What Is Data De-Identification?

So, what's the deal with data de-identification? It's a computing standard in which sensitive medical information in electronic health records (EHR) can be de-identified so that unauthorized users cannot read the actual content since it is no longer in its original state. There are two types of data de-identification: the statistical method and the anonymization method. The statistical approach is when the EHR is disconnected from the individual by removing all personal identifiers. This way, unauthorized users cannot read the actual content since it is no longer in its original state. Many people think that de-identifying data is a simple process. Just take the last four digits of a person's social security number or date of birth. But this is not true! De-identifying data can be more complicated than you think. With EHRs growing in usage throughout various healthcare institutions, facilities and private practices, privacy groups and legislators are akin to confidentiality assurance for all patients. Not considered the most straightforward programming technique, the data de-identification process can ironically make the designated data unprotected by current privacy laws. Once the data is de-identified, it is not considered protected medical data. Keeping your data safe can be done without expertise in statistics. It's true: the Department of Health and Human Services has put together a helpful guide for de-identifying data using statistical methods. It's easier than you might think! The best way around this is to include a re-identification code. However, the information can only be re-identified if the code is not identifiably connected to the individual to whom the EHR pertains. It should have no way of being translated to reveal their identity. The other strategy is to keep the code ultra-confidential. The person assigning the code must not use nor disclose the actual code for different reasons than to re-identify the information. The Department of Health and Human Services has these on its website for further instructions on de-identifying via statistical methods. Appropriately done, statistical de-identifying should not be obvious.