Abramson Cancer Center Support Grant

点击翻译按钮获取中文摘要

PROJECT SUMMARY/ABSTRACT Colorectal cancer is the 4th most common cancer diagnosed and the 2nd most common cause of cancer death in the United States. The age-adjusted incidence of CRC in Philadelphia County – a persistently high poverty area– is nearly 25% above the national average. Guideline-based screening for CRC via colonoscopy or fecal immunohistochemistry (FIT) reduces CRC-associated mortality. Despite the proven benefit of regular colonoscopy, CRC screening completion rates remain only around 50% in Philadelphia County and are consistently 10-15% lower for African-Americans than White individuals living in Philadelphia County. Adverse social determinants of health such as high poverty contribute to CRC screening nonadherence disproportionately for African-American and other minority communities. Recognizing this, in 2011, Penn Medicine created a navigation program to increase access to screening colonoscopies for patients in underserved areas of West, South, and Southwest Philadelphia by providing services that reduce barriers to cancer screening, including transportation assistance and detailed instructions on bowel prep. Despite initial success in increasing colonoscopies, a key challenge in scaling this navigation program is identifying patient populations at increased risk of CRC, who may benefit most from timely navigation. Automated machine learning (ML) algorithms based on routine electronic health record (EHR) data accurately estimate a patient’s relative risk of CRC. High-risk individuals may be particularly motivated to comply with disease screening recommendations and be targeted with an effective but resource-constrained navigator program. The overarching goals of this Administrative Supplement is to support the Abramson Cancer Center (ACC) mission to increase colorectal cancer (CRC) screening completion among high-risk individuals living in a persistent poverty county by designing, conducting, disseminating and evaluating an electronic health record- based automated identification program to target effective, culturally-sensitive CRC screening navigation to individuals who have not completed an ordered colonoscopy or fecal immunochemical test (FIT). Specifically, the goals of this supplement are to: 1) Adapt a previously validated EHR-based machine learning algorithm to predict CRC detection by retraining the model using data from patients seen in primary care clinics serving zip codes with a high proportion of racial and ethnic minorities living in Philadelphia County, a persistent poverty county; and 2) Implement and evaluate the feasibility and effectiveness of an algorithm-based CRC navigation program to increase colorectal cancer screening among 344 patients seen at one of 7 primary care practices within Philadelphia county who are at high risk of CRC, have uncompleted colonoscopies. Together, these projects aim to increase evidence-based screening in order to reduce the burden of CRC among high-risk individuals living in a persistent poverty county by utilizing evidence-based, targeted, culturally-sensitive CRC screening navigation that in part addresses social factors that prevent colorectal cancer screening.