people publications research teaching jobs
contact publications research teaching opportunities

Avian Influenza (bird flu)

Can vaccination of poultry against bird flu mask the spread of the infection?

Experimental vaccination of poultry, particularly chickens and turkeys, shows that it is effective; birds are much less likely to become infected, and if they do, will excrete much less virus in their faeces (the main source of infection of bird flu). But, it does hide infection. People have argued that this may lead to 'silent spread' of infection within and between poultry flocks.

When we were asked to look at this phenomenon in UK poultry flocks by the Department of Environment, Food and Rural Affairs (Defra), we discovered that no studies had been done to quantify silent spread.

We therefore set about developing an individual-based mathematical model to see how vaccination affected the spread of infection within poultry flocks that were infected by highly pathagenic . We trawled the literature and asked experts for their opinions on what should be put into the model and for values for appropriate parameters.

The model, simulated on a computer, tracks the infection status of each and every bird in a flock over the course of an outbreak. What we found was unexpected. Figure 1 illustrates our results for a caged system of 10,000 birds (though the qualitative results are robust to changes in many of the parameters), and assume that the vaccine is 100% effective, in other words, a vaccinated bird will never be infected.

We vary the number of birds within the flock that are protected from 0% (unvaccinated flock) to 100% (completely protected flock). As protection rises in the flock the probability that an outbreaks occurs drops (Fig. 1A, solid line), but results in outbreaks going undetected (Fig. 1A, dashed line).

In uninfected commercial poultry flocks, birds die at a rate of about 0.05% per day for a variety of reasons. So bird flu is detected (or notified) by farmers when the death rate is usually much higher than this. The Dutch and Italian authorities have established certain detection thresholds. When these thresholds are passed poultry farmers must notify the authorities of suspected bird flu. The thresholds are:
  • mortality greater than 0.5% on two consecutive days
  • mortality greater than 3% over a week
  • feed or water intake or egg production down by 5% over two consecutive days
  • one dead infected unvaccinated sentinel bird
So, as protection levels rise, there are fewer birds that become infected and die. This slows down the spread of infection and reduces death rate. Thus it takes longer to detect bird flu, or it may never be detected at all.

At the end of a flock's production cycle, birds are removed from their barns and transported to abattoir, and their housing units cleaned. Biosecurity tends to break down, so this time is thought to be critical in the transmission of bird flu between flocks and farms. Because bird flu is harder to detect as protection rises it becomes more likely that undetected infection may be present at the end of the production cycle (Fig. 1B). If enough birds are protected (more than about 95% for this case) then only a few birds become infected and the outbreak never takes off.

During an outbreak, infective faeces accumulate in the housing units of flocks. The infectiousness of faeces does decay rapidly over a few days, and the dryer the conditions the faster the decay. As protection levels rise in a flock, outbreak duration rises, but fewer birds become infected. These two counter-acting forces cause a drop then a peak and finally a drop in the cumulative infectiousness of the infective faeces over an outbreak (Fig. 1C). This effect is still present when we take into account the reduced probability of an outbreak occuring. So what was most surprising about this work was that a flock with about 80% of its birds vaccinated may be more infectious (in terms of chance of undetected outbreak at end of production cycle and cumulative infectiousness of faeces) than an unvaccinated flock.

Another surprising result is that the effectiveness of a vaccine in protecting individual birds from infection makes little difference to silent spread of bird flu. Our work suggests protection at the level of a flock is a more important risk factor of silent spread than vaccine effectiveness in individual birds.

Vaccination should never be done without the use of unvaccinated 'sentinel' birds. These birds are the best method for rapid detection of bird flu in a vaccinated flock. Figure 2 shows the effects of placing 100 sentinel birds into a caged flock.


They reduce the chance of undetected outbreaks at the end of the production cycle and the cumulative infectiousness of infective faeces during an outbreak. Figure 2C shows an interesting result; the use of sentinels even in unvaccinated flocks speeds up the detection of bird flu (compare to Fig. 1C). This is because bird flu is detected when just a single bird (a sentinel) dies, rather than having to wait for a many bird deaths to pass a detection threshold.

This research is published in
Nature. Supplementary material, podcast, podcast transcript and UoE Press release.

Document made with Nvu  Created 24 August 2006. Updated 2 April 07.