OBJECTIVES

To determine the relationship between the number of delirium days experienced by intensive care patients and mortality, ventilation time, and intensive care unit stay.

DESIGN

Prospective cohort analysis.

SETTING

Patients from 68 intensive care units in five countries.

PATIENTS

Three hundred fifty-four medical and surgical intensive care patients enrolled in the SEDCOM (Safety and Efficacy of Dexmedetomidine Compared with Midazolam) trial received a sedative study drug and completed at least one delirium assessment.

INTERVENTIONS

Sedative drug interruption and/or titration to maintain light sedation with daily arousal and delirium assessments up to 30 days of mechanical ventilation.

MEASUREMENTS AND MAIN RESULTS

The primary outcome was all-cause 30-day mortality. Multivariable analysis using Cox regression incorporating delirium duration as a time-dependent variable and adjusting for eight relevant baseline covariates was conducted to quantify the relationship between number of delirium days and the three main outcomes. Overall, delirium was diagnosed in 228 of 354 patients (64.4%). Mortality was significantly lower in patients without delirium compared to those with delirium (15 of 126 [11.9%] vs. 69 of 228 [30.3%]; p<.001). Similarly, the median time to extubation and intensive care unit discharge were significantly shorter among nondelirious patients (3.6 vs. 10.7 days [p<.001] and 4 vs. 16 days [p<.001], respectively). In multivariable analysis, the duration of delirium exhibited a nonlinear relationship with mortality (p=.02), with the strongest association observed in the early days of delirium. In comparison to 0 days of delirium, an independent dose-response increase in mortality was observed, which increased from 1 day of delirium (hazard ratio, 1.70; 95% confidence interval, 1.27-2.29; p<.001), 2 days of delirium (hazard ratio, 2.69; confidence interval, 1.58-4.57; p<.001), and ≥3 days of delirium (hazard ratio, 3.37; confidence interval, 1.92-7.23; p<.001). Similar independent relationships were observed between delirium duration and ventilation time and intensive care length of stay.

CONCLUSIONS

In ventilated and lightly sedated intensive care unit patients, the duration of delirium was the strongest independent predictor of death, ventilation time, and intensive care unit stay after adjusting for relevant covariates.