BALANCING PRIVACY AND USABILITY IN REAL-WORLD DATA FOR OLDER ADULTS
Author(s)
Amy Price, PhD1, Dena Jaffe, PhD2;
1Oracle Health, Kansas City, MO, USA, 2Oracle Health, Alon Shvut, Israel
1Oracle Health, Kansas City, MO, USA, 2Oracle Health, Alon Shvut, Israel
OBJECTIVES: Older adults represent a growing and important segment of the U.S. population, accounting for a disproportionate share of healthcare utilization and expenditures due to their unique and often complex needs. Generating real-world evidence (RWE) to better support healthy aging relies on real-world data (RWD) that reflect the diversity and clinical profiles of older individuals. However, curating fit-for-purpose RWD for this population is challenged by privacy-driven data modifications, which can limit fidelity. This study examines the impact of age top coding on the relevance of RWD for evidence generation in older adults.
METHODS: The Oracle Health RWD electronic health record (OHRWD EHR) dataset (7/2024-7/2025) and U.S. Census Bureau data (2024) were used to describe adults aged ≥65 years. Data in the OHRWD EHR are top coded (aggregated) at 89 years.
RESULTS: In the OHRWD and U.S. Census populations, 25.2% and 16.9% of adults were aged ≥65 years, respectively. Demographic characteristics varied across older age sub-groups. The percentage of female adults aged ≥89 years was 61.7% (OHRWD) and 67.9% (U.S. Census). Using U.S. Census non-aggregated age data, the percentage of females increased from 63.4% to 81.1% from ages 89 to 100. For race, the percentage of Black and Asian adults aged ≥89 years was 4.6% and 2.5% (OHRWD) and 8.3% and 6.1% (U.S. Census), respectively. Using U.S. non-aggregated age data, these percentages remained similar among elderly Black and increased among elderly Asian adults from 5.6% to 10.3% from ages 89 to 100.
CONCLUSIONS: The heterogeneity of demographic characteristics in older age groups indicates that aggregation into a single, broad age category may bias risk estimates and mask within-category heterogeneity. Study designs that avoid this potential bias by using less restrictive age inclusion criteria can limit generalizability. Transparency about privacy-protecting techniques and awareness of data limitations are essential for producing relevant and reliable RWE.
METHODS: The Oracle Health RWD electronic health record (OHRWD EHR) dataset (7/2024-7/2025) and U.S. Census Bureau data (2024) were used to describe adults aged ≥65 years. Data in the OHRWD EHR are top coded (aggregated) at 89 years.
RESULTS: In the OHRWD and U.S. Census populations, 25.2% and 16.9% of adults were aged ≥65 years, respectively. Demographic characteristics varied across older age sub-groups. The percentage of female adults aged ≥89 years was 61.7% (OHRWD) and 67.9% (U.S. Census). Using U.S. Census non-aggregated age data, the percentage of females increased from 63.4% to 81.1% from ages 89 to 100. For race, the percentage of Black and Asian adults aged ≥89 years was 4.6% and 2.5% (OHRWD) and 8.3% and 6.1% (U.S. Census), respectively. Using U.S. non-aggregated age data, these percentages remained similar among elderly Black and increased among elderly Asian adults from 5.6% to 10.3% from ages 89 to 100.
CONCLUSIONS: The heterogeneity of demographic characteristics in older age groups indicates that aggregation into a single, broad age category may bias risk estimates and mask within-category heterogeneity. Study designs that avoid this potential bias by using less restrictive age inclusion criteria can limit generalizability. Transparency about privacy-protecting techniques and awareness of data limitations are essential for producing relevant and reliable RWE.
Conference/Value in Health Info
2026-05, ISPOR 2026, Philadelphia, PA, USA
Value in Health, Volume 29, Issue S6
Code
RWD168
Topic
Real World Data & Information Systems
Topic Subcategory
Data Protection, Integrity, & Quality Assurance
Disease
No Additional Disease & Conditions/Specialized Treatment Areas