Influenza-attributable mortality in Australians aged more than 50 years: a comparison of different modelling approaches

• Published in: Epidemiology and infection Vol. 138; no. 6; pp. 836 - 842
• Main Authors: NEWALL, A. T., VIBOUD, C., WOOD, J. G.
• Format: Journal Article
• Language: English
• Published: Cambridge, UK Cambridge University Press 01.06.2010
• Subjects: Age groups; Aged; Aged, 80 and over; Australia; Australia - epidemiology; Biological and medical sciences; Cardiovascular Infections - mortality; Cardiovascular Infections - virology; Confidence intervals; Death; Disease models; Epidemiology; Estimates; Fundamental and applied biological sciences. Psychology; Humans; infectious disease; Infectious diseases; influenza; Influenza A virus; Influenza B virus; Influenza, Human - mortality; Methods; Microbiology; Middle Aged; Modeling; Modelling and methods; Models, Biological; Mortality; Poisson Distribution; Regression Analysis; Respiratory syncytial virus; Respiratory syncytial viruses; Respiratory Tract Infections - mortality; Respiratory Tract Infections - virology; Studies; Surveillance; time-series models; Australia; influenza; mortality; time-series models; Australia; infectious disease; Age groups; Infection; Epidemiology; Modeling; Viral disease; Influenza
• ISSN: 0950-2688; 1469-4409
• DOI: 10.1017/S095026880999118X

This study aimed to compare systematically approaches to estimating influenza-attributable mortality in older Australians. Using monthly age-specific death data together with viral surveillance counts for influenza and respiratory syncytial virus, we explored two of the most frequently used methods of estimating excess influenza-attributable disease: Poisson and Serfling regression models. These approaches produced consistent age and temporal patterns in estimates of influenza-attributable mortality in older Australians but some variation in the magnitude of the disease burden. Of Australians aged >50 years, average annual estimated influenza-attributable deaths (all cause) ranged from 2314 to 3457 for the Serfling and Poisson regression models, respectively. The excess influenza-attributable disease burden was substantial under all approaches.