Routine delirium monitoring is independently associated with a reduction of hospital mortality in critically ill surgical patients: A prospective, observational cohort study

Abstract

Purpose

Although delirium monitoring is recommended in international guidelines, there is lacking evidence for improved outcome due to it. We hypothesized that adherence to routine delirium monitoring would improve clinical outcome in adult critically ill patients.

Material and methods

We present the results of a prospective, noninterventional, observational cohort study that was conducted on 2 intensive care units (ICUs) of a tertiary care medical center between July and October 2007 (International Standard Registered Clinical Trial Record identifier: 76100795). We assessed delirium-monitoring and outcome parameters on a daily basis. Besides multivariate logistic and robust linear regression to analyze the relationship between delirium monitoring and outcome, we used the doubly robust augmented inverse probability weighting method for observational data to estimate effect sizes.

Results

Of 355 screened patients, we included 185 surgical ICU patients into our final analysis, of which 87 were mechanically ventilated. We found an independent association between delirium-monitoring adherence and in-hospital mortality for ventilated patients (odds ratio, 0.973; P= .041). Estimating the effect size, delirium monitoring indicated a reduction of 22% of in-hospital mortality if conducted 50% or more of ICU days per patient. The average ICU length of stay of 46 days was estimated to be reduced by 19 days (P= .031) if patients were sufficiently monitored.

Conclusion

Our data suggest an improved outcome for mechanically ventilated patients being screened for delirium in clinical routine.

Introduction

Delirium is defined as a disturbance in attention accompanied by a change in either cognition or consciousness that fluctuates over the course of the day and results from an underlying medical condition [1]. The incidence of delirium in critically ill patients has a wide variability depending on the mode of diagnosis, screening, and the patients under observation [2], [3], [4], [5]. Various studies have shown that delirious patients have an increased length of hospitalization [6], as well as a higher risk for long-term cognitive impairment [7] and functional impairments [8]. There are also studies stating a higher risk for mortality, which is a matter of current discussion [2]. The practice guideline of the Society of Critical Care Medicine [9] and national societies [10] recommend a frequent screening for delirium with particular assessment tools. These tools have been developed to allow a valid and reliable screening for delirium in clinical routine [3]. In comparison to a subjective, clinical evaluation, the use of validated scores improves the physician's and nurse's ability to detect delirium [11].

Delirium monitoring is part of the evidence-based organizational approach referred to as the “ABCDEF bundle” (Awakening and Breathing Coordination, Choice of sedatives, Delirium monitoring, Early mobility, Fast sleep) [12]. In this respect, feasibility and effectiveness have been shown for the implementation of parts of this bundle [13]: Considering the single features of the bundle, there is a body of literature favoring protocol-based sedation and showing negative effects of a continuous benzodiazepine-driven sedation [14], compared with a regime favoring nonbenzodiazepine sedation [15] and favoring less sedation [16]. In addition, early mobility has proven benefits for the patient [17]. In contrast, the distinct value of delirium monitoring in clinical routine has not been addressed so far and remains hypothetical. This might be one potential cause for the low implementation of delirium screening in clinical practice [18].

We set up a prospective cohort study to address this issue. We hypothesized that adherence to delirium monitoring, as an additional level of care, would result in a reduction of mortality in ventilated and nonventilated patients.