I am not dealing with this for many years, so may be I could miss some recent advance. Any way here is my comment.
Repeated eigenvalues bring associated different eigenvectors, which means that sensitivities of parameters are somehow different from sensitivity in case of a single eigenvalue.
In some situations, repeated eigenvalues are associated with higher sensitivity to perturbations in the system parameters but such characterization is not easy to do (at least for me). I assume it is a open problem but also quite difficult one specially due to the difficulties in experimental validation of it.
Unfortunately, I do not remember of one simple good text with this subject. May be others could recommend here something on it.