Attrition in longitudinal studies: advancing and evaluating statistical methods (2009-2011)

Longitudinal studies offer a unique opportunity to increase knowledge of the causes and mechanisms behind disease and mortality. Because of their length, these studies not only have rich scientific potential, but are likely to have cost a significant amount of time and money. Studies that have collected data over many years often have high levels of attrition, or loss to follow-up, which threatens their validity and generalisability. Our proposal aims to regain the full potential of longitudinal studies by determining the optimal designs for minimising attrition and methods for correcting resulting bias, and to produce recommendations for the analysis and reporting of attrition in longitudinal studies.