Dr Simon Cauchemez

My main research interest is to improve understanding of the ways diseases spread in human and animal populations. In particular, I am interested in the development of sound methods for the analysis of epidemic data, with the aim to i) get insights on transmission dynamics, ii) support decision making and iii) optimize control policies.

 Analysis of epidemic data - Investigating the determinants of  transmission

I am interested in the analysis of epidemiological data where members of small communities (e.g. households or schools) are followed-up during an epidemic. This type of data can provide unique insights on the determinants of transmission; but analysis is challenging because it is rarely possible to determine exactly who got infected by whom. I rely on sophisticated statistical methods to probabilistically reconstruct the pattern of spread and estimate the rates of transmission in different settings.
Recently, in collaboration with CDC and the Pennsylvania Department of Health, I used those methods to study the spread of 2009 H1N1 pandemic influenza in an elementary school in Pennsylvania (PNAS, in press).  The analysis shows how social networks shape the spread of diseases.

In the past, I used similar techniques for example to analyze transmission of influenza in households in collaboration with CDC (NEJM, 2009) and Inserm (Stat Med, 2004) or to study the heterogeneity in S. pneumoniae transmission in schools in collaboration with Pasteur Institute (JASA, 2006).

In collaboration with Peter Horby and colleagues in Vietnam, I am currently developing new methods to analyze serological data which are commonly used to measure influenza infection attack rates in populations.

Impact of school closure on flu pandemics and epidemics

In 2009, in a collaborative work with ECDC and European colleagues, we reviewed pros and cons for prolonged school closure during pandemics (Lancet ID, 2009). We found that although health benefits can be expected (in particular a reduction of heath care demand at the peak of the outbreak), the intervention is associated with high economic, social and educational costs. In addition, the transmission reduction benefits may be undermined if children simply meet together elsewhere. And while school closure may reduce peak demand on health care systems in the pandemic, it can also disrupt healthcare provision via increased absenteism since a lot of doctors and nurses are parents. Two years after this first paper, we are currently reviewing national and local experiences on school closure during the 2009 H1N1 pandemic

In 2008, with Neil M Ferguson and colleagues from Inserm, I evaluated the impact of school closure on flu transmission from the analysis of surveillance Sentinel data and the timing of holidays in France (Nature, 2008). It was an opportunity to developed new statistical methods to estimate highly structured epidemic models from aggregated data. The method, which is based on Sequential Monte Carlo techniques, relies on the concept of “constrained simulation”; that is, epidemics are simulated in such a way that they are consistent with the observed epidemic curves.

 Monitoring the efficacy of control measures in real time

Another research interest is to develop statistical methods that can be used to inform the public health community and support decision making in real-time, during an outbreak. In real-time, key summary statistics are biased. I have developed relatively simple techniques that can be used in real-time to monitor the efficacy of control measures from relatively raw data. Some of those methods were recently used to analyze epidemic data during the 2009 H1N1 influenza pandemic (Science, 2009).