At the height of the fear last September, when Ebola was sweeping across West Africa and seemingly threatening the world with a deadly pandemic, the professionals tasked with battling such perils issued a warning that resonated with chilling effect: Absent effective action, the globe could wind up confronting as many as 1.4 million Ebola cases by late January. So estimated the researchers at the Centers for Disease Control and Prevention in Atlanta.
But as January came and went, the CDC reported a minute fraction of that scary estimate. There were only 22,369 cases, the agency said last week. The World Health Organization (WHO) said new cases were emerging at a rate of fewer than 100 a week — the slowest the disease has spread since June.
In a sign of the growing sense that Ebola has been effectively contained, the pharmaceutical company Chimerix last week halted trials of one of two drugs that were being tested on actual patients to treat Ebola in Liberia, saying only “a handful” had enrolled in the trial.
How communities in West Africa and beyond dodged the worst prognosis for Ebola is in part reflective of what some experts describe as a highly effective public health campaign -- albeit, following a tragically fumbled beginning. Yet the CDC also now acknowledges that the worst-case scenarios it described last fall appear to have amplified, prompting the agency to reevaluate the models it employs to produce its estimates, and the way it communicates data to the public.
Some epidemiologists now fret that the enormous disparity between the worst estimates and the reality could leave the world complacent in the face of the next potential pandemic, as the public tunes out necessary warnings.
“You may start to lose the public’s trust next time we have an epidemic,” says Gerardo Chowell, an expert in infectious disease modeling at Georgia State University. “When we have avian flu, for example, we may have trouble trying to communicate to the world and to donors that this may be a catastrophe.”
By the CDC's reckoning, the warnings it issued played a vital role in provoking regional governments and international agencies to come forward with a tremendous amount of aid, with this infusion of energy and resources ultimately playing a key role in keeping the disease at bay.
Ebola was indeed raging last September when the CDC released its estimates for the long-term growth of the epidemic. On the same day the CDC issued its warning that Ebola cases could burgeon to as many as 1.4 million in two of the most affected countries, the WHO reported that the number of cases had doubled to nearly 7,000 within the previous month alone. The CDC projected that, without a much-intensified campaign, the world might face a low estimate of 550,000 cases of Ebola in Liberia and Sierra Leone alone by Jan. 20, 2015.
“When we first created the model, we all hoped that we would not see that,” says Michael Washington, who leads the CDC’s modeling task force.
The headline numbers in these detailed projections quickly swept through the media, encouraging a sense that a potential crisis was at hand. Still, the CDC's report also included caveats that either did not make their way into media accounts or were buried beneath a wave of alarming talk: The outbreak might be contained with appropriate interventions in place.
“As these measures are rapidly implemented and sustained, the higher projections presented in this report become very unlikely,” the CDC researchers spelled out in their report. They added that aid agencies could effectively end the epidemic by admitting more patients to hospitals for appropriate care -- about 70 percent of existing Ebola patients, they estimated.
Chowell, the Georgia State University expert, says this context was largely ignored by reporters who covered the outbreak. But he also thinks the outsized attention to the worst-case scenario underscores a need for the CDC and other infectious disease authorities to reconsider how they engage with the public and whether they ought to emphasize more realistic ranges in the future.
The authorities also need to examine the basics of their models, says Chowell, asserting the CDC's worst-case estimate failed to take one important factor into account: that as communities in affected areas began mobilizing, the epidemic would slow.
“There was no consideration in that work of that fact that people would change behavior,” Chowell says. “Once the epidemic started to hit their towns, the population would start to realize that, indeed, there is a virus that is spreading and killing their family members.”
A reconsideration of how the authorities ought to assess the dangers of an epidemic and how they ought to engage with the public is now underway. Officials with the CDC recently spoke with representatives from WHO to discuss the lessons they learned from Ebola. Separately, they met with other modelers and disease researchers during a workshop at Georgia Institute of Technology to do the same.
None of this is likely to be easy, for the simple reason that epidemics are by their nature fast-changing. The trajectory of transmission spreads as people hear about and react to the dangers: The degree to which the authorities sow fear has direct influence over the course and magnitude of the epidemic itself. And all of this is hard to capture in the models wielded by epidemiologists to predict what comes next.
“It's easier to predict a number based on everything will stay the same, but if you want to include some changes in your model, that's usually difficult to answer,” says Hasan Guclu, a biostatistician at the University of Pittsburgh, who specializes in infectious diseases. “You really don't know what's going to happen.”
The CDC was constrained by the quality of its data. Modelers were working mostly from case numbers in situation reports published by WHO, which only reported cases at the country level. Chowell says it would have been far more valuable to look at how the number of cases changed in districts over time. WHO, however, failed to start making this information available to the international scientific community until October.
“The reality is that WHO had the data and they used it, but in the early stage of the epidemic when the predictions were published by the CDC, those data were not available to the international research community,” Chowell says. “It was available too late.” Tarik Jašarević, a representative of WHO, denies that district-level data was available before October, saying the organization did not have enough resources on the ground for data collection and reporting until then.
Guclu says there is still a critical lack of high-resolution data in rural versus urban areas, where the spread of the disease could look quite different. Washington and Manoj Gambir, who is second-in-command on the CDC’s task force, admit that modeling a disease as it breaks is no easy task.
“We tried to make some reasonable assumptions,” Washington says about working with the preliminary data on the Ebola. Gambir adds that a model is, by definition, a way to look at a situation in its developing stages before all the information emerges. “Modeling has its forte in a situation where there aren't necessarily a lot of data available. When there are a lot of data available, modeling takes a little bit of a backseat,” Gambir says.