Detection of delirium

Delirium is best detected by the use of tools. There are two main types of tool used in clinical practice: episodic tools and monitoring tools.

Episodic tools are used at the front door (e.g. ED, medical admissions), around the time of surgery (e.g. immediately post-operatively, and at other times when delirium is suspected.

Episodic tools provide a good balance of sensitivity and specificity. They always require some bedside assessment including cognitive testing, and include an item that notes if there is a change from the patient’s baseline. This kind of testing is needed to inform DSM-5 criteria. Episodic tools give information that is close to providing a formal diagnosis, though clinical judgement is always required.

The 4AT is an example of an episodic tool. The Confusion Assessment Method (CAM) is an episodic tool when preceded by bedside cognitive testing and interview (it is sometimes used without cognitive testing and interview as a monitoring tool but is not validated for this purpose). The bCAM, UB-CAM and 3D-CAM, which like the 4AT have built-in cognitive assessments, are other episodic tools.

Episodic tools involve cognitive testing and interview and are too long and burdensome (for both patients and staff) to be performed effectively multiple times per day for several days. Daily use of an episodic tool for a small number of days in high-risk situations or when assessing recovery from delirium is reasonable. However, completion rates and/or detection performance are low when episodic tools are used repetitively for longer periods. Effective longer-term ongoing inpatient monitoring requires use of shorter monitoring tools.

Monitoring tools are used in inpatients to detect new onset delirium arising after hospital admission. They are done daily or more in inpatients. Because they are done frequently, monitoring tools are generally observational only, or involve a few seconds of interaction only.

Validated examples includes the National Early Warning Score – 2 (NEWS2) used in UK hospitals, the Single Question in Delirium (SQiD: ‘Is this patient more drowsy or confused than usual?’), the Modified Richmond Agitation Sedation Scale (mRASS), DOSS, NuDESC and the Recognizing Delirium as Part of your Routine (RADAR) tool.

If a monitoring tool gives a score suggestive of possible delirium, this should trigger assessment using an episodic tool such as the 4AT. This will then provide a clearer picture of the presence or absence of delirium.

The CAM-ICU and the ICDSC are widely used in practice, though almost exclusively in ICU settings. They are not validated for use in general settings. There are many other delirium assessment tools, but apart from the above list most are used for research. Single-item bedside tests such as Months Backwards or asking the patient the current year have been evaluated and show variable sensitivity to delirium. They have some value as initial assessment tests but lack the specificity required to function as standalone tests. They are not suitable for repetitive use for several days on end because of practice effects. Like with monitoring tests, single-question tests if positive should trigger a more definitive assessment with an episodic tool.

Profiles and evidence summaries for these and around 130 other tools are in the delirium assessment tools directory at deliriumtools.com.