Geospatial approaches for surveillance of disparities in clinical trial enrollment (058)

Abstract

<b>Objectives:</b> We aimed to describe a geospatial approach that provides surveillance for sociodemographic, economic, and environmental diversity in clinical trial enrollment. <b>Methods:</b> This is a descriptive proof of concept study analyzing retrospective records of endometrial carcinoma patients eligible for a clinical trial at our institution. Each patient's home address was geocoded in a Geographic Information System (GIS). Then, patients were categorized into the following groups: enrolled, declined, ineligible after the additional screening, and inactive. The latter group included patients who were offered the trial but could not be assigned into one of the previous categories due to various reasons, such as missed follow-up appointments. The Social Vulnerability Index (SVI) was utilized to capture multiple dimensions of social determinants of health and diversity and overlaid to patient-level data in the GIS. Geospatial analysis was performed to detect patterns associated with low recruitment using Nearest Neighbor Index (NNI) to quantify the presence of clustering and Kernel Density Estimation to visualize locations. Mean SVI for each group of patients was calculated. <b>Results:</b> The study included 385 patients (enrolled=211, declined=24, ineligible=13, inactive=137). NNI applied to the enrolled. Inactive cohorts displayed clustering (p=0.00), and across most of the study area, these clusters were in the same places, indicating that there was no geographic disparity at a regional level. However, one notable difference existed where inactive patients were concentrated in a group of neighborhoods with a majority Black/African American population. Due to small numbers, declined or ineligible groups were mapped for visual assessment of patterns and determined to be dispersed across the region. The mean SVI was similar across patient groups, except for those deemed ineligible having a higher SVI (enrolled=0.39, declined=0.41, ineligible=0.60, inactive=0.44). SVI ranges from 0-1, with higher values indicating higher disadvantage. These findings require further investigation but also point to the potential benefit of using geospatial surveillance prospectively to detect emergent patterns and direct recruitment. <b>Conclusions:</b> Disparities in clinical trial recruitment raise concerns for scientific generalizability and treatment equity. Recent studies have questioned the role of the neighborhood and social vulnerability on biological processes that impact health outcomes. If a race is more sociopolitical than biological, clinical trial recruitment should reflect diversity in the same geopolitical factors, such as racism, poverty, and neighborhood deprivation. While we did not identify consistent neighborhood-level geographic bias in this study, refinement with individualized risk factors is ongoing. The geospatial analysis offers a method to monitor in real time and direct recruitment efforts to ensure equitable and diverse populations are included.Fig. 1