Anticipation of age at onset in multiple sclerosis: methodologic pitfalls