Design: systematic review and meta-analysis of eight observational studies using validated anticholinergic burden scores
Population: patients with Parkinson's disease, higher versus lower anticholinergic burden
Primary outcome: dementia, cognitive decline, fracture and emergency department visits
Effect: fracture OR 1.53 (95% CI 1.28-1.84); emergency department visits OR 1.25 (1.15-1.36); dementia OR 1.71 (0.93-3.17, nonsignificant); cognitive decline SMD -0.71 (-1.51 to 0.08, nonsignificant)

Eight studies comparing higher with lower anticholinergic burden in Parkinson's disease, scored by validated systems, were pooled. The two outcomes most often assumed — dementia and cognitive decline — did not reach significance: an odds ratio of 1.71 for dementia with an interval crossing 1.0, and a standardised mean difference of -0.71 for cognitive decline with an interval crossing zero. A European subgroup did show a significant association with dementia, odds ratio 2.91.

What was significant across the whole analysis was physical harm. Fracture carried an odds ratio of 1.53 and emergency department visits 1.25, both with intervals comfortably clear of 1.0.

That pattern is worth taking seriously rather than dismissing as a partly negative result. Anticholinergics cause blurred vision, orthostatic symptoms, urinary retention and confusion in a population that already has postural instability. Fracture is the outcome that pathway predicts, and it is the outcome that showed up.

In practice the burden is rarely one drug. It accumulates from a trihexyphenidyl started years ago for tremor, an oxybutynin for urinary urgency, an amitriptyline for sleep and a promethazine bought over the counter — the last of which is freely available in India and will not appear on any prescription list you review.

Score anticholinergic burden formally at review rather than judging drug by drug
Ask specifically about over-the-counter antihistamines and sleep aids; they will not be on the prescription record
Treat a fall or fracture in Parkinson's disease as a prompt to recalculate the burden
Reconsider trihexyphenidyl in any patient over 65 or with cognitive symptoms
Deprescribe one agent at a time and document why, so the next clinician does not restart it

Don't overread it

These are pooled observational associations — sicker patients receive more anticholinergic drugs, so confounding by indication cannot be excluded as an explanation for the fracture signal.

The statistics, in plain English

The dementia odds ratio of 1.71 has a 95% confidence interval of 0.93 to 3.17 — it crosses 1.0, so no association is established, though the interval is wide enough that a real and substantial effect has not been excluded either. The significant European subgroup finding, OR 2.91 with an interval of 1.02 to 8.28, barely clears 1.0 at its lower bound and comes from a subgroup analysis, which is the weakest form of evidence in a meta-analysis and should not be read as a regional truth. The fracture and emergency-visit results are different in kind: intervals of 1.28-1.84 and 1.15-1.36 are both narrow and clear of 1.0.