A SMOOTHING DYNAMIC MODEL FOR IRREGULARLY TIME-SPACED LONGITUDINAL DATA

摘要

In nondesigned longitudinal observational studies, irregularly spaced measurements are commonly present over a period of follow-up time. We propose a smoothing dynamic model, based on the idea of varying coefficients, to analyze this highly unbalanced longitudinal data. The estimate of model parameters can be obtained by implementing a well-developed Bsplines technique. Our method is illustrated with data from a primary care based longitudinal cohort of rheumatoid arthritis patients. The results show that the effects of some risk factors might be underestimated by an intention-to-treat analysis using a last-value-carried-forward method.