Aspects of Surveillance of Outbreaks

Outbreaks, defined as changes from a constant level to a monotonic trend, are of interest in connection with epidemic diseases but also in financial and other contexts. It is important to detect the onset of the outbreak as soon as possible to allow timely action to be taken. New observations are made sequentially, and decisions whether the outbreak has started or not are made after each new observation. Statistical surveillance, which continuously evaluates the evidence, can be used to ensure a quick and reliable detection. In paper I, “Robust outbreak surveillance of epidemics in Sweden”, a semi-parametric method of surveillance was suggested and studied. The properties of the method were evaluated both by application to Swedish data on influenza and tularemia and by Monte Carlo simulations. The method was compared to other statistical methods as well as to surveillance based on subjective judgment. In paper II, “Explorative analysis of spatial patterns of influenza incidences in Sweden 1999—2008”, regional influenza data were studied. Two different types of data were examined and some quality problems were identified. The emphasis was on the onset phase of the seasonal outbreak, and the aim was to examine possible spatial patterns in the spread of influenza. Such information could be used for efficient multivariate surveillance. A new spatial model was suggested as a basis for surveillance.