Abstract:
|
Introduction. The association between two variables are often assessed by correlations. Repeated measurements data have become more prevalent in medical research. However, estimating the correlation coefficient under this type of setting has been challenging. Methods. We examined the relationship between age at magnetic resonance imaging (MRI) visits and beta stiffness index in 49 patients with bicuspid aortic valve using data from the first two visits. The naïve approach calculated the standard Pearson correlation coefficient assuming the observations were simple random samples while the subject means approach estimated the coefficient using averages of two variables for each subject. To overcome the inherent problems with these approaches, parallel regression with varying intercept and linear mixed models (LMM) have been explored. Results. The parallel regression with varying intercept approach suggested that there were no significant correlations between age at MRI visit and beta stiffness index. However, the standard Pearson correlation, subject mean approach, as well as LMM approach indicated a significant positive correlation.
|