Modeling the effects of health care practices on Ebola epidemic behavior

Science Spotlight

Modeling the effects of health care practices on Ebola epidemic behavior

Oct. 17, 2016

Boxplots of the number of cases assuming different intervention strategies for the Ebola outbreak in Guinea, from September 2014 through February 2015. The red dots represent recorded numbered and the colored boxes represent the model output taking into account the different routes of transmission: Ebola treatment units (ETU), contact tracing, and safe burials.

During the recent outbreak of Ebola virus in West Africa many metrics were collected as parts of epidemiological studies and health care reporting. These data lend themselves to further analysis using computational modeling of the epidemic. The epidemic behavior of three sites, Guinea, Serra Leone, and Liberia, can be compared to establish the effects of varying intervention practices and infection determinants. Guinea differed from the other countries in that it has a fairly developed public health system that allowed for early setup of Ebola treatment units (ETU) and good health practices, in contrast with the other sites. The Guinea outbreak has also differed from the other sites in that it has had fewer cases, but has had more flare-ups over a prolonged time period. This site was also the location of a cluster randomized phase 3 trail of the recombinant vesicular stomatitis virus vector vaccine (rVSV).

In their recent study published in BMC Medicine, Dr. Halloran and colleagues (Vaccine and Infectious Disease Division) developed a data-driven stochastic agent-based model to help understand disease dynamics and the effects of health care measures and vaccination on spread. The model also projected the effects of vaccination using the rVSV vaccine in the context of the already established control measures. The model took in to account many factors involved in Ebola infection and transmission including contact tracing, safe burial practices, ETU bed availability, age based risk, and heterogeneity of transmission. Data was obtained from the Guinea Ministry of Health and the WHO.

Model validation was performed on data from August 2014-May 2015 and showed that the average simulated epidemic over that period matched the outbreak data and predicted the decrease in cases seen starting in March 2015. The model also accounted for the geographic distribution of the epidemic and correctly modeled the spatial heterogeneity of the epidemic. To further dissect the epidemic and transmission, the model looked at the distribution of transmission occurring between household/extended family, unsafe burials, and hospital/ETUs and predicted that family has the highest rate of infection due to transmission events. Funeral attendants, hospital personnel and ETUs followed family as the most vulnerable to infection. In simulating transmission trees, the data correlated with known reports, but showed differences in the number of secondary infections, with a small number of individuals responsible for the majority of infections. In looking at contact tracing, unsafe burials, and ETU admissions, the group found that the number of contact tracings is strongly negatively correlated with the incidences of cases observed 10-30 days later. The negative impact of unsafe burial ceremonies on preventing disease spread was constant. As expected, the scientists found a positive correlation between the admittance into ETUs and Ebola incidences.

Overall the authors found that in the absence of intervention, infection rates would have continued to increase and that contact tracing and ETUs were key in controlling the epidemic. Lastly the group simulated a ring vaccination strategy, which is the targeted vaccination of a ring of  people around an infected individual.  The model found that ring vaccination could be most helpful in regions where contact tracing was poor. In conclusion, the model developed in this study was able to accurately explain the dynamics of the Ebola outbreak in Guinea. From tracking transmission cause, the study identified contact tracing and ETU availability as key drivers of outbreak dynamics. The model also provides a way to test effectiveness of planned approaches to disease prevention in future outbreaks.

Ajelli M, Merler S, Fumanelli L, Pastore Y Piontti A, Dean NE, Longini IM, Halloran ME, Vespignani A. 2016. Spatiotemporal dynamics of the Ebola epidemic in Guinea and implications for vaccination and disease elimination: a computational modeling analysis. BMC Med 14:130.

This research was funded by the National Institutes of Health and the EU Cimplex Grant agreement Framework program.