Analysis and sample size calculation for split-mouth studies

Greetings!
I would like to see if a type of dental restoration on temporary teeth has the side effect of accelerating the exfoliation of the temporary tooth. For this purpose, I am designing an observational split-mouth study where a tooth with the restoration will be with its counterpart, that is, the tooth on the other side without the restoration (split-mouth design).

It would be fantastic to receive some expert comments about:

  1. is it OK to make an estimate of the sample size as shown with R ?

     gsDesign::nSurvival(lambda1=.1 , ## hazard rate control (no restoration)
                 lambda2=.25, ## hazard rate restoration group, less since I expect "less" survival, i.e. earlier exfoliation time
                 eta = .1, ## equal drop out rate for both groups
                 Ts = 9, ## maximum study duration in months
                 Tr = 3, ## accrual (recruitment) duration in months
                sided=1, alpha=.025, ratio = 1, beta = .2); ss # default values
    
  2. Which approach do you recommend for the analysis considering the fact that the units (teeth) are not independent? My idea is to perform a survival analysis and determine whether or not the restoration with this particular material accelerates the exfoliation of the temporary teeth.

Thanks in advance!

2 Likes