Feedback on the use of race and ethnicity data in general for which this value in the value set for Observation.method provides an example: • "administrative" - The definition for this value is: Identifies that the information about the subject was collected by an organization for administrative reasons not directly related to obtaining healthcare or health insurance.
Comment: The American Medical Association generally supports voluntary data collection efforts in attempts to improve health equity and outcomes and reduce disparities. However, patient privacy, autonomy, and personal preferences are of the utmost importance and must be preserved; even in recognizing the need to share data for public health purposes (including public health emergencies) we are careful to recognize the need to narrowly tailor such disclosures. The AMA is particularly wary of race and ethnicity data being shared with individuals or entities (including those within the health care system) for non-treatment purposes, including payers, Big Tech, and government entities. Prior to collecting race and ethnicity data, implementers and health care providers (including clinicians) must be prepared to answer questions around why race and ethnicity data are being collected, how such data will be used and shared, and by whom. Data collection and privacy protection practices should be developed concurrently to prevent the creation and exacerbation of health inequities. Part of this involves careful attention to how terms are defined and who defines them. For example, do we want tech companies or public health authorities and clinicians defining what "public health" is for purposes of data disclosure? When should race and ethnicity data be disclosed to government agencies or law enforcement, particularly as we all become more aware of the disparate treatment of individuals by certain governmental entities based on the color of their skin or national origin? When is individually identifiably data truly necessary to accomplish a stated purpose as opposed to aggregate or disaggregated data? Bearing in mind that algorithms are built on data sets that embed the racist structures in which they were created, will the collected race and ethnicity data be used to fuel machine learning and augmented intelligence tools that perpetuate inequities?