I have a question about whether something could be considered acceptable or not (I have feeling I really shouldn't do this but I do not have tons of experience with this particular statistical topic).

I have a project I am working on involving creating a matched cohort (using MATCHIT in R). I played around with several matching methods (genetic, optimal, nearest neighbor....) to find which method seemed to product the best match (Genetic was best in my case). However, as time goes on, more people will be added to my treatment group which means I will need to match those people to some sort of control. I do not want to continuously re-match everyone, however (once someone has been matched that is there match forever).

When I did the initial match there were 130 or so people I needed to match. Yet, new members are only added in very small numbers and since I am doing an exact match on a certain variable there is likely only 1 treatment person in a particular group. The problem I'm running into is that the MATCHIT package for genetic matching doesn't seem to like when there is only 1 treatment in a group. However, Nearest Neighbor seems to handle this alright.

My question: even though I matched my initial 130 or so people using a genetic method, is it acceptable practice to use a different matching method for future people (again, my guess is no).

Any thoughts/opinions would be helpful.