Modified 4AT and culturally adapted versions

The 4AT is now used across many languages, cultures and health systems. As it has spread, clinicians and researchers have adapted the attention task, the mode of delivery and the surrounding clinical workflow. This page describes those versions, why they have arisen, what the research shows and how we can learn from them.

What “modified 4AT” means

The term “modified 4AT” has been used for several kinds of change:

• Attention-task substitution: replacing Months Backwards with Days Backwards, counting backwards or another brief attention task.
• Language and cultural adaptation: changing an item because a literal translation would make it unusually easy, unfamiliar or linguistically awkward.
• Delivery-mode adaptation: giving the assessment remotely, for example through an interphone in an isolation room.
• Workflow adaptation: embedding the 4AT in an electronic record, repeating it at set intervals or redesigning the paper form.
• Scoring adaptation: using a different item rule or total-score threshold.

These changes answer different clinical questions. A shorter attention task may make repeated screening more acceptable. A cultural substitution may restore the intended cognitive demand of the item. An electronic build may make the 4AT part of routine care without changing the test itself. Calling each of these simply “modified 4AT” hides useful information, so this page names the exact change wherever the source allows it.

Why adapt the attention task?

Months Backwards works well in English and in many other languages. It requires attention, working memory and an overlearned sequence, while remaining quick to give. But calendar language is not culturally neutral. In Korean, for example, the months are commonly expressed as month 1, month 2 and so on. Reciting them backwards can become close to counting from 12 to 1. Similar reasoning led teams in Malaysia and Singapore to use counting from 20 to 1 instead.

Workflows create another reason to adapt. Houston Methodist uses Days of the Week Backwards in a nursing programme in which older patients may be assessed every 12 hours. Days Backwards is shorter and easier than Months Backwards, which can reduce burden when staff repeat the assessment for a limited period. It may also be more workable when illness, fatigue or the clinical environment makes a longer task difficult.

The task still needs to do useful cognitive work. An adaptation can improve linguistic fairness or clinical fit while changing difficulty. Record the task and study what happens.

Attention tasks now in use

Days of the Week Backwards versus Months of the Year Backwards

Days Backwards is easier. This matters, but the effect is likely to differ across the two jobs performed by the 4AT.

For detecting possible delirium, the effect may be fairly small. The 4AT does not depend on one attention task alone. Abnormal alertness and evidence of acute change or fluctuation each carry substantial weight, and inability to undertake the cognitive items is also strongly weighted. This makes the architecture of the 4AT relatively tolerant of a shorter attention task.

The trade-off is clearer in the cognitive testing range. Months Backwards gives more scope to distinguish full, partial and very limited performance. Because Days Backwards is easier, more patients are likely to complete it correctly. A local version may therefore lose some of the information that an English version 1.2 score of 1 to 3 provides about possible cognitive impairment.

That can still be a reasonable choice. A ward that uses the 4AT every 12 hours for a short period may value speed, completion and acceptability. A service using a single 4AT as a richer cognitive screen may prefer Months Backwards. The intended use should guide the choice.

Language and cultural adaptation

A literal translation does not always preserve the difficulty or meaning of a cognitive task. Numbered month names are the clearest example. The instruction may be translated correctly while the task itself becomes easier than intended. Cultural adaptation aims to preserve the function of the item, not only its words.

We audited every current instrument in the 4AT translation collection on 28 July 2026: 26 PDF files and 26 matching editable Word files across 23 language pages. Three hosted versions use an attention task other than Months Backwards:

• Chinese, Putonghua: repeated subtraction from 30 by 3, ten times.
• Chinese, Cantonese: the same repeated-subtraction task.
• Japanese: saying the word sekaichizu (“world map”) in reverse.

The other 21 language pages retain Months Backwards, including the Korean translation currently hosted here. This is separate from the published modified Korean K-4AT, which uses Days Backwards. None of the current downloadable translations uses Days Backwards or counting from 20 to 1.

What the evidence shows

A 2021 systematic review identified two early US studies in which Days Backwards replaced Months Backwards while the delirium-positive threshold remained the same.¹ Later work has described the adaptation directly and at much greater scale.

The Agarwal and Taffet programme

The longest-running large clinical programme comes from Houston Methodist and colleagues including Kathryn Agarwal and George Taffet. The team changed Months Backwards to Days Backwards in collaboration with the 4AT developers and embedded the result in nursing electronic flowsheets. Nurses assessed patients aged 70 years or older every 12 hours. In the seven-centre stroke study, completion reached 94% to 97%. A positive screen was associated with higher mortality, longer hospital stay and discharge to a skilled nursing facility.²

The later eight-hospital study included 260,200 older adults with determinate screening results. It linked the routine modified 4AT programme with detailed data on delirium at admission, hospital-acquired delirium, mortality, length of stay and discharge destination across demographic and socioeconomic groups.³ This is strong evidence that a clearly defined adaptation can work at health-system scale and produce clinically meaningful information.

Modified Korean K-4AT

Hur and colleagues used Days Backwards because Korean month names are numerical. They studied 286 general inpatients and then gave the tool remotely to 108 patients in COVID-19 isolation rooms. The areas under the curve were 0.992 and 0.996. The best-performing total threshold was 3 in both samples.⁴ The study shows how language, delivery and score calibration can be examined together rather than treated as afterthoughts.

Malaysia and Singapore

The Malaysian My4AT replaced Months Backwards with counting from 20 to 1. A conference study of 208 older inpatients reported sensitivity of 88.9%, specificity of 86.3% and an area under the curve of 0.93.⁵ Changi General Hospital in Singapore chose the same task because local month names are commonly numerical and the count-back task was already familiar from a local mental-status test. The modified 4AT became part of a wider delirium pathway with electronic prompts, clinical review and a care bundle.⁶

Evidence for Days Backwards as an attention task

In a US study of 234 general medical patients, any error on Days Backwards had 84.0% sensitivity and 81.9% specificity for delirium.⁷ Other work found that frailty affected Days Backwards and numerical count-back tasks more than Months Backwards.⁸ These findings make sense together. Days Backwards is a credible brief attention task, and its performance is not identical to Months Backwards.

Using and reporting an adapted version

Clear description makes local learning cumulative. A useful minimum is to:

• give the version a specific name and date;
• state the exact instruction used for the attention task;
• record how partial performance and inability to undertake the task are scored;
• state the total-score threshold and whether it differs from English version 1.2;
• identify any remote, interpreter-assisted or electronic mode of delivery;
• label the version in the electronic record and in research papers, rather than storing every variant as an undifferentiated “4AT”; and
• share the version and its evaluation where possible.

The 4AT’s Creative Commons licence allows free use and adaptation with attribution. Precise naming allows clinicians and researchers elsewhere to understand the work, reproduce it and improve it.

Where the 4AT goes next

The most useful next study would compare Months Backwards, Days Backwards and counting from 20 to 1 within the full 4AT in the same patients. Randomising the order would reduce practice effects. The study should examine time, completion, refusal, inability to engage, item scores and the full 4AT result, alongside a blinded delirium reference assessment. Results should be reported by dementia, frailty, education, first language, aphasia and hearing impairment.

We also need a public record of adapted versions: the exact wording, language, task, scoring, date, authors and supporting evidence. The translation collection on this website is a natural place to build it.

My view

I have been too slow to address modified versions of the 4AT directly. That was a mistake. Clinical tools change as they move between languages, cultures and health systems. Some of these adaptations are sensible, well reasoned and supported by encouraging data. The Houston experience shows that a defined local version can become part of a large and durable clinical system. The Korean, Malaysian and Singapore work shows why literal transfer is not always the best way to preserve a cognitive task.

The useful response is openness. We should describe adaptations, learn from them and let the evidence shape future versions of the 4AT. The tool needs enough consistency for scores to retain meaning, and enough flexibility to work well for the patients and clinical teams who use it.

References

1. Tieges Z, MacLullich AMJ, Anand A, et al. Diagnostic accuracy of the 4AT for delirium detection in older adults: systematic review and meta-analysis. Age and Ageing. 2021;50:733-743. doi:10.1093/ageing/afaa224.
2. Pan AP, Agarwal K, Taffet GE, et al. Delirium leads to poor in-hospital and 90-day outcomes among patients with acute ischemic stroke with and without intravenous thrombolysis or intraarterial therapy. Stroke: Vascular and Interventional Neurology. 2022;2(6):e000338. doi:10.1161/SVIN.122.000338.
3. Pan AP, Agarwal K, Potter T, et al. Social disparities, delirium occurrence, and related outcomes among hospitalized older adults. Journal of the American Geriatrics Society. 2025;73:3729-3737. doi:10.1111/jgs.70137.
4. Hur HJ, Jang YN, Park HY, et al. A prospective study of remote delirium screening using the modified K-4AT for COVID-19 inpatients. Frontiers in Psychiatry. 2022;13:976228. doi:10.3389/fpsyt.2022.976228.
5. Sirimanne D, Khor H, Lim JF, et al. The effectiveness of modified 4AT for rapid delirium screening in hospitalized older adults. Australasian Journal on Ageing. 2018;37(S2), abstract OR56. doi:10.1111/ajag.12550.
6. Lim SC, Chiam WM, Goh CH, et al. Initiatives to improve awareness of delirium in a teaching hospital in Singapore: for better patient care. Archives in Neurology & Neuroscience. 2019;2(3). doi:10.33552/ANN.2019.02.000536.
7. Marra A, Jackson JC, Ely EW, et al. Focusing on inattention: the diagnostic accuracy of brief measures of inattention for detecting delirium. Journal of Hospital Medicine. 2018;13:551-557. doi:10.12788/jhm.2943.
8. Bellelli PG, Biotto M, Morandi A, et al. The relationship among frailty, delirium and attentional tests to detect delirium: a cohort study. European Journal of Internal Medicine. 2019;70:33-38. doi:10.1016/j.ejim.2019.09.008.
9. Günther U, Wolke M, Hansen H-C, et al. Bedeutung der Desorientierung bei der Delireinschätzung. Medizinische Klinik – Intensivmedizin und Notfallmedizin. 2021. doi:10.1007/s00063-021-00850-z.

Page reviewed: 28 July 2026. We welcome details of other well-described modified or culturally adapted 4AT versions through the contact page.