Correction of Bias in Estimating Risk of AD and Cognitive and Mobile Decline Using Auxiliary Information

Abstract Evaluating risk of Alzheimers' disease (AD) and rate of decline in cognitive and mobile performance and possible risk factors is an important goal in many longitudinal aging studies as AD and cognitive/mobile decline are global public health problems of enormous significance. An important challenge facing many longitudinal aging studies is non-random missing data. Participants with poorer health may be more likely to drop out or miss a visit, therefore missing data is likely not at random. This can result in biased estimates of change in cognitive and mobile performance and risk of AD, as all statistical analyses are based on the assumption that the data is missing at random. Misleading scientific conclusions can be obtained as a result. Auxiliary data, measures that are associated with the outcome, allow us to evaluate the random missing assumption and to eliminate or reduce bias from non-random missing data. We have shown that auxiliary data have the potential to correct for the bias caused by informative missing data in preliminary works. However, details of how the auxiliary data and missing data process affect the results from utilizing auxiliary information remain unclear, and other sources of incomplete data including ceiling and floor effects further complicate the problem. Research on utilizing auxiliary information to handle non-random censoring for time to AD and dementia is also lacking. In this study we plan to evaluate the impact of auxiliary data and missing data on the estimation of risk of disease and change in longitudinal outcomes, through extensive simulation studies, and with application to incidence of AD and the decline of cognitive and mobile performance in aging cohorts. This project will result in better understanding and application of auxiliary data, which has the potential to provide better approach to handle missing data using all available information, and improved study designs of clinical trials and observation studies in aging.