Texas Antiviral Release by Travis Chambers

Travis completed his thesis in March 2015.  His thesis studies optimal antiviral release strategies to combat the spread of influenza in Texas.

Complete Thesis

Chambers, T. L. (2014).  Optimization of Influenza Antiviral Response in Texas (Masters Thesis).  Naval Postgraduate School.

Executive Summary

Pandemic influenza poses a significant threat to the populations of the world. Through the past century, four global influenza pandemics and a severe influenza epidemic have shown that the world is not fully prepared to respond to pandemic influenza, especially a severe strain such as the 1918 Spanish flu. Over one-fifth of the world’s population is believed to have been infected with Spanish flu (Billings, 2005), and, 50 to 100 million people are thought to have died from Spanish flu or secondary complications incurred from the disease (U.S. Department of Health & Human Services, 2014). As recently as 2009, Swine flu caused approximately 180,000 deaths worldwide (Simonsen, Spreeuwenberg, & Lustig, 2013).

The World Health Organization (WHO) provides guidance to countries and international organizations and the Centers for Disease Control and Prevention (CDC) coordinates the national effort to respond to pandemic influenza (CDC, 2013). However, the burden of pandemic response falls on the state and local agencies to implement pandemic controls and interventions. Vaccinations are the most effective intervention, but they need time to develop and distribute. Until vaccines are available, the best response to pandemic influenza is a combination of social control measures, such as school closures and quarantine, and antivirals (Texas Department of State Health Services, 2008). Even in developed countries, antivirals are limited in number when compared to the requirements of a severe strain of influenza (Texas Department of State Health Services, 2008). The Texas Department of State Health Services (DSHS) has outlined a response strategy that includes antivirals, however it needs improvement from updated disease spread and intervention models. The Texas Pandemic Flu Toolkit provides decision makers in the state of Texas with critical disease spread information and response advice. Included in the toolkit, the Texas Pandemic Flu Simulator is a powerful compartmental disease spread model with time, location and demographic dimensions.

In our thesis, we develop an optimization program that is implemented in an online tool called Texas antiviral release scheduling (TAVRS), included in the Texas Pandemic Flu Toolkit, to provide the optimal allocation of antivirals to decision makers. The optimization program includes time, location, and age-demographic dimensions. The optimization program targets the average treatable population during an influenza pandemic. The treatable population is the number of individuals at a certain location and time that have been symptomatic with influenza for less than 24 hours. Stochastic variations of historical influenza strains and geographic origins in Texas from the Texas Pandemic Flu Simulator are averaged to create this input into the optimization program. We formulate a mixed integer linear program to determine geo-temporal antiviral release schedule to maximize the benefit of available antivirals. We consider three benefits of antivirals: lives saved, hospitalizations avoided, or quality adjusted life years (QALY) saved.

We first consider a base case scenario, which consists of a 1918-like randomorigin influenza pandemic with unlimited antivirals. The release schedule maximized the lives saved to roughly 26,500 by releasing a large amount immediately before the rise in treatable people. Although 30 million antivirals are available, TAVRS releases only 14 million. Antivirals are released to the highest populated counties during the weeks that precede the fast rise in treatable population. Antivirals released earlier may be consumed by the worried-well and antivirals released later are not available for the rise in treatable people.

Next, we examine several variants from the base model. When the pandemic originates near the border, the optimal schedule releases antivirals to the counties immediately prior to when the disease significantly spreads to them. When the pandemic originates in a highly populated county, the pandemic can spread quickly. A less virulent pandemic strain generates an extended release schedule duration and fewer antivirals released. A smaller supply of antivirals results in a delay in the release of the bulk of the antivirals. Finally, the specific objective function (lives saves, hospitalization, QALYs) had no apparent impact of the release schedule of antivirals.

Our analysis concludes with a comparison of the lives saved between the TAVRS antiviral release schedules and a simpler population-proportionate population distribution. The comparison found that in response to a mild pandemic, like the 2009 swine flu, the population-proportionate antiviral release schedule worked comparably the TAVRS antiviral release schedule. However, in response to a severe strain of influenza, like the 1918 Spanish flu, the TAVRS antiviral release schedule performed drastically better saving roughly 10,000 more lives, three to four times greater, than the population proportionate release schedule.

© Copyright 2004-2019 - Ned Dimitrov