Example: Modelling survival during a real life outbreak of Ebola

The ebola virus causes a zoonotic disease that is transmitted to people from wild animals. It was first identified in 1976 in 2 simultaneous outbreaks, one in what is now, Nzara, South Sudan, and the other in Yambuku, Democratic Republic of Congo. The latter occurred in a village near the Ebola River, from which the disease takes its name. Fruit bats may be the natural Ebola virus hosts, but Ebola is introduced into the human population through close contact with the blood, secretions, organs or other bodily fluids of a range of other animals such as primates (chimpanzees, gorillas and monkeys) and forest antelope found in the rain forest. Hunting for bush meat is implicated in this process.

The animal/human contact route of transmission is usually responsible for the initial “signal” cases. However the disease then spreads in the human population through human-to-human transmission. Unlike zombies, Ebola patients are usually too sick to move. However their bodily fluids are highly contagious, even after death.

Fatality rates vary from between 80% and 50%, depending to some extent on the initial health of affected individuals.

A useful introductory fact sheet on the disease is available on the WHO website.

http://www.who.int/mediacentre/factsheets/fs103/en/

The virus does undergo an incubation period of between 2 and 21 days between first contact and symptoms developing, during which the patient is not contagious. However to simplify modelling this can be ignored, as cases are not usually detectable until symptoms occur.

In May 1995, an international team characterized an outbreak of Ebola hemorrhagic fever (EHF) in Kikwit, Democratic Republic of the Congo. Data derived from

A.S. Khan, et al. 1999. The reemergence of Ebola hemorrhagic fever, Democratic Republic of the Congo, 1995. J Infect Dis 179:S76-S86.

Map showing the location of Kikwit and other Ebola outbreaks

Modelling the outbreak

The key data from this outbreak have been included in the modelling tool linked below.

The total population of Kikwit is around 200,000. However the population at risk of exposure through contact with the initial cases was between 200 and 500 people, as some general isolation measures were fairly quickly put in place such as closing schools and preventing all movement into the area directly affected.

At the time of this outbreak no vaccination was available. Relatively little was still known about the illness.

The data set therefore serves as a baseline “worst case” scenario. Our increased knowledge of the nature of the disease should allow more effective containment of future outbreaks. However general isolation measures were less effective initially in the 2014-2016 outbreak that affected seven West African countries and led to over 10,000 deaths in multiple clusters of cases.

Model example

During this early outbreak the only real intervention used by workers on the ground involved limiting the size of the population at risk.

A simple SIR model (ignoring asymptomatic incubation period) has been set up and starting parameters estimated for you using a Bayesian framework in order to match the observed pattern during the 1999 Kikwit outbreak as closely as possible. This is the baseline model.

You can experiment adjusting the model parameters and observe the match between the model and the empirical data using this link.

http://r.bournemouth.ac.uk:3838/Epidemics/Ebola/

This is the model which will be used for the assignment

Mapping the epidemic in west Africa in 2014-2016

The 2014-2016 epidemic in West Africa raised public awareness of Ebola. You can track the geographical spread of the 2014- 2016 outbreak using this mapping application.

http://r.bournemouth.ac.uk:3838/Epidemics/Ebola_spatial/