(109) Measuring Health Trends of Transgender People Using Real World Data in the United States

Background: Lack of standardized gender identity data collection in real-world data sources prevents researchers from being able to quantify and meaningfully address mental health care trends and unmet health care needs among transgender people. To date, it remains unclear what the mental health care utilization trends over time are for transgender (TG) individuals, in part because of the difficulties systematically identifying TG individuals in large administrative data sources.

Objectives: We evaluate the performance of previously defined computational phenotypes (CPs) in a health insurance claims database to identify TG individuals and use a combined CP to assess changes over time in the population identified as TG and the prevalence of common mental health disorders (anxiety and depression) in that population.

Methods: We first applied Jasuja et. al and Proctor et. al CPs independently and compared CPs by measuring their concordance and Cohen’s kappa. Concordance, Cohen’s kappa results, and substantive area knowledge informed our decision to combine the CPs and establish a retrospective cohort of TG individuals from 2007 to 2021 from Merative MarketScan commercial insurance claims. We then measured population level trends over time and sub-group trends with depression and anxiety using join-point regression.

Results: Proctor et. al and Jasuja et. al CPs identified 67,794 (0.2%) and 45,323 (0.13%) transgender patients from 2007 to 2021, respectively. The Cohen’s kappa comparing both CPs was 78.9%. We combined both CPs to reach our final sample size of 67,809 unique patients. The transgender population increased from less than 1 transgender patient per 10,000 enrollees in 2007 to 22 transgender patients per 10,000 enrollees in 2021. Gender-related ICD codes increased from 59% of all TG-related claims in 2007 to 97% in 2021. After adjusting for age and region, no major trend changes among the total TG sample were found, as there was a continuous percentage increase in transgender patients by 38.2% per year (95% CI: 32.3, 44.4). There were 44.5%, 44.4%, and 57.3% of transgender patients diagnosed with depression, anxiety, or both, respectively. Among transgender patients diagnosed with depression, anxiety, or both, a 2.0% (1.2, 7.2) increase per year from 2007 to 2019 was followed by an 8.4% (-16.8, 0.7) decline per year from 2019-2021.

Conclusions: The combined CP identified the largest transgender sample in commercial insurance claims data to date. Almost all TG individuals were identified through TG-related ICD codes for both CPs, leading to large overlap. More TG individuals had claims for gender-related procedures and HRT prescriptions in Proctor et. al. than Jasuja et. al. We chose to combine CPs as Proctor uses a conservative approach while Jasuja has a more inclusive approach for those who may not have a TG-related ICD code. Our final CP may be specific but not sensitive hence underestimating the true TG population. Large increases in the total TG population may represent an increase in access to insurance-covered gender-affirming services. Persistently high prevalence of depression and anxiety may represent the ongoing need to reduce the burden of psychiatric disorders among this population.