The Medications That Never Come Off

Every visit adds something. Almost no visit takes anything away. That asymmetry is how a person ends up on fourteen medications and nobody can say why.

Is it actually safe to stop medications?

In the situations that have been tested, yes — and stopping sometimes improves quality of life rather than shortening it.

The clearest evidence comes from a randomized trial published in JAMA Internal Medicine in 2015. Adults with an estimated life expectancy between one month and one year, who had been on a statin for at least three months and were not taking it for active cardiovascular disease, were randomly assigned to stop the statin or continue it.

Stopping did not increase 60-day mortality. The group that stopped reported better quality of life, took fewer other medications, and spent less on drugs.

That trial is about statins in advanced illness specifically, and it should not be stretched into a general claim that any drug can be stopped safely. What it does establish is the principle: a medication that made sense at 62 for the prevention of a heart attack in twenty years' time is being taken for a benefit that now sits beyond the person's remaining life, while its side effects are being experienced today.

Why does the list only ever get longer?

Because adding is a low-friction decision and removing is a high-friction one, and nothing in a normal appointment corrects for that.

Every clinician who sees your person can add something. Very few will remove something another clinician started — it feels like overruling a colleague, it takes longer to explain, and if something goes wrong afterward the person who stopped it owns the outcome.

So the list accumulates. A drug started for a hospital stay in 2019 is still being dispensed in 2026. A cardiologist adds; a urologist adds; a neurologist adds; the primary care doctor inherits all of it and has fifteen minutes.

The person best placed to interrupt this is usually the one who watches the whole picture. Frequently that is you.

What is a prescribing cascade?

A side effect gets mistaken for a new medical problem, and a second drug is prescribed to treat the first drug.

This is the most preventable form of harm in the whole medicine cabinet, and dementia produces a textbook example.

Cholinesterase inhibitors — donepezil, rivastigmine, galantamine — increase acetylcholine. One common consequence is urinary urgency and incontinence. That gets reported as a new problem, and the response is often a bladder antimuscarinic such as oxybutynin or tolterodine, which works by blocking acetylcholine.

The two drugs are now pulling in opposite directions, and the second one is an anticholinergic, a class independently associated with worsened cognition.

This is not a theoretical concern. A population study of 44,884 older adults with dementia found that those dispensed a cholinesterase inhibitor were significantly more likely to subsequently receive an anticholinergic bladder drug — 4.5 percent versus 3.1 percent, an adjusted hazard ratio of 1.55. And a 2026 study in the Journal of the American Geriatrics Society examined that specific pairing in people living with dementia and linked it to increased risk of delirium and falls.

Other cascades worth knowing:

• A calcium channel blocker causes ankle swelling; a diuretic is added.
• A drug causes tremor; a Parkinson's medication is added.
• An antipsychotic causes stiffness and slowing; that gets read as the dementia progressing, and the dose goes up.

The question that catches all of these: "When did this symptom start, and what was added just before it?" Bring a timeline, not a symptom.

Which medications are worth reviewing?

The ones whose benefit is years away, the ones that worsen cognition, and the ones nobody can remember starting.

The reference clinicians use is the American Geriatrics Society Beers Criteria — a list of medications that are potentially inappropriate in adults 65 and over. The current version is the 2023 update, with a further revision anticipated. In late 2025 the AGS added an Alternatives List, which is the more useful companion for families: it names what to use instead, rather than only what to avoid.

Categories that most often deserve a conversation:

• Anticholinergics. Bladder drugs, older antihistamines, some antidepressants, some sleep aids. This class directly opposes the mechanism of dementia treatment and carries a cumulative burden most people have never had added up.
• Benzodiazepines and Z-drugs — lorazepam, diazepam, zolpidem. Sedation, falls, confusion, and a tolerance that means the original problem returns while the harms remain.
• Antipsychotics started during a crisis and never reviewed. These carry a mortality warning and are frequently continued for years past the episode they were meant to cover.
• Statins for primary prevention in someone with limited life expectancy — the situation the trial above addressed directly.
• Tight diabetes control. Sulfonylureas and aggressive targets in a frail older adult trade a distant complication for a real risk of hypoglycemia today, which in a person with dementia presents as confusion nobody attributes to the blood sugar.
• Proton pump inhibitors started for a short course years ago and never stopped.
• Bisphosphonates, vitamins, and supplements accumulated over a decade with no current indication.
• The cholinesterase inhibitor itself, in advanced disease, where the goal it was prescribed for may no longer be reachable.

None of these is an instruction. Every one of them is a legitimate question.

How is a medication actually stopped?

One drug at a time, tapered where the drug requires it, with something written down beforehand so you can tell what changed.

The way deprescribing goes wrong is stopping four things at once, watching the person get worse, and having no idea which one mattered.

A workable approach:

• Change one thing at a time, and leave enough time between changes to attribute the result.
• Taper what needs tapering. Benzodiazepines, antidepressants, gabapentinoids, PPIs and beta blockers can all produce withdrawal or rebound effects. Abrupt stops are how deprescribing gets a bad reputation.
• Know what withdrawal looks like so it is not misread. Rebound insomnia after a sleep aid, or rebound reflux after a PPI, tends to peak and then settle. Stopping the taper at the peak means the drug is restarted permanently for a problem that would have resolved.
• Write down two or three observable things first — sleep, appetite, steadiness on the feet, how many prompts it takes to get dressed — with the date. Otherwise the assessment is memory versus memory.
• Agree a review date at the same time as the change, not afterward.
• Ask specifically what to watch for and when to call.

Who does a proper medication review?

Ask for a pharmacist. They are the most under-used clinicians in this entire process.

A pharmacist-led medication review is a different thing from a doctor glancing at the list. Pharmacists are trained specifically in interactions, cumulative anticholinergic burden and dosing in renal impairment, and they generally have more than fifteen minutes.

Routes worth asking about:

• Medicare Part D Medication Therapy Management, which many people with multiple chronic conditions and multiple drugs qualify for without knowing it.
• The annual wellness visit, which includes a medication review.
• A consultant pharmacist through a home health agency, hospice, or the facility if the person is in one.
• Your community pharmacist, who can often run an interaction and burden check on the spot if you bring the complete list.

Bring everything, including over-the-counter drugs and supplements, in the original containers — including the ones from the other doctor, the ones bought at the grocery store, and the ones you suspect are not being taken. The gap between the prescribed list and what is actually swallowed is itself important information.

What if the doctor says no?

Ask what would need to change for it to be worth stopping, and put the answer in the record.

A refusal is not the end of the conversation. Two questions usually move it:

• "What is this one treating, and what would we expect to happen if it came off?"
• "If not now, what would have to change for it to be worth revisiting?"

The second question converts a no into a condition, and conditions can be met. It also creates a written marker that someone — possibly a different clinician, possibly in six months — can act on.

Nobody is likely to trim this list on their own. The review happens because a family member brings the whole bag to an appointment and asks what could come off. That question is not an imposition. For a person with dementia, it often leads to the most useful change of the visit.