[Time series analysis of non-typhoid salmonella infections].

The use of time series modeling for non-typhoidal salmonella infections is important for improving epidemiological surveillance. Specific techniques which were necessary for our objectives are detailed in an accompanying paper. For each series analysed, we discuss how to choose an appropriate statistical model taking into account both the data structure and the specific objectives. For salmonella infections, we have quantified a "media coverage effect", constructed and alert threshold and estimated a link between two series. We have thus shown that the analysis of temporal variations is of relevance in public health and can contribute to epidemiological knowledge.