The importance of being urgent: The impact of surveillance target and scale on mosquito-borne disease control

Abstract

With the emergence or re-emergence of numerous mosquito-borne diseases in recent years, effective methods for emergency vector control responses are necessary to reduce human infections. Current vector control practices often vary significantly between different jurisdictions, and are executed independently and at different spatial scales. Various types of surveillance information (e.g. number of human infections or adult mosquitoes) trigger the implementation of control measures, though the target and scale of surveillance vary locally. This patchy implementation of control measures likely alters the efficacy of control. We modeled six different scenarios, with larval mosquito control occurring in response to surveillance data of different types and at different scales (e.g. across the landscape or in each patch). Our results indicate that: earlier application of larvicide after an escalation of disease risk achieves much greater reductions in human infections than later control implementation; uniform control across the landscape provides better outbreak mitigation than patchy control application; and different types of surveillance data require different levels of sensitivity in their collection to effectively inform control measures. Our simulations also demonstrate a potential logical fallacy of reactive, surveillance-driven vector control: measures stop being implemented as soon as they are deemed effective. This false sense of security leads to patchier control efforts that will do little to curb the size of future vector-borne disease outbreaks. More investment should be placed in collecting high quality information that can trigger early and uniform implementation, while researchers work to discover more informative metrics of human risk to trigger more effective control.

(Meta)population dynamics determine effective spatial distributions of mosquito‐borne disease control

“…The model presented here builds on a model previously employed in Schwab et al (2017) to explore the efficacy of different types of surveillance data to trigger effective control, and similarly builds on a robust existing literature of models of mosquito populations incorporating density-dependent feedback in larval populations (Braks et al 2004, Magori et al 2009). All simulations were run using Matlab version R2017b (MathWorks, Natick, MA, USA).…”

“…Any time an arbovirus outbreak occurs, emergency vector control measures need to be implemented quickly; the time it takes to identify hot spots can postpone the implementation of control measures, leading to larger outbreaks (Schwab et al. ). Also, due to complex oviposition site selection dynamics, applying control measures only in hot spots may leave a sufficient abundance of immature mosquitoes in the remaining larval pools to maintain transmission of mosquito‐borne diseases in some ecological contexts (Wasserberg et al.…”

Coordination among neighbors improves the efficacy of Zika control despite economic costs

“…Mosquito population dynamics and disease transmission dynamics are described by a series of discrete time, SIR-type difference equations. These equations are the same as in [28] (reproduced in Appendix 1 for ease of readership). During the period of surveillance and control, participating patches are treated the following day when a threshold for control is exceeded (Table S1).…”

“…Costs can also take the form of time delays since time spent communicating with other regions or waiting for other regions to be ready for coordinated treatment can delay the start of treatment in an affected patch. Previous findings suggest that surveillance resulting in earlier treatment after an outbreak is most effective in reducing human infections [28]. Even if coordinated treatment has the potential to improve disease control, the reduction in efficacy from delaying treatment may outweigh the benefits from coordination.…”

“…To answer these questions, we build on a spatially explicit SIR model first presented in [28]. We examine two different possible costs of coordination: costs in time or costs in resources.…”