Bostonia is published in print three times a year and updated weekly on the web.
It’s a persistent cliché in films dramatizing deadly epidemics: that chart with the ominously multiplying paper-doll cutouts—first 2 cases, then 4, then 16…then an entire city under statistical siege. But for real-life disease trackers like Laura Forsberg White, a School of Public Health associate professor of biostatistics, the monitoring and prediction of the course of calamities from avian influenza to Ebola rely on several important variables. And unlike diseases like cholera, which is caused by contaminated water, or diseases borne by mosquitos, Ebola virus disease is spread only through direct contact with body fluids, and only after an infected person begins to show symptoms, which can be as long as 21 days after exposure. When it is not confined to an isolated community, the difficulties of tracking Ebola, which in past outbreaks has been fatal in 60 to 90 percent of cases, are many. Identifying and isolating infected individuals in more densely populated areas—such as the Liberian who succumbed to Ebola in Lagos, Nigeria—present a different, more urgent set of challenges, says White.
Bostonia talked to White about how the current Ebola epidemic compares to earlier ones, and how scientists in her field follow the spread of the virus with the goal of stopping it in its tracks.
White: This outbreak is notable for several reasons. First, it is in a part of Africa that has never had a documented Ebola outbreak. Second, it is in urban areas. Third, it is covering a very large geographical area, currently three countries. These are all unusual in the 40 or so years that we have been having Ebola outbreaks.
Determining the origin of any outbreak requires understanding how it spreads. Ebola is transmitted person to person through contact with an infected person’s bodily fluids or from objects contaminated with these fluids. For these types of outbreaks, interviewing those close to victims—family members, friends, etc.—will allow investigators to trace back the outbreak to the initial case. In most cases with Ebola, that initial case had contact with an infected animal. Cholera is primarily transmitted through infected water, so it is important to determine what water source is contaminated and how and when the contamination occurred.
Public health surveillance systems are used to detect and track Ebola outbreaks. However, given that Ebola typically occurs in rural areas with limited resources, this has posed substantial challenges. Past experience, particularly in 2000 with a large Ebola outbreak in Uganda, lead to the implementation of improved surveillance systems with labs in areas of Africa that have been more frequently impacted by Ebola outbreaks. Since outbreaks of Ebola have typically occurred in very rural areas of Africa with limited health care and public health resources, it has been a challenge to detect them. Ebola outbreaks have typically been relatively small, relatively contained geographically, and in mostly rural areas. This current outbreak is noteworthy for its size, impact on urban areas, and geographic extent.
There is a growing movement in public health surveillance to use information outside the health care system to detect disease outbreaks. For instance, people have looked into monitoring activity on Twitter feeds, Google searches, or even texting individuals through their mobile phones and asking them a few brief questions. Strategies like the last one could potentially be useful in outbreaks in resource-poor areas where there are limited public health and health care resources.
There really is not much of a distinction between how we track diseases based on the type of pathogen. The key distinction is often based on the mode of transmission—person-to-person, waterborne, etc.
There has been very little work done to predict the spread of Ebola. Ebola outbreaks have tended to be small and localized, making the need for sophisticated methods to model the illness relatively unnecessary. The current outbreak will likely push the field of infectious disease modeling to consider how to handle problems such as the one we are seeing now.
Statistical models have often been of greater interest for outbreaks where the disease is hard to identify and can move through large portions of the population. For example, influenza outbreaks impact large portions of a population, and the majority of cases are never seen in a traditional health care setting where public health surveillance systems will be able to be notified of them. Models help us understand how they are spreading and what their likely impact could be.
Cholera is also a disease that we want to model and follow because it arises from water contamination. If we know where people are getting sick, we can intervene to find the source of contamination and limit further spread of the disease.
Ebola is an illness that has been monitored through contact tracing, which is more feasible when the outbreak is impacting no more than a few hundred people. In other words, when someone becomes sick, health workers find out who has been in contact with them, and they can closely follow those individuals before they infect anyone else, and also provide them more rapid care if they become ill. Outbreaks of Ebola usually impact at most a few hundred people and this type of approach, coupled with safe burial practices and infection control in health care settings, is very effective at stopping the outbreak.
The current outbreak is much more complex and is impacting a much larger number of people. This requires enormous resources to track and contain.
One challenge is that there is a dramatic increase in the number of people who can potentially be infected. Mathematically, we can show that in the absence of effective control measures, the final size of the outbreak will increase. There is a lot of interest in infectious disease epidemiology with better understanding the nature of the type of contacts people make and how they relate to the chance of an infection being passed between them.
When people are moving in and out of an outbreak, that clearly increases the chance of the outbreak spreading more broadly, as it is now in western Africa. In larger-scale diseases, specifically influenza, modelers have developed models to predict how they will spread based on travel patterns. This is a substantial issue with illnesses that do not tend to make people sick enough to halt travel plans. Ebola outbreaks have not tended to be so impacted by travel and migration. When people begin to get sick they tend to be so sick that they are not traveling. Since Ebola appears to be transmitted only by people who have symptoms, travel has not been a major factor in the spread of Ebola in the past.
This is a bit challenging to answer in a simple way. If an outbreak is contained in a geographical area and no one is traveling in or out and there are no public health interventions, then it will run its course. Models predict that it will grow exponentially at first and then taper off. However, this and many outbreaks typically do not play out that way: people travel and public health gets involved. And there is always the role of random chance that can cause an outbreak to be bigger or smaller than models would predict.
There are also mathematical and statistical models that describe how an infectious disease outbreak will progress. For Ebola, we consider models where the disease is spread by person-to-person contact. These models provide our basic idea of how an outbreak will progress when there is no one moving in or out of the affected area and people are more or less interacting with each other in a random way. Most of the assumptions of these models are not realistic, but it is still a useful starting point. We can create more complex disease-specific models as we learn more about the characteristics of the illness.
These models generally predict that an outbreak will grow exponentially fast at first and then as more people recover from the illness and are no longer susceptible to getting sick again, the outbreak will begin to subside. The rate at which it grows is impacted by two things: 1) how many people an infected individual infects—the reproductive number—and 2) how long it takes for someone to become infectious once they are infected. We can impact the rate of growth of the epidemic and work to bring it under control by trying to decrease the reproductive number. Things like vaccines, drugs that increase immunity or decrease infectiousness of the illness, or measures to keep people from coming in contact with each other—think school closures and quarantine—are often used. Without intervention, an outbreak will theoretically die out once there are little or no more remaining individuals who are still susceptible to getting sick.
Ebola in a rural setting often follows the predictions of these models reasonably well because there is not a lot of movement in and out of the population. However, there are a lot of ways for an outbreak to defy the typical course the model predicts. On one hand, public health intervention can decrease the amount of contact people have with each other and improve people’s immunity through vaccines or drugs. On the other hand, having people travel and take the illness with them to other areas can prolong it.
The current Ebola outbreak is much more complex than these basic models. It is spreading through a very fluid population where there is a lot of travel.
A pandemic occurs when the scope of an outbreak begins to cover a large region. How big that region needs to be depends somewhat on the particular disease and what its expected scope is.
In the case of Ebola, to my knowledge, there is not a lot of involvement from modelers, but the degree to which modelers are being involved in decision making is increasing. In the 2009 H1N1 pandemic, the modelers were consulted frequently in this country as policy makers were trying to determine the best course of action.
In terms of the level of agreement between modelers, this also depends a lot on the disease. I have not seen modelers weighing in on the current Ebola outbreak. This is challenging to predict with sparse information and so many variables.
The sources I have been reading have been pretty accurate and seem careful. I think that some articles and headlines I have seen have been trying to over-sensationalize the transport of two sick Americans to Atlanta for care. The threat to the US public is negligible. The key take-home point is that transmission requires contact with bodily fluid from an infected person or contact with an object contaminated with this fluid. The special units in which these individuals are cared for are designed to eliminate this risk.
I think that any model considering this would show that Ebola would not spread here. And I think that most would agree that the likelihood of this outbreak spreading outside of Africa is very small.
Tomorrow: Working with a deadly virus. Read all stories in our series, “Battling Ebola” here.