I have been looking for models that may have driven policy decisions regarding the predictive risks of Ebola spreading to other countries, the impact of interventions, and the quarantine period. I am not sure if these have been shared previously but I read through them in order to better understand the advice, if these were considered, that officials may have formulated from these reports. They are not the easiest things to read and I am sure there are other interesting variables I did not catch so I encourage you to read through them, too.
In the model for assessing the International Spread, I found the graph depicting air travel from outbreak countries revealing. It depicted the US as receiving 3000-6000 travelers from those countries weekly, not the 150 a day figure I have seen used in many articles. I also found it interesting that it described the limited access to air travel in those countries as reducing the risk of spread, with only Guinea connecting to the US.
The report noted that if Nigeria were to have an outbreak, which they did not include as an outbreak nation spreading in this model, it poses the greatest risk of spread because of the density of their population and because more people have the means/cause to travel, in addition to being an international hub connected to most other countries. They noted that Ghana was the next greatest risk of contributing to the spread if it should experience an outbreak that wasn't contained for the same reasons. In other words, they wanted to make the point that if they had outbreaks in either of those two countries, the model would dramatically change as would the risk of spreading to other countries. That would suggest to me, that if either of those two countries import a single infection, everyone better help them contain it. That may be more critical of an intervention, and at least equal to assisting in stopping the spread in current outbreaks.
I am not a math person so help me out with this to see if I am understanding their methods correctly. From what I can determine they built their spread risk model based on the assumption of an 80% reduction in airline traffic but that is not the actual experience. The air flight out of those countries hasn't been reduced by 80% has it? They explained there are several countries that have stopped direct flights but noted there are still airlines from other countries flying out, and the non-booking airlines are assisting them to book with those other airlines. In other words, in the model they presented, they assumed that air traffic would be dramatically reduced.
With that in mind, had the air traffic been reduced to 80%, they projected only a 3-4 week delay in reducing the risk of spread to other nations. They also assumed that there would be no mitigation reducing the ratio of transmission during the delay of the 3-4 week period. They assumed the 80% reduction on the front end of their projections but since that has not really happened, all of their international spread predictions are less than they should/would be/are. Does that make sense?
They had to have modeled projections with lesser reductions in air flight to arrive at the 3-4 week delay, with the 80%, and any lesser reductions in air flight would have shown an increase in the risk of spread. What I am saying is, if the air traffic matched actual experience, which I am pretty sure is no where near the 80% used, the risk of spread is/would be much greater than what they presented.
What they should have done was show actual experience, which I am guessing at most has been reduced to 50%, by the few airlines that have stopped flying out of those countries. This model isn't for the purpose of showing us the real risks we are experiencing. It had another purpose.
Maybe I am seeing too much in this, but was the purpose of this model to prove to policy makers or provide proof in support of a policy, that even if we reduced air flight by 80%, it would only delay the risk of spread by 3-4 weeks as a pre-set target someone had in mind? I would really like to see the real risks projections of what we are actually facing rather than the assumption of a 80% reduction. I would like to see a 95% reduction, too, but that might be unrealistic, because now we are flying people in there to work, and any reduction in risk by reducing the native population would be offset by the increase of flying aide workers in and out, as they would also have the potential of importing an infection.
Their assumption that there would be little mitigation during the 3-4 week delay is probably correct if the Impact of Interventions model is correct. The impact model basically determined that the outbreaks are past the point where mitigation can slow it down, predicting it will rise unabated to a peak before it decreases and that the epidemic is only in its beginning phase. In other words, the virus is in control, there is no way to stop its path of spread in those populations.
If these models are correct, in the current outbreaks in Africa, the only thing we can impact is survival rates and prevent its spread to other countries. The only prevention of spread to other countries is to reduce air flight. Looking at these three models together, that fact is inescapable. It would be interesting to see a vaccine model impact. Surely it would impact the spread if enough vaccines could be distributed to the population, especially if the claims Canada is making about their vaccine, in that it can prevent and treat an infection. The only impacts they assumed were that of diagnosis, isolation and support. I think they left in the risks in hospitals and funerals as unchanging.
The 80% reduction in air flight model showcasing the small gain of 3-4 weeks delay seems to be the major point they were trying to make in the spread model. The spread model assumed the hospital and funeral rate of infection in the spread of outbreak African countries but assumed zero risk of spread in hospitals or funerals in other countries that imported an infection. We all know that the US didn't do a very good job of preventing the spread in the hospital, so that assumption was not accurate and would have to be remodeled. It appears they did build in two scenarios into the Nigerian risks, one where they contained it, and one where they did not, but they did not correlate their experience to other countries.
The spread model was not projected beyond a month. This is because they would need to access additional outbreaks and their ability to contain those as part of the experience in the spread risks. The ratio of reproduction was a range of 1.5 to 2.0, basically doubling as we have seen used in other reports. Looking at the graph of their results it was interesting to see that they predicted the UK and the US to have a greater than, or equal risk to other African countries, with the US having twice the risk as South Africa and almost equal to Nigeria. That I am sure is primarily because we have not stopped air traffic from those countries. We have a greater risk of importing Ebola than countries that border the outbreak areas.
They also assumed that asymptomatic infected people do not infect other people. I am not sure that has been proven. They assumed singular imports of infection, no cluster imports, in other words, people traveling alone and did not factor in direct transmissions on planes. Once the import reached their destination, they assumed that person would infect less than 10 people, high end of 6 people and average of 4. They also assumed, in words, importing countries would contain the spread, but that wasn't modeled because the model was only for a short period of a month.
They did not build in any social factors of travel, in other words they assumed only local community transmission. I don't think that assumption can be applied to the culture of travel in the US. In the US we have already seen the potential of intrastate, interstate and international travel risk of spread with just two infected individuals that we know of, there could be more. Again, I am not a math person, but in explaining how they arrived at their math they referenced other models, and I am guessing those models were built from previous outbreak experience, and those outbreaks were in third world countries.
I think the report on the quarantine period was shared here already, but it basically concluded that 21 days is probably not enough.
Please feel free to criticize my analysis. I openly admit I am not a math person.
http://currents.plos.org/outbreaks/...ed-with-the-2014-west-african-ebola-outbreak/
http://currents.plos.org/outbreaks/...pidemic-of-ebola-in-sierra-leone-and-liberia/
http://currents.plos.org/outbreaks/article/on-the-quarantine-period-for-ebola-virus/#ref17
In the model for assessing the International Spread, I found the graph depicting air travel from outbreak countries revealing. It depicted the US as receiving 3000-6000 travelers from those countries weekly, not the 150 a day figure I have seen used in many articles. I also found it interesting that it described the limited access to air travel in those countries as reducing the risk of spread, with only Guinea connecting to the US.
The report noted that if Nigeria were to have an outbreak, which they did not include as an outbreak nation spreading in this model, it poses the greatest risk of spread because of the density of their population and because more people have the means/cause to travel, in addition to being an international hub connected to most other countries. They noted that Ghana was the next greatest risk of contributing to the spread if it should experience an outbreak that wasn't contained for the same reasons. In other words, they wanted to make the point that if they had outbreaks in either of those two countries, the model would dramatically change as would the risk of spreading to other countries. That would suggest to me, that if either of those two countries import a single infection, everyone better help them contain it. That may be more critical of an intervention, and at least equal to assisting in stopping the spread in current outbreaks.
I am not a math person so help me out with this to see if I am understanding their methods correctly. From what I can determine they built their spread risk model based on the assumption of an 80% reduction in airline traffic but that is not the actual experience. The air flight out of those countries hasn't been reduced by 80% has it? They explained there are several countries that have stopped direct flights but noted there are still airlines from other countries flying out, and the non-booking airlines are assisting them to book with those other airlines. In other words, in the model they presented, they assumed that air traffic would be dramatically reduced.
With that in mind, had the air traffic been reduced to 80%, they projected only a 3-4 week delay in reducing the risk of spread to other nations. They also assumed that there would be no mitigation reducing the ratio of transmission during the delay of the 3-4 week period. They assumed the 80% reduction on the front end of their projections but since that has not really happened, all of their international spread predictions are less than they should/would be/are. Does that make sense?
They had to have modeled projections with lesser reductions in air flight to arrive at the 3-4 week delay, with the 80%, and any lesser reductions in air flight would have shown an increase in the risk of spread. What I am saying is, if the air traffic matched actual experience, which I am pretty sure is no where near the 80% used, the risk of spread is/would be much greater than what they presented.
What they should have done was show actual experience, which I am guessing at most has been reduced to 50%, by the few airlines that have stopped flying out of those countries. This model isn't for the purpose of showing us the real risks we are experiencing. It had another purpose.
Maybe I am seeing too much in this, but was the purpose of this model to prove to policy makers or provide proof in support of a policy, that even if we reduced air flight by 80%, it would only delay the risk of spread by 3-4 weeks as a pre-set target someone had in mind? I would really like to see the real risks projections of what we are actually facing rather than the assumption of a 80% reduction. I would like to see a 95% reduction, too, but that might be unrealistic, because now we are flying people in there to work, and any reduction in risk by reducing the native population would be offset by the increase of flying aide workers in and out, as they would also have the potential of importing an infection.
Their assumption that there would be little mitigation during the 3-4 week delay is probably correct if the Impact of Interventions model is correct. The impact model basically determined that the outbreaks are past the point where mitigation can slow it down, predicting it will rise unabated to a peak before it decreases and that the epidemic is only in its beginning phase. In other words, the virus is in control, there is no way to stop its path of spread in those populations.
If these models are correct, in the current outbreaks in Africa, the only thing we can impact is survival rates and prevent its spread to other countries. The only prevention of spread to other countries is to reduce air flight. Looking at these three models together, that fact is inescapable. It would be interesting to see a vaccine model impact. Surely it would impact the spread if enough vaccines could be distributed to the population, especially if the claims Canada is making about their vaccine, in that it can prevent and treat an infection. The only impacts they assumed were that of diagnosis, isolation and support. I think they left in the risks in hospitals and funerals as unchanging.
The 80% reduction in air flight model showcasing the small gain of 3-4 weeks delay seems to be the major point they were trying to make in the spread model. The spread model assumed the hospital and funeral rate of infection in the spread of outbreak African countries but assumed zero risk of spread in hospitals or funerals in other countries that imported an infection. We all know that the US didn't do a very good job of preventing the spread in the hospital, so that assumption was not accurate and would have to be remodeled. It appears they did build in two scenarios into the Nigerian risks, one where they contained it, and one where they did not, but they did not correlate their experience to other countries.
The spread model was not projected beyond a month. This is because they would need to access additional outbreaks and their ability to contain those as part of the experience in the spread risks. The ratio of reproduction was a range of 1.5 to 2.0, basically doubling as we have seen used in other reports. Looking at the graph of their results it was interesting to see that they predicted the UK and the US to have a greater than, or equal risk to other African countries, with the US having twice the risk as South Africa and almost equal to Nigeria. That I am sure is primarily because we have not stopped air traffic from those countries. We have a greater risk of importing Ebola than countries that border the outbreak areas.
They also assumed that asymptomatic infected people do not infect other people. I am not sure that has been proven. They assumed singular imports of infection, no cluster imports, in other words, people traveling alone and did not factor in direct transmissions on planes. Once the import reached their destination, they assumed that person would infect less than 10 people, high end of 6 people and average of 4. They also assumed, in words, importing countries would contain the spread, but that wasn't modeled because the model was only for a short period of a month.
They did not build in any social factors of travel, in other words they assumed only local community transmission. I don't think that assumption can be applied to the culture of travel in the US. In the US we have already seen the potential of intrastate, interstate and international travel risk of spread with just two infected individuals that we know of, there could be more. Again, I am not a math person, but in explaining how they arrived at their math they referenced other models, and I am guessing those models were built from previous outbreak experience, and those outbreaks were in third world countries.
I think the report on the quarantine period was shared here already, but it basically concluded that 21 days is probably not enough.
Please feel free to criticize my analysis. I openly admit I am not a math person.
http://currents.plos.org/outbreaks/...ed-with-the-2014-west-african-ebola-outbreak/
http://currents.plos.org/outbreaks/...pidemic-of-ebola-in-sierra-leone-and-liberia/
http://currents.plos.org/outbreaks/article/on-the-quarantine-period-for-ebola-virus/#ref17
Last edited:
She has discovered that data used has some problems...
Caitlin Rivers @cmyeaton