“Anonymization” and “De-identification” of Medical Records
This post is from the ‘you’re not in Kansas anymore, Dorothy’ perspective. It seems now that we’ve created the momentum for every patient in the country to have an electronic medical record (EMR), we’ve got to figure out an effective way to gut millions of records of any individual identifying information. The broader social value of EMRs, beyond avoiding medical errors for individual patients, is to make them usable for health systems and medical research purposes. The trick, however, will be to make that important use conform to the requirements of federal law (HIPAA) protecting the privacy of health information, while at the same time retaining the data of value to research.
So it’s not without surprise that we see being reported software programming attempts to do just that. Thanks to the International Council on Medical Care & Compunetics (ICMCC) we hear of several projects (abstracts only) which show promise (here, here, and here). Finding the confidential identifiers goes beyond just the usual suspects — name, address, doctor, etc. — but to info embedded in the EMR which could traced to what HIPAA has targeted as links to personal identification. That will be the challenge.
The impact of the social and technical engineering attendant to EMRs is just starting. One can imagine a Kafkaesque postmodern dystopia emerging as ordinary folks try to figure out who-has-what when it comes to their virtual personal medical information. I think most of us will need a little help.
