Spatial multi-criteria decision analysis for the selection of sentinel regions in tick-borne disease surveillance

Abstract

Background: The implementation of cost-effective surveillance systems is essential for tracking the emerging risk of tick-borne diseases. In Canada, where Lyme disease is a growing public health concern, a national sentinel surveillance network was designed to follow the epidemiological portrait of this tick-borne disease across the country. The surveillance network consists of sentinel regions, with active drag sampling carried out annually in all regions to assess the density of Ixodes spp. ticks and prevalence of various tick-borne pathogens in the tick population. The aim of the present study was to prioritize sentinel regions by integrating different spatial criteria relevant to the surveillance goals.

Methods: We used spatially-explicit multi-criteria decision analyses (MCDA) to map priority areas for surveillance across Canada, and to evaluate different scenarios using sensitivity analyses. Results were shared with stakeholders to support their decision making for the selection of priority areas to survey during active surveillance activities.

Results: Weights attributed to criteria by decision-makers were overall consistent. Sensitivity analyses showed that the population criterion had the most impact on rankings. Thirty-seven sentinel regions were identified across Canada using this systematic and transparent approach.

Conclusion: This novel application of spatial MCDA to surveillance network design favors inclusivity of nationwide partners. We propose that such an approach can support the standardized planning of spatial design of sentinel surveillance not only for vector-borne disease BDs, but more broadly for infectious disease surveillance where spatial design is an important component.

Keywords: Lyme borreliosis; Lyme disease; Multi-criteria decision analysis; Public health decision-making; Sentinel surveillance; Vector-borne diseases.

Fig. 1

Diagram of general steps in multi-criteria analysis

Fig. 2

Maps projecting MCDA Phi scores from scenario a) risk-based, b) population, c) environment, and d) weighted. The shading indicates the relative performance across the set of alternatives – a higher score represents a better performance according to the criteria and weighting used within the models. Specifically, scores represent how performance is distributed across space at the provincial level according to the MCDA, depending on criteria weightings. These maps were presented back to stakeholders to support decision-making for selection of sentinel regions

Fig. 3

Sentinel regions (n = 37) for the Canadian Lyme Sentinel Network (CaLSeN) (pink dots)