Diseases such as HIV and chicken pox spread in very different social networks depending on the mechanism of transmission. For many diseases we have an idea of the typical number of contacts people may have in the relevant social network. Until recently most of the predictive tools we have had for infectious diseases spread have assumed that the networks are randomly wired so that people are as likely to infect someone across town as their next-door neighbour.

Reality, as always, is messier than theory. People tend to have contacts in groups, and so the networks along which disease spread have significant clustering. We have not had good mathematical tools to predict how clustering impacts the spread of an infectious disease.

This paper shows that clustering can significantly reduce the early growth of an epidemic. However, if the typical number of contacts is not small (perhaps at least 5), then although clustering slows the early growth, it does not significantly change the final number infected in an epidemic - it will eventually reach all the "pockets" in the community.

These results help us to plan strategies to respond to a new emerging disease such as SARS. Our ability to better predict the early growth gives a better understanding of how long we have to prepare a response, while having more certainty about the final size of an epidemic helps us choose the appropriate size of the response.

Journal of the Royal Society Interface

Journal of the Royal Society Interface is the Society's cross-disciplinary publication promoting research at the interface between the physical and life sciences. It offers rapidity, visibility and high-quality peer review and is ranked fifth in JCR's multidisciplinary category. The journal also incorporates Interface Focus, a peer-reviewed, themed supplement, each issue of which concentrates on a specific cross-disciplinary subject.

http://rsif.royalsocietypublishing.org/site/misc/focus.xhtml