Delirium in critically ill COVID-19 patients: incidence, predictors, and outcomes in invasive mechanical ventilation with versus without ECMO support

Delirium is a frequent and serious complication in critically ill COVID-19 patients, especially those requiring prolonged mechanical ventilation (IMV) or extracorporeal membrane oxygenation (ECMO). Given its negative impact on morbidity and mortality in the ICU, understanding the prevalence, risk factors, and long-term effects of delirium in cardiothoracic ICU patients remains crucial. This study aimed to determine delirium prevalence, identify independent predictors, and assess its association with short- and long-term outcomes in a cohort with a high proportion of prolonged weaning and ECMO support. In this single-center cohort study, adult COVID-19 patients admitted between March 2020 and December 2021 requiring ≥ 72 h of invasive mechanical ventilation (IMV) and/or ECMO were included. Delirium was assessed retrospectively using daily CAM-ICU scores, Richmond Agitation-Sedation Scale (RASS) documentation, and structured chart review to capture hypoactive and hyperactive forms. Clinical data, sedation, analgesia exposure, and ICU management parameters were collected. Multivariable logistic regression identified predictors; a second model included ICU length of stay to account for potential bidirectional associations. Time-dependent Cox regression was used in mortality analyses to reduce survivorship bias. Long-term psychological outcomes were evaluated 6–12 months post-discharge. Among 102 patients (58 ECMO, 44 IMV only), 53.9% developed delirium. Prevalence was higher in the IMV-only group (63.6%) than in the IMV–ECMO group (46.6%; p = 0.065). Multivariable analysis showed that lack of ECMO support, male sex, and longer duration of invasive ventilation were associated with increased delirium risk. When ICU length of stay was included, only absence of ECMO and ICU length of stay remained significant, while male sex lost significance. Thirty-day survival was higher in delirious patients (p = 0.039), but this was not confirmed for overall survival or in time-dependent analyses. No statistically significant differences were found in long-term psychological outcomes, although no statistically significant differences were observed between patients with and without delirium. Delirium is common among prolonged IMV and ECMO COVID-19 patients in cardiothoracic ICUs. Differences in incidence between ECMO and IMV-only patients may reflect differences in care protocols rather than illness severity alone, but no causal conclusions can be drawn from our data. Therefore, we suggest modifying potentially preventative factors such as sedation practices, and emphasize the need for future prospective studies.