Prognostic significance of delirium subtypes in critically ill medical and surgical patients: a secondary analysis of a prospective multicenter study.

Smit L, Wiegers EJA, Trogric Z et al

Journal of Intensive Care (2022) 10:54 

Aim of Study
To determine how delirium subtypes are associated with hospital mortality and other clinical outcomes including length of stay, coma duration, and use of sedative medications.

Design (and Location)
Secondary analysis on a prospective multicentre implementation study (iDECePTIvE – ICU Delirium in Clinical Practice Implementation Evaluation).
Rotterdam, The Netherlands (2012 – 2015).

iDECePTIvE was a prospective multicentre implementation study broken down into 3 parts (and 4 phases):

  • Part 1: Baseline measurements of data (phase I).
  • Part 2: Implementation of delirium assessment methods (CAM-ICU or ISDSC)
  • Part 3: 3 further data collection periods (phases II-IV) to assess the effects of the implementation methods.

This study analysed the data collected from phases II-IV.

  • All patients >18 yrs admitted to ICU.


  • Patients admitted with a primary neurological diagnosis.
  • Persistent coma during ICU stay (defined as RASS –4 or –5).
  • ICU readmissions.

Primary outcome
Hospital mortality

Secondary outcomes
ICU mortality
ICU length of stay (days)
Presence of coma during ICU stay, number of coma days
Use of mechanical ventilation during ICU stay, number of days ventilated.

If delirium was deemed to be present then it was subtyped into hyperactive, hypoactive and mixed type. The subtypes of delirium were also compared the above outcomes along with the following secondary outcomes:

  • Use of IV sedation, and number of days.
  • Use of benzodiazepines, and number of days.
  • Use of antipsychotics, and number of days.


Continuous data (eg length of stay, dose of medications etc…) were summarised as means with standard deviations.

Categorical data (delirium or no delirium) were shows as frequencies and percentages.

Analytic methods included χ2 test, t-test and Mann Whitney U tests.


There were 5057 ICU admissions during the data collection period. After the inclusion and exclusion criteria were applied 1564 patients were included in the study, of which 381 (24.4%) suffered with delirium during their ICU stay.

Patients admitted to intensive care, who suffered with delirium were more likely to have a higher hospital mortality than non-delirious patients (21.4% vs 11.9%; p 0.001).

Delirious patients were also more likely to have higher ICU mortality (13.4% vs 8%; p 0.001) and longer ICU stays (7 days vs 3 days; p 0.001).

In terms of the different types of delirium, hypoactive delirium was the most common (52.5%), and mixed-type delirium had the highest mortality.

Delirium incidence was associated with use of IV sedation (73.8% vs 35.1%; p 0.001), benzodiazepines (29.4% vs 10%; p 0.001), and opioid use (74.3% vs 41.7%; p 0.001).


Delirium was associated with higher hospital and ICU mortality as well as longer ICU and hospital stays. The incidence of delirium was also associated with increased and longer duration of use of IV sedative mediations, benzodiazepines and opioid medications.

Delirium measurement, as well as stratifying delirium by its subtypes is a useful prognostic tool in ICU populations.

Stated limitations from the study

It was a secondary analysis of previously collected data, so not all pieces of information were available to the researchers.

The use of benzodiazepines and sedative medications was not clear whether this was a causative factor for delirium or as part of the treatment for delirium.

Unknown mortality rates after discharge from hospital. Would be useful to know long term mortality and morbidity.

Discussions from journal club (?change of practice)

Delirium measurement and subtyping would be useful to use as a prognostic tool and can be measured alongside RASS.

Whilst reducing the use of IV sedative medications, benzodiazepines and opioids is obviously beneficial to patients, it is difficult to manage restless and delirious patients in hospitals without these medications. More research into the link is probably required before a change of practice.

Summary by Dr N McWilliams. Journal Club Meeting 16 February 2023.

Comments are closed.

Blog at

Up ↑

%d bloggers like this: