By looking at 15 years of mosquito-trapping data in Chicago, researchers have developed a model to pinpoint the best locations for traps to detect West Nile virus earlier and more accurately. Ranked by the Centers for Disease Control and Prevention as the leading cause of mosquito-borne disease in the U.S., West Nile virus is spread by mosquitoes in the genus Culex, such as the southern house mosquito (Culex quinquefasciatus), shown here. (Photo by Lauren Bishop, CDC Public Health Image Library)

By Andrew Porterfield

Andrew Porterfield

The West Nile virus has spread rapidly across the United States, infecting more than 52,000 people since its 1999 emergence. The U.S. Centers for Disease Control and Prevention (CDC) ranks West Nile virus as the leading cause of mosquito-borne disease in the country. There is no cure or prevention method, leaving mosquito control the only way to manage the infection’s spread.

West Nile virus is spread in Culex species mosquitoes, and birds bitten by these mosquitoes can amplify viral spread by re-infecting yet more mosquitoes. Most people bitten by carrier mosquitoes show no symptoms, indicating that the virus’s true prevalence is under-reported.

To monitor mosquitoes carrying West Nile, public health officials and mosquito abatement agencies set mosquito traps, which capture mosquitoes in large numbers, and then pools of up to 50 mosquitoes are tested at once for the virus. But there has been scant research on determining the best locations to place these important traps. A team of researchers at the University of Illinois Urbana-Champaign has developed a novel statistical model that can predict the most likely areas of mosquitos, and recommend where to place traps. Their study was published this month in the Journal of Medical Entomology.

Quite a few factors play into a trap’s ability to accurately predict West Nile virus prevalence. These include surrounding terrain, local mosquito and bird populations, and human demographics (income, education, population density).

By looking at 15 years of mosquito-trapping data in Chicago, Rebecca Smith, Ph.D., associate professor of epidemiology at the University of Illinois Urbana-Champaign, and colleagues have developed a model to pinpoint the best locations for traps to detect West Nile virus earlier and more accurately.  (Photo by Fred Zwick, UIUC News Bureau)

The researchers looked at 1,062 traps placed in Chicago and surrounding suburbs from 2004 to 2018, by 19 mosquito abatement agencies. The researchers then worked in three phases to constructed statistical models to determine how to predict the best placement for traps.

First, they created a logistical model to quantify each trap’s historical performance. They determined the risk of West Nile as using a method known as Maximum Likelihood Estimation. They also arrived at average sensitivity and specificity values for each trap.

Second, they developed a scoring for each trap, using average specificity and sensitivity values developed during the first step.

Third, and finally, they characterized locations with higher trap scores, which pointed to the best places to locate traps.

The researchers found that the best locations for traps in the Chicago area had more than 10,000 people within a 1.5 kilometer radius, high mosquito populations, and historically variable West Nile test results. “While these findings may confirm what mosquito abatement districts have long observed in practice, this is the first analysis the quantitatively substantiates those insights using empirical data,” the researchers write.

The study also revealed some impacts of socioeconomic factors in trap placement. Their data suggest that, in regions of high poverty and low levels of education, traps predicted the incidence of cases, even if cases were never reported there. This indicates possible underreporting of West Nile virus in these areas.

The statistical models don’t have to be specific to West Nile virus, the researchers say. “For instance, this could be useful to decide which parts of a field to monitor for crop pests, which water bodies to monitor for invasive species, or where to place pollution monitors to predict a bad air day,” says Rebecca Smith, Ph.D., associate professor of epidemiology at the University of Illinois and senior author of the study. “I’d love to see it applied more broadly.” Her team is now creating an app to make it easier for people to use their models on specific spatial sampling problems.

Andrew Porterfield is a writer, editor, and communications consultant for academic institutions, companies, and nonprofits in the life sciences. He is based in Camarillo, California. Connect with him via LinkedIn or via email at [email protected].