Low frequency of structured documentation for cancer genetic testing in a large electronic health record dataset: A brief report using All of Us Research data

Abstract

Annually, over two million individuals in the United States are diagnosed with cancer, with 10–20% attributed to hereditary cancer syndromes. Genetic testing for pathogenic variants is a standard component of cancer care guided by tumor pathology and family history. Despite this, access to and completion of cancer genetic testing remains suboptimal. This study aims to understand the rates and factors associated with genetic testing completion among individuals with cancer. Utilizing data from the All of Us Research Program, which includes over one million Americans, we examined the documentation of genetic testing in electronic health records. Participants diagnosed with breast, ovarian, colon, endometrial, or pancreatic cancer were selected using the All of Us Workbench cohort builder tool. Descriptive and univariate analyses were conducted within the integrated Jupyter Notebook. Out of 60,135 individuals with a diagnostic code for the eligible cancers, over 73% reported a family history of cancer. However, only 281 individuals had a diagnosis or procedural code for a cancer genetic test, with 82% completing the test post-cancer diagnosis. While the All of Us data is a robust resource for large-scale research, challenges in data acquisition and interpretation arise due to the reporting structure of genetic findings in most data sources, such as electronic health records. To effectively utilize large-scale data for addressing issues in cancer genetic testing, shared data elements and standardized documentation are essential.