Ethics statement

Our study protocol was approved by the Institutional Review Boards (IRBs) of PRISMA (CE0882.21) providing local Peruvian approval and the Tulane School of Public Health and Tropical Medicine (2021–1749). The University of California, Davis IRB provided an exemption. The protocol was also reviewed and approved by the Loreto Regional Health Department which oversees health research in Iquitos. Written consent was obtained from all participants prior to the initiation of focus group discussions and to surveys.

Study design

Our mixed-methods study started with focus group discussions (FGDs) to identify barriers to community acceptability of a dengue vaccine and motivators for uptake based on the 5C’s Framework of Vaccine Hesitancy [21]. Informed by the focus group discussions, we created and administered a quantitative survey to assess attitudes toward a hypothetical dengue vaccine and to identify factors associated with being “unsure” about getting vaccinated (includes individuals who might accept vaccines but have some doubts, to those who might delay their acceptance or refuse some vaccines but not all; basically, individuals who could be “moved” on the vaccine hesitancy continuum to accept a vaccine), compared with “acceptors” to vaccination.

Setting

We worked in two regions in Peru characterized by high dengue transmission and severe outbreaks: the state of Piura in northwestern coastal Peru and Iquitos city located in the northeastern Peruvian Amazon basin. Piura reported the highest number of dengue cases in the country over the past 15 years (111,889 cases from 2007 to 2022) and ranks fifth in cumulative incidence of dengue with warning signs and severe dengue (722 cases per 100,000 inhabitants) [23]. In 2017, the El Niño phenomenon led to increased temperatures, rainfall, and flooding, triggering a major outbreak that accounted for 65% of all reported cases nationwide that year (68,290 cases) [24]. In 2023 and 2024, Piura became the epicenter of the largest dengue outbreaks in Peru, with 79,304 and 33,612 cases, respectively. In contrast, Iquitos bears the highest dengue burden in the Loreto region [6], which ranks second nationally in cumulative incidence of dengue with warning signs and severe dengue (1,518 cases per 100,000 inhabitants) from 2007 to 2022 [23]. In 2023 and 2024, dengue also spread to Lima, southern regions, and rural regions of Loreto; all historically considered low-incidence areas [6].

We started our study in Iquitos, the capital of Loreto Department, with a population of approximately 400,000 residents; surrounded on three sides by the Itaya, Nanay, and Amazon rivers, it is accessible only by boat or plane, forming a socio-epidemiological island [25]. The city is divided into four districts, each of which also have rural sections along the Iquitos-Nauta highway and adjacent rivers. Additionally, the city has been a dengue research hub since the early 1990s, contributing significantly to scientific understanding of dengue and its transmission and prevention, including the spatiotemporal dynamics, behavior, and genetic structure of the Aedes aegypti vector [26–34]; the community-level understanding and lived experience of dengue illness [35,36]; the evaluation of targeted and city-wide vector control strategies [37–41]; the longitudinal and interepidemic transmission patterns of dengue virus [42]; the feasibility of implementing community-based surveillance tools for febrile disease detection [43–45]; and the role of human mobility in shaping transmission risk and spatial spread [46–51]; among other relevant contributions to the field [52,53]. Based on our studies, dengue virus activity remains elevated from August to April, peaking between November and January [41,54]. From late 2010, until late 2019, DENV-2 was the dominant serotype; thereafter, DENV-1 emerged, triggering an outbreak that was overshadowed by the COVID-19 pandemic [55]. According to the Peruvian National Institute of Statistics (INEI), 30% of the urban population in the Amazon rainforest has at least one unmet basic need, and 18.2% lives in poverty [56]. Despite these challenges, literacy rates remain relatively high, ranging from 88% to 92% [25]. Our study was carried out in the Punchana and Iquitos districts of Iquitos city, selected due to previous participation (half of the sample) in dengue research projects.

We then moved to Piura in the northern tropical coastal region of Peru, which has shown a markedly different pattern of dengue virus behavior compared to Iquitos. In the state of Piura, with an estimated population of 1,856,809 residents [25], we worked in Chulucanas and La Matanza districts (Morropón Province) and La Unión district (Piura Province). These sites were selected due to their participation (also in half of the sample) in previous research on the tetravalent Dengvaxia vaccine [11].

Data collection instruments

To facilitate discussion in the FGDs, we used the recent and tangible experience with the COVID-19 vaccination as a starting point to examine acceptance towards a potential (but at the time, still hypothetical) dengue vaccine, as participants had all recently been engaged in this COVID-19 national vaccination program. Participants were then asked to envision a scenario where the Ministry of Health had approved a dengue vaccine and addressed three key questions: 1) What information would you need to accept vaccination? 2) How can public confidence in the vaccine be improved? 3) What is the best way to distribute a new vaccine quickly in your community? The Iquitos study team developed and piloted the FGD guides in Spanish [S1 File].

A survey was then developed using the preliminary results from the FGDs [S1 File]. The main outcome of interest was dengue vaccine acceptability, but at the time of data collection, no validated instrument was available to specifically measure dengue vaccination acceptability, particularly in pre-implementation contexts. We therefore adapted the Oxford COVID-19 Vaccine Hesitancy Scale [57], a tool designed to capture gradients of vaccine acceptance, uncertainty, and refusal prior to large-scale vaccine rollout. This framework is appropriate for dengue vaccination given the strong conceptual parallels between both scenarios, including evaluation of a novel vaccine under conditions of incomplete information, uncertainty regarding safety and effectiveness, widespread circulation of risk-related narratives, and decision-making in advance of routine implementation.

The seven Oxford hesitancy items were linguistically and contextually adapted to the dengue setting and administered to all participants [see details of process in S2 Text, S3 Text, S4 Text, S5 Text]. Following the analytical recommendations proposed by Freeman et al. [57], and to account for a minor modification introduced during pilot testing of one item, dengue vaccine acceptability was operationalized using four complementary outcome specifications: (i) a latent continuous hesitancy construct estimated using confirmatory factor analysis; (ii) an observed continuous mean hesitancy score; (iii) a three-level categorical outcome (“acceptors” to vaccination, “unsure” about getting vaccinated, and “refusers”); and (iv) a dichotomous outcome contrasting acceptors versus unsure about getting vaccinated (since the literature suggests focusing on moving those in the unsure category to the acceptor category [58,59]).

The survey also covered sociodemographic characteristics; experiences with COVID-19 and dengue; COVID-19 vaccination use and perceptions; vaccine attitudes (VAX scale and 5Cs Framework); participation in vector control activities; and knowledge, attitudes, and practices related to dengue prevention. Correct knowledge of dengue transmission was defined as identifying the bite of an infected mosquito as the mode of transmission. The Vaccine Attitudes Examination (VAX) scale, previously validated in Spanish, measured four constructs: “1) mistrust of vaccine benefit, 2) worries about unforeseen future effects, 3) concerns about commercial profiteering, and 4) preference for natural immunity” [60]. After piloting this instrument with 20 individuals, the response format was adapted from a 7-point to a 5-point Likert scale for comprehension, tested for internal consistency (Cronbach’s α = 0.73; standardized α = 0.72). The survey was developed, collected, and managed using REDCap (version 14.0.8) electronic data capture tools [61,62] and applied by our team on tablets [S1 File].

In the main manuscript, descriptive analyses are presented using the three-level categorical outcome to preserve meaningful gradients of vaccine acceptability. However, multivariable regression analyses focus on the dichotomous outcome, selected to identify determinants of vaccine uncertainty among individuals potentially amenable to intervention. Detailed descriptions of outcome construction and sensitivity analyses across all specifications are included [S2 Text, S3 Text, S4 Text, S5 Text].

Recruitment and procedures

Both qualitative (FGDs) and quantitative (survey) components of our study were stratified by recruiting from intervention and control areas. In the city of Iquitos, the intervention group was defined as clusters (grouped blocks) that had previously participated in a vector control trial [38] whereas the control groups were selected from clusters where residents had no prior exposure to community-based research studies conducted by our group. Both the intervention and control groups were in Iquitos and Punchana districts of the city. We replicated this process in Piura, with the intervention group defined as districts where the Dengvaxia vaccine trial took place, located in the Chulucanas and La Matanza districts [11]. Each intervention group was matched to a control area in La Unión district, which was not involved in the vaccine trial. We hypothesized that a community’s previous interaction with dengue research teams – and in Piura, a dengue vaccine trial – might influence responses, so we wanted to ensure some participants were selected from “intervened” areas.

We recruited FGD participants from a 5x5 block radius through door-to-door convenience sampling one day prior to the FGDs, limiting participation to two participants per block. Eligible participants were adults (>20 years old) who made household decisions, were mentally competent to provide consent, and met the age and gender criteria for the selected FGD. The research team returned 1–2 hours before each FGD to assist with transportation to the site. The FGD team was led by a senior qualitative researcher (VPS) and three research assistants: one co-facilitated (ASVS), one recorded participant feedback on butcher paper to facilitate discussion (EJRL in Iquitos, CC in Piura), and the third took notes (JLCC). We conducted 16 focus group discussions (FGDs) between February and March 2022 to reach saturation—eight in Iquitos, followed by eight in Piura. In both study sites, four FGDs were conducted with residents from intervention areas and four with residents from control areas [38]. We stratified all FGDs by gender (female or male) and age (20–39 or 40–60 years). All FGDs were audio-recorded.

For the quantitative phase, a base sample of ≥400 participants per study site was defined by selecting city blocks randomly, followed by systematic door to door recruitment of subjects with a maximum number of enrollees per block. Only one participant per household was permitted (see below). This target ensured stable site-specific estimates, adequate precision for a binary outcome under a conservative assumption (±5% margin of error at 95% confidence), and feasibility for door-to-door household surveys. We used random sampling to select blocks in both the intervention and control areas to target recruitment in every household within those blocks. Recruitment was door to door until target enrollment of eligible adult participants (aged 20–60) who were household decision makers was reached. ASVS oversaw training and field monitoring; JLCC managed quality control as data was entered on tablets and uploaded at least twice a day. Before starting data collection, the interviewers were trained in survey administration and research ethics. The surveys were conducted in August 2022 in Iquitos and a month later, in September, in Piura.

Data analysis

Audio recordings of our FGDs were transcribed into Spanish using the Trint platform [63] and our research team conducted quality control (editing, verification, de-identification) of the transcripts within the Trint software [63]. We conducted thematic analysis using the 5C’s – confidence, complacency, convenience, communication, and context – as deductive codes, and emerging themes were added as subcodes inductively. Data from the FGDs was managed, coded, and analyzed using Dedoose software (Version 9.0.107) [64].

For our quantitative survey, dengue vaccine hesitancy was operationalized using multiple outcome specifications, as detailed above. Descriptive analyses were conducted using the three-category outcome (vaccination “acceptors”, “unsure” about getting vaccinated, and vaccination “refusers”) to characterize participant attributes and response distributions across groups that reflect differing positions along the acceptance–hesitancy continuum. Categorical variables were summarized using proportions and compared using chi-square tests.

For multivariable analyses, the primary outcome was a dichotomous dengue vaccine hesitancy variable contrasting “acceptors” versus with those who were “unsure”. This dichotomy was chosen to identify determinants of vaccine hesitancy among individuals who had not expressed unequivocal refusal that might provide insights about moving them from the “unsure” to the “acceptor” category. Multivariable logistic regression models were estimated to assess the association between predictors derived from the 5C’s framework and the likelihood of being unsure about vaccination. Adjusted odds ratios (ORs) and 95% confidence intervals were reported.

Selection of predictors for the final multivariable model followed a structured, theory-driven approach grounded in the 5C’s framework. All candidate variables were classified a priori into the five conceptual domains (Confidence, Complacency, Convenience, Communication, and Context). Within each domain, multicollinearity was assessed using variance inflation factors (VIF) and overlapping or highly correlated variables were evaluated to guide retention or exclusion. Crude and domain-specific adjusted models were examined to assess the stability of effect direction and relative magnitude within each domain. In addition to domain-specific predictors, a core set of a priori sociodemographic covariates (sex, age group, education level, and occupation) was included in the final multivariable model to control for potential confounding. These covariates were retained irrespective of statistical significance. Final inclusion of predictors in the multivariable model was guided by conceptual relevance, absence of problematic multicollinearity, consistency of effects across domain-specific analyses, and model parsimony. Variables were not selected solely based on statistical significance; conceptually important predictors were retained even when not statistically significant to ensure adequate representation of key dimensions of vaccine hesitancy. All variables evaluated within each domain, the set of predictors retained in the final model, and the results of multicollinearity diagnostics are presented [S1 Text]. Additionally, sensitivity analyses were conducted using alternative outcome specifications (three-level categorical, latent continuous, and observed continuous outcomes), applying the same predictor set [S4 Text].

Data was cleaned using Stata 15.0 (Stata Corp., College Station, Texas, United States), whereas analysis and graphics were created in R version (Version 4.4.3; R Foundation for Statistical Computing).

Study population

A total of 147 individuals participated in the 16 FGDs: 72 from Iquitos and 75 from Piura. Over half of participants were female (53%), and 52% were aged 20–39. In 15 of the 16 FGDs, at least one participant had experienced dengue. Most participants had received at least one dose of a COVID-19 vaccine [Table 1]. The full table of quotes in English and the original Spanish can be found in the supplemental files [S1 Table].

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Focus Group Discussion Sample Characteristics.

https://doi.org/10.1371/journal.pntd.0013572.t001

A total of 883 of 1,240 invited participants (71%) were enrolled in the survey (see Fig 1). Of the survey respondents, 122 (13.8%) were excluded from the final multivariable analysis – 42 from Piura (9.3%) and 80 from Iquitos (18.5%) – due to classification as “refusers” of our outcome variable of dengue vaccine acceptability or missing covariate data under a complete-case approach. The final regression models were estimated using the 761 participants. Nearly half of participants were aged 40–60 (48.7%), the majority were women (71.1%), and around 83.5% had completed secondary education or higher [Table 2].

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Sociodemographic Characteristics of Survey Participants.

https://doi.org/10.1371/journal.pntd.0013572.t002

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Flow diagram of survey participant enrollment and inclusion in data analysis.

A characterization of the 90 participants excluded from the multivariable model due to missing covariate data are available in S6 Text.

https://doi.org/10.1371/journal.pntd.0013572.g001

5C’s framework

Below we present a summary of both qualitative (relevant quotes) and survey findings, organized under the 5C’s framework.

Confidence

Both qualitative and quantitative data indicated generally positive attitudes towards a potential dengue vaccine, and the scales showed relatively low vaccine hesitancy [Tables 2 and 3]. The following statements from FGD reflected this sentiment: “We assume that a vaccine is good for your health. Who wouldn’t want to get it? But as long as we know what symptoms [side effects] we will get. Right?” (FG1). Survey results were consistent with FGDs findings, with 81.9% (723) of respondents indicating they were willing to receive a new dengue vaccine when available [Table 2]. Two main themes emerged in FGDs regarding confidence about a new vaccine: 1) the need for clear information about the vaccine to increase people’s confidence in it, including information on its efficacy, number of doses required for protection, and possible side effects, and 2) vaccine information should come from trusted institutions, naming the Ministry of Health and health professionals as such [Table 3].

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Themes Identified in the FGDs and Representative Quotes.

https://doi.org/10.1371/journal.pntd.0013572.t003

Most survey participants highly trusted the benefits of vaccines (69.6%) [Table 4]. There was mild preoccupation about potential future vaccine effects both among FGD and survey participants. FGD participants said they wanted information on the vaccine’s components, duration of protection, potential side effects, and whether people in other areas or countries had already been vaccinated – or if they would be first: “I will wait for others to get it, but I am not going to risk getting it first” (FG11). There were also questions regarding vaccine eligibility, ranging from whether it was available to all or if there were conditions such as having had dengue before, not being pregnant or not having co-morbidities: “What problems would it cause in a pregnant woman or a diabetic [person] or any other disease a person may have?” (FG10).

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. 5C’s by dengue vaccine acceptability category: vaccination acceptors, those who are unsure, and vaccination refusers.

https://doi.org/10.1371/journal.pntd.0013572.t004

In the vaccination attitudes examination scale (VAX), the top three statements that participants agreed with focused on trusting vaccines against severe infectious diseases (83%), feeling protected after being vaccinated (81%), and feeling safe after vaccinations (78%) [Fig 2]. In our adjusted multivariate model showed that participants who expressed doubts about the COVID-19 vaccine also expressed doubts about the dengue vaccine (OR 2.17, 95% CI 1.23-3.79) [Table 5]. Likewise, those who had reduced confidence in the COVID-19 vaccine due to its rapid development were twice as likely to express uncertainty about getting the dengue vaccine than acceptors (OR 2.17, 95% CI 1.17, 3.98) [Table 5].

Download:

• PNG
  larger image
• TIFF
  original image

Table 5. Unadjusted and adjusted multivariate analysis of factors associated with being unsure about dengue vaccination, compared to acceptors (reference).

https://doi.org/10.1371/journal.pntd.0013572.t005

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Participant responses to Vaccination Attitudes Examination (VAX) Scale.

Note: Percentage on the left represents the sum of “Strongly disagree” and “Disagree” responses, whereas the percentage on the right represents the sum of “Agree” and “Strongly agree” responses”.

https://doi.org/10.1371/journal.pntd.0013572.g002

Complacency

Complacency was assessed based on a person’s perceived risk of getting infected and, if infected, their perception of disease severity. Most survey participants believed dengue can lead to death (94.4%) yet there were differences between the groups, as only 81.3% of refusers and 88.3% of “unsures” believed this compared to the 96.1% of acceptors [Table 4]. The perceived severity of dengue seemed to influence FGD participants’ attitudes toward vaccination, with some expressing a strong desire for protection and high interest in a future dengue vaccine, particularly to prevent severe disease: “When you get severe dengue, then you start to worry, then you start to become more conscious that dengue is no joke!” (FG 13). There was also recognition that one could get dengue multiple times in these discussions, worthy of prevention: “I would get the vaccine, because… there are four types of dengue… I would fear hemorrhagic dengue the most. Besides that, dengue comes every year or when the river grows in our region. I would get it [vaccine]” (FG3). Only 59.3% of survey participants believed you can get dengue more than once [Table 4], and in the crude multivariate model, unsure participants were less likely than acceptors to believe they could get dengue again if previously infected, but this was not significant in the adjusted model (OR 0.71, 95% CI 0.43-1.16) [Table 5]. Mothers in the FGDs showed interest particularly in vaccinating their children, with one suggesting that vaccine eligibility should start with children. Even parents hesitant to vaccinate themselves acknowledged the importance of protecting their children as part of their parental duty. Sex was found to be a significant predictor for being unsure about getting vaccinated in the crude model but not significant in the adjusted model: men were more likely to be unsure about getting vaccinated than women (OR 1.63, 95% CI 0.91, 2.93) [Table 5].

In the FGDs, people asked whether getting vaccinated would mean they would no longer need to participate in vector control activities: “As for me, what I want to know is if I get vaccinated, then I no longer need to kill my mosquitoes?” (FG2).

Convenience

Participants in FGDs reported long waiting lines as a barrier to accessing the COVID-19 vaccine; some specified queuing for hours in the morning, which was particularly challenging for individuals with work or caregiving responsibilities. Some admitted to a “known practice” of hiring others to hold their place in line, though this option was not affordable for everyone: “Yes, some people got up very early, took their chairs and sat at the front of the line, and then would sell their spot for 10-20 soles (US$3-6)” (FG1). However, none stated long waiting lines as a reason not to get the vaccine.

Most survey respondents believed the dengue vaccine should be free, as with the COVID-19 vaccine, with less than half (40.2%) of those surveyed reporting willingness to pay for it [Table 4]. Those unsure of getting vaccinated reported less willingness to pay for the vaccine than the acceptors (OR 0.40, 95% CI 0.23, 0.67) [Table 5]. In the FGDs, it was pointed out that the vaccine should be covered by fees paid in taxes or part of government healthcare coverage. These participants also suggested prices ranging from ~$1.40-$13.60 USD, if necessary. Some argued that the vaccine cost was justified compared to the expenses of treating severe or hemorrhagic dengue. One individual noted that even the cost of over-the-counter drugs for mild dengue could exceed the vaccine cost: “because one might not only be in a hospital… Sometimes you will need to buy something [medication] for the headaches. For prevention, I might as well get a vaccine that costs 30 soles [~US$ 8] and then I won’t have other expenses” (FG1). Participants in FGDs also highlighted that large families would struggle to afford multiple doses and proposed sliding scale costs or discounts, such as ~$8.15 USD per person or ~$13.60 USD for two people.

Context

Among participants who reported receiving a third or fourth dose of the COVID-19 vaccine (n = 737), 86.4% were classified as acceptors for a vaccine against dengue, compared with 71.9% among those who were unsure and 62.5% among refusers [Table 4]. Reasons given for the COVID-19 vaccine in the FGDs included having high-risk household members, community pressure to do so, and feeling sufficiently informed to decide. Some FGD participants cited obligation due to Peruvian laws requiring proof of vaccination for travel, banking, grocery shopping, or work as their main reason for getting vaccinated – with some saying it was the nudge they needed to get it.

In our study sites, all FGD and survey participants had either had dengue or knew someone who had it. Additionally, 15.2% of survey participants reported close experience either through personally having or knowing someone with severe dengue [Table 4]. Study site and previous participation in dengue research did not have an impact on survey participants’ vaccine acceptability [Tables 4 and 5]. There was also a higher level of dengue vaccine hesitancy among survey participants with a technical or higher education level, a wealthier socioeconomic status, and those living in an urban area. Specifically, participants with a technical or higher education level were more likely to express being unsure about getting a dengue vaccine compared to those with a primary level education (OR 3.33, 95% CI 1.31, 8.97) [Table 5]. No differences in hesitancy were found were found between age groups, occupation of the participant, and close experience with dengue/COVID-19. Individuals who were unsure about a vaccine were also less likely than acceptors to know how dengue is transmitted (via the bite of an infected mosquito) (OR 0.45, 95% CI 0.23, 0.93) [Table 5].

Drawing on their experience with COVID-19 vaccination, FGD participants recommended establishing vaccination posts in central, frequented areas, such as city centers. Despite these suggestions, FGD and survey respondents also identified challenges in reaching certain populations, including individuals with specific religious affiliations (e.g., evangelical), indigenous communities; and, only mentioned in the survey, people living in peri-urban areas and of lower socioeconomic status [Table 6].

Download:

• PNG
  larger image
• TIFF
  original image

Table 6. Targeted Strategies for Vaccine Outreach.

https://doi.org/10.1371/journal.pntd.0013572.t006

Communication

At both sites, when asked about most trusted sources of communication about a new dengue vaccine, FGD participants expressed trust in their physicians, healthcare friends, and health professionals or community health workers from local facilities to endorse a vaccine. They also stated they would be more likely to get vaccinated for dengue if the Ministry of Health approved it. In Iquitos, one name came up in every FGD: most valued the opinion of a local physician and parish priest who had been influential and demonstrated commitment to the community at the height of the COVID-19 pandemic. In contrast, Piura participants expressed distrust in their local leaders and felt abandoned by the local government, and yet, despite these negative perceptions about health facilities in both sites, FGD participants expressed wanting approval from these same organizations and health professionals before receiving a future dengue vaccine. Participants who were unsure about getting a dengue vaccine were less likely to identify health centers or local primary care facilities as a source of dengue vaccine information than acceptors (OR 0.55, 95% CI 0.29, 1.02) [Table 5]. Non-health care sources were also mentioned: FGD participants described receiving and believing information about COVID-19 vaccines from non-healthcare sources such as radio, television (local and national programs), news channels, internet sources (including social media), religious organizations, and family and friends.

To promote a dengue vaccine, FGD and survey respondents recommended used the radio (including regional programs such as La Voz de la Selva, Radio Loreto, Radio Saby Star, and national programs, such as Radio Exitosa), TV programs, social media like Facebook, and using a truck with loudspeakers to drive around neighborhoods to share information (more commonly suggested in Piura where this method is used routinely in some communities) [Table 6]. They also suggested press releases. For healthcare engagement, participants suggested holding open forum townhalls where health professionals could educate and answer questions.

Survey data showed that 50% of those unsure and 43.8% of refusers were influenced by their experience with negative public information about the COVID-19 vaccination [Table 4]. Participants in FGDs noted that clear messaging is crucial, describing how changing COVID-19 vaccine recommendations had eroded their trust. Importantly, in most FGD, participants mentioned that health promotion messages must be tailored since sub-populations may face different barriers, specifying certain religious groups that had refused COVID-19 vaccination and indigenous populations that ‌‌may need different outreach strategies due to linguistic, cultural, and geographic barriers.

Comparison of qualitative and quantitative findings

Qualitative and quantitative findings were largely convergent [S2 Table]. Both data sources demonstrated high acceptability of a potential dengue vaccine, with confidence—shaped by trust, transparency, and prior vaccination experiences—emerging as a central determinant. Strong convergence was also observed in the domain of communication: participants emphasized the importance of disseminating information through multiple media channels (e.g., radio, loudspeakers, social media) and from trusted sources. FGD participants also emphasized high levels of trust in health professionals, although this did not quite translate into a significant finding in the quantitative analysis. Misinformation and residual mistrust related to the COVID-19 vaccination experience emerged in both datasets, underscoring the need for clear and consistent messaging. FGDs, however, provided more detailed insight into informational needs, including safety, side effects, eligibility criteria, number of doses, and prior vaccine use. In the domain of convenience, logistical barriers such as waiting times were not described as deterrents in either FGDs or surveys, yet cost emerged as a salient factor; willingness to pay was lower among participants classified as unsure, and FGD participants generally expected vaccination to be free. For complacency, qualitative findings emphasized lived experience with dengue as a motivator, whereas survey data highlighted perceived risk and knowledge of dengue transmission as factors associated with acceptance. Finally, contextual findings were complementary: FGDs identified Indigenous and certain religious communities as potentially harder to reach, while survey analyses identified sociodemographic correlates of vaccine uncertainty.

Limitations

Several limitations must be considered. First, this study took place in specific urban and peri-urban areas of two dengue endemic sites, which may not reflect views across other geographical and socio-economic settings in Peru or Latin America. Second, this study took place two years after the start of the pandemic, and two years prior to the Qdenga rollout in Peru. Findings reflect attitudes towards a hypothetical dengue vaccine which do not always translate into behaviors. Third, we focused on individuals who were unsure about a vaccination for programmatic reasons; our sample of refusers was small, and we did not explore their perspectives in depth. Future research could explore what is driving the perspectives amongst vaccine refusers. Fourth, self-reported data on vaccine acceptance and hesitancy may be subject to social desirability bias; this may be exacerbated by discussing a dengue vaccine which is not yet available. In other words, the study’s focus on attitudes toward a hypothetical dengue vaccine may not capture actual behaviors and decision-making processes once a vaccine becomes available. Finally, almost three quarters of our survey participants were women, which is not representative of the whole population and may affect generalizability of the study, yet in our experience women tend to be the decision makers about health and vaccines in their families in Peru.