A scoping review of ethical data sharing in critical care: A global framework for distributed Trusted Research Environments

Abstract

Clinical data, including diagnoses, laboratory results, outcomes, and social determinants of health, are crucial for advancing healthcare delivery, particularly in intensive care units (ICUs), where real-time decisions have a direct impact on survival. Artificial intelligence (AI) holds promise for supporting clinical reasoning and improving outcomes in critical care, but its safe and equitable implementation depends on access to diverse, high-quality data. However, ICU data often remain fragmented in institutional silos, limiting research opportunities, especially for investigators in low- and middle-income countries (LMICs). Trusted Research Environments (TREs) have emerged as a secure and privacy-preserving infrastructure for sharing sensitive health data, ensuring compliance with ethical standards. This scoping review analyzed 37 TREs across high- and middle-income countries to evaluate their features, accessibility, and inclusion of critical care datasets. We found that TREs remain largely inaccessible to researchers from LMICs due to cost, infrastructure limitations, and technical barriers. Moreover, only a subset of TREs—primarily in the UK, the USA, and Brazil—provide ICU data, despite their central importance for high-impact critical care research. To address these disparities, we propose the framework of “democratized TREs”, which prioritize equitable access, data security, and researcher inclusion. Recommendations include reducing financial barriers through credit-based or subsidized models, enhancing usability and interoperability, and fostering global collaboration to expand shared infrastructure. Special attention is given to supporting LMIC researchers and building the capacity for AI-driven research in critical care. Democratizing TREs can accelerate innovation, promote fair representation in AI development, and ultimately reduce health disparities in high-stakes environments such as ICUs, where timely, data-driven insights are most needed.

Ähnliche Arbeiten