The vicious cycle: Delirium, sedation, and patient outcomes in critical care
Shibu Sasidharan, Harpreet DhillonABSTRACT
Background:
Delirium is a prevalent neuropsychiatric complication in critically ill patients, associated with increased morbidity, mortality and long-term cognitive impairment. Sedation practices are increasingly recognised as modifiable risk factors for delirium; however, the optimal approach remains debatable.
Objective:
This narrative review examines the bidirectional relationship between sedation management and delirium in the intensive care unit (ICU), by synthesising evidence from randomised controlled trials (RCTs), meta-analyses and systematic reviews published between January 2000 and December 2025.
Methods:
A structured literature search was conducted in PubMed, Embase, the Cochrane Library and Web of Science using predefined search terms related to ICU delirium and sedation. English-language RCTs, systematic reviews, meta-analyses, cohort studies and practice guidelines addressing sedation practices and delirium in adult ICU patients were included.
Results:
Contemporary evidence supports a paradigm shift towards lighter sedation targets and non-benzodiazepine approaches. Dexmedetomidine has been associated with lower delirium rates compared to benzodiazepines in several RCTs, though the magnitude of benefit varies across populations and comparator agents. Multi-component prevention bundles, including the ABCDEF bundle, have demonstrated consistent delirium reduction in observational and quasi-experimental studies. However, important heterogeneity exists across trials, and several key studies have produced null or conflicting results for the primary endpoints.
Conclusions:
Sedation minimisation and avoidance of benzodiazepines represent reasonable evidence-based strategies; however, the strength of evidence varies by population and context. Significant knowledge gaps persist regarding optimal strategies for specific subgroups, and the implementation of evidence-based practices remains suboptimal.