Optimizing the location of vaccination sites to stop a zoonotic epidemic

TL;DR

This paper presents a data-driven method to optimize vaccination site locations, aiming to increase participation and coverage in rabies control campaigns by minimizing travel distances.

Contribution

It introduces a computational algorithm using the p-median approach to strategically place vaccination sites, improving coverage and spatial distribution based on survey data.

Findings

Predicted increase in vaccination participation with optimal site placement

More even vaccination coverage across regions

Potential increase in workload and waiting times at some sites

Abstract

The mainstay of canine rabies control is fixed point mass dog vaccination campaigns (MDVC). However, in some regions, ideal vaccination coverage in dogs is not obtained due to low participation in the MDVC. Travel distance to the vaccination sites has been identified as an important barrier to participation. We aim to increase MDVC participation by optimally placing fixed point vaccination locations to minimize walking distance to the nearest vaccination location. We quantified participation probability based on walking distance to the nearest vaccination point using a Poisson regression model. The regression was fit with survey data collected from 2016-2019. We then used a computational recursive interchange technique to solve the facility location problem to find a set of optimal placements of fixed point vaccination locations. Finally, we compared predicted participation of optimally placed vaccination sites to historical participation data from surveys collected from 2016-2019. We identified the p-median algorithm to solve the facility location problem as ideal for fixed point vaccination placement. We found a predicted increase in MDVC participation if vaccination locations are placed optimally. We also found a more even vaccination coverage with optimized vaccination sites; however, the workload in some optimized locations increased significantly. We developed a data-driven computational algorithm to combat an ongoing rabies epidemic by optimally using limited resources to maximize vaccination coverage. The main positive effects we expect if this algorithm is to be implemented would be increased overall vaccination coverage and increased spatial evenness of coverage. A potential negative effect could be the presence of long waiting lines as participation increases.