The study, conducted by a team of psychiatrists and infectious disease specialists affiliated with Seoul National University College of Medicine and its affiliated hospitals, followed 1,030 patients admitted with COVID-19 to two tertiary hospitals in Seoul and Gyeonggi Province between January 2021 and December 2022. The findings, published in BMC Psychiatry, offer one of the most detailed long-term pictures yet of what delirium means for survival in this population, with follow-up extending well beyond the acute hospitalization period.

Delirium is notoriously difficult to study. It is defined clinically as an acute disturbance in attention and awareness that develops over a short period, tends to fluctuate throughout the day, and cannot be explained by a pre-existing neurocognitive disorder. To bring rigor to the diagnosis, the Korean team excluded patients with dementia diagnosed before admission, eliminating one of the most common sources of diagnostic ambiguity. Delirium was assessed using the Confusion Assessment Method, a widely validated bedside screening tool, and diagnoses were confirmed according to the criteria of the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders. This dual approach—systematic screening combined with formal diagnostic criteria—strengthens confidence that the 119 patients classified as delirious (11.6 percent of the cohort) genuinely met the threshold.

The demographic and clinical profile of the cohort was typical of a seriously ill hospitalized COVID-19 population in the pandemic’s later waves. Patients had a mean age of 62.9 years, and 44.4 percent were women. Over the course of the study, 264 patients—25.8 percent of the entire cohort—died. The raw numbers told a striking story: mortality among patients who had experienced delirium reached 47.1 percent, compared with just 23.0 percent among those who had not, a difference so large that the probability of it arising by chance was less than one in a thousand. Unadjusted survival analysis reinforced the same pattern. Patients in the delirium group survived a median of 18.0 months, while those without delirium survived a median of 25.6 months, again a statistically significant gap.

But raw associations in observational data can mislead. Older, sicker patients are both more likely to become delirious and more likely to die, raising the classic question of whether delirium is a cause of death or merely a fellow traveler of frailty. To disentangle this, the researchers deployed logistic regression to identify independent risk factors for delirium and multivariable Cox proportional hazards regression to test whether delirium predicted mortality once other variables were held constant. These models adjusted for a comprehensive set of potential confounders, including age, comorbidity burden as measured by the Charlson Comorbidity Index, the severity of COVID-19 illness, and medication exposures during hospitalization.

The results of the risk-factor analysis were revealing on their own. Three characteristics independently predicted which patients would develop delirium. Each additional year of age raised the odds by roughly 3 percent, reflected in an odds ratio of 1.03 with a 95 percent confidence interval of 1.02 to 1.05. Greater COVID-19 severity carried an odds ratio of 3.41 (95 percent CI 2.56–4.53), meaning patients with more severe disease had more than triple the odds of becoming delirious. Most strikingly, patients who had been taking antipsychotic medications before admission faced odds 3.52 times higher (95 percent CI 1.44–8.63)—a finding that underscores how the pharmacological history of a patient’s brain shapes its resilience during acute systemic illness. Pre-admission antipsychotic use may serve as a proxy for pre-existing neuropsychiatric vulnerability, and it may also alter neurotransmitter systems, particularly dopaminergic and cholinergic pathways, that are central to the pathophysiology of delirium.

When the researchers turned to survival, the picture changed dramatically. After full multivariable adjustment, delirium was no longer a statistically significant independent predictor of death, with a hazard ratio of 1.33 and a confidence interval spanning unity (0.89–2.00, p = 0.16). In plain terms, once the analysis accounted for the fact that delirium tends to occur in older patients with more comorbidities and more severe COVID-19, the apparent survival penalty of delirium largely evaporated. What did independently predict mortality were the underlying forces that produce delirium in the first place: older age, a heavier burden of chronic disease, greater COVID-19 severity, and the use of opioids and antiepileptic medications during hospitalization. These drugs, often necessary for pain control and seizure management in critically ill patients, are themselves known to depress central nervous system function and are well-established pharmacological contributors to delirium.

The authors’ interpretation reframes the clinical conversation. Delirium, they argue, is best understood not as a direct determinant of mortality but as a barometer of systemic vulnerability—a manifestation of a brain under siege from infection, inflammation, hypoxia, metabolic derangement, and sedating medications. This distinction matters because it changes what clinicians should do with the information. If delirium were an independent killer, the priority would be treating the delirium itself as a life-threatening condition. If instead it is a sentinel, the priority becomes identifying, early in the hospitalization, the patients most likely to develop it—and intervening on the modifiable factors that drive both delirium and death.

That practical implication is central to the study’s conclusion. The identified risk factors—advanced age, severe disease, and pre-admission antipsychotic use—are largely recognizable at or near the time of admission, which creates a genuine window for proactive care. Strategies with evidence in general hospitalized populations include minimizing sedating medications where clinically feasible, ensuring adequate oxygenation, maintaining sleep-wake cycles, early mobilization, and involving family members in reorientation. In the COVID-19 context, where isolation measures, intensive care, and mechanical ventilation or high-flow oxygen therapy complicate every aspect of care, such non-pharmacological approaches require deliberate effort. The finding that opioids and antiepileptics were associated with mortality suggests that careful medication stewardship—regularly reviewing whether each sedating drug remains necessary—could be a particularly high-yield target.

The study’s strengths lie in its size, its multicenter design, its systematic delirium screening, and its extended follow-up, which distinguishes it from earlier work that captured only short-term in-hospital or 30-day outcomes. By demonstrating that the association between delirium and mortality persists in raw data over many months but dissolves under multivariable adjustment, the study provides a methodological caution for the field: many published claims that delirium “causes” death in acute illness may partly reflect inadequate confounder control. At the same time, the authors acknowledge limitations inherent in a retrospective design. Delirium detection depended on clinical documentation and screening practices, and hypoactive delirium—the quiet, withdrawn form that is easily missed—may have been underrecognized, potentially blurring the contrast between groups. The exclusion of patients with pre-existing dementia, while methodologically valuable, also means the findings may not generalize to the very population in which delirium is most common.

Funded by the Patient-Centered Clinical Research Coordinating Center under the Ministry of Health and Welfare of the Republic of Korea, the study adds an important piece to the evolving understanding of neuropsychiatric complications of COVID-19. Years after the acute phase of the pandemic, researchers are still mapping how the virus and the measures taken against it reshape the trajectory of vulnerable patients. This work suggests that the confused, disoriented patient in a hospital bed is not simply a psychiatric problem to be suppressed with medication, but a whole-body alarm—a signal that the patient’s physiology is failing on multiple fronts and that aggressive, thoughtful, multidisciplinary intervention may still change the long-term outcome. For clinicians managing hospitalized patients with severe infections, the message is clear: recognize delirium early, treat it as a marker of danger, and direct that vigilance toward the underlying systemic illness it reflects.