The Differential Impact of Violence on Ebola Virus Disease Transmission: A Mathematical Modeling Study of the 2018-2019 Outbreak in the Democratic Republic of the Congo

Background: Violence can impact Ebola virus disease (EVD) transmission in the current outbreak in the eastern Democratic Republic of the Congo (DRC). We hypothesized that violent events targeted against the Ebola response will be associated with more EVD transmission than untargeted events. Methods: We used a dynamic cohort of individuals who lived in Ebola-affected and unaffected health zones in DRC from April 2018 to August 2019. The time-varying exposure was the number of violent events, as defined by the Armed Conflict Location & Event Project (ACLED) database, that occurred in each health zone over a series of days. These violent events were categorized into vs. Ebola-untargeted, and further sub-categorized by civilians, military or politics, or protests. The outcome variable was the estimated daily reproduction number (Rest) by health zone, which was estimated from daily EVD case counts (from DRC Ministry of Health) by the Wallinga-Teunis method. We fit a linear time-series regression to model the relationship of violent events and R comparing EVD-affected and unaffected health zones. Findings: The average Rest was 1.06 (95% confidence interval [CI]: 1.02-1.11). We found an overall change in R of 0.035 (95% CI: 0.020-0.050) among Ebola-affected compared to unaffected health zones. Violent events targeting the Ebola response were associated with an increase in Rest of 0.098 (95% CI: 0.064-0.132) while untargeted, violent events had a smaller effect (0.022, 95% CI: 0.005-0.038). Additional analyses showed the increase in Rest was primarily driven by Ebola-targeted civilian events and, to a lesser extent, by Ebola-untargeted military, political, or protest events. Interpretation: These findings suggest that civilian acts of violence directly targeted against the Ebola response efforts had the largest impact on EVD transmission. To a lesser extent, untargeted military, political, or protest events also had an impact. Funding Statement: This work was supported by National Institute of Allergy and Infectious Disease (K08 grant number AI139361 to ETR; K23 grant number AI135037 to JDK) and National Institute of General Medical Sciences (R01 grant number GM130900 to TCP, JDK, and ETR). Declaration of Interests: The authors stated: None to declare. Ethics Approval Statement: The authors stated: Data were publicly available and de-identified, so ethics committee approval was not needed.