no equal distances in time between visits

[Agnieszka Kulesza]

Hi all,

I am trying to run a longitudinal analysis. In a nutshell, I have two groups (experimental: 18 subjects; control: 19 subjects) with 6 visits (time points). The experimental group was under specific training for 7 months.

More precisely, the first time point (tp0) was pre-exposure, tp1 was after one week of training, tp2 after 6 weeks of training, tp3 after 12 weeks of training, tp4 after 7 months of training, the last one (tp5) was a follow-up, it means in 5 months after finishing the training. As you can see, the intervals between timepoints were not equal.

The control group had the same design of visits, but no training.

I would like to model an interaction effect between time and group, the main effect of group and the main effect of time. 

So far, I modeled the simpler situation, looking only at the first 5 timepoints (without the follow-up). Taking into account your advice on the topics already discussed here (cases when there are fewer than ~25 subjects per group)  I didn’t specify the groups in the group indicator. 

Moreover, I run a model with stronger assumptions about the changes over time in the hopes of obtaining increased sensitivity. The design generated with:

nVis=5;

nGrp=[18 19];

X=[];

Xv = [ones(nVis,1) [1:nVis]'-mean(1:nVis)];

for ig = 1:length(nGrp)

  X0 = kron(ones(nGrp(ig),1),Xv);

  X  = [X                           zeros(size(X,1),size(X0,2));

        zeros(size(X0,1),size(X,2)) X0];

end

The F-test for a time interaction I used: 

0 1 0 -1

0 0 0 1

Questions: 

1. Do you think my approach is correct? 
2. How would you model no equal distances in time between visits?
3. How would you model the follow-up after finishing the training?

Looking forward to hearing from you. Thanks in advance.

Regards,

Agnieszka