Hi Lael,
only have found your post today! This sounds interesting - we’ve found shorter CDR3s in T1D (PMID:29176645), although clearly it was related to the different insertion/deletion pattern. I would look into VDJ usage, as Jamie suggested.
Also try to eliminate confounders - what was your repertoire source, PBMCs? Sorted cells? As different cell subsets might have different repertoire and a difference in ratio of CD4s to CD8s might explain a difference.
Also, I assume that patients/controls were processed together, so nothing like a batch effect?