Long-Term Outcomes after Delirium in the ICU: Addressing Gaps in Our Knowledge

Delirium occurs in up to 50% of all critically ill adults (1). This common ICU phenomenon is associated with a substantial burden to patients and families and has serious ICU and post-ICU sequelae (1). Mortality is an important concern among ICU survivors and their families (2). The relationship between ICU delirium and post-ICU mortality is unclear (3). Cohort studies evaluating the association between delirium and mortality over 1 to 12 months of follow up have discordant results (1, 3, 4). Among these studies, there is important variability in ICU patient populations, methods of delirium detection and evaluation (e.g. incidence vs. prevalence, duration, severity), and how potential confounding has been considered.

In this issue of the Journal, Fiest and colleagues (pp. 412–420) make an important contribution via their population-based study evaluating the association of ICU delirium and mortality over up to 2.5 years of follow up in 12,137 adults consecutively admitted >24 hours to any of the 14 medical-surgical ICUs in the province of Alberta, Canada (population: 4.4 million) (5). This study also explored the association between ICU delirium and subsequent hospital readmissions and emergency department visits, including mortality as a competing risk. Using five province-wide databases, the authors evaluated comprehensive data, including patient demographics, ICU clinical variables, mortality, hospitalizations, and emergency department visits. Using propensity scoring, the “ICU delirium” and “no ICU delirium” patient cohorts were matched on five baseline variables and four ICU variables. The statistical methods considered time dependence of the outcome measures with delirium, patient clustering within ICUs, and different methods of evaluating delirium (e.g., duration and severity).

Among the 5,936 propensity-matched critically ill adults who survived to hospital discharge, the incidence of delirium in the ICU was associated with greater mortality (hazard ratio [HR], 1.44; 95% confidence interval [CI], 1.08–1.92) up to 30 days after hospital discharge (5). Beyond 90 days after hospital discharge, a significant association was not found (HR, 1.09; 95% CI, 0.91–1.16). During the 2.5-year study period, delirium occurrence was associated with an increased risk for emergency department visits, hospital readmissions, or death after the index hospitalization (HR, 1.12; 95% CI, 1.07–1.17).

These results are an important building block in better understanding the long-term outcomes of ICU delirium and in reflecting on clinical care in the ICU (5). The survivorship experience is a critical concern for ICU patients and families (6). The incidence and duration of ICU delirium may be a potentially modifiable risk factor for post–intensive care syndrome (6). Notably, a longer duration of delirium in the ICU is independently associated with worse global cognition at the 3- and 12-month follow up (7). Although not evaluated in the paper by Fiest and colleagues (5), multicomponent ICU quality-improvement interventions (e.g., the ABCDEF bundle), supported by the Pain, Agitation/Sedation, Delirium, Immobility, and Sleep clinical practice guidelines (3), are associated with reductions in ICU delirium, hospital mortality, and ICU readmissions (8). However, the impact of such interventions on long-term mortality and patient outcomes requires more evaluation.

To further build on the analysis by Fiest and colleagues (5), future studies should evaluate interrelationships between ICU sedation status (including coma), sedative choice, and delirium occurrence and their effect on post-ICU mortality and patient outcomes. Herein, we provide some recommendations for future research in this area. First, explicit consideration of a sedative-induced coma is important given its association with mortality and given that coma is a competing risk in evaluating delirium in the ICU (3, 4, 9). Second, an evaluation of specific classes of medications in the ICU (e.g., benzodiazepines, propofol, dexmedetomidine, and opioids) is important to better understand associations of delirium with post-ICU mortality and patient outcomes (3, 10, 11). Third, given that critically ill adults are frequently discharged on psychoactive medications, further exploration of associations of post-ICU medications and patient mortality and outcomes is recommended (12). Fourth, given that preexisting frailty and cognitive function are important predictors of ICU delirium and associated with increased mortality and deleterious post-ICU patient outcomes (13, 14), these baseline variables are important to evaluate in future research. Finally, given that post-ICU exposures (e.g., rehabilitation services and hospital readmissions) and variability in patient recovery trajectories impact survivors’ post-ICU outcomes, their consideration is warranted. Figure 1 proposes key baseline, ICU, and post-ICU risk factors and research considerations for delirium and long-term outcome studies.

Figure 1.

Risk factors and research considerations for delirium and long-term outcomes studies. Various risk factors and research considerations may influence reported associations between in-hospital incidence, duration, and/or severity and postdischarge outcomes. Readmission to the ICU or hospital may further negatively impact delirium and long-term patient outcomes. *Two separate Core Outcome Sets exist for research on delirium in the ICU and on long-term outcomes after acute respiratory failure (15, 16). Please also see https://www.improvelto.com/. TIA = transient ischemic attack.

In conclusion, via this new population-based retrospective study (5), important progress has been made in better understanding the association of delirium with post-ICU mortality and healthcare resource use. To continue advancing the field, future prospective studies should embrace a recent Core Outcome Set for ICU delirium research that recommends inclusion of seven outcomes: delirium occurrence (prevalence or incidence), delirium severity, time to delirium resolution, health-related quality of life, emotional distress, cognition, and mortality (15). Future prospective studies also should consider addressing key knowledge gaps via evaluating established delirium risk factors, post-ICU mortality, and patient-important outcomes while taking into account the complexities of competing risks in assessing delirium and long-term outcomes.