I have decided to start reporting, hopefully weekly, developments in influenza, especially focusing on avian flu. I plan to continue these reports as long as a potential pandemic is lurking out there, with the possibility of it affecting the health and well-being of the human population. While I do not do research on influenza, I am familiar with the field, and hope to distill and translate the latest information about influenza and potential outbreaks. This information is cross-posted at my microbe blog
In this initial writing I am going to access the risks of a pandemic. This is something that I have not seen really discussed in the press with any precision. An interesting article by Ferguson et. al. was written in Science Magazine (Science. 2004 May 14;304(5673):968-9) on this subject and I will use it as a jumping off point. The article itself is behind a subscription wall, but the salient section is below.
How do we quantify the possible risks of a pandemic strain's emerging, and how can we rapidly but reliably detect the emergence of such a strain? Simple mathematical analysis can provide some insights. At any time point, the risk of a reassortment event is proportional to the number of people coinfected with human and avian strains.
The flu virus genome consists of 8 separate pieces of single-stranded RNA. Early in an infection, it is possible for two viruses to both enter a host at the same time. This is called coinfection. If the two viral particles enter the same cell, they are both replicated. During the formation of progeny viruses, the 8 RNAs are packaged inside a viral coat (called a capsid). Each of the 8 viral RNAs may come from either of the infecting viruses and the particles formed are new combinations of the two infecting viruses. This is called reassortment. Humans and pigs can be coinfection vessels where new flu strains are born. Here the authors are focusing on the infection of a human with both H5N1 and a influenza subtype that circulates more commonly in the human population.
Making the reasonable assumptions that 10% of the population are infected with human influenza over the 12 weeks of a typical influenza season, and that there is a 1-day window in early infection where coinfection with an avian strain is possible, then 0.12% of the population are susceptible to coinfection with an avian strain at any one time. Hence, even if reassortment is certain following coinfection, the probability of a reassortment event having occurred after n cases of avian influenza in humans is 1 - (1 - 0.0012)n; so 600 human infections would be required for a 50% chance of reassortment, and around 45 for a 5% chance.
The authors here show that on any given day, only 0.12% of the human population has a human influenza strain and are a potential target for coinfection with avian virus. Grinding through their equation shows that if 600 cases of avian influenza occur during the normal flu season, there is a 50-50 chance of coninfection with human influenza and subsequent reassortment.
If reassortment is a rare outcome of coinfection or other processes reduce the chance of coinfection, then the number of infections required for reassortment to become likely increases substantially. The risk of a pandemic is therefore currently small but not negligible. As the risk increases directly with the number of human H5N1 infections, it is critical that the avian epidemic is contained as rapidly as possible and that the risk of human-avian contact is minimized in affected areas.
The first analysis is a worst case scenario. Now, if we make the reasonable assumption that coninfection does not always lead to reassorment, there is some probability less than 1 and that there may be other barriers that prevent coinfection. For example, someone who is coming down with flu may not be shopping at the market that day. Or that the strains may not infect that same subset of cell in a human host. A second factor to throw in here that the authors don't mention, is that mixing of these flu strains may not lead to infectious flu particles or at least the probability of having infectious flu particles from this coinfection is again less than 1. The conclusion is, formation of a pandemic strain is going to be a rare event. Right now governments should be doing everything possible to limit the number of avian flu cases to prolong the time period we have to get ready for this pandemic. Realize that the risk is small, but at some point we are going to be unlucky and this illness is going to crossover to humans. Hopefully, with a little luck it will be later than sooner.
One other interesting note from the article is worth explaining. For any disease outbreak to be self-sustaining, the basic reproduction number, R0, must be larger than 1. What this means is if you are sick, you infect at least one more person. Another way to think about is, if there are 100 sick individuals, they will transmit the disease to more than 100 other hosts. So if R0 is 1.3, then 100 ill people will infect 130 new hosts. Ferguson et. al. suggest that public health officials should pay attention to R0 for avian flu outbreaks and they provide a formula for doing just that. They then use that formula to examine the most recent outbreak of influenza at that time (The Vietnam outbreak of 2004) and calculate an R0 value of 0.06. Clearly we are a long way from a transmissible influenza outbreak with H5N1 as the pathogen.