Australasian Science Magazine

Credit: CDC/Cynthia Goldsmith

By Robert Cope, Joshua Ross & Phillip Cassey

Improving outbreak control in West Africa resulted in reduced risk to Australia.

In March 2014, the World Health Organization declared an Ebola outbreak in Guinea, West Africa. The outbreak gained global media attention when healthcare workers from the US and UK were infected in August, and after infected individuals entered the US and Spain during October. These events stimulated much debate in Australia about the country’s response.

We decided to model the risk to Australia, and how the events in West Africa affected the chances of an Australian outbreak. We found that although an outbreak was unlikely, the best strategy for Australia was to support measures to assist healthcare in West Africa and contain the outbreak there rather than focusing on trying to restrict potential carriers from entering Australia.

This outbreak is thought to have begun with the death of a single young boy in a rural village in December 2013. The disease spread quickly to neighbouring Liberia and Sierra Leone, and by August 2014 there were about 200 new cases per week observed in the Liberian capital Monrovia alone. In total more than 20,000 reported cases had occurred by the end of 2014, primarily in West Africa.

Ebola is a devastating virus. Approximately 60% of hospitalised patients within West Africa have died. Control of the existing outbreak, and prevention of new outbreaks elsewhere, is of paramount concern.

Researchers have investigated which destinations might be most at risk of a new Ebola outbreak. While Australia was not identified as one of the most at-risk locations, we considered it valuable to determine the risk that a case might enter Australia, and how this might change under different outbreak scenarios. To do this we constructed mathematical models based on the observed outbreak spread, international flight data and historic data on the number of individuals who travel from West Africa to Australia.

When we examined the risk, based on the best data available in mid-October 2014, it seemed almost certain that an individual with Ebola would travel to Australia sometime within the following 6 months. However, based on new data in early December 2014, the risk that an infected individual would enter Australia was drastically reduced, primarily due to decreases in disease proliferation within West Africa.

In the meantime, the Australian government restricted visas for individuals travelling from West Africa to Australia. We modelled the impact of this policy, and although it reduced the risk in the short term, this reduction was still less so than what by implementing on-the-ground improvements to healthcare conditions in West Africa.

Even if a case entered Australia it would likely be contained through surveillance, rapid response, public awareness and contact tracing.

Calculating the Risk

We developed mathematical models for the spread of Ebola within infected countries and its possible transport to Australia, and used these models to calculate the chance that a case would enter Australia within a given time period.

To model the spread of Ebola within each country, we used an “SEIR” epidemic model whereby each individual was assigned to one of four categories based on their infection status: they were either susceptible (S), exposed (E), infected (I) or removed (R). When a susceptible individual comes into contact with an infected individual, they can become exposed to the disease. The probability that they do depends on the “contact rate”. When exposed, they don’t show symptoms and are not infectious. After 5–6 days these exposed individuals start showing symptoms, at which point they are classed as infected (and can subsequently infect others). After a further 5–6 days, infected individuals are removed, either through recovery or death.

All of these transitions are random. If only one individual is infected there is a chance they might not actually infect anyone else, so these small outbreaks can die out. But if an outbreak is already well established then on average more people will become infected. The contact rate in this model was determined based on the observed “doubling time” of these Ebola outbreaks – the time it takes for the number of infected individuals to double.

We considered two ways that a case might enter Australia:

• someone infected in West Africa could travel directly to Australia before showing symptoms; or

• the outbreak could spread from West Africa to elsewhere, and someone exposed in a new location could then travel to Australia.

To date the small outbreaks worldwide have occurred via direct travel from West Africa to other nations. However, only a handful of people travel directly from West Africa to Australia each day.

When modelling the direct travel scenario we assumed that susceptible or exposed individuals were eligible to travel, given that exposed individuals are not yet showing symptoms of Ebola and that each were as likely to travel as the other. Based on the proportion of eligible travellers exposed to Ebola, we calculated that there was a small (random) chance that one of the travellers entering Australia each day would be carrying the disease. However, over a 6-month period these small chances can accumulate so that the risk that a case is eventually seen is quite high.

For the indirect travel scenario we used international flight data to model transport of the disease to other countries. For each infected country we knew how many passengers departed to each possible destination every day. If 0.1% of eligible passengers were exposed to Ebola on a particular day, and 1000 passengers departed, then on average one of those passengers would transport the disease. Again, this is random: on some days no exposed passengers might depart, and on other days it could be many passengers.

While infected individuals did reach other countries, these small outbreaks tended to die out or at least take a long time to establish, typically not resulting in outbreaks that were transported to Australia.

Why the Risk Was Reduced

The data we examined suggests that Ebola slowed considerably between mid-October and early December 2014. During the preceding months, the outbreak was most serious in Liberia but by early October it appears that it was past its peak. The outbreak continued to grow in Sierra Leone.

A significant change was in the estimated doubling time of the disease. When we initially performed this analysis, estimates of the doubling time for this Ebola outbreak varied considerably, but were typically between 15 and 40 days. We performed our initial modelling based on a relatively conservative doubling time of 30 days.

By early December, more precise data allowed us to estimate directly the doubling time of the disease. For the fastest-growing outbreak in Sierra Leone, the doubling time was approximately 45 days – much slower than what we had been using previously.

This change in doubling time makes sense. Initial estimates were based on early outbreak conditions when the disease was growing quickly and few control measures were in place, but the outbreak slowed when aid and public health measures were introduced. In addition to this, the quality of available data improved dramatically over the period due to increased reporting and capacity for rapid laboratory testing within West Africa. As a result, our December estimates were able to use confirmed new cases only, and these were lower than our previous estimates.

Over the second half of 2014, measures to contain Ebola within West Africa resulted in very significant changes in conditions. At the start of the outbreak, no hospital in Liberia had an isolation ward, so initial hospitalisations caused further transmission. Traditional burial practices resulted in the spread of the disease due to physical contact with the deceased, possibly causing 60% or more of cases.

But by January 2015 there were enough beds to keep every reported patient isolated, and sufficient capacity to safely bury each reported death. Community awareness measures were also put in place so that people were aware of what to look for and to seek medical help. Contact tracing, which monitors the health of every person who comes into contact with an infected person, was instituted for the majority of infected individuals.

While there is a lot of work still to be done, public health measures towards ending the Ebola outbreak within West Africa are showing positive signs. The effect of this success has a knock-on effect, subsequently reducing the risk that transmission will occur to other nations, including Australia.

Robert Cope is a postdoctoral research associate at The University of Adelaide, working with A/Profs Joshua Ross (School of Mathematical Sciences) and Phillip Cassey (School of Biological Sciences). The authors acknowledge the significant contribution of Prof Graeme Hugo (The University of Adelaide), who passed away following the completion of this study.