Background : Health research for transgender and gender non-binary (TGNB) populations has been severely limited by lack of measures to identify TGNB people. Use of a two-step method (sex assigned at birth and current gender identity) may improve data collection going forward. However, questions remain about optimal implementation of gender identity data collection in different settings and populations. Alternate methods to identify TGNB patients may be valuable to assess the performance of two step data collection measures and to access years of existent data. Data abstracted from electronic medical records (EMR) has been successfully used to identify other uncounted populations. The current study aims to assess use of EMR data collected in routine clinical care to retrospectively identify a cohort of TGNB patients among patients living with HIV seeking care at a large urban health system.
Methods : Search criteria including ICD codes, combinations of gender marker and medications, free text search terms, and fields specific to HIV care data collection were applied to a pool of 18,009 HIV positive patients seeking care at a Montefiore Medical Center between 2008-2015. Chart review was completed for all patients meeting ≥1 of these criteria. A reference standard was applied to classify patients as TGNB or unlikely TGNB.
Results : Of the 18,009 total patients, 262 patients met at least one search criteria identifying them as potentially transgender or gender non-binary. Of the patients meeting ≥ 1 search criteria, 166 (63%) were confirmed TGNB by chart review. Of the TGNB confirmed patients, 98% were transfeminine spectrum identified. ICD code variables and combination of gender marker with hormone variables together captured 130 (78%)TGNB confirmed patients, with free text key words and HIV specific demographic variables accounting for the remainder.
Conclusions : Use of EMR data can be used to identify TGNB patients in large health systems, a step toward examining TGNB health outcomes. Further research is needed to assess use of EMR based data, which may identify qualitatively different TGNB patients than those captured with claims data.