Why The CDC Thinks 1.4 Million People Might Contract Ebola By February
A crash course in modeling the outbreak of a deadly disease
Last June, the Kenema Government Hospital in Sierra Leone built a temporary ward in anticipation of an influx of Ebola patients. The makeshift building was covered with thin, tin metal sheets, haphazardly held to the walls by wire twist-ties. Robert Garry says that when the structure fell, it seemed inevitable.
But finding physical space for patients wasn’t the only problem at the hospital. Even after the ward was rebuilt, Garry and his staff have been overworked, and many have contracted the disease. Another doctor at the Kenema Hospital, Sheik Humarr Khan, was often left as the only person caring for 80 people. But he told his sister: “If I refuse to treat them, who would treat me?”
Western Africa is struggling to prevent collapse during the worst Ebola outbreak ever. This week the CDC released a study predicting that by January of 2015, 1.4 million people may have contracted the virus. With new studies suggesting that the virus's mortality rate may be close to 70 percent, in the next six months 980,000 more people might die from the disease. Thomas Gift, one of the people who worked on the CDC report, explained that their model was based on daily case counts from the current outbreak, as well as estimates of previous ones.
To develop these types of predictive models, epidemiologists break individuals up into four groups: people who are at risk, people who are infected but not contagious, people who are infected and contagious, and people who have recovered. Ultimately, researchers try to determine how those groups interact with each other and predict movements between the groups. These movements comprise what is known as the “effective reproductive rate,” or the number of people each infected person will give the disease to. This number is called R. When R is less than 1, the outbreak is declining; if it’s more than 1, the disease is spreading.
Reports vs. Predicitions Graphs show how model predictions and actual case reports match up.
Gift says the parameters in the CDC study were set to minimize the differences between the model and what is being reported on the ground in Sierra Leone and Liberia. But he says, “Like all models, they are simplifications of reality.” Any model is only as good as the data that is fed into it, and it's possible that as many as 75 percent of all Ebola cases are going unreported.
For this reason, the model uses a correction factor to account for underreporting. The CDC study decided that factor should be based on the number of beds in use, since counting physical objects is somewhat more reliable than depending on case reporting. So researchers compared the number of patients in beds predicted by the model on August 28th to the number of actual patients in beds. But the authors only had access to the bed counts in Liberia, so it’s still just an estimate.
CDC scientists aren't the only ones trying to develop models of Ebola transmission. Christian Althaus, an epidemiologist at the University of Bern, has also developed an estimate of the outbreak. He says that the CDC’s estimate of 1.4 million is based on an assumption that the real number of cases is 2.5 times higher than reported. However, Althaus says that way before Ebola patients number in the millions, "we will see a dramatic change in behavior and probably a collapse of the countries. That might limit the spread to some extent, but it will be very important to maintain social order.”
He says it’s hard to quantify the impact of international aid, but Caitlin Rivers, a PhD student of computational epidemiology at Virigina Tech, says her modeling results suggest that contact tracing and improving infection control can help but not halt the progress of the epidemic. “But there's no question it needs to be a top priority for the global public health community,” Rivers says.
Despite the uncertainty involved in developing these models -- including the chronic underreporting, which has made solid data about the outbreak difficult to come by -- even conservative estimates put the number of people infected by Ebola in the tens of thousands by the end of the year.
As of September 24th, WHO has reported 6,263 cases of infection, with 2,917 deaths. “The number of lives already lost is staggering, and grows every day,” Rivers says.
