- Design
- Retrospective population-based cohort using linked medical and long-term care insurance claims, Japan, 2019-2023
- Population
- 8117 patients aged 65 or over with Alzheimer's disease, institutionalised or at care needs level 4 or 5
- Primary outcome
- Duration of treatment with the initial cholinesterase inhibitor formulation
- Effect
- Mean duration 1.5 years oral vs 0.6 years transdermal (log-rank P < 0.001); adjusted hazard ratio for discontinuation with patch 11.3 (95 per cent CI 10.2 to 12.4)
Linked medical and long-term care insurance claims from one Japanese prefecture identified 8117 people aged 65 or over with Alzheimer's disease who were either institutionalised or certified at care needs level 4 or 5 - complete dependence in activities of daily living. The question was how long the cholinesterase inhibitor they started stayed running.
Oral formulations, donepezil or galantamine, were continued for a mean of 1.5 years; transdermal patches, rivastigmine or donepezil, for 0.6 years. The adjusted hazard ratio for discontinuing a patch rather than an oral drug was 11.3 (95 per cent CI 10.2 to 12.4). Hospitalisation and mortality rates were both lower in the oral group (236 against 272, and 210 against 272 per 1000 person-years).
The comparison between formulations is heavily confounded and should not be read as the finding. Patches are chosen for people who cannot reliably swallow, which is a marker of more advanced disease, and that alone would produce shorter treatment, more hospitalisation and higher mortality without the patch doing anything. The finding that survives is the one about the oral group: a mean of 1.5 years of cholinesterase inhibitor treatment in people with complete functional dependence, where the evidence for continued benefit is thin and the burden of an anticholinesterase - bradycardia, falls, incontinence, weight loss - is not. That is a deprescribing prompt, and the right time to have the conversation is at the review where nothing else has changed.
- Ask at every review of advanced dementia whether the cholinesterase inhibitor is still doing anything identifiable.
- Do not read the patch-versus-oral difference as a drug effect; confounding by swallowing ability explains most of it.
- Look specifically for bradycardia, syncope, falls, urinary frequency and weight loss when reviewing.
- Agree a stopping trial with the family rather than an abrupt discontinuation, and record what would prompt restarting.
- Note this is Japanese claims data on institutionalised patients; the prescribing culture may differ from yours.
The statistics, in plain English
A hazard ratio of 11.3 with an interval of 10.2 to 12.4 is extraordinarily large and extraordinarily precise, which in observational data usually signals that the two groups differ fundamentally rather than that the exposure is powerful. Here the difference is almost certainly who gets a patch. Rates per 1000 person-years account for differing follow-up length, which matters when one group is treated more than twice as long, but they cannot adjust for the underlying difference in illness severity.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for geriatrics, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free