Thursday, October 9, 2014

Traveling With Ebola Is Not Traveling With Influenza

With everyone away at IDWeek, I've had time to think about Ebola and things like airport screening programs. I'm not supportive of outright travel bans. As many have eloquently said, bans will do more economic harm than good and hinder efforts in West Africa. However, I'm worried that we might be equating Ebola with Influenza. Many of the discussions concerning travel restrictions and Ebola spread have centered around models and estimates derived from respiratory viruses epidemics like the 2009 H1N1 influenza pandemic. Yes, influenza and Ebola are both viruses, but that's like equating a sparrow and an Allosaurus because they're both dinosaurs.

The first and perhaps most important difference between the current Ebola outbreak and the the 2009 H1N1 pandemic is that Ebola it is very slow moving. For example, the first case of Ebola is thought to have occurred 307 days ago on December 6th in a two-year old boy. Since that time there have been an estimated 8,032 cases (granted these could be underestimates). If you compare a similar 307-day period for 2009 H1N1, April 12, 2009 to February 12, 2010 CDC estimated that between 42 million and 86 million cases occurred in the US with a mid-level estimate of 59 million people infected. Think about that - 7300 times more cases of H1N1 using the mid-level estimate during the same 307 days.

Another difference between influenza and Ebola is the incubation period (time from exposure to symptoms). Generally, the incubation period for influenza is 1-4 days (2-day average). For Ebola symptoms appear 2 to 21 days after exposure with an average of 8 to 10 days.

A final difference between Ebola and 2009 H1N1, which seems to be overlooked in discussions of airport screenings and other control measures, is infectivity during the incubation period. Put another way, can you transmit the virus without knowing you are sick? With Ebola, humans are not infectious until they develop symptoms. In comparison, with 2009 H1N1 it's reported "that pre-symptomatic influenza transmission occurred via both contact and respiratory droplet exposure before the earliest clinical sign, fever, developed" in a ferret model. This finding has been confirmed in humans. Interestingly, SARS is not infectious prior to symptom development (see CDC and Zeng et al), which may explain why we were ultimately able to contain SARS (unlike influenza).

To summarize, Ebola is slower moving, has a much longer incubation period (especially compared to the duration of a transcontinental flight), and is not contagious before symptoms develop. What does this mean? It means that if Ebola was as infectious as influenza, millions would have already died - apocalypse. It also means that since Ebola is not transmissible during its long incubation period, it may be possible to quickly isolate patients when symptoms develop. Thus, airport screening on exit or entry could limit transmission and perhaps through early diagnosis allow Ebola infected patients to receive life saving treatment more quickly.

A more concrete example: Imagine a person infected yesterday with influenza but still asymptomatic during their two day-incubation period. This person would screen negative overseas and in the US. However, it's highly likely that they are already infectious or will become infectious during their flight. Thus, many other passengers in the airport and plane would take influenza home with them as a vacation souvenir. Screening doesn't work for influenza. This would not be the case with Ebola because they will detect their symptoms as they become infectious and only spread it though blood exposure - something unlikely so early in the infection even on a long transcontinental flight. And think about how many times another person has bled on you even when they were bleeding (i.e. the Ebola condition) versus how many times someone has coughed or sneezed on you when they were coughing and sneezing (the influenza condition). **cough**

Airport screening for Ebola symptoms may still be ineffective, but I would like to see a few more mathematical models analyzing the epidemiology of Ebola and the impact of specific screening programs. In the meantime, let's focus our attention and resources on the horrible plight in West Africa.

image source: xkcd


  1. great post Eli. Interesting to think how flu and MERS/SARS are so different from Ebola. An influenza pandemic with a high mortality rate would be devastating.

  2. Interesting post. It would seem that screening should be done prior do departure, if at all. Perhaps flight attendants could then be attuned and report ill-appearing passengers for additional screening after arrival. Think of all the good we could do in Africa with the millions of dollars saved. Better use of those funds, I think.

  3. Great points, Eli. In addition to incubation period, I think it's useful to consider the serial interval (the period between infection and transmission) and basic reproduction ration (R0) as well. As has been pointed out previously, the estimates for R0 for Ebola in this event so far have been similar to the estimates of R0 for pandemic influenza events (i.e., roughly in the neighborhood of 1-2). White and Pagano wrote a paper in 2008 in which they describe a likelihood-based method for estimating the serial interval and R0 of an epidemic. They demonstrated their method for Ebola in Congo, swine flu, and avian influenza in the Netherlands, estimating R0 for the flu epidemics to be 1.1-1.3 and for the Ebola outbreak to be 1.4-2.8, and serial intervals of 1.7-4.7 days for flu versus 5.4-7.6 days for Ebola ( Recently, the WHO Ebola Response Team estimated for the current epidemic the incubation period to be near 11 days and serial interval to be near 15 days. This is difference in serial interval between flu and Ebola is consistent with your observations of the dramatic difference in spread between pandemic H1N1 2009-2010 and Ebola in West Africa 2013-2014.

    What do we take away from this? One interpretation is that longer serial intervals essentially give more time to institute control whereas, as White and Pagano state, the short serial interval of influenza necessitates aggressive strategies for control. Obviously we cannot conclude that Ebola doesn't necessitate aggressive control, but it may mean that, as you describe, in the case of Ebola where the serial interval is substantially longer than the incubation period, there is a higher likelihood of detecting an infectious traveler in an airport than there is for a disease like influenza.

    1. Excellent points David. It seems that serial interval discussions have been somewhat ignored and I hope you will follow up these comments with a post on your blog.

      I do have a question in regards to Ro. Ro is thought of as the # of other people who contract the virus from the initial case when it initially enters a population (e.g country) - when there is no immunity and no interventions have taken place. It would seem that with viruses with a long serial interval, the Ro becomes less important since the subsequent Ri would be substantially lower as interventions take place. Thus, wouldn't equating influenza and Ebola through the Ro be misleading (to none modelers)? Or rather, should discussions focus more on serial interval?

      Another question: How dependent is the Ro on initial conditions? My sense is that the Ro for Ebola might be 2 in West Africa, but would be substantially lower in the US, say less than 1. I say this because transmission with Ebola is non-uniform with respect to time, with most transmissions occurring very late in the course (near and after death). In the US all of the patients would be isolated in the hospital when they become highly infectious so that community transmission will be close to zero and hospital transmission will be very low. I ask this because equating Ebola and influenza Ro using Africa data may not apply to the situation in the US. Saying this could calm the public in some regards.

      Appreciate your thoughts.

    2. Thanks, Eli. My comments above now appear, refined and clarified, in a full post at, where I also address the issue of equating influenza and Ebola through R0 -- which, as you point out, is misleading.

      How dependent is R0 on initial conditions? It isn't, but it is dependent upon human behavior and, as you suggest, I think this could result in different values of R0 for different outbreak venues. Thinking about mathematical models for a moment, in the case of SIR-like models, expressions for R0 involve parameters that are "biological" in nature (e.g., periods of infectiousness, rates of death, incubation periods) and thus not likely to change across outbreaks but also a parameter that is dependent upon human behavior: the adequate contact rate. The contact rate in Europe or North America, for example, may be different than it is in Africa due to the different societal norms and customs and also the different healthcare practices. So, there isn't necessarily one R0 for a pathogen, but rather R0s are estimated for different outbreaks.