Characteristics of intensive care unit registries - findings from the Global Registry ICU Datasets (GRID) survey

TL;DR

This study maps the characteristics and research activities of ICU registries worldwide, highlighting areas of convergence and potential for future research.

Contribution

The study provides a comprehensive survey of ICU registries, identifying commonalities and opportunities for shared data standards and interventional research.

Findings

Most ICU registries use severity of illness scores like SOFA and record organ support measures such as mechanical ventilation.

Over half of ICU registries have structured quality improvement initiatives, and nearly half support observational research.

There is significant heterogeneity in coding systems and additional datasets across registries.

Abstract

Intensive care unit registries, which aim to improve the quality of intensive care unit care through benchmarking and quality improvement initiatives, are active worldwide, with considerable dishomogeneity. We aimed to map core datasets, additional variables, and research activities of these registries.

A cross-sectional survey was disseminated to registry leads between October 2023 and June 2024. The survey was structured into four main topics: registry characteristics and coverage, core dataset features, additional modules, and registry-enabled research.

Leads of 34/42 national registries responded (response rate 81%), covering 3,337 intensive care units, with a larger representation from South America. Systematized nomenclature of medicine, clinical terms, and customized categorical classifications were the main nomenclatures used. All registries except one employed a severity of illness score/risk prediction model. The SOFA score was reported by 88% of registries. Organ support measures were often recorded, including mechanical ventilation (97%), vasopressor administration (86%) and renal replacement therapy (86%). Three out of four intensive care unit registries coded interventions such as intubations, intravenous lines and tracheostomies. Additional datasets differed, with many use cases for nosocomial infection burden, bed availability and staffing resources. Over half of intensive care unit registries had current structured quality improvement initiatives. Registry-enabled observational research was reported in 46% of registries, while interventional studies were reported in only 22%.

Over three thousand intensive care units in 35 countries participate in an intensive care unit registry. Despite heterogeneity in coding systems, risk models, and additional datasets, we identify several areas of convergence that may inform a future shared core dataset. There is potential for further intensive care unit registry-based research, particularly interventional.

Figures