Considerations of response bias and value in linking an existing longitudinal cohort study with national health record data. (2017)