Considerations for conducting and interpreting long-term follow-up of intervention studies: avoiding spoiled milk.