Longitudinal designs are frequently encountered in epidemiologic research, particularly in the cardiopulmonary field. Many different longitudinal models have been proposed in the statistical literature including the general linear, autoregressive and random effects models, and simple models based on an analysis of slopes over time. Complex models are not widely used in epidemiology, due mainly to a lack of understanding of their underlying utility and the questions that could be answered with complex models that cannot be addressed using simple models and a lack of available software. We propose to perform a comparative study of these models on datasets from 4 large epidemiologic studies in the cardiopulmonary field. The models will be compared as regards goodness of fit, ease of implementation, and interpretability. In addition, new statistical methods will be developed to model phenomena which seem poorly-fitted by existing models, including adult longitudinal bp and pulmonary function data. Our overall goal is to develop tools for identifying appropriate classes of longitudinal statistical models. This has important public health implications, since longitudinal data continue to accumulate rapidly and no guidelines exist as to appropriate methods of analysis. Furthermore, it is often only through modelling of longitudinal data rather than through cross-sectional or separate two time-point analyses that underlying processes pertaining to change can be understood. This application is a continuation of an existing grant whose goal was to perform comparative longitudinal analyses of existing data sets as well as to develop new methods of analysis. Under this grant, we have performed (i) an important longitudinal bp analysis of the Fels longitudinal bp cohort with followup from ages 15-40 and (ii) an analysis of adult pulmonary function data on the Vlagtwedde-Vlaardingen longitudinal cohort. We propose to confirm and extend these analyses using the Normative Aging Study database (ages 40-70). We have also developed innovative methods for fitting AR-1 and AR-2 models if missing or unequally spaced data are present, have begun to address the measurement error issue in longitudinal data and have developed a new family of models (the damped AR model) which often fit existing cardiopulmonary outcomes better than AR models.
Showing the most recent 10 out of 32 publications
