# Leeds team's ACMI scores 88 anticholinergic drugs by delirium and fall risk in older adults

- Published: 2026-11-06 03:05 UTC
- Section: Life & Health › Medicine
- Tags: anticholinergic medication index, anticholinergic burden, delirium and falls risk, prediction model, structured medication review, Connected Bradford, SAIL databank, overprescribing, BMJ Medicine
- URL: https://realtech.media/news/rt-7593ce84dc457910ede88359_leeds_team_s/

**UK researchers derived a score for 88 anticholinergic medicines from a model validated on 678,134 older adults, to help prioritise medication reviews in primary care.**

![Leeds team's ACMI scores 88 anticholinergic drugs by delirium and fall risk in older adults](https://sgzwpqqelzmajkcubgkl.supabase.co/storage/v1/object/public/rt-media/2026/11/2b6d969ab48dc278.jpg)
*U.S. Navy NMRTC by Emily McCamy, Wikimedia Commons, Public domain*

Researchers from the University of Leeds, the University of Liverpool and the Bradford Institute for Health Research developed the Anticholinergic Medication Index (ACMI), a measure of anticholinergic burden derived from a prediction model for falls and delirium, according to a paper in BMJ Medicine (bmjmedicine.bmj.com) [1]. The index scores 88 anticholinergic medicines prescribed in primary care [1].

The development cohort comprised 71 757 residents of Bradford and Craven aged over 65 from the Connected Bradford database. In 2019, 4147 of them (5.8%) had the composite outcome: hospital admission for delirium, or for a fall or fracture [1]. External validation used 678 134 Welsh residents of the same age from the SAIL databank, of whom 43 675 (6.4%) had the outcome [1].

## Model performance and prescribing data

Anticholinergic medications recorded in electronic health records served as the model's predictors [1]. Of the 88 medicines, codeine, prednisolone and furosemide were prescribed most often: 8.3%, 6.5% and 6.0% of patients in the development cohort received them at least once [1].

In the development cohort, the C index was 0.794, R2 0.06 and the calibration slope 0.97 [1]. In external validation the C index was 0.721 (95% CI 0.718 to 0.734), R2 0.04 and the calibration slope 0.69, which the authors describe as some evidence of miscalibration [1].

## Scores and use in primary care

Model coefficients were transformed into ACMI scores that range from 0.1 for desloratadine to 2.3 for amantadine [1]. The authors note that existing scales usually weight medicines by expert consensus on biological potency rather than by prediction of defined adverse outcomes, and that such tools are not integrated into electronic health record systems [1].

The ACMI was designed for automatic calculation in primary care. Combined with the model's predicted risk of delirium or falls, it allows clinicians to prioritise patients for structured medication reviews to reduce overprescribing [1].

The team also piloted the tool at GP surgeries in Bradford. It generated a list of patients aged 65 and over with their ACMI score, the medicines contributing to it and a one-year predicted risk of fall or delirium. Without the tool, this information would have had to be worked out patient by patient.

## Sources

1. [bmjmedicine.bmj.com](https://bmjmedicine.bmj.com/content/5/1/e002835)
2. [EurekAlert](https://www.eurekalert.org/news-releases/1150253)

---

Source: RealTech, realtech.media — https://realtech.media/news/rt-7593ce84dc457910ede88359_leeds_team_s/