Dengue risk perception and public preferences for vector control in Italy and France: utility and regret-based choice experiments

BackgroundDengue is rapidly emerging in parts of Europe. How households value vector control attributes, and whether inferences depend on decision models or message framing, is unclear. MethodsWe conducted a split-ballot online experiment among adults in Italy and France, as well as a hotspot subsample from Marche, Italy. National samples included 1,505 respondents in Italy and 1,501 in France; 183 respondents were recruited in Marche. Participants were randomised to a discrete choice experiment (random utility maximisation) or a regret-based choice experiment (random regret minimisation) and to one of three pre-task messages (control, loss aversion, community values). Each respondent completed 12 choice tasks comparing two dengue control programmes and an opt-out. We estimated mixed logit and mixed random-regret models with random parameters and treatment effects. ResultsAcross frameworks, nearby cases and high mosquito prevalence were the dominant drivers of programme uptake, whereas cost and operational burden were secondary. In pooled analyses, loss-aversion messaging increased the weight on high mosquito prevalence in both models (from 0.483 to 0.547 in the utility model; from 0.478 to 0.557 in the regret model). Cost effects were small nationally but larger in the hotspot subsample. ConclusionsRisk salience dominates preferences for dengue vector control in these European settings. Random utility and random regret models yield consistent rankings of attributes but differ in behavioural interpretation and some secondary effects; messaging effects were modest and context dependent. Key messagesO_LIPublic preferences for dengue vector control are primarily driven by perceived epidemiological risk, especially proximity to cases and mosquito prevalence, rather than cost or inconvenience. C_LIO_LIRandom utility and random regret models produce consistent rankings of key attributes, indicating robust findings across behavioural decision-making frameworks. C_LIO_LIBehavioural messaging (e.g., loss aversion or community framing) has only modest and context-dependent effects on preferences compared to the strong influence of local risk context. C_LI