How to establish and run a national ICU benchmarking registry

National ICU benchmarking systems are critical infrastructure for evaluating and improving the quality of critical care across all settings. This review summarizes the recent literature on establishing, operating, and sustaining such systems, drawing on evidence from registries across high-income countries (HICs) and low-and-middle-income countries (LMICs).

Recent evidence not only confirms the continued expansion of national ICU registries globally but also reveals persistent heterogeneity in data architecture and severity adjustment approaches. Key advances include the development of common data models enabling international collaboration (LOGIC consortium), the use of federated analysis for score development (SOFA-2), and the validation of simplified severity scores for resource-limited settings. Calibration drift remains universal, requiring periodic reassessment of risk-adjustment models. New methodological work highlights the impact of atypical patients and potential organ donor admissions on standardized mortality ratios. While 46% of registries support observational research, only 22% currently enable interventional studies - representing a major opportunity for the field.

Establishing a national benchmarking system requires deliberate choices about data architecture, severity adjustment, governance, and feedback mechanisms. Challenges vary across the registry lifecycle, from securing initial buy-in and ensuring data quality at scale to maintaining engagement and model accuracy at maturity. The ultimate measure of a benchmarking system is not only the volume and coverage of the data but also the improvements in care it enables.