LAUSR

Every year vector-borne diseases (VBDs) result in significant illness and death in the United States. Between 2004 and 2019, >800,000 cases of VBDs were reported to the U.S. Centers for Disease Control and Prevention, with the number of reported disease cases more than doubling over this period.1 Furthermore, in the summer of 2021, the largest local outbreak of West Nile virus (WNV) disease on record occurred in Arizona.2 This event, which was largely obscured by the COVID-19 pandemic, was likely influenced by a wetter-than-average monsoonal season.3 In an article in this issue of Environmental Health Perspectives, Gorris et al.4 used a random forest model to explore how seasonal weather variables influence the spatial distribution and incidence of WNV disease in humans. Their model successfully identified the high-incidence regions of the United States and linked high WNV disease incidence to dry and cold winters in addition to wet mild summers. Prevention of VBDs is highly challenging and complicated by numerous factors, including the lack of licensed vaccines for VBDs endemic in the United States, insecticide resistance, and public opposition to the use of pesticides, to name a few.5 For many outbreaks, particularly outbreaks of WNV disease, by the time public health agencies and municipalities respond, the outbreak may already be subsiding. For example, in the WNV disease outbreak that occurred in the summer of 2012 in Texas, a sharp decline in reported cases occurred immediately following aerial spraying of insecticides.6 On closer observation, however, by the time spraying was completed, >90% of the human cases had already occurred, and the outbreakwaswaningwith the onset of fall.6 The deficiencies of a public health response to a VBD outbreak are easy to point out retrospectively; however, in our experience, the path forward is not nearly as clear during an actual event. There can be a lag between when surveillance data are collected and when they can be acted upon by those empowered to make key decisions.6 The process is multifaceted and complex, and involves multiple teams of specialists (entomologists, virologists, diagnostic specialists, epidemiologists, and policy experts). In addition, because response options often spark controversy, extensive communication efforts and public discourse are required prior to implementation efforts.5 Decision support tools that use a combination of historical trends, recent surveillance data, and environmental variables, if available, could provide a potentially powerful resource for predicting and preparing for VBD outbreaks, with the ultimate goal of facilitating a timely and effective response. Disease models have been developed for numerous purposes, including understanding the why and how of past outbreaks or disease trends, forecasting where and when cases are most likely to occur, predicting how bad an outbreak might be, and determining how best to intervene.7–9 Models that predict the risk of WNV disease have previously7 been reviewed extensively and grouped into three categories: a) those that predict relative risk associated with specific geographic regions, b) those that typically use data on weather and other variables to provide early warning signals, and c) those that provide early virus detection and can be integrated with current surveillance data to predict risk, usually in a conscribed area. Such models can be valuable public health tools, assisting municipalities in their efforts to prepare for and forecast disease outbreaks. In addition, models may be integrated with surveillance data to predict the effectiveness of a repertoire of available interventions and to support decisions, such as whether to use mosquito larvicides or adulticides or to use truck-based or aerial spraying, as well as where and when to apply these tools to optimize results.8 In their new paper, Gorris et al.4 provided an exceptional description of the ecological and epidemiological variables that drive VBD outbreaks where numerous factors contribute collectively to the risk of human infection. These complexities are the primary challenge to developing effective decision support tools that can be applied in time to halt an emerging epidemic. Identifying weather-related factors associated with the observed spatial structuring ofWNV disease is an important step in elucidating the ecological mechanisms that result in persistent trends in annual incidence.10 This, in turn, can lead to improved predictive capacity and control measures. Gorris et al. emphasized that their findings related to the spatial structure of WNV disease incidence are independent from, but complementary to, studies that focused on interannual variability in WNV cases.4 The use of a random forest modeling approach allowed the authors to consider interactions between weatherrelated factors to better describe complex transmission dynamics without mechanistically prescribing the nature of the relationship. In addition to providing information that may be helpful in preparing for an outbreak, the authors pointed out that identifying weather-related factors associated with WNV disease spatial structuring may also provide insight into predicting the potential effect of climate change on future WNV disease incidence and distribution, an observation that could be extended to many other VBDs. Identifying the major drivers, especially with an eye to climate drivers, provide public health officials with vital insight for predicting and combatting VBDs in a complex and changing world.