Need help in choosing appropriate modeling method

Dear TalkStats!

I'm having a bit of a trouble, as I'm quite new to working with longitudinal data.
I'm analyzing longitudinal change in relationship quality, with 4 measurement points in 4 waves. The problem is that the relationship duration is very heterogeneous in the sample, as some started the survey period in new relationships, some in established, long marriages.
I'm looking to examine the changes in relationship quality through time, and an appropriate time measure would be relationship duration, however, i am unsure how to proceed with my analysis, as there is individual heterogeneity in time (no one is measured at exactly the same time into their relationship).
Is there a good method to analyze growth and change with this heterogeneity in time?
I was thinking about mixed effects growth modeling. (suggested in

- I used group based trajectory (traj in stata, by Nagin) to examine possible trajectories, using time passed since first wave as a time variable - there are interpretable results, but it would be much easier to understand with relationship duration.
- Separating the sample is not an option, as relationship duration is too heterogeneous, resulting in very small subsamples (<100 respondents).

Thank you for any input, and sorry for my english.