse event)	<30%	≥30%	–	–

A full factorial design would generate $4^4\times 2^2=1024$ possible alternatives and $C_{1024}^2=\mathrm{523,776}$ possible choice sets if each choice task contained 2 alternatives. We employed a Bayesian D-optimal efficient design with uniformly distributed priors, implemented using SAS V.9.4 software, to generate 16 choice tasks, each comprising 2 alternative choices.¹⁹ These tasks were randomly divided into two blocks.¹⁹ Each respondent was randomly assigned to one of these blocks and completed eight choice tasks, which helped balance learning effects and mitigate respondent fatigue. Additionally, to further minimise any learning effects, each respondent was required to complete a warm-up task prior to the set of eight experimental tasks.¹⁰ The framework of the warm-up task was the same as that of the experimental tasks. Respondents were then asked which vaccine they chose in the warm-up task; those who selected an inconsistent vaccine may not have fully understood the task, and in such cases, the following task was terminated to ensure careful completion. Each choice task consisted of two vaccine alternatives. Before data collection, a pilot study was conducted with five individuals to assess their understanding of the survey and to refine the experimental design. Participants completed the survey using the same online platform. One-to-one interviews were subsequently conducted to gather feedback, which mainly focused on the layout and presentation of the survey content, such as font size and whether to bold information for emphasis, to enable participants to clearly see the task. Amendments were made to the experimental design based on this feedback. The final survey instrument is shown in the online supplemental Methods. The first section presented the DCE. An example of the framework is provided in the online supplemental Methods. The subsequent section collected sociodemographic information and mpox-related data from the respondents, including age, gender, educational level, household income, whether they live with children, the number of chronic diseases and mpox infection status.

Sample size

The sample size was calculated using Orme’s thumb rule²⁰: $N\S amp;gt;500\times c/\left(t\times \mathrm{a}\right)$, where N represents the minimum sample size, c denotes the largest number of levels for any attribute (c=4), t indicates the number of choice tasks (t=8) and a signifies the number of alternatives (a=2). The minimum sample size required for this DCE study was calculated as $125\times 2=250$. To ensure adequate statistical power, we aimed to recruit 300 respondents from each country.

Data analysis

The mixed logit model was primarily employed to assess respondents’ preferences regarding each attribute of mpox vaccination. Mixed logit models, which address issues of preference heterogeneity and the independence of irrelevant alternatives, are widely used to analyse preferences in DCEs.21,23 All attribute levels were dummy coded, except vaccination cost, which was treated as a continuous variable in the model. Although four cost levels were included in the design, treating cost as continuous allows for the estimation of the marginal utility of cost and the calculation of willingness to pay (WTP), an approach that is widely recommended and applied in DCE studies.^{22 24 25} Furthermore, in this multicountry study, there is no common standard to categorise vaccine prices as ‘low’, ‘medium’ or ‘high’ across different settings; treating cost as continuous provides a more realistic and generalisable specification of how respondents value vaccine price in diverse economic contexts. Individual utility was estimated based on factors including the distance from home to vaccination facilities, vaccination cost, vaccine effectiveness, duration of protective effect, vaccine supply and side effects of the vaccine. The utility (U) of respondent i in the choice task t, related to the choice of alternative j can be expressed as follows21,23:

$${\displaystyle {\displaystyle \begin{array}{l}{U}_{ijt}={\beta }_0+{\beta }_{1}cost+{\beta }_2\mathrm{d}\mathrm{i}\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{c}{\mathrm{e}}_1+{\beta }_3\mathrm{d}\mathrm{i}\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{c}{\mathrm{e}}_2\\ +{\beta }_4\mathrm{d}\mathrm{i}\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{c}{\mathrm{e}}_3+{\beta }_{5}effectivenes{s}_1+{\beta }_{6}effectivenes{s}_2\\ +{\beta }_{7}effectivenes{s}_3+{\beta }_{8}duratio{n}_1+{\beta }_{9}duratio{n}_2\\ +{\beta }_{10}duratio{n}_3+{\beta }_{11}supply+{\beta }_{12}side\mathrm{\_}effect+{\epsilon }_{ijt}\end{array}}}$$

where ${\beta }_0$ is the alternative-specific constant (ASC), ${\beta }_{1~12}$ are the attribute level estimates that indicate the relative importance of each attribute level, and ${\epsilon }_{ijt}$ is the error term. A positive (negative) β coefficient indicates that whether the attribute-level combination was preferred (not preferred) relative to the reference level. The greater the absolute value of the β coefficient, the stronger the preference for that level compared with the reference level.

The WTP for a specific change in an mpox vaccination scenario is defined as the amount of money (or its equivalent) that represents an individual’s marginal payment for the altered attribute levels in a new alternative scenario. The WTP was estimated using a mixed logit model by assessing the ratio of the coefficients of other attributes (${\beta }_i$) to the coefficient of vaccination cost (${\beta }_{cost}$): ${WTP}_i=-\frac{{\beta }_i}{{\beta }_{cost}}$.²² Furthermore, the probability of choosing alternative i rather than alternative j is given by: $P_i=\frac{e^{\beta `x_i}}{\sum e^{\beta `x_j}}$, where x is a vector of attribute coefficients.²² Therefore, the change in the probability of opting for the baseline mpox vaccine due to a change in one of the vaccine attributes (assuming all other attributes remain constant) is calculated by: $P_c=P_i-P_j$.²² The baseline mpox vaccination scenario is defined as follows: a vaccination cost of 200 dollars, walking time from home to vaccination facilities of 60 min, vaccine effectiveness <60%, duration of vaccine protective effect <6 months, limited vaccine supply and side effects of vaccine ≥30%.

Subgroup analyses were conducted using mixed logit models across various demographic groups, including age, gender, education level, living with children (<18 years old) or not, household income level, existing health conditions, mpox infection status and country. The WTP and changes in probability were also calculated for each subgroup. To further identify underlying patterns of preference heterogeneity among respondents, we conducted a latent class analysis.²⁶ The optimal number of latent classes was determined based on model fit criteria, including the Akaike Information Criterion and the Bayesian Information Criterion, as well as qualitative assessment of the interpretability of class-specific preference patterns.²⁷ The latent class conditional logit models were fitted using the expectation–maximisation algorithm.²⁸

All analyses were performed using the Stata V.15.0 software (StataCorp). Statistical significance was defined as a two-sided p<0.05.

Patient and public involvement

Patients and/or the public were not involved in the design, or conduct, reporting or dissemination plans of our research.

Results

Respondent characteristics

A total of 1823 individuals (ranging from 302 to 307 at the national level) participated in the study. Online supplemental Table S2 presents the characteristics of the respondents. Among these, 935 (51.29%) were female, and the majority (1267/1823, 69.50%) were aged 44 years or younger. The age and sex distribution in the study sample for each country was similar to that of the national population.

Main effects from the mixed logit model and deterministic preference heterogeneity

Table 2 presents the primary coefficients of the attributes derived from the mixed logit model. Ranking the attributes by the magnitude of their coefficients shows that vaccine effectiveness was the strongest driver of mpox vaccine uptake (≥90% vs <60%: b=1.196, 95% CI 1.089 to 1.303), followed by the duration of protective effect (lifetime vs <6 months: b=1.053, 95% CI 0.920 to 1.186). Moderate but significant preferences were observed for a lower risk of side effects (<30% vs ≥30%: b=0.495, 95% CI 0.427 to 0.562) and sufficient vaccine supply (sufficient vs limited: b=0.417, 95% CI 0.360 to 0.475). Additionally, compared with a walking distance of 60 min from home to vaccination facilities, a walking distance of 45 min was significant (b=0.402, 95% CI 0.296 to 0.508). However, there was no significant effect for walking distances of 15 or 30 min. Vaccination cost had a modest but significant effect (per US$1, b=−0.00289, 95% CI −0.00369,–0.00208), indicating that higher costs reduced the probability of mpox vaccination uptake. Notably, although the coefficient was relatively small due to cost being modelled as a continuous variable, the impact of vaccine price on uptake remains meaningful and should not be overlooked.

Table 2. Mean preference and willingness to pay estimates for mpox vaccine attributes across mixed logit model.

Attribute	Mean preference (b coefficient to 95% CI)	P value	Willingness to pay to US$ (95% CI)	P value
Vaccination cost (US$)	−0.00289 (-0.00369 to 0.00208)	<0.001	NA	NA
Distance from home to vaccination facilities (walking time)
60 min	Reference		Reference
15 min	−0.00266 (−0.0812 to 0.0759)	0.947	−0.92 (−28.09 to 26.25)	0.947
30 min	−0.0846 (−0.170 to 0.000309)	0.051	−29.31 (−62.06 to 3.45)	0.079
45 min	0.402 (0.296 to 0.508)	<0.001	139.30 (70.55 to 208.04)	<0.001
Effectiveness of vaccines
<60%	Reference		Reference
60%–79.99%	0.529 (0.444 to 0.614)	<0.001	183.17 (119.74 to 246.6)	<0.001
80%–89.99%	1.136 (0.998 to 1.275)	<0.001	393.58 (247.02 to 540.13)	<0.001
≥90%	1.196 (1.089 to 1.303)	<0.001	414.08 (286.34 to 541.82)	<0.001
Duration of vaccine protective effect
<6 months	Reference		Reference
6–11 months	0.290 (0.175 to 0.405)	<0.001	100.39 (66.33 to 134.46)	<0.001
12–36 months	0.522 (0.413 to 0.631)	<0.001	180.64 (137.67 to 223.62)	<0.001
Lifetime	1.053 (0.920 to 1.186)	<0.001	364.74 (281.56 to 447.92)	<0.001
Vaccine supply
Limited	Reference		Reference
Sufficient	0.417 (0.360 to 0.475)	<0.001	144.54 (95.13 to 193.95)	<0.001
Side effects of the vaccine (any adverse event mainly including local adverse event)
≥30%	Reference		Reference
<30%	0.495 (0.427 to 0.562)	<0.001	171.32 (127.24 to 215.41)	<0.001

Table 3, online supplemental Table S3-S9 present the coefficients of the attribute levels across different populations based on interactions. Overall, preferences for vaccine attributes varied according to age group, gender, educational level, household income, existing health conditions, whether respondents lived with children, history of mpox virus infection and country. Notably, older adults did not express a preference regarding the distance from their homes to vaccination facilities (b=0.303, 95% CI −0.0365 to 0.643) and the duration of vaccine protective effect was considerably less important (table 3). Among respondents living with children, point estimates suggested stronger preferences for higher vaccine effectiveness, sufficient supply and a low risk of side effects, although these differences were based on overlapping confidence intervals (online supplemental Table S7) and figure 1). Additionally, the positive impacts of all vaccine attributes appeared somewhat more pronounced among respondents who had never been infected by MPXV (figure 1 and online supplemental Table S8). Meanwhile, the negative association with higher vaccination costs tended to be greater among respondents living with children and those with a history of mpox infection (figure 1 and online supplemental Table S7 and S8). Country-specific analyses revealed that, in Uganda, the lifetime duration of vaccine protection was the strongest driver of uptake (b=1.149, 95% CI=0.828, 1.470), whereas in the other five countries, higher vaccine effectiveness was consistently the most influential attribute. Additionally, preferences regarding vaccination cost were not statistically significant in Uganda and Nigeria, in contrast to the significant negative effect observed in the other settings (online supplemental Table S9).

Table 3. Subgroup analysis of mean preference estimates for mpox vaccine attributes across mixed logit model by age.

Attribute	18–44 years old		45–64 years old		≥65 years old
	Mean preference (b Coefficient to 95% CI)	P value	Mean preference (b Coefficient to 95% CI)	P value	Mean preference (b Coefficient to 95% CI)	P value
Vaccination cost (US$)	−0.00283 (−0.00382 to 0.00184)	<0.001	−0.00331 (−0.00501 to 0.00162)	<0.001	−0.0025 (−0.00517 to 0.000161)	0.066
Distance from home to vaccination facilities (walking time)
60 min	Reference		Reference		Reference
15 min	−0.0207 (−0.117 to 0.0758)	0.674	0.1 (−0.0631 to 0.263)	0.229	−0.113 (−0.382 to 0.155)	0.409
30 min	−0.134 (−0.238 to 0.030)	0.012	0.0798 (−0.0985 to 0.258)	0.380	−0.117 (−0.408 to 0.175)	0.432
45 min	0.404 (0.273 to 0.536)	<0.001	0.458 (0.238 to 0.678)	<0.001	0.303 (−0.0365 to 0.643)	0.080
Effectiveness of vaccines
<60%	Reference		Reference		Reference
60%–79.99%	0.568 (0.463 to 0.674)	<0.001	0.497 (0.321 to 0.673)	<0.001	0.34 (0.0755 to 0.604)	0.012
80%–89.99%	1.173 (1.000 to 1.346)	<0.001	1.124 (0.841 to 1.406)	<0.001	1.057 (0.595 to 1.519)	<0.001
≥90%	1.194 (1.063 to 1.325)	<0.001	1.241 (1.015 to 1.468)	<0.001	1.236 (0.876 to 1.596)	<0.001
Duration of vaccine protective effect
<6 months	Reference		Reference		Reference
6–11 months	0.321 (0.177 to 0.464)	<0.001	0.208 (−0.0276 to 0.444)	0.084	0.269 (−0.0912 to 0.63)	0.143
12–36 months	0.547 (0.411 to 0.683)	<0.001	0.51 (0.285 to 0.735)	<0.001	0.385 (0.0547 to 0.715)	0.022
Lifetime	1.058 (0.893 to 1.222)	<0.001	1.214 (0.929 to 1.499)	<0.001	0.694 (0.288 to 1.1)	0.001
Vaccine supply
Limited	Reference		Reference		Reference
Sufficient	0.481 (0.408 to 0.553)	<0.001	0.336 (0.225 to 0.446)	<0.001	0.178 (0.0186 to 0.338)	0.029
Side effects of the vaccine (any adverse event mainly including local adverse event)
≥30%	Reference		Reference		Reference
<30%	0.559 (0.474 to 0.643)	<0.001	0.423 (0.282 to 0.563)	<0.001	0.2 (0.0188 to 0.381)	0.031

Figure 1. Subgroup analysis of mean preference estimates, willingness to pay estimates and changes in probability for mpox vaccine attributes across mixed logit model, by whether living with children and mpox virus infection.

WTP analysis and heterogeneity

The analysis results of WTP (table 3) presented that respondents had the highest WTP for the attribute of vaccine effectiveness. If the effectiveness improved from <60% to ≥90%, the estimated WTP was about US$414.08 (95% CI US$286.34 to US$541.82). Additionally, the WTP for the duration of protective effect was also considerable. If the duration changed from <6 months to a lifetime, the WTP was US$364.74 (95% CI US$281.56 to US$447.92). Respondents were also willing to pay for a reduced risk of side effects (WTP=US$171.32, 95% CI US$127.24 to US$215.41), sufficient vaccine supply (WTP=US$144.54, 95% CI US$95.13 to US$193.95), and shorter distance to vaccination facilities (WTP=US$139.30, 95% CI US$70.55 to US$208.04). According to the subgroup analyses results (online supplemental Table S10-S17), males, respondents with secondary education or below, those with an average level of household income, individuals with at least one existing health condition, those living without children, individuals who have never been infected by MPXV, and respondents from Egypt, Kenya, South Africa and Morocco were willing to pay more for the vaccine effectiveness and the duration of protective effect. Respondents living without children and who have never been infected by MPXV were willing to pay more for both the vaccine effectiveness and the duration of protective effect compared with the overall sample (figure 1).

Forecast on vaccine uptake

Figure 2 and online supplemental Table S18 illustrate the changes in mpox vaccine uptake probabilities as vaccine attributes vary. We defined a base case: vaccination cost of US$200, a walking time of 60 min from home to vaccination facilities, vaccine effectiveness of <60%, duration of protective effect of <6 months, limited vaccine supply and side effects of <30%. In comparison to the base case, higher vaccine effectiveness (≥90%: change in the probability=53.55%, 95% CI 49.73% to 57.37%; 80%–89.99%: 51.41% (95% CI 46.30% to 56.51%); 60%–79.99%: 25.85% (95% CI 21.88% to 29.81%)), longer duration of protective effect (lifetime: change in the probability=48.28%, 95% CI 43.17% to 53.38%; 12–36 months: 25.51% (95% CI 20.41% to 30.60%); 6–11 months: 14.39% (95% CI 8.75% to 20.04%)), lower vaccine cost (US$0: change in the probability=28.10%, 95% CI 20.66% to 35.53%; US$20: 25.42% (95% CI 18.62% to 32.21%); US$20: 14.34% (95% CI 10.39% to 18.29%)), the risk of side effects at less than 30% (change in the probability=24.24%, 95% CI 21.06% to 27.42%), sufficient vaccine supply (change in the probability=20.57%, 95% CI 17.84% to 23.31%), and walking time of 45 min from home to vaccination facilities (change in the probability=19.84%, 95% CI 14.74% to 24.95%) all increased the probability of mpox vaccine uptake.

Figure 2. Effects of changing the attribute levels on the probability in choosing an mpox vaccination. Notes: The scenario from which these attribute levels were varied is: vaccination cost (US$200), walking time from home to vaccination facilities (60 min), vaccine effectiveness (＜60%), duration of vaccine protective effect (＜6 months), limited vaccine supply and side effects of vaccine (<30%).

Online supplemental Table S19-S26 reports the changes in the probabilities of mpox vaccine uptake as vaccine attributes vary across different populations. Vaccine cost did not affect the probabilities among older adults aged 65 years or older (online supplemental Table S19) and those with secondary education or below (online supplemental Table S21). The walking time from home to vaccination facilities did not influence the probabilities among individuals who had been infected by MPXV (figure 1 and online supplemental Table S25).

The latent class analysis identified three latent preference classes, accounting for 22.2%, 30.4% and 47.4% of respondents, respectively (online supplemental Table S27). Class 1 showed strong preferences for sufficient supply (sufficient vs limited, b=1.604, 95% CI 1.334 to 1.873), and class 2 placed pronounced emphasis on vaccine effectiveness and safety, preferring higher effectiveness levels (≥90% vs <60%, b=2.208, 95% CI 1.829 to 2.586) and a low risk of side effects (<30% vs ≥30%, 2.079, 95% CI 1.875 to 2.282); and class 3, exhibiting strong preferences for shorter distance (15 min vs 60 min, b=0.835, 95% CI 0.642 to 1.028) and longer-lasting or lifetime protection (lifetime vs <6 months, b=2.018, 95% CI 1.849 to 2.186). Analysis of class membership indicated that respondents with tertiary education were more likely to belong to class 2, whereas other characteristics were not significantly associated with class membership (online supplemental Table S28).

Discussion

The access and allocation mechanism for mpox vaccines established by the WHO and other partners prioritises individuals at high risk of infection.⁹ While except for the distribution of these critical resources, how to encourage the population to accept resources is also urgent. One study reported that a vaccine uptake of 5% and a high vaccine hesitancy rate of 58.1% in 2023 in the WHO African region.¹¹ Rather than predicting uptake levels, the main aim of our study was to identify specific vaccine attributes preferred by the population, with a view to assessing which of these attributes could be amenable in policy.

Our study found that both the vaccine effectiveness and the duration of protective effect significantly influence vaccine preferences. The general population in Africa is willing to accept vaccines with an effectiveness of 90% or higher and a lifetime duration of protection. All else being equal, a vaccine with such high effectiveness and a longer duration of protection could increase uptake by approximately 54% and 48%, respectively. Additionally, this study sheds new light on another concern regarding mpox vaccination in the African region: vaccine supply. When vaccine supply is sufficient, the acceptance increased by 21% compared with situations with limited vaccine supply.

To a lesser extent, other factors that may promote vaccine uptake include a lower risk of side effects and a walking distance of 45 min from home to vaccination facilities. The high vaccine cost presents a significant barrier to vaccine coverage in the African region.⁸ The cost of vaccination negatively affects mpox vaccine uptake; if the mpox vaccine were provided free of charge, the probability of acceptance could increase by 28%. Conversely, when a fee of US$20 is required, this probability may decrease by half. Our study also reported that the increased vaccine effectiveness and the duration of protective effect could enhance the WTP for vaccination, with maximum values of US$414.08 and US$364.74, respectively. While the WTP estimates offer useful insights into the relative strength of preferences, it is important to note that some values (eg, US$414 for ≥90% vs <60% effectiveness) exceed the maximum vaccination cost level (US$200) presented in the choice sets. This occurs because vaccination cost was modelled as a continuous variable, and WTP was calculated as the ratio of attribute coefficients to the cost coefficient. As a result, the estimates may reflect extrapolations beyond the range of the presented cost levels. Similar findings have been reported in other DCEs, where extrapolated WTP values primarily serve to highlight the relative importance of attributes rather than absolute affordability thresholds.^{25 29} Therefore, the seemingly high WTP values in our study should be interpreted mainly as indicating the strong preference respondents place on vaccine effectiveness and long-term protection.

Vaccine effectiveness and the duration of protective effect were the two most influential factors affecting preferences for mpox vaccination. Individuals were willing to accept and pay more for vaccines with higher effectiveness and longer-lasting protection. A low risk of side effects could also increase the possibility of mpox vaccination. Although four studies reported mpox vaccination hesitancy in Nigeria,^{30 31} Algeria³² and Ghana³³ in 2023, none of them investigated the influence of vaccine attributes. Previous DCEs in other countries have reported similar concerns regarding COVID-19 vaccine effectiveness and safety.¹⁰ Our previous study reported the safety and effectiveness of the mpox vaccine were the most important aspects to hesitancy among the Chinese MSM population.³⁴ Additionally, our study addressed two critical topics in mpox vaccination policy in the African region: vaccine cost and supply. Although the impact of these two attributes was smaller compared with vaccine effectiveness, they can be significantly influenced by policymakers. Providing sufficient and free vaccines could potentially increase the acceptance rate by nearly 20%. Interestingly, we found that the proximity of vaccination sites did not correlate with a greater increase in vaccination probability or WTP. Positive effects were only observed when the walking distance from home to vaccination facilities was 45 min, not at 15 or 30 min. Furthermore, older adults did not express a preference for the distance from home to vaccination facilities within 45 min. Therefore, these results do not support the notion that shorter distance increases vaccine uptake; in other words, shorter distance was not associated with higher mpox vaccination rates. This may be because the mpox epidemic is more severe in the African region. Given the limited availability of the mpox vaccine, when a vaccine is available, people are more inclined to get vaccinated regardless of the distance, compared with other vaccine attributes.

The high case fatality rate of mpox among children was a critical concern during this outbreak in the African region.^{35 36} Therefore, in addition to common demographic characteristics, we also analysed preference heterogeneity among respondents living with children. Populations living with children preferred to choose vaccines with higher vaccine effectiveness, sufficient supply, a low risk of side effects and lower cost. However, distance from home to vaccination facilities and the duration of vaccine protective effect were not more sensitive among them. Furthermore, those living with children were less willing to pay more for the vaccine effectiveness and the duration of protective effect compared with those living without children. These findings suggest that, in practical vaccination strategies aimed at reducing the risk for children, healthcare professionals should recommend free vaccination for their family members by providing information on vaccine effectiveness, supply and risk of side effects.

To the best of our knowledge, this study is the first in the African region to investigate public preferences for mpox vaccination using a DCE. Importantly, the study may assist policymakers in fine-tuning their vaccination strategies. However, our study has several limitations. First, as socioeconomic characteristics do not vary within a choice set, they could not be directly included in the mixed logit model.²² Therefore, subgroup analyses were conducted to explore sources of preference heterogeneity; however, these findings may be subject to residual confounding and should be interpreted with caution. We further conducted a latent class analysis, which allowed preference heterogeneity to be identified in a data-driven manner without imposing a priori subgroup definitions, thereby providing complementary insights into underlying preference phenotypes. second, we relied on self-reported information, which may have introduced recall bias. Finally, although we attempted to recruit participants using an age-gender stratified design, selection bias remained due to the constraints of the online survey. Since the online survey was a more tech-friendly audience, the representativeness of the sample may be compromised. We speculated that individuals with higher educational levels who chose to complete the DCE might be more attentive to mpox-related information and may pay more attention to the vaccine effectiveness and side effects. In fact, our latent class analysis found that those with tertiary education were more likely to belong to participants in class 2 (30.4%) who focused on vaccine effectiveness and safety, which supported above speculations. Additionally, while we aimed to include all African countries, the platform was only able to investigate the six countries. Consequently, caution should be exercised when generalising these findings to Africa as a whole.

Conclusions

The effectiveness, duration of protective effect and side effects significantly influenced decisions to receive vaccines against mpox. Meanwhile, vaccine supply and cost also had minor effects on preferences for the mpox vaccine. The effectiveness of vaccines, supply, risk of side effects and vaccine cost are critical considerations for African populations, particularly those with children. In Africa, in addition to providing more reliable vaccines, offering a freely available and sufficient supply, particularly to populations living with children, would encourage mpox vaccine uptake.

Data availability statement

Data are available on reasonable request.