Anticholinergic Drugs and Incident Dementia, Mild Cognitive Impairment and Cognitive Decline

A Meta-analysis

Nina T. Pieper; Carlota M. Grossi; Wei-Yee Chan; Yoon K. Loke; George M. Savva; Clara Haroulis; Nicholas Steel; Chris Fox; Ian D. Maidment; Antony J. Arthur; Phyo K. Myint; Toby O. Smith; Louise Robinson; Fiona E. Matthews; Carol Brayne; Kathryn Richardson


Age Ageing. 2020;49(6):939-947. 

In This Article


Observational studies report, on average, a 20% greater incidence of dementia associated with the use of drugs with definite anticholinergic activity. Greater associations with dementia are also reported with longer durations of exposure. Observational studies also report anticholinergic drug use is associated with long-term global cognitive decline. However, there was no randomised evidence, study findings are heterogeneous and all but one study had a serious or critical risk of bias. No reported factor could explain the heterogeneous study findings.

To our knowledge, this is the most comprehensive systematic review and meta-analysis on the long-term cognitive effects associated with anticholinergic drug use. We excluded cross-sectional studies and studies with insufficient follow-up. We used recognised methods for performing an updated search[19] and minimised the risk of missing studies. We provided a thorough assessment of risk of bias by utilising the latest rating scales. The included studies used different scales to measure anticholinergic burden, and although they differ in how they classify anticholinergic burden, they are typically consistent in which drugs they classify as strongly or definitely anticholinergic.[56]

A major limitation in drawing substantive conclusions from this review is that all studies were observational and most were rated at serious/critical risk of bias.[24] In particular, 88% included participants who began taking anticholinergic drugs before the study period. As such, the ROBINS-I tool rates these study estimates at serious risk of selection bias[24] and therefore a serious overall risk of bias at best. For short-term effects, examining only new users of therapy reduces biases by capturing early events and confounders for prescribing therapy.[57] However the impact of excluding prevalent users on selection bias when examining long-term effects is unclear.[58] The largest threat to study validity is likely residual confounding, given that anticholinergics are indicated for conditions also associated with cognitive decline such as depression and Parkinson's disease.[59–61] There may also be residual confounding by frailty, as only one study explicitly adjusted for frailty.[40] Definitions of incident dementia and MCI were generally consistent. Studies varied, however, on how they assessed and reported cognitive function; therefore we could only pool results from six studies. There was moderate risk of exposure misclassification, particularly for studies relying on self-reported use or prescription records. Self-reported use of anticholinergic drug classes has only moderate agreement with pharmacy dispensing records.[62] Publication bias is possible, but difficult to assess owing to the lack of registration requirements for observational studies. Unfortunately, we were unable to pool absolute risks, as only one study provided these.[9]

Our findings concur with previous reviews. A previous meta-analysis reported a stronger association between anticholinergic use and incident dementia (OR 1.43; 95% CI 1.16–1.73) but only pooled results from three cohort studies.[17] Whether the observed associations between anticholinergic drug use and dementia is a causal relationship or reflects risk factors or early symptoms of dementia (such as depression or bladder instability) remains unclear from the available evidence. Some studies reported an increasing dementia risk with greater exposure to anticholinergic drugs, consistent with a causal link.[9,10,26] However, few studies excluded drug exposure in the period just prior to dementia,[9,10,41] and so drugs may have been prescribed for early symptoms of dementia. Two large UK studies report differences in dementia risk according to anticholinergic drug class, inconsistent with a causal link.[9,55]

Mechanistic evidence is limited, although some neuropathological studies in humans and mice support a role of anticholinergics in affecting neurodegenerative pathology,[63] and anticholinergic use has been associated with increased brain atrophy and reduced glucose metabolism.[64] Evidence from RCTs on anticholinergic cessation has failed to show improvements in cognition, but has generally been underpowered and has focussed on short-term outcomes.[65]

This study adds to the weight of evidence supporting current recommendations that prescribers should be cautious about using medications with anticholinergic activity in older people. Prescribers should discuss the risks and benefits of anticholinergic drugs with patients and their carers and consider either avoiding medication or using alternatives where appropriate.[66] However, the current evidence is insufficient to warrant aggressive deprescribing of these drugs due to long-term cognitive effects. Nevertheless, other known effects such as dry mouth and eyes, confusion, constipation and urinary retention remain a concern.[58,59] Due to our findings of stronger associations with long-term use, clinicians should regularly review and consider stopping anticholinergic drugs if there is no clear evidence of benefit. There is a need for consensus on the management of the long-term use of anticholinergic drugs, particularly on developing clear timeframes regarding proposed durations of use.

The decision to prescribe anticholinergic drugs needs careful weighing up of the risks and benefits, and this review highlights the lack of reliable research evidence on cognitive risks.

Although such evidence will be difficult to obtain, we offer suggestions in Appendix 3. There is also a need to regularly update anticholinergic assessment tools and agree consensus on their ratings. Due to the growing number of assessment tools, we recommend a systematic harmonisation of current tools specifically aimed at examining central anticholinergic effects, taking into consideration blood–brain barrier permeability, to facilitate care and research.[67]