Deprescribing in Polypharmacy: Sequencing Tapers in Older Adults

Polypharmacy in older adults is one of the most common — and most under-managed — clinical scenarios in primary care and geriatric psychiatry. A patient in their seventies presenting with five or more concurrent CNS-active medications (an antidepressant, a benzodiazepine, a Z-drug, gabapentin, and an antipsychotic added during a hospital admission a decade ago) is not a rare case; it is a representative one. Deprescribing in this setting is not simply "tapering" — it is a sequencing problem, with overlapping withdrawal syndromes, drug-drug interaction unmasking, and shifting pharmacokinetics that make the order of operations as important as the rate of reduction.

Why sequencing matters more than rate

The pharmacology of withdrawal in older adults differs from younger patients in three clinically meaningful ways. First, hepatic mass and CYP450 enzyme activity decline with age, lengthening effective half-lives — particularly for diazepam, clonazepam, fluoxetine, and other long-half-life drugs with active metabolites. Second, renal clearance reductions extend exposure to drugs and metabolites that are renally eliminated, including gabapentin, pregabalin, and the hydroxylated metabolites of risperidone. Third, the aging brain has reduced receptor reserve and impaired homeostatic compensation, meaning that what would be a tolerable abrupt reduction in a younger patient can precipitate delirium, falls, or seizures in an 80-year-old.

If two CNS depressants are tapered in parallel, the patient experiences additive withdrawal — and the clinician loses the ability to attribute any emergent symptom to a specific agent. Sequencing tapers serially, with adequate stabilization periods between agents, preserves diagnostic clarity. When a tremor or a worsening of anxiety emerges, the prescriber can interrogate the most recently reduced agent rather than guessing across four.

The Maudsley Deprescribing Guidelines (Horowitz & Taylor, 2024) emphasize that pharmacological reasoning, not pill-counting convenience, should drive sequencing. NICE guidance on multimorbidity (NG56) and the American Geriatrics Society Beers Criteria (2023 update) provide complementary frameworks for identifying which agents are highest-priority for removal, but neither prescribes the order in a polypharmacy regimen — that judgment remains with the treating clinician.

A decision framework for ordering tapers

Before initiating any reductions, the prescriber should complete three preparatory steps:

1. Reconcile and rank. Build a complete medication list including over-the-counter sedatives (diphenhydramine, doxylamine), herbals (valerian, kava), and PRN agents the patient actually takes daily. Rank each by anticholinergic burden (using the Anticholinergic Cognitive Burden scale), fall risk contribution, and STOPP/START flags.
2. Establish baseline function. Document cognition (Montreal Cognitive Assessment), gait and balance (Timed Up and Go), sleep architecture by patient report, mood, and the specific target symptoms each agent was originally prescribed to treat. Without baseline data, neither the patient nor the clinician will be able to distinguish withdrawal from relapse three months in.
3. Confirm indication and goals of care. Many psychotropics in older adults have outlived their indication. A mirtazapine started for post-bereavement insomnia in 2014 may have no current target. Conversely, an SSRI started for recurrent unipolar depression may still be load-bearing. The deprescribing conversation begins with whether the drug should still exist in the regimen at all.

Once these are in hand, sequencing follows clinical risk logic. Several principles apply.

Principle 1: Remove the drug with the worst risk-benefit profile first, unless withdrawal is uniquely dangerous

Anticholinergics, first-generation antihistamines used as hypnotics, and antipsychotics prescribed off-label for dementia-related behaviors are among the highest-yield targets. The CATIE-AD trial and subsequent meta-analyses (Schneider et al., NEJM) demonstrated that the absolute benefit of antipsychotics for behavioral and psychological symptoms of dementia is small, while the mortality signal is substantial. The Beers Criteria classify these as high-priority for removal.

However, this principle has an important exception: when a drug has a dangerous abrupt-withdrawal profile, it cannot simply be "first off the list" — it must be tapered, even if it is also the highest-risk agent. Benzodiazepines and Z-drugs both fall into this category, as do high-dose gabapentinoids and clozapine.

Principle 2: Taper the drug with the longest half-life last among any class

If two agents within a similar pharmacologic class are being removed (for example, two anticholinergics, or a benzodiazepine plus a Z-drug), tapering the shorter-half-life agent first allows the longer-half-life drug to provide background receptor occupancy that softens withdrawal. This is the rationale behind the Ashton Manual's classical recommendation to cross-titrate a short-half-life benzodiazepine to diazepam before tapering — though that approach is increasingly questioned in older adults where diazepam itself carries substantial accumulation risk and contributes to falls and delirium.

Principle 3: Avoid concurrent tapering of two agents with overlapping withdrawal profiles

Two GABAergic agents (a benzodiazepine and gabapentin), two serotonergic agents (an SSRI and trazodone), or a benzodiazepine plus an alcohol cessation effort should not be tapered simultaneously. Stabilize one before initiating the next. The receptor adaptations that drive withdrawal are not additive in a linear sense — they interact, and overlapping reductions can produce symptom intensity disproportionate to either single change.

Principle 4: Stabilize for weeks, not days, between agents

The prescriber should plan to hold the regimen stable for an interval long enough to confirm that the patient has returned to a new pharmacologic baseline before initiating the next taper. For most psychotropics, this means several weeks at minimum after the previous agent is fully discontinued — longer for drugs with protracted withdrawal syndromes such as benzodiazepines, SSRIs, and SNRIs. The Maudsley Deprescribing Guidelines describe these protracted syndromes in detail, and the prescriber should not interpret weeks of relative quiescence as confirmation of full neurochemical re-equilibration in every case.

A worked sequencing example

Consider a 78-year-old woman taking sertraline (started 2012 for major depressive disorder, currently in sustained remission), lorazepam (started 2016 for generalized anxiety, taken nightly), zolpidem (started 2018 for sleep-onset insomnia), gabapentin (started 2020 for diabetic neuropathy, partially effective), and quetiapine (started during a 2022 hospitalization for delirium, never reassessed).

Applying the framework:

• Quetiapine has the weakest current indication (the delirium has resolved), carries cardiovascular and metabolic risk, contributes to anticholinergic burden via its M1 affinity, and has a relatively benign withdrawal profile at the low doses typical of behavioral prescribing. It is the highest-priority removal, and can plausibly go first.
• Zolpidem and lorazepam both act at GABA-A, with overlapping withdrawal profiles. They should not be tapered simultaneously. Conventional practice tapers the Z-drug first because its specific α1 selectivity makes it the "less broadly anxiolytic" of the two; removing it first leaves the benzodiazepine to attenuate GABAergic withdrawal during that phase.
• Lorazepam has a shorter half-life than diazepam but a longer half-life than zolpidem. Once zolpidem is off and a stabilization interval has elapsed, the benzodiazepine taper begins. This is typically the slowest and longest phase of the entire deprescribing sequence — frequently many months to over a year, consistent with the Maudsley and Ashton frameworks.
• Gabapentin has an active neuropathic indication. Whether to taper it depends on a renewed pain assessment, not on the deprescribing goal in isolation. If pain is well-controlled and the patient is willing to attempt reduction, it follows the benzodiazepine taper.
• Sertraline is last. The patient has remitted depression on it, and there is an evidence base (Lewis et al., NEJM 2021 — the ANTLER trial) showing relapse rates above 50% in primary-care patients discontinuing maintenance antidepressants. The deprescribing conversation here is genuinely shared-decision-making, and the SSRI may be retained indefinitely.

The total sequence may take a year or more. This is not a flaw of the framework — it reflects the underlying pharmacology and the compounding fragility of the older adult nervous system.

Special considerations in dementia

In patients with established cognitive impairment, two additional factors apply. First, the patient cannot reliably self-report subtle withdrawal symptoms; the clinician must rely on caregiver observation and structured behavioral assessment. Second, abrupt regimen changes can precipitate delirium even when the individual drug-level change would be tolerated in a cognitively intact patient. Slower-than-standard tapers are appropriate, and the threshold for pausing or reversing a step is lower.

Cholinesterase inhibitors and memantine deserve specific mention. The DOMINO trial (Howard et al., NEJM 2012) showed measurable cognitive and functional decline on discontinuation of donepezil in moderate-to-severe Alzheimer's disease. Deprescribing these agents is reasonable when the patient has progressed to a stage where the original treatment goal is no longer achievable, but it is not a routine "polypharmacy cleanup" step and should be reasoned through separately from the psychotropic taper sequence.

Communication with the patient and family

The deprescribing conversation in an older adult almost always involves a family member or caregiver. Three communication points consistently improve adherence to the plan:

• Name the sequencing rationale explicitly. "We are removing the quetiapine first because it has the least benefit and the most risk, and the lorazepam last because it will take the longest and is most likely to cause problems if we move too fast." This counters the common expectation that the most "obvious" drug (often the sleeping pill) should come off first.
• Pre-discuss what withdrawal might feel like and how it is distinguished from relapse. Withdrawal symptoms typically appear within days to weeks of a reduction and resolve over a similar timeframe; relapse of the original disorder typically emerges later and resembles the original syndrome the drug was prescribed to treat. This framing is endorsed in the Royal College of Psychiatrists' 2023 guidance on stopping antidepressants.
• Explicitly authorize pauses. Patients and caregivers should be told in advance that holding at the current step or returning briefly to the previous step is not failure — it is a feature of the protocol.

Drug-interaction unmasking

A frequently overlooked complication of sequential deprescribing is the unmasking of interactions that the now-removed drug was buffering. Removing a CYP3A4 inducer can elevate plasma levels of remaining agents; removing a strong anticholinergic can unmask cholinergic side effects from a cholinesterase inhibitor that were previously offset; removing a sedating agent can reveal that a different agent in the regimen has been contributing more to daytime somnolence than recognized. The prescriber should review the remaining regimen's interaction profile at each transition, not only at the start of the sequence.

When to refer

Several scenarios warrant referral to a clinical pharmacist, geriatric psychiatrist, or specialist deprescribing service:

• Complex benzodiazepine tapers beyond approximately a year of daily use, particularly in patients with prior failed attempts.
• Concurrent substance use (alcohol, opioids) with the prescribed regimen.
• Severe baseline frailty where any destabilization carries high fall or delirium risk.
• Patients on three or more psychotropics with overlapping mechanisms.
• Patients with significant cognitive impairment where caregiver reporting is the primary symptom signal.

Several health systems now have dedicated deprescribing pharmacy clinics; where available, these are appropriate consult partners and can take ownership of the longest and most labor-intensive phase of the sequence.

Clinical pearls

• Sequence, do not parallelize. Only one CNS-active agent should be in active reduction at any given time. Stabilize fully before initiating the next.
• Pick the drug with the weakest current indication first, unless its withdrawal profile is uniquely dangerous (benzodiazepines, high-dose gabapentinoids, clozapine). In that case, taper a lower-priority but safer-to-withdraw agent first to simplify the regimen before tackling the hard one.
• Plan for the full sequence to take many months in a typical five-drug polypharmacy case — often well over a year. Compressed timelines are the most common cause of failed deprescribing in older adults.
• Document baseline cognition, gait, sleep, mood, and target symptoms before the first reduction. Without this, withdrawal versus relapse becomes unresolvable.
• Re-review drug-drug interactions at every transition, not only at initiation. Removing one agent can change the kinetics and clinical profile of those remaining.
• Authorize pauses and partial reversals in advance. Tell the patient and caregiver that returning to the prior step is a designed feature of the protocol, not a setback.
• Refer early for complex benzodiazepine tapers, concurrent substance use, or severe frailty. Specialist deprescribing services exist for a reason.

---

For more clinician resources on safe deprescribing and tapering, visit tapermeds.com.