Hong Kong Confirms First Human Case of Bird Flu
reuters.com
reuters.com
The games goes together with the online MOOC "Epidemics - the Dynamics of Infectious Diseases" https://www.coursera.org/course/epidemics
1. Yes, this thing has a mortality rate of 30%. What that means is that out of all detected cases 30% of the people have died. The keyword here is detected. At this stage, and without a massive surveillance plan looking for specifically for serological evidence of the virus, authorities are only aware of a tiny fraction of the actual cases (the most severe) which tend to skew the mortality rate. The actual mortality rate for the 2009 H1N1 has just now been reevaluated with more accurate data and techniques [1, 2]
2. So far it doesn't seems to have crossed the threshold to be able to spread quickly from person to person. Doing that will imply a few more mutations that will likely make it less aggressive and deadly (it's hard to spread if you kill your host too quickly).
3. Better than watching Contagion to get an idea of what a spread looks like at a global scale is to watch [3] which is an accurate representation of the H1N1 spreade in 2009. Each edge you see represents the (likely) first infected person traveling from an infected to an uninflected city. You can play with the (client) software we used for these simulations over at [4]
4. Attempts to use real time proxy data like mentions in Twitter and searches in Google to monitor the spread of infectious diseases haven't really been very successful. Google flu trends had some early success but it seems to be breaking down recently. As usual Twitter seems to be much better at predicting the past than the future. In fact the CDC has recently launched a competition to try and do just this [5]
5. In summary, don't panic. Much smarter people than myself are keeping an eye on this and it's likely not as bad as the link baity headlines would have you believe.
[0] I worked directly with the official surveillance data and used it to model And forecast the worldwide spread of the virus. We were actually able to predict (and publish) the epidemic peak for dozens of countries I'm advance You can access all the relevant publications here: http://www.bgoncalves.com/publications.html
[1] http://www.npr.org/blogs/health/2013/11/26/247379604/2009-fl....
[2] http://www.plosmedicine.org/article/info%3Adoi%2F10.1371%2Fj...
[3] http://m.youtube.com/watch?v=YAf1aXCvvdU
[4] http://www.gleamviz.org/simulator/
[5] https://www.federalregister.gov/articles/2013/11/25/2013-281...
Edit: spelling
In the case of SARS, even though it killed more people in Hong Kong than anywhere else -many of them doctors and nurses-, the secondary effects of the disease affected everyone here: travel to Hong Kong stopped, the economy tanked, companies and shops closed and people lost their livelihood, no-one wanted to invest, everyone was afraid.
We should always be careful with the headlines.
I couldn't agree more. Even SARS, concerning as it was when its pathology was unknown, was associated with fewer deaths in Hong Kong than influenza during the same year.
I've done some epidemiology, but I'm not a biologist. I don't understand your line of reasoning there. I suspect it stems from a logical mix-up, but its not detailed enough to say for sure.
I understand that if a disease kills its host, before it can infect more people, then it will 'burn itself out' and fail to spread. Therefore, all other things being equal, a disease that kills more slowly, or kills less, will spread more.
However, that does not mean that a disease must become less aggressive and deadly to spread quickly from person to person - or even has pressure to.
We can use the terminology of compartmental epidemiology modelling to make this clearer:
- The rate which describes how contagious the disease is, between an infected person, and a non-infected person: 'B' (beta).
- The rate a which someone infected becomes non-infective (recovered or dead): 'v'.
- Lets also talk about the mortality rate, the % of infected people who become dead, rather than recovered.
You say:
>(it's hard to spread if you kill your host too quickly).
I interpret that as a statement about 'v': you are saying that if the disease kills too quickly, it won't spread too fast.
That's true - if 'v' is much shorter than a normal flu - let's say its minutes, as an illustrative unrealistic extreme - then clearly even a very contagious disease (high Beta) will burn itself out.
However:
> So far it doesn't seems to have crossed the threshold to be able to spread quickly from person to person.
That is a statement about 'B'.
You imply that it is necessary, or likely, that 'v' changes as 'B' increases.
But what reason do you have to think that?
There's no reason to think that a disease couldn't have a normal 'v' (couple of weeks; but even a couple of days would probably still be incredibly dangerous), but a really high 'B' and a really high mortality rate, if we are very unlucky.
The mutation of the disease isn't guided, or intelligent; it doesn't care if it reaches equilibrium with its host population, or kills us all.
The evolutionary pressure on the disease to become less lethal is a populations-of-diseases level pressure (i.e. the specific disease will die, if it kills all humans); whereas the evolutionary pressure to spread more applies at the level of individual members of this specific disease (which is normally how we think about evolutionary pressures).
Why think there's a casual link from the former to the latter? I believe your argument mixes the two up, leading to a faulty conclusion, and an unwarranted assumption of an extra safety factor.
I've seen a number of people working in the area of epidemiology make what I think is this mistake, and I'd like to know whether I'm missing something or they are.
You are correct, of course. I might have tried to say too much in a single sentence, I did say "likely make it less agressive" because that's what usually happens with the flu (even true for the 1918 Spanish flu). Of course, this is not a requirement, and you have the example of Smallpox with a 30% mortality rate and an R0=B/v in your notation of around 5 (flu is usually ~1.2-1.5).
There's usually a problem in terms of notation and confusion when moving from "intra-host" populations to "inter-host" populations. Starting with intra-host, a "good" quasi-species argument for why this is likely is:
If you assume that mortality is one of the dimensions of the high dimensional "genetic vector" space and person to person transmissibility another, by randomly exploring the space through mutations and reproducing the mutations that produce more offspring you are likely to move away from the "high mortality" area unless that's somehow important for viral survival. Any virus that is able to keep the host alive for longer with have more time to explore it's genetic space for longer to find the region where it is able to spread more easily and will also likely generate hosts with higher viral load leading to easier inter-host transmission. Or, as you say,
a disease that kills more slowly, or kills less, will spread more.
It's not easy to talk about these things just considering Beta because Beta hides many factors, such has number of contacts, probability of exposing another person to the virus, host viral load (which affects how many virions a person is in contact with), etc...
A simpler and perhaps not completely correct way of saying this is that "it's hard to spread if you kill your host too quickly" because you have less time to generate a high viral load, less generations between transmissions to look for beneficial mutations, etc...
As an extreme example, you have HIV which has a generation time of a few hours and a huge mutation rate, so each few hours your immune system has to deal with a "new" virus until it can no longer cope. As I've heard several people say, HIV wins the war by losing every battle.
Fortunately, it doesn't seem to spread human-human particularly well - yet.
Mortality source: http://en.wikipedia.org/wiki/Influenza_A_virus_subtype_H7N9#...
Transmitability source: http://en.wikipedia.org/wiki/Influenza_A_virus_subtype_H7N9#...
How they do it: http://www.google.org/flutrends/about/how.html
Also, I saw a presentation given at an infectious disease conference...less than a month ago. Google Flu Trends appears to really break down at small spatial scales, like a single city. Unfortunately, Google is also super opaque about how it works, so it's hard to tune.
Inclusion of an "its" would help: "Hong Kong reports its first human case of bird flu".
What? No!
This virus doesn't just magically show up in a bird, the bird has to catch it from another bird.
Stay away from large flocks, sure. But home pets? No problem at all.
Indeed, there's some evidence to suggest avian strains of influenza (vs. pig strains) have a little bit of a disadvantage when it comes to kicking off pandemics.
Also, early mortality estimates are super unstable.
Has anyone else noticed a pattern here?
That said, I'm imagining the small possibility that people of the future look back on this comment going like "this guy knew and everyone ignored him."