Model outcomes were stratified by age (<6 months, 6 months to 4 years, 5–14 years, 15–44 years, 45–64 years and 65+) and clinical risk group. Due to the small number of deaths in hospital in patients under 65 years patients were grouped into <15 years and 15–64 years to estimate influenza-attributable deaths in hospital. AZD2281 chemical structure Seasonal variations in the numbers of laboratory reports for the 8 pathogens likely to cause acute respiratory illness are

shown in Fig. 1 for two key age groups: 6 months–4 years (panel A) and 65 years and over (panel B). Respiratory syncytial virus dominates reports in young children during winter, while S. pneumoniae dominates reports in older people throughout the year, but especially

during winter. For influenza, there is substantial variation between seasons in the number Nintedanib of laboratory-confirmed cases by age group and strain ( Fig. 2) There was an annual average of over 300,000 admissions for acute respiratory illness among those without co-morbidities and almost 520,000 among those in a clinical risk group; the majority of the admissions and the highest case fatality rates were in 65+ year olds (Table 1). In all age groups, the incidence per 1000 population of admission for acute respiratory illness was higher in those with a clinical risk.

For those under 65 years of age, the risk of dying in hospital was much higher for those in a clinical risk group, declining from 35.1 times higher in <6 month olds to 5.9 times higher in 45–64 year olds. In 65+ year olds the case fatality rate was similar in those with and without a clinical risk. The best fitting model to the weekly number of episodes leading to hospital admissions, consultations in general practice and deaths reproduces the observed annual averages to within 1% (Supporting Text – Section 4). This model was one that incorporated these a moving average to smooth out laboratory reports, and a linear increase in the number of hospitalisations not attributable to specific respiratory pathogens. Separation of influenza A into subtypes, allowing for interactions between co-circulating pathogens and incorporating a temporal offset did not improve model fit. Detailed results of the fitting process, and examples comparing the best fitting model results with data on the weekly number of hospital admissions, GP consultations and deaths for various age and risk groups are presented in the Supporting Text (Sections 1–3). The contributions of the various pathogens to the attributed disease burden are shown in the Supporting Text – Section 4. In both risk and non-risk groups, S.