I'm using the library nlme to assess longitudinal change in volume of a given brain region (Region X) in a group of patient versus controls. I read on forums (ex: here) that log-transforming the VD can help the interpretation of the coefficients. However, I wanted to make sure that I'm correct with my interpretation. Here is my model:
model_lme = lme(log(Volume_RegionX) ~ Time*Group + AgeAtBaseline + Gender,
random = ~1 | SubjectID, method = "REML", data= mydata)
Time is in year, Group has 2 levels (Patient vs Controls), AgeAtBaseline corresponds to the age at the first visit, and Gender has obviously 2 levels (F/M). I included subject as random effect. My results are the following:
round(summary(model_lme)$tTable,3)
Value Std.Error DF t-value p-value
(Intercept) -2.143 0.464 139 -4.617 0.000
Time -0.001 0.012 139 -0.081 0.936
GroupPatient -0.289 0.058 67 -4.938 0.000
AgeAtBaseline -0.002 0.004 67 -0.593 0.555
GenderM -0.119 0.081 67 -1.475 0.145
Time:GroupPatient -0.104 0.018 139 -5.731 0.000
Is it correct to interpret this result by saying: "In RegionX, patients show an annual decrease in volume of 10.4% compare to controls"? Does this interpretation involve anything as regard to the other variables? (example: "... 10.4% compare to controls at the age at baseline of XXX and for Female [which is the reference group for Gender]).
Thanks in advance for the help.
