ABSTRACT

Based on our experience, the de-identification methods that have the most acceptability among data recipients are generalization and suppression. In some instances sub-sampling is also an acceptable option.