A privacy challenge to longitudinal study methods: patient-derived codes.

Recent changes to privacy legislation in Australia have resulted in more stringent requirements with respect to maintaining the confidentiality of patient health information. We describe a method employed to de-identify health information collected in a longitudinal study using codes. Using a patient-derived code that did not change during the life of the study follow-up resulted in errors in a quarter of the follow-up surveys. This may introduce bias that could compromise the validity of the study. Alternative methods of coding may alleviate some of these issues. However, removal of some of the constraints imposed by interpretations of privacy legislation may be the best way forward.