What can we learn from the relationship between gait deviations and clinical impairments when comparing two databases?