Modelling influenza epidemics in the relation between black smoke and total mortality. A sensitivity analysis.

3have been studied, but not diVerent methods for modelling influenza epidemics. In previous work influenza epidemics were controlled for by means of only one dummy variable, “1” denoting the occurrence of an epidemic and “0” otherwise. 13 Using this approach we cannot control for the possible variability in total daily mortality within influenza epidemic periods between years. Other authors have suggested using quantitative information about influenza epidemics, say the daily number of cases of influenza. 4 Our objective is to conduct a sensitivity analysis examining the eVect of diVerent methods of modelling influenza epidemics on the estimated association between black smoke and total mortality in Barcelona, in the period 1991‐1995.