Using Electronic Health Records from a Large Clinical Data Research Network to Understand Cancer Burden and Cancer Risks Among Transgender and Gender Nonconforming (TGNC) Individuals

摘要 变性人和性别不一致(TGNC)人群面临着不成比例的不利健康负担 结果。尽管关于TGNC中独特的健康问题的文献越来越多 在人口方面，由于关于TGNC健康的现有数据稀少，他们仍然严重得不到充分的服务。少报 由于与社会和经济边缘化、耻辱和歧视有关的问题很常见，导致 在获得基于人口的估计方面面临挑战，因为TGNC个人往往不愿意自我识别 不愿参与传统调查。此外，过去的TGNC研究主要集中在 精神健康、药物使用和滥用以及性传播感染和疾病。数据是有限的 可用于与年龄相关的慢性疾病，如癌症，这是美国第二大死因 各州。尽管如此，癌症是TGNC人群的首要研究重点之一。以迅速的速度 日益老龄化的TGNC人口，迫切需要描述其中的癌症负担 并了解癌症对他们的影响与非TGNC患者有何不同。论 另一方面，电子健康记录(EHR)系统的快速采用使临床数据成为纵向的 可供研究。电子病历不仅包含重要的结构化数据，如人口统计、诊断、 手术和药物，但也有非结构化的临床叙述，如医生的笔记。超过80人 百分比的临床信息记录在临床叙述中，其中包含更详细的患者 包括性别认同和癌症风险因素在内的信息。受这些观察结果的激励，并以此为基础 我们以前的研究是关于1)TGNC性别认同术语的充分性，2)临床自然语言 信息提取的处理方法，以及3)基于EHR的队列研究，我们建议进行 以人群为基础的队列分析，以检查TGNC人群的癌症负担和危险因素 来自大型EHR网络的唯一数据来源-One佛罗里达州，13个PCORI资助的临床数据之一 为PCORnet做出贡献的研究网络(CDRN)。同时使用结构化和非结构化的One佛罗里达 数据，我们将首先开发可计算的表型来识别TGNC个体，并随后评估其 癌症风险。我们的研究意义重大，因为：1)没有基于人群的癌症风险队列研究 在TGNC人群中进行了调查。我们的结果将支持开发量身定制的证据- 基于TGNC人群的癌症筛查计划；2)我们的研究将创建一组TGNC人群， 不仅可以在电子病历中进行纵向跟踪，还可以招募用于未来的临床研究；以及3)使用 PCORnet CDRN使我们的分析框架可以概括到整个PCORnet。整体而言，建议的 研究将促进我们对老年TGNC人群中癌症的了解。

ABSTRACT Transgender and gender nonconforming (TGNC) people face a disproportionate burden of adverse health outcomes. Although there is a growing body of literature on the unique health issues among TGNC populations, they remain severely underserved as existing data on TGNC health are scarce. Under-reporting is common due to issues related to social and economic marginalization, stigma, and discrimination, leading to challenges in obtaining population-based estimates since TGNC individuals are often unwilling to self-identify and reluctant to participate in traditional surveys. Further, past TGNC research has primarily focused on mental health, substance use and abuse, and sexual transmitted infections and diseases. There is limited data available on age-related chronic conditions such as cancer, the second leading cause of death in the United States. Nonetheless, cancer is one of the top research priorities among the TGNC population. With a rapidly growing aging TGNC population, there is an urgent need to characterize the cancer burden among these individuals and understand how cancer impact them differentially compared to non-TGNC individuals. On the other hand, rapid adoption of electronic health record (EHR) systems has made longitudinal clinical data available for research. EHRs contain not only important structured data, such as demographics, diagnoses, procedures, and medications, but also unstructured clinical narratives such as physician’s notes. More than 80 percent of the clinical information is documented in clinical narratives, which contain more detailed patient information including gender identity and cancer risk factors. Motivated by these observations and built upon our previous studies on 1) the adequacy of TGNC gender identity terms, 2) clinical natural language processing methods for information extraction, and 3) EHR-based cohort studies, we propose to conduct a population-based cohort analysis to examine the cancer burden and risk factors among TGNC people using a unique data source from a large network of EHRs—OneFlorida, one of the 13 PCORI-funded clinical data research networks (CDRNs) contributing to the PCORnet. Using both structured and unstructured OneFlorida data, we will first develop computable phenotypes to identify TGNC individuals and subsequently evaluate their cancer risk. Our research is significant because: 1) no population-based cohort studies on cancer risk have been conducted among the TGNC population. Our results will support the development of tailored, evidence- based cancer screening programs for TGNC people; 2) our research will create a cohort of TGNC people that can be not only tracked longitudinally in EHR but also recruited for future clinical studies; and 3) working with a PCORnet CDRN makes our analysis framework generalizable to the overall PCORNet. Overall, the proposed research will advance our knowledge in cancer among the aging TGNC population.