Ethics statement

To achieve continuous consent, written informed consent was obtained electronically from all panelists in their preferred language at the start of each questionnaire. Confidentiality and anonymity were ensured by using participants’ study codes. Ethical approval was obtained from the Hamilton Integrated Research Ethics Board (HiREB) at McMaster University, Hamilton, Ontario, Canada, to conduct this study (Project Number: 15321).

Study design

We used a sequential, multi-method study design. We conducted a qualitative scoping review of the literature [17], which informed the subsequent electronic Delphi (e-Delphi) stage (quantitative and qualitative). The e-Delphi method is a recommended tool for synthesizing professional opinion [18], and developing conceptual frameworks [19]. We reported this study according to the ACCORD (ACcurate COnsensus Reporting Document) guidelines [20].

Population of interest

The target population in this study were English-, Spanish-, French- and Portuguese- speaking researchers and practitioners from LAC with responsibilities and/or technical profiles in entomology, infectious disease epidemiology, sociology, and ecology related to dengue.

e-Delphi panelists were of three primary participant groups:

1. (i) Researchers actively/previously involved with research on dengue who are/were conducting research on the eco-bio-social determinant(s) of dengue transmission affiliated with LAC institutions.
2. (ii) Grantees of the World Health Organization’s Special Programme for Research and Training in Tropical Diseases (TDR) project funded by Canada’s International Development Research Centre (IDRC) titled: “Eco-Bio-Social Approach on Dengue in Latin America and the Caribbean” (IDRC Project ID: 104951).
3. (iii) Previous/acting policy and program practitioners involved with dengue interventions or strategic decision-making in governmental or non-governmental capacity.

Sampling strategy and participant recruitment

The participants were obtained from the target population by employing a non-randomized, purposive sampling strategy using two recruitment methods. First, we recruited authors of papers from the scoping review stage via email. Second, we recruited grantees of the IDRC-funded TDR project via email. We employed the snowballing strategy by asking invited participants to recommend colleagues. The recruitment period for this study began on January 1, 2023, and ended on April 30, 2023.

Sample size

The e-Delphi panel size relies on group dynamics in reaching an agreement among panelists rather than on statistical power [21]. We aimed for the recommended sample size of 20 panelists [22], since previous literature suggests that this number is sufficient to ensure moderate replicability [23].

Data collection

Between May 2023 to November 2023, panelists answered online questionnaires on the web-based e-Delphi platform, Welphi (https://www.welphi.com/). Participation was quasi-anonymous, where panelists’ identities were known to the research team but not to one another. This study contained 3 rounds of iteratively developed questionnaires, which were designed in English and translated to Spanish and Portuguese, and then back translated to English to maintain accuracy. Questionnaires were pilot tested in the target languages for comprehensibility with researchers in LAC who were not participants of this study.

After each round, the responses from the questionnaires were aggregated into a report and shared with the panelists. Panelists had the opportunity to adjust their answers and add further comments.

Data analysis

Data were analyzed quasi-anonymously. We defined consensus as 70% or more of panelists agreeing on all categories of the framework (i.e., percent agreement). We chose 70% as an indication that a substantial number of panelists agreed with what was proposed, while retaining room for dissenting views. There is no consensus as to what is the “best” criterion to use, as these cut-offs are arbitrary, however this threshold is well-established in e-Delphi literature for public health research [19,26–28]. Eq (1) shows the calculation for percent agreement.

(1)

Characteristics of panelists

Out of 50 individuals invited to participate in this study, 11 agreed to participate as panelists. Panelists were from Argentina, Bolivia, Brazil, Colombia, Mexico, Panama, and Uruguay. Two of the panelists were previous grantees of the IDRC-funded TDR project. Some panelists had experience with dengue policy and programming in LAC. All 11 panelists participated in Round 1; 10 participated in Round 2; and 9 participated in Round 3. Panelists’ characteristics are described in Table 1.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics of the panelists.

https://doi.org/10.1371/journal.pgph.0004115.t001

Ranking of relevance

According to panelists, seasonal weather and urban microclimatic oscillations were the most relevant ecological factors in determining dengue transmission. The most relevant biological factor was vectorial capacity and competence. The most relevant social factor was urbanization and land use (Fig 1). The least relevant factors were topographical characteristics, pathogen evolution, age, and sex and gender, and were removed from the subsequent concept mapping exercise.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Ranking the relevance of the eco-bio-social factors determining dengue transmission in LAC.

https://doi.org/10.1371/journal.pgph.0004115.g001

Concept mapping exercise

The average importance value for the ecological factors was 4.71 out of 5, and the average operability value was 2.66 out of 5, for 5 ecological factors. On average, panelists viewed that the most important ecological variable was seasonal weather oscillations and meteorological factors, although, this variable was the least operable on average. The only variable in the ‘go-zone’ was climate change.

The average importance value for the biological factors was 3.94 out of 5, and the average operability value was 2.48 out of 5. On average, of the 5 biological factors ranked, panelists viewed that the most important and operable biological variable was the EIP of the virus. The least important variable was serotype competition, and the least operable variable was vectorial capacity and competence.

The average importance value for the social factors was 4.28 out of 5, and the average operability value was 3.83 out of 5. On average, of the 10 social factors, the variables in the ‘go-zone’ were urbanization and land use, human population density, and public health infrastructure and services. The variables in the no-zone were human mobility and migration, community cultural practices, and individual socioeconomic status. The operability average of the social factors was higher compared to the operability averages of the ecological and biological factors.

The bivariate plot from the concept mapping exercise is shown in Fig 2.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Bivariate ‘go-zone’ plot showing importance and operability rankings for the ecological, biological, and social factors determining dengue transmission in LAC.

The quadrants were constructed using the average importance (x) and operability (y) values. The ‘go-zone’ quadrant on the upper right shows the variables that were ranked above average for both importance and operability indicators in the category. The ‘no-zone’ quadrant on the bottom left shows the variables that were ranked below average in both importance and operability indicators in the category. Importance indicator scale: 1 = not important at all; 2 = not important; 3 = somewhat important; 4 = important; 5 = very important. Operability indicator scale: 1 = not operable at all; 2 = not operable; 3 = somewhat operable; 4 = operable; 5 = very operable.

https://doi.org/10.1371/journal.pgph.0004115.g002

Conceptual framework development

Over the 3 consultation rounds, panelists evaluated an eco-bio-social conceptual framework and achieved over 70% agreement (i.e., consensus) on each of the items in a 10-item Likert scale (Table 2). The final framework included 16 ecological factors, 11 biological factors, and 28 social factors (Fig 3).

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Summary of questionnaire items and consensus scores (% agreement) from panelists on an eco-bio-social conceptual framework for dengue transmission in LAC over three consultation rounds^a,b.

https://doi.org/10.1371/journal.pgph.0004115.t002

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Final visual, comprehensive, eco-bio-social conceptual framework for dengue transmission in LAC.

The directivity of the arrows represents the course of action and interaction between variables; dashed arrows represent external and complex pathways of actions and interactions between components; peripheral, political and programmatic factors are signified by dashed boxes; and underlined terms are further expounded in adjacent closed boxes. (G × G – genotype-by-genotype interactions).

https://doi.org/10.1371/journal.pgph.0004115.g003

According to panelists, the audiences that may benefit from the application of the conceptual framework include researchers and academics; governmental policy and decision-makers; health institutions, program managers and health authorities; and communities. Table C in S1 Text describes panelists’ views regarding how utilizing the consensus-based eco-bio-social conceptual framework may support the research, and policy and program agendas described below.

Research agenda development

Research agenda.

The research themes in the research agenda developed by panelists included: sociocultural research; entomological, viral, and immunological research; vaccine development research; and epidemiological risk modeling research (Table E in S1 Text). Overall, 89% of panelists agreed that the 4 research themes accurately represented what is needed to address the research gaps.

The research gaps and themes contemplated by the panelists were integrated into a research agenda that included 3 research themes: ecological and environmental research; biological and immunological research; and social and cultural research (Fig 4).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Integrated research agenda for eco-bio-social research on dengue in LAC.

https://doi.org/10.1371/journal.pgph.0004115.g004

Policy and program agenda development

Policy and program agenda.

The policy and program agenda was categorized into 4 categories, including: government investments, integrated programs, intersectoral approaches, and innovative practices (Fig 5).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Policy and program agenda for dengue prevention and control in LAC.

https://doi.org/10.1371/journal.pgph.0004115.g005

Panelists evaluated the policy and program agenda based on 4 criteria: applicability, generalizability, feasibility, and acceptability (Table F in S1 Text; S2 Fig). Overall, 88% of panelists either agreed or strongly agreed that the agenda can improve dengue prevention and control in LAC. Additionally, 44% of the panelists either disagreed or somewhat disagreed that the agenda can be equitably implemented in the region. A group of panelists either disagreed or somewhat disagreed (44%) that the policy and program agenda can be generalizable to many or all countries in LAC. The political feasibility of the agenda was met with a degree of variability among panelists, however most panelists (55%) agreed or strongly agreed that the agenda was politically feasible.

Conceptual framework for dengue transmission in LAC

While conceptual models for understanding dengue transmission have previously been proposed, existing socioecological and biophysical-ecological frameworks often lack an integrative structure that uses a systems thinking approach to organize diverse components of transmission. Although these frameworks are grounded in empirical evidence, they overlook the perspectives of key stakeholders, limiting their relevance in real-world settings. To integrate the eco-bio-social factors that affect dengue transmission and emergence, Tana et al. (2012) adopted a Venn diagram to illustrate factors situated within the episystem [29]. However, this eco-bio-social framework is generic, as described by the authors, and lacks depth and interrogation of how and at which scale the factors impact parameters of dengue transmission dynamics. Our eco-bio-social framework addresses these gaps and provides unique contributions to the literature due to its structured design, integration of socioeconomic, biological, and climatological determinants, and validation through professional consensus, rendering it regionally grounded.

Utility of the conceptual framework by key audiences

Policy recommendations for dengue prevention and vector control in LAC

In our consultations, panelists considered investment domains for dengue prevention and control which included investments in research, vaccine development, and public health infrastructure. Training health personnel to deliver community education through participatory approaches, such as clean-up campaigns, was emphasized to empower community members to take ownership of vector control at home and in their communities [30]. Building resilient monitoring and early warning information systems and urban ecosystem health assessments, which have been useful control strategies implemented by governments in Asia [31,32], were also recommended. Moreover, intersectoral collaboration between government sectors, such as sanitation, urban management, health services, and surveillance was a vital tool highlighted by panelists to develop locally tailored prevention programs.

Panelists mentioned the need for integrated vector surveillance and control programs throughout administrative levels, with cost-effectiveness and ecological sustainability evaluated through community-government partnerships. Although community cultural practices were ranked low on both importance and operability by panelists, possibly due to the observation that cultural behaviors are deeply ingrained and require long-term, trust-based efforts, panelists noted that culturally sensitive community-based approaches integrating entomological surveillance and participation may contribute to better dengue outcomes. Furthermore, innovative practices, such as releasing Wolbachia-infected mosquitoes to hinder DENV transmission, were viewed as promising, cost-effective biological control strategies. Given its proven benefits [33–35], expanding cost-effective biological vector control strategies could enhance prevention efforts [36].

Overall, panelists diverged in opinion regarding the feasibility of scaling dengue prevention strategies equitably across LAC. Dengue policies can only exist when they are supported by capable urban infrastructure [37]. However, the enormous investment required for urban infrastructure alone in developing countries poses a significant equity challenge, rendering the adaptability of intersectoral approach unfeasible. Long-term political will and cross-border collaboration in LAC is necessary, especially since the introduction of a widely available vaccine is increasingly likely [38]. Further, sustainability of dengue policies and programs depends not only on political commitments [39], but also on social mobilization and contextually tailored contributions. For instance, although gender was ranked as “less relevant” by panelists, emerging policy perspectives suggest that the consideration of gender norms as a contextual factor in transmission is necessary for achieving equity and, by extension, sustainability [40].