Characterisation of adverse events leading to unscheduled intensive care admissions

Unscheduled admissions to intensive care units (ICUs) pose major challenges for patient safety, hospital organisation, and healthcare costs. A substantial proportion of these admissions are linked to adverse events associated with care (AEAC) occurring outside the ICU, making them a critical target for quality improvement initiatives.

We conducted an exploratory retrospective single-centre study in an 800-bed regional hospital, including all patients aged ≥18 years who experienced unscheduled ICU admissions between January and December 2022. Eligible admissions occurred 24-144 hours after hospitalisation. A multidisciplinary expert panel reviewed patient records to identify AEACs using Wilson's criteria. Events were assessed for causality (6-point scale), severity (NCC MERP classification), and avoidability. AEACs with a causality score ≥4 were retained. Root cause analysis was performed using an Ishikawa diagram. Relationships between causality, severity, and avoidability were analysed using Spearman correlation and simple linear regression.

Among 858 ICU admissions recorded in 2022, 122 patients met the inclusion criteria for unscheduled ICU admissions. Among these, 89 cases (73%) were associated with at least one adverse event associated with care (AEAC), while the remaining 33 reflected the natural progression of the underlying condition. Overall, these AEAC-related unscheduled admissions represented approximately 10% of all ICU admissions during the study period. Organisational failures (36%), therapeutic (25%), diagnostic (19%), and medication errors (12%) were the most frequent AEAC categories. Most events were severe (categories G-I) and preventable. Admissions peaked between 6 p.m. and midnight, a period characterised by reduced clinical supervision. Spearman correlation demonstrated strong associations between causality and severity (r = 0.78, P < .001) and between causality and avoidability (r = 0.72, P < .001). Linear regression confirmed these relationships (adjusted R2 = 0.61 and 0.55, respectively).

AEACs substantially contribute to unscheduled ICU admissions and are frequently preventable. The strong correlations observed highlight the need for improved early detection systems and targeted preventive strategies, particularly during low-supervision periods. Organisational factors are prominent and actionable. These findings support the implementation of enhanced patient monitoring, structured clinical workflows, and preventive safety measures. Future research should include multicentre validation and assess the economic burden and cost-effectiveness of interventions to reduce preventable ICU transfers.