69 Incidence Peaks and the Role of Influenza We again separately analysed NT1 from the combined group of NT2/IH. Detailed yearly influenza data was extracted for each included country from the Global Influenza Programme of the World Health Organization (WHO) [165]. This platform combines FluNet and fluID data by the Global Influenza Surveillance and Response System (GISRS) and national epidemiological institutions to provide world-wide yearly monitoring of influenza season severity with specification of different strains (type A: H1N1 and H3, type B: Victoria and Yamagata). The Global Influenza Programme is dependent on monitoring policies of individual countries and data is therefore not uniformly acquired over countries. After the pH1N1 many countries only implemented consistent subtyping for type A influenza infections, and type B influenza subtyping data was only available for the Netherlands in this study. Since the Global Influenza Programme reports sentinel and non-sentinel specimens that were collected after medical consultation, it also does not provide reliable estimations for mild influenza infections that generally do not require medical consultation. The Dutch National Institute for Public Health and the Environment has since 2012 annually released generalized estimates of symptomatic influenza incidence for the whole population [166, 167]. These estimates provide more reliable influenza incidence rates in the whole population instead of only reporting laboratory-confirmed test results, especially for infections less frequently requiring medical attention such as type B influenza. This measure is calculated by multiplying the national influenza-like illness incidence rate with the distribution of influenza strains (derived from laboratory testing of a representative random subsample of people with influenza-like illness), while correcting for healthcare seeking behaviour. Linear mixed models were implemented to investigate the relationships between hypersomnolence disorder onset and preceding influenza season severity. Data from the Netherlands, France, Italy and Czech Republic were first normalised per country considering the population sizes among these countries. Normalisation was performed by indexing the number of influenza infections in each year to the total infections over years in each country. Linear mixed models with country as random effect were separately performed with type A H1N1 and type A H3N2 influenza incidence rates as predictors, to predict the incidences of different CDH phenotypes. Analyses were separately performed for NT1 and the combined group of NT2/IH, and within these groups for all individuals, only children and only adults. Additional non-parametric Spearman correlation analyses were implemented (data were not normally distributed) to respectively correlate the NT1 incidence rates with influenza season severity in individual countries. 3
RkJQdWJsaXNoZXIy MjY0ODMw