Survival Analysis: 1995 Outbreak of Ebola Haemorrhagic Fever (EHF) in Kikwit,
Democratic Republic of the Congo
Abstract
Ebola Haemorrhagic Fever (EHF) is a severe viral disease which is usually fatal in humans,
transmitted between humans and other mammals such as fruit bats and primates. The 1995
outbreak of Ebola in the city of Kikwit, DRC affected over 300 people and killed 245. Using SIR
modelling, the effects of three intervention methods (vaccination, isolation and treatment) were
tested, giving insight into the effective management of future outbreaks.
Methods
Analysis was carried out using an SIR model, which was constructed using data from the Kikwit
EHF outbreak in 1995. The model incorporated three factors which were: healthy individuals,
individuals who were infected with the virus, and deaths as a result of EHF manifestation. The
model also incorporated several parameters which affected those factors mentioned above,
including mortality, population at risk, rate of transmission (β-parameter) and recovery rate (γ-
parameter).
In order to evaluate the impact of human intervention on the epidemiology of Ebola, three
parameters were independently changed to see how this affected the proportion of the
population which is healthy, sick and dead. The three parameters each corresponded to one
intervention method to be tested: population size (related to isolation of infected individuals),
treatment (related to mortality, as treatment results in reduced mortality) and vaccination or
natural immunity to infection. The β- and γ-parameters were not changed as they are constants
and changing these would lead to a misrepresentation of the disease dynamics.
Results and Discussion
The first intervention method which was tested using the model was isolation. This was inferred
from population size, with a smaller population size being associated with a lower rate of contact.
To investigate whether population size affected the SIR model, population size was reduced from
330 to 180. Figure 1 shows the normal model (above) and with a population of 180 (below). As can
be seen in figure 1, there was very little change in proportional death rate, with 65.7% dead by
August with a population of 330, compared to 65.5% in a population of 180. The proportion of
healthy individuals remaining by August did not change with a decreased population size, with
33.3% remaining healthy in both cases.
The original hypothesis for testing population size as an intervention method was based on the
idea of contact frequency and transmissibility. Larger populations tend to have a higher rate of
contact between individuals due to increased population density over an area, and therefore the
transmission rate of a pathogen would be higher in larger populations and in lower in smaller
populations (McCallum et al. 2001). The results do not correspond to this trend and there may be
several explanations to explain this; for one the Ebola virus is a highly transmissible pathogen,
being spread via multiple routes including through the bodily fluids of those infected or killed as a
result of EHF, also through objects such as needles and syringes, and also through zoonotic
, transmission by mammals such as primates and fruit bats (Leroy et al. 2005). In addition, the
model does not acknowledge the socio-economic conditions of the area in question. Kikwit is a
highly rural area, located next to the Kwilu River and strongly associated with forested areas, and
so close contact with wild animals including primates and fruit bats would be frequent (Leirs et al.
1999).
Figure 1: Comparison of the proportion of the population which were healthy, sick and dead with a
population size of 330 (above) and 180 (below). Proportionally, the changes observed with a reduction in
population size were very minimal, reflecting the highly infective nature of the disease.
Democratic Republic of the Congo
Abstract
Ebola Haemorrhagic Fever (EHF) is a severe viral disease which is usually fatal in humans,
transmitted between humans and other mammals such as fruit bats and primates. The 1995
outbreak of Ebola in the city of Kikwit, DRC affected over 300 people and killed 245. Using SIR
modelling, the effects of three intervention methods (vaccination, isolation and treatment) were
tested, giving insight into the effective management of future outbreaks.
Methods
Analysis was carried out using an SIR model, which was constructed using data from the Kikwit
EHF outbreak in 1995. The model incorporated three factors which were: healthy individuals,
individuals who were infected with the virus, and deaths as a result of EHF manifestation. The
model also incorporated several parameters which affected those factors mentioned above,
including mortality, population at risk, rate of transmission (β-parameter) and recovery rate (γ-
parameter).
In order to evaluate the impact of human intervention on the epidemiology of Ebola, three
parameters were independently changed to see how this affected the proportion of the
population which is healthy, sick and dead. The three parameters each corresponded to one
intervention method to be tested: population size (related to isolation of infected individuals),
treatment (related to mortality, as treatment results in reduced mortality) and vaccination or
natural immunity to infection. The β- and γ-parameters were not changed as they are constants
and changing these would lead to a misrepresentation of the disease dynamics.
Results and Discussion
The first intervention method which was tested using the model was isolation. This was inferred
from population size, with a smaller population size being associated with a lower rate of contact.
To investigate whether population size affected the SIR model, population size was reduced from
330 to 180. Figure 1 shows the normal model (above) and with a population of 180 (below). As can
be seen in figure 1, there was very little change in proportional death rate, with 65.7% dead by
August with a population of 330, compared to 65.5% in a population of 180. The proportion of
healthy individuals remaining by August did not change with a decreased population size, with
33.3% remaining healthy in both cases.
The original hypothesis for testing population size as an intervention method was based on the
idea of contact frequency and transmissibility. Larger populations tend to have a higher rate of
contact between individuals due to increased population density over an area, and therefore the
transmission rate of a pathogen would be higher in larger populations and in lower in smaller
populations (McCallum et al. 2001). The results do not correspond to this trend and there may be
several explanations to explain this; for one the Ebola virus is a highly transmissible pathogen,
being spread via multiple routes including through the bodily fluids of those infected or killed as a
result of EHF, also through objects such as needles and syringes, and also through zoonotic
, transmission by mammals such as primates and fruit bats (Leroy et al. 2005). In addition, the
model does not acknowledge the socio-economic conditions of the area in question. Kikwit is a
highly rural area, located next to the Kwilu River and strongly associated with forested areas, and
so close contact with wild animals including primates and fruit bats would be frequent (Leirs et al.
1999).
Figure 1: Comparison of the proportion of the population which were healthy, sick and dead with a
population size of 330 (above) and 180 (below). Proportionally, the changes observed with a reduction in
population size were very minimal, reflecting the highly infective nature of the disease.
