Ms Hannah Murdoch

Honours thesis

Estimating influenza-associated mortality in Australia

Of all vaccine preventable diseases, influenza is the number one cause of death. It is becoming increasingly recognised that influenza has a signi?cant impact on cardiac and respiratory deaths. Studies such as those by Nichol et al [1] and Madjid et al [2] have shown that influenza vaccination is associated with signi?cant reductions in hospitalisations and deaths due to cardiac and respiratory diseases. For the last forty years 'excess-above-baseline' methods [3, 4] have been used to estimate influenza-associated mortality and hospitalisation. During an influenza epidemic the numbers of hospitalisations and deaths due to pneumonia and influenza increase above a baseline. The intensity of the epidemic is measured by how much these numbers exceed this baseline. These methods do give reliable information on the intensity of influenza activity but do not provide information on the different influenza subtypes circulating in human populations (some cause more severe illness than others). Also, the impact influenza has on cardiovascular and respiratory disease is ignored. This is partly because of the difficulty in measuring this hidden cause of death. Prior to 1999, in Australia, only one cause of death was recorded on a death certificate. So if, for example, influenza has exacerbated an existing heart condition leading to a fatal heart attack, the underlying cause of death is recorded as a heart attack. Since 1999, it has been possible to record multiple causes of death to include associated causes or conditions. However, it is still possible for influenza to be overlooked because it may not be laboratory-confirmed, as is often the case. Also, it can be several weeks between infection with influenza and death due to complications, and the virus will no longer be detectable. Thompson and colleagues [5] at the Center for Disease Control and Prevention (CDC) in the United States have attempted to address this problem by developing a novel model to estimate the number of influenza-associated deaths per year using Poisson regression incorporating age and information on influenza virus type and subtype. They have also investigated the role of influenza in cardiac and respiratory deaths. The model by Thompson et al was the first attempt to quantify the relationship between influenzaassociated deaths and circulation of influenza viruses [6]. In Australia, there has been very little recent published research on the health burden of influenza, particularly on the number of deaths associated with influenza. The aim of this Project is to develop a model to estimate influenza-associated mortality using Australian mortality and influenza surveillance data. A further aim is to investigate the effects of influenza on cardiac and respiratory mortality in Australia. The model is based on the Thompson approach but has been suitably modified and extended to the Australian context and for the available data. Chapter 2 gives an introduction to influenza, discussing symptoms of influenza, the nature of the virus and methods of prevention and treatment. Sections on pandemic influenza and the potential for influenza as a bioweapon are included to highlight the importance of understanding the effects of influenza and the need for control. Chapter 3 is concerned with the theory of Generalised Linear Models (GLMs), since this is the basis of the approach by Thompson et al and the Poisson regression model developed in this project. Poisson regression belongs to the class of models known as GLMs. Chapter 4 describes previous work in estimating influenza-associated mortality. In particular, 'excess-above-baseline' methods and the Thompson et al model are discussed in more detail. A brief overview of recent studies aimed at estimating influenza-associated mortality is also included here. Chapter 5 sets out our investigation and analysis of influenza-associated mortality in Australia. The data, model development and estimates of number of deaths are discussed in detail. Chapter 6 provides a brief discussion of further research work to be undertaken on this important topic.