Assessing real-time Zika risk in the United States

Here, the authors of an article published recently in BMC Infectious Diseases talk about their model of Zika transmission in Texas, which aims to help public health officials make informed decisions and assess epidemic risk.

Spring rains and rising temperatures came early this year, signaling the start of mosquito season and the threat of new Zika outbreaks in the United States. Imagine that you are a public health official in Houston, Texas. You know that parts of the city may be vulnerable to Zika spread and that at least 81 people have arrived in the metropolitan area infected with Zika since 2015. The challenge is that most Zika infections are asymptomatic and therefore not diagnosed by doctors or reported to public health agencies.

Although most Zika cases are quite mild, a Zika outbreak is cause for serious concern and immediate intervention because of the established link between Zika infection during pregnancy and neurological complications in unborn babies.

Zika could spread silently in Houston long before it’s detected. Although most Zika cases are quite mild, a Zika outbreak is cause for serious concern and immediate intervention because of the established link between Zika infection during pregnancy and neurological complications in unborn babies.

You receive notification that a single individual with no travel history to Zika affected regions has tested positive for Zika, signaling that local transmission has occurred. Should you initiate costly interventions, including mosquito control and educational campaigns, or wait for more cases to arise? How would a second case change your decision? Would your response be different if you were a public health official in Lubbock, Texas instead of Houston? Depending on your location, the presence of two cases can reflect an anomalous low probability event or indicate an imminent or ongoing outbreak.

In our recent article: “Assessing real-time Zika risk in the United States”, we present a computational model intended to help public health officials navigate such challenges and assess epidemic risks as cases accumulate within their jurisdiction.

Our model considers the two prerequisites for a Zika outbreak: the arrival of infected travelers and the prevalence of Aedes mosquito species capable of transmitting disease to uninfected people. If infected travelers are the “sparks”, then mosquito bites provide the fuel for a full-blown outbreak.

Using Texas as a high risk case study, we find that the Zika importation rate can be predicted from local tourism spending, and that mosquito-borne transmission depends on climatic and socioeconomic conditions, with the highest risks occurring in low GDP, high temperature regions.  By simulating these processes, in conjunction with the unpredictable nature of disease transmission and Zika’s low reporting rate, we found that the 254 counties of Texas vary enormously in risk. Most counties have almost no chance of a Zika outbreak while a few high risk counties have 60-fold greater risk.

So, when and where should public health officials trigger interventions? In 2016, the Centers for Disease Control and Prevention (CDC) prudently recommended a two case threshold and Texas Department of State and Health Services (DSHS) conservatively opted to reduce the threshold to a single reported locally-transmitted case of Zika. Indeed, during the small November 2017 Zika outbreak in Cameron County, Texas DSHS intervened immediately.

Our model is designed to help officials determine appropriate intervention triggers, depending on local environmental and socioeconomic conditions.

Our model is designed to help officials determine appropriate intervention triggers, depending on local environmental and socioeconomic conditions. High risk regions, like the Houston Metropolitan region and Rio Grande Valley should certainly have early (one or two case) triggers.  However, given economic and societal costs of Zika intervention, lower risk regions like the Dallas or Austin metropolitan areas might consider less stringent thresholds.

As we enter mosquito season in the Southern US, this model can be used readily for public health planning and real-time risk assessments. Zika prevalence across the Americas is considerably lower today than this time last year; during the first 15 weeks of 2016, the weekly prevalence peaked around 40,000 cases per week compared to less than 2,000 cases per week this year.

This has translated into fewer infected travelers arriving in Texas, down from 33 by May of 2016, to only 11 as of May in 2017.   Thus, as we optimistically look towards a mosquito season with fewer Zika ‘sparks’, this framework can provide data-driven insight for anticipating and managing risk in vulnerable regions of the US.

View the latest posts on the BMC Series blog homepage

Comments