If you’re planning a chart review study, before submitting, please give some thought to whether you could get your study data from the AR-CDR rather than directly from medical records. The AR-CDR (it stands for Arkansas Clinical Data Repository) is a vast collection of data extracted from EPIC and other, older, medical records systems. It can be searched for specific diseases, date ranges, diagnoses, procedures, you name it — you can get exactly the data you want or need out of it.
Best of all, the AR-CDR can provide you data without any identifiers, i.e. anonymized, if that would work for your study. Even things like dates can be altered so you can still measure when events occurred in relation to each other, but without getting dates, which are identifiers (just ask HIPAA!).
Why would you want to get anonymized data? First, the less identifiable data we have out there, the lower the likelihood of a HIPAA breach. Also, a study that uses only completely anonymous data, provided to you without anything allowing reidentification of subjects, does not involve human subjects as defined in regulations, and so would not require IRB review.
We hasten to add that even if you plan to use only anonymous AR-CDR data, you will still want to submit a human subject research determination form through CLARA to get an official letter indicating your study does not require IRB review.
Of course, there are times when anonymized data just won’t work for your study. For example, you may need to supplement what you’re able to extract from AR-CDR with information from handwritten notes in patient charts, requiring you to manually search the medical record. Note that AR-CDR can also give you data that includes identifiers. Or your area may have its own specialized clinical database that better suits your needs than AR-CDR data would.
Last but not least, when you submit a chart review study to the IRB, please specifically describe the source of your data, and indicate why you need identifiers for it, if indeed you do need identifiers. Otherwise, we’ll send back a contingency asking you to justify why you can’t use the study using anonymized data from AR-CDR instead (which we can now type without even thinking about it, because of how often we’ve sent that one back).