Validating Data-driven Methods to Identify Transgender Individuals in the Veterans Affairs.

We sought to operationalize and validate data-driven approaches to identify transgender individuals in the U.S. Department of Veteran Affairs (VA) health care system through a retrospective analysis using VA administrative data from 2006 to 2018. Besides gender identity disorder (GID) diagnoses, a combination of non-GID data elements were used to identify potential transgender veterans, including: 1) endocrine disorder, unspecified or not otherwise specified codes, 2) receipt of sex hormones not associated with the sex documented in the veteran's records (gender-affirming hormone therapy), and 3) change in the administratively recorded sex. Both GID and non-GID data elements were applied to a sample of 13,233,529 veterans utilizing the VA healthcare system between January 2006 and December 2018. We identified 10,769 potential transgender veterans. Based on a high positive predictive value of GID (83%, 95% Confidence Interval (CI)=77-89%) versus non-GID-coded veterans (2%, 95% CI=1-11%) from chart review validation, the final analytical sample comprised of only veterans with a GID diagnosis code (n=9,608). In the absence of self-identified gender identity, findings suggest that relying entirely on GID diagnosis codes are the most reliable approach to identify transgender individuals in the VA. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of Public Health 2021.