Respiratory syncytial virus seasonality in Brazil: implications for the immunisation policy for at-risk populations

Freitas, André Ricardo Ribas; Donalisio, Maria Rita

Time series -Wavelet and Fourier analysis -Initially, plots were made to visualise time parameters and seasonal trends. The IIF positivity proportions of nasopharyngeal secretion from laboratory data and hospitalisation rates were analysed using the wavelet technique for identifying periodic patterns of RSV occurrence and associated morbidity. Wavelet analysis provides information on the time-scale domain for stationary and nonstationary events, allowing for simultaneous localisation of time and of high and low frequencies. Subtle time patterns of occurrence of phenomena of different nature can be detected, as well as changes in its periodicity. Wavelet analysis permits inferences regarding the way continuous variables relate to different frequencies and whether this relationship changes over time (Torrence & Compo 1998 , Cazelles et al. 2007 ). The algorithms described by Torrence and Compo (1998) , available in free software were used (Alonso & McCormick 2012) . This analysis allows establishing whether a specific periodic phenomenon has a stationary frequency, a basic assumption required in Fourier analysis.