Multilevel analysis with both repeated measures and paralell groups

Analysis
1

Hi community,

I have a clinical study with the following design:

  • 60 patients studied
  • 30 patients in the untreated group (group 1)
  • 30 patients in the treated group (group 2)

Each patient was measured et 3 times D0, D1,D2:

  • D0 is the measure before any treatment (for the two groups)
  • D1/D2 are the measures after 1 (resp. 2) months for group 1
  • D1/D2 are the measures after 1 (resp. 2) months of treatment for group 2

The questions are:

  • For each time: compare the two groups
  • For each group : compare the three times (by pairs)

I wanted to apply a multilevel analysis on the patient to remove the inter-patients variability (and better see the effects between times). But I think it is not correct for comparing the two groups for each time (multilevel analysis introduce bias in this comparison). Do you agree with that ?

What do you recommend in this case ? Perform 3 analyses ?

  • The first one on the group 1 only after multilevel
  • The second one on the group 2 only with multilevel
  • The third to compare the groups at each tile (without applying multilevel)

Thanks a lot for your help,
Best regards,
Eléonore

2

hi @graviee,

We have a nested multilevel option that might suit your needs, see Multilevel Liver Toxicity Case Study | mixOmics for an example (note: not many users use it, so let me know if it works …)

I also recommend you do as you propose, just to get a better understand of that is happening in this non trivial case. More info about multilevel: Multilevel | mixOmics

Kim-Anh