Using Laboratory Data to Aid Early Warning in Prospective Influenza Mortality Surveillance

Abstract

Many countries prospectively monitor influenza-attributable mortality using a variation of the Serfling seasonal time series model. Our aim is to demonstrate use of routine laboratory-confirmed influenza surveillance data to forecast predicted influenza-attributable deaths during the current influenza season. The two models provided a reasonable forecast for 2012. The model forecasts of weekly deaths during 2012 were compared against observed deaths using root mean squared error (RMSE). The results shown that the model including influenza type A and B provided a better fit. Here, we demonstrated a time series model for influenza-attributable mortality surveillance based on laboratory surveillance information.