# Precision blood pressure control, cognition and adverse events in older adults: detecting heterogeneity in treatment effects in randomized trialsusing machine-learning approaches > **NIH NIH R03** · HARVARD UNIVERSITY D/B/A HARVARD SCHOOL OF PUBLIC HEALTH · 2024 · $302,960 ## Abstract Project Summary Hypertension is a major modifiable risk factor for Alzheimer's disease and related dementias (AD/ADRD). However, there is inconsistent evidence regarding the optimal blood pressure (BP) treatment targets in older adults. While the SPRINT MIND trial demonstrated that intensive BP control reduced mild cognitive impairment compared to standard BP control, other trials focusing on intensive BP control have produced conflicting results. Furthermore, several studies suggest that intensive BP control may also elevate the risk of serious adverse events. Given the substantial variation in health status and functional level among older adults, it is likely that certain individuals may benefit from intensive BP treatment, while others may experience adverse events related to overtreatment. This suggests that adopting a one-size-fits-all approach to BP control in older adults may be insufficient. Thus, a personalized approach to BP control in older adults, tailored to their unique characteristics, is needed. However, the current evaluation of heterogeneous treatment effects on cognitive outcomes and serious adverse events is largely limited by insufficient statistical power in individual trials. Advanced causal heterogeneity analyses have yet to be fully employed to improve the detection of heterogeneous treatment effects. To bridge these gaps and inform personalized BP control strategies, we propose to use advanced statistical methods and pooled individual data from multiple large-scale randomized trials to detect heterogeneity in treatment effects of intensive BP control on cognitive outcomes and serious adverse events. By employing conventional subgroup analyses, risk score analyses, and effect score analyses in pooled trial data, this proposal aims to identify individuals who would benefit the most from intensive BP treatment for preventing cognitive decline and dementia (Aim 1), as well as those who are most likely to have serious adverse events (Aim 2). Specifically, risk score and effect score analyses offer a more comprehensive and clinically relevant way to capture the heterogeneity of treatment effects compared to traditional subgroup analyses. Risk score analyses integrate multiple characteristics into a single risk score, enabling the identification of clinically meaningful subgroups that may be overlooked in conventional subgroup analyses. Effect score analyses based on machine-learning methods directly estimate the extent to which the benefits and potential harm of intensive BP control differ across multidimensional individual characteristics and may further outperform risk score analyses. In summary, by identifying individuals who would benefit most from intensive BP control for dementia prevention and those at risk of harm from serious adverse events utilizing advanced statistical methods in pooled trial data, the findings will inform personalized BP targets for older adults. This proposal may also serve as a model for develop... ## Key facts - **NIH application ID:** 11057202 - **Project number:** 1R03AG087481-01A1 - **Recipient organization:** HARVARD UNIVERSITY D/B/A HARVARD SCHOOL OF PUBLIC HEALTH - **Principal Investigator:** Yuan Ma - **Activity code:** R03 (R01, R21, SBIR, etc.) - **Funding institute:** NIH - **Fiscal year:** 2024 - **Award amount:** $302,960 - **Award type:** 1 - **Project period:** 2024-09-30 → 2026-08-31 ## Primary source NIH RePORTER: https://reporter.nih.gov/project-details/11057202 ## Citation > US National Institutes of Health, RePORTER application 11057202, Precision blood pressure control, cognition and adverse events in older adults: detecting heterogeneity in treatment effects in randomized trialsusing machine-learning approaches (1R03AG087481-01A1). Retrieved via MORVS 2026-10-26 from https://api.morvs.ai/grant/nih/11057202. Licensed CC0. --- *[NIH grants dataset](/datasets/nih-grants) · CC0 1.0*