The integration of independent national HIV surveillance datasets and application of statistical methods to enhance their public health utility

• Main Author: Chadborn, T. R.
• Published: University College London (University of London) 2012
• Subjects: 614.4
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.625948

The aims of this thesis were two-fold. Firstly, to develop a robust method to create a fully integrated, national surveillance dataset for human immunodeficiency virus (HIV) infections by linking records from three independent, national HIV case reporting systems. Secondly, to apply statistical techniques, more commonly used in cohort study research, to the integrated dataset to yield more of the potential from the constituent information and increase the public health utility of the data. This demonstrated that an integrated dataset can reduce missing information in each surveillance system and improve information use by combining different data that was previously available only in separate databases. Using the integrated dataset, I achieved the following: accounted for missing information; described the frequency of CD4 count measurements and associated factors; determined characteristics associated with late diagnosis and consequent mortality; estimated the national incidence of acquired immunodeficiency syndrome (AIDS) and death and the influencing factors; assessed information about the date of starting therapy for each individual; and assessed factors associated with immune defence recovery after the start of treatment. These analyses will be/have been integrated into national surveillance processes as appropriate and used to publish academic papers. Lessons have been learnt for surveillance methodology with regards sharing information and ensuring that data are representative of the whole population. The large size, wide coverage and prospective nature of the integrated dataset mean that national (and local) policy decisions can be based on information that reflects the national picture rather than unrepresentative and time-bound studies. The dataset also has the power to unpick differences within small population groups. For example, evidence about late diagnoses and mortality has been used to promote the need for earlier HIV diagnosis and is updated annually and used to target local needs and to monitor improvements.