Cohort studies were found to be frequently biased by missing disease information due to death.