Why use the 4at?
The 4AT was designed specifically for routine clinical use. Its features mean that it can readily be implemented in day to day practice:
short
simple
no special training needed
built-in cognitive testing
gives a clinically usable 4AT score, not an Unable To Assess (UTA), in patients too sleepy or restless for cognitive testing or interview
excellent sensitivity and specificity
effective in practice: detects delirium at the expected level
no registration or permissions needed
extensively validated in multiple settings
Since introduction in 2011 the 4AT has grown to become the main delirium detection tool used in practice in many countries.
4AT VALIDATION STUDIES
The 4AT has the most diagnostic test accuracy data of all delirium tools in the literature: >27 validation studies involving >5000 patients.
The first general validation study was published in 2014. A 2020 meta-analysis of 17 studies found a pooled sensitivity of 88% and a pooled specificity of 88%. Further validation studies have since been published.
For context, two other well-validated tools in the literature are the Nursing Delirium Screening Scale (Nu-DESC), with 13 diagnostic test accuracy studies involving 3039 patients, and the Confusion Assessment Method (CAM), with 23 diagnostic test accuracy studies involving 2629 patients.
A large (N=785), high quality STARD-compliant diagnostic RCT comparing the 4AT and the CAM found that the 4AT had higher sensitivity than the CAM, and a similar specificity.
A recent systematic review covering various delirium assessment tools showed that the 4AT performs well when used at scale in routine clinical practice.
Validation studies span general medical settings, geriatrics wards, the emergency department, and surgical populations. A further diagnostic test accuracy study in a palliative care population will shortly be published.
global USE OF THE 4AT
The 4AT is recommended in multiple clinical guidelines and pathways internationally (UK NICE & SIGN Guidelines, USA, Canada, Ireland, Australia, New Zealand, Germany, Malaysia, etc.). It has been introduced into several EMRs/EHRs including EPIC, TrakCare, etc.
Clinical implementation
The 4AT is proven as a tool that can succesfully be implemented into clinical work, with (a) good rates of completion, and (b) delirium detection at the predicted level. See here and here for specific examples.
origin of the 4at
In 2010 the authors had tried to introduce delirium screening using the tool recommended at the time. We encountered several challenges.
We found that it was problematic to provide the necessary training on the original tool because of the staff time involved.
We also found surprisingly low rates of delirium detection compared with the sensitivity figures reported for the original tool in research studies. This was because the original tool (a) was not being completed, (b) was not scoring positively when delirium was present, or (c) commonly had ‘Unable to Assess’ (UTA) ratings in patients with delirium unable to answer cognitive testing or interview questions.
So we decided to design a new tool for use in routine clinical practice. The aim was to make a tool that was both validated and practical for day to day use.
Early versions were tested many times in real world clinical practice along with extensive user feedback. Ongoing improvements were made, then the 4AT was published on this website in 2011.
design of the 4AT
When first designing the 4AT, the questions we asked ourselves were:
(1) What does a practitioner do at the bedside when quickly assessing a person for delirium?
(2) Can we turn this process into a simple, easy tool?
ITEM 1: Level of alertness
We thought that the first clinical feature a practitioner would look for on approaching the bedside would be altered level of alertness: sleepiness/unresponsiveness, or agitation.
Rather than being an obstacle, altered level of alertness is a very valuable sign clinically, because it is highly specific to delirium.
So the first item of the 4AT measures level of alertness, and if positive gives 4 points, indicating ‘possible delirium’.
Note that it is common for patients with delirium to be so drowsy or hyperactive as to make cognitive testing or even a basic interview difficult or impossible. The 4AT is designed that such patients still receive a score. No patients with delirium are ‘Unable to Assess’ (UTA) with the 4AT.
ITEMS 2 & 3: Orientation & attention
We also wanted to build in some brief cognitive tests to assist in the assessment of mental state as well as provide some basic testing of cognition. We decided that the tests should be built in to the 4AT rather than needing a separate process.
We chose the Abbreviated Mental Test - 4 (AMT4), which consists of 4 simple orientation questions, and the Months Backwards test as a test of attention.
Patients who are unable to give a response to both of these questions because of severely altered level of alertness are given points on the 4AT as being ‘untestable’. This gives 2 points each, that is, a total of 4 points. This allows the test to continue rather than the patient being considered ‘UTA’.
item 4: acute change or fluctuating course
Item 4 assesses acute change or fluctuating course. This is an essential feature of delirium which is present in all diagnostic criteria. It is a highly specific indicator of delirium. If there is evidence of this feature this gives 4 points on the 4AT, indicating ‘possible delirium’.
Note that Item 4 incorporates the ‘Single Question in Delirium’ (SQiD).
Authors
The 4AT was originally designed in 2011 by Prof Alasdair MacLullich (Edinburgh Delirium Research Group, University of Edinburgh, Scotland), and Dr Tracy Ryan and Dr Helen Cash (NHS Lothian, Scotland). Updates since then have been carried out by and are the responsibility of Alasdair MacLullich.
We thank the following colleagues for generously contributing their time and expertise during the initial development and testing of the 4AT: Dr Atul Anand, Prof Daniel Davis, Dr Fergus Doubal, Dr Duncan Forsyth, Dr Kyle Gibson, Dr Aditya Hosakere, Prof Bree Johnston, Mr Colin MacDonald, Dr Katie Marwick, Dr Johanna Paddison, Prof Emma Reynish, Dr Jeyakumar Selwyn, and Dr Susan Shenkin.
The authors welcome any comments, suggestions or criticisms. Contact: Prof Alasdair MacLullich (a.maclullich @ ed.ac.uk; @A_MacLullich).