Seasonality and Seasonal Decomposition Analysis of hMPV
HMPV epidemics alternated years (Fig. 4). Years with the highest hMPV incidence were 2011, 2013, and 2015. Cases tended to peak during July­–August, however there was additional variation of incidence throughout the year, which was particularly notable in 2015 (Fig. 4 and Supplemental Fig. 2). In 2012 and 2014 there was no major epidemic. There was a rise in case numbers in 2016 in September­–January which differs from peaks in previous years, however it is possible that the peak epidemic period occurred later into 2017 (Fig. 4, Supplemental Fig. 2).
When the trend (Tt ) was isolated from the seasonal and remainder components in Classical Additive Decomposition, there was no overall observable increase or decrease in the number of monthly symptomatic hMPV cases (Fig. 5B). When the seasonality and remainder were extracted from the trend there was a distinct decrease of the number of cases in 2012 and 2014 (Fig. 5B), and an increase in cases in 2013 and 2015. This trend was also noted in the remainder component where there were significant increases and decreases that were not explained by the trend (Tt ) or seasonal component (St ) (Fig. 5C, D). The increases in symptomatic hMPV cases when annual seasonal patterns (St ) and longitudinal trend were extracted (Tt ) revealed a biennial residual pattern.