Extension of Disease Burden Modeling from Seasonal Influenza to 2009 Pandemic Influenza

This dissertation, "Extension of Disease Burden Modeling From Seasonal Influenza to 2009 Pandemic Influenza" by Xiling, Wang, 王锡玲, was obtained from The University of Hong Kong (Pokfulam, Hong Kong) and is being sold pursuant to Creative Commons: Attribution 3.0 Hong Kong License. The content of this dissertation has not been altered in any way. We have altered the formatting in order to facilitate the ease of printing and reading of the dissertation. All rights not granted by the above license are retained by the author. Abstract: Reliable quantification of disease burden associated with influenza is critical to formulating proper public health prevention and intervention strategies. Direct counting of influenza infections on laboratory-test results, hospitalization records or death certificates grossly underestimates the true disease burden because of under-ascertainment and underreporting of influenza cases. To tackle the limitation of counting approach, statistical modeling approach utilizing routine public health surveillance data has been developed rapidly since last decade. However, few studies have properly discussed on how to adequately adjust for confounders in influenza disease burden modeling, such as in the Poisson model. Therefore, I conducted a simulation study to assess the performance of four model selection criteria in selecting the best Poisson model with adequate adjustment for Reliable quantification of disease burden associated with influenza is critical to formulating proper public health prevention and intervention strategies. Direct counting of influenza infections on laboratory-test results, hospitalization records or death certificates grossly underestimates the true disease burden because of under-ascertainment and underreporting of influenza cases. To tackle the limitation of counting approach, statistical modeling approach utilizing routine public health surveillance data has been developed rapidly since last decade. However, few studies have properly discussed on how to adequately adjust for confounders in influenza disease burden modeling, such as in the Poisson model. Therefore, I conducted a simulation study to assess the performance of four model selection criteria in selecting the best Poisson model with adequate adjustment for confounders. Generalized cross-validation (GCV) criterion was selected as the best criterion as it consistently provided the smallest bias and root-mean-square error in estimating disease burden of seasonal influenza. The Poisson model has been validated to provide reliable estimates for disease burden of seasonal influenza, but whether it could be extended to pandemic influenza remains unclear given that the epidemiological profiles of pandemic strains usually differ from the seasonal viruses. In this thesis, I extended the Poisson model from seasonal influenza to the 2009 pandemic influenza with proper adjustment for changes of hospital admission thresholds. The Poisson model estimated that in Hong Kong there were 10,377 excess hospitalizations for acute respiratory disease (subcategory: 7,204 for pneumonia and influenza), 1,676 for cardiovascular disease (subcategories: 848 for ischemic heart disease and 359 for stroke) and 1,550 for diabetes associated with the 2009 pandemic influenza. Compared with those of seasonal viruses, the hospitalization burden of pandemic strain clearly shifted towards children and young adults. By combining the influenza disease burden modeling with previous serology studies, the hospitalization risk of the 2009 pandemic cases was the highest in people aged 60 years or older (17.5%). Stratified by age and gender, influenza disease burden modeling revealed statistically significant gender difference of excess hospitalization for A(H3N2) in children and adolescents, but not in other age-virus categories. Influenza vaccination remains the most effective way to prevent influenza infection. The influenza vaccin