The Ashton Manual in Practice: Diazepam Substitution for Benzodiazepine Withdrawal

Long-term benzodiazepine prescribing remains widespread despite three decades of guidance against it, and abrupt discontinuation in physically dependent patients carries seizure, autonomic instability, and protracted withdrawal risk. Professor C. Heather Ashton's Benzodiazepines: How They Work and How to Withdraw — the Ashton Manual — has been the most widely used clinical taper protocol since its 1999 release and 2002 revision, and its diazepam-substitution method continues to anchor outpatient deprescribing practice. This article translates the Ashton method into prescriber-facing guidance, integrates more recent evidence (Horowitz & Taylor, Maudsley Deprescribing Guidelines 2024), and addresses common practical failure modes.

Why diazepam substitution at all

Most patients presenting for benzodiazepine taper are on a short- or intermediate-half-life agent: alprazolam (t½ 6–12 h), lorazepam (t½ 10–20 h), oxazepam (t½ 4–15 h), temazepam (t½ 8–22 h), or clonazepam (t½ 18–50 h, intermediate in this group despite its long parent half-life). Three pharmacokinetic problems make these agents difficult to taper directly:

1. Inter-dose withdrawal. With alprazolam in particular, plasma levels fall below the receptor-occupancy threshold for symptom suppression several times per day, producing rebound anxiety, autonomic surges, and reinforcement of the dose schedule.
2. Granularity. A 0.25 mg alprazolam tablet is the smallest commercially available unit; a 5–10% reduction from 1 mg/day cannot be achieved with whole or even half tablets without compounding or liquid preparation.
3. Receptor pharmacology. High-potency agents (alprazolam, clonazepam, lorazepam) exhibit steeper occupancy-vs-dose curves at the GABA-A α1 subunit, so a small absolute dose change is a large fractional receptor change.

Diazepam (t½ 20–50 h; active metabolites desmethyldiazepam t½ 36–200 h, oxazepam, temazepam) produces a stable plasma level with once- or twice-daily dosing, is available in 2 mg and 5 mg scored tablets and a 1 mg/mL oral solution, and has a flatter dose-response curve in the clinically relevant range. Substituting it for a high-potency short-acting agent uncouples the dose schedule from symptom rebound and creates a substrate that can be tapered in small fractions.

The Ashton method does not claim diazepam is the only acceptable carrier — Ashton herself noted that patients stable on clonazepam or on the long-half-life agents may be tapered directly. The argument for substitution is specifically for patients on alprazolam, lorazepam, oxazepam, or temazepam, or for those experiencing inter-dose withdrawal on any short-acting agent.

Ashton equivalence table

The Ashton equivalences differ slightly from those in the Maudsley Deprescribing Guidelines and the BNF. They were derived from clinical experience at the Newcastle Withdrawal Clinic over more than two decades; they are conservative (i.e., they tend to over-estimate diazepam-equivalent dose, which is safer in substitution because under-substitution risks acute withdrawal).

| Benzodiazepine | Ashton-equivalent dose to 10 mg diazepam | || | Alprazolam (Xanax) | 0.5 mg | | Bromazepam (Lexotan) | 5–6 mg | | Chlordiazepoxide (Librium) | 25 mg | | Clobazam (Frisium) | 20 mg | | Clonazepam (Klonopin, Rivotril) | 0.5 mg | | Flurazepam (Dalmane) | 15–30 mg | | Lorazepam (Ativan) | 1 mg | | Nitrazepam (Mogadon) | 10 mg | | Oxazepam (Serax) | 20 mg | | Temazepam (Restoril, Normison) | 20 mg | | Triazolam (Halcion) | 0.5 mg | | Zolpidem (Ambien) | 20 mg | | Zopiclone (Imovane) | 15 mg | | Zaleplon (Sonata) | 20 mg |

The Z-drugs are included because cross-tolerance at the GABA-A α1 subunit is sufficient that diazepam will suppress withdrawal in patients dependent on zolpidem, zopiclone, or zaleplon, even though the parent compounds are not benzodiazepines.

A 1 mg alprazolam BID prescription is therefore approximated by 40 mg diazepam daily — a dose that surprises clinicians unfamiliar with the equivalence and which often prompts the (incorrect) reaction that the patient must be on "too much" alprazolam. The patient is on the dose they are tolerant to; the diazepam dose is what is required to occupy the same fraction of GABA-A receptors.

Two substitution strategies

Stepwise substitution (preferred for most outpatients)

Rather than swap the full dose at once, replace one dose at a time, beginning with the night-time dose. A typical schedule for a patient on alprazolam 0.5 mg three times daily (= 30 mg diazepam equivalent):

• Week 1–2: Continue alprazolam 0.5 mg morning and afternoon; replace bedtime alprazolam 0.5 mg with diazepam 10 mg at bedtime.
• Week 3–4: Continue alprazolam 0.5 mg morning; replace afternoon alprazolam with diazepam 10 mg at midday or early evening; continue diazepam 10 mg at bedtime.
• Week 5–6: Replace morning alprazolam with diazepam 10 mg.

The endpoint is diazepam 10 mg three times daily, or — preferable for adherence and pharmacokinetic stability — consolidate to twice-daily dosing once full substitution is complete. Tapering does not begin until substitution is complete and the patient has stabilised, typically a further 1–2 weeks.

Mixed regimen (when full substitution is poorly tolerated)

Some patients — particularly those on alprazolam or clonazepam who have been on the agent for many years — experience subjective destabilisation when the parent compound is removed entirely, even at correct equivalence. A mixed regimen retains a small dose of the original agent while the bulk of the receptor occupancy is provided by diazepam. The original agent is then itself tapered later in the schedule. This is more pharmacokinetically complex and is generally a second-line approach.

Tapering after substitution: the Ashton schedule and its modern revisions

The original Ashton schedules tapered diazepam in 1–2 mg decrements every 1–2 weeks, beginning at doses of 30–40 mg/day and reaching zero over 6–18 months. These schedules use linear decrements — 2 mg from 40 to 38, then 2 mg from 38 to 36, and so on.

The Horowitz & Taylor Maudsley Deprescribing Guidelines (2024) and prior 2019 Lancet Psychiatry paper argue, on the basis of receptor-occupancy data, that linear dose reductions produce non-linear changes in receptor occupancy at low doses. The dose-occupancy curve for benzodiazepines at GABA-A is hyperbolic: at 20 mg diazepam, occupancy is near plateau; at 2 mg, a 1 mg reduction is approximately a 20–30% drop in occupancy. The clinical implication is that linear schedules are well tolerated at high doses and become disproportionately difficult at low doses — exactly the pattern Ashton documented.

The contemporary modification ("hyperbolic taper") is to reduce by a fixed percentage of current dose rather than a fixed number of milligrams:

• Above 20 mg diazepam-equivalent: 10% of current dose every 2–4 weeks.
• 20–10 mg: 5–10% of current dose every 2–4 weeks.
• Below 10 mg: 2.5–5% of current dose every 2–4 weeks.
• Final cut to zero: at 0.5–1 mg, the final reduction may need to be a final small step rather than a sub-mg fraction; alternatively, dose can be tapered to ≤0.5 mg via liquid before discontinuation.

Concretely, on a hyperbolic schedule a patient at diazepam 20 mg/day tapering at 10% every 2 weeks will reach 10 mg in approximately 14 weeks, 5 mg in a further 14 weeks, 2 mg in a further 18 weeks, and 0.5 mg approximately one year after starting. Schedules of 12–24 months for patients on long-term high-dose benzodiazepines are appropriate; rushing produces relapse, dropout, and protracted withdrawal.

Compounding, liquid diazepam, and small fractions

Below 2 mg, tablet splitting is no longer accurate. Three options:

1. Diazepam oral solution 1 mg/mL (commercially available in most jurisdictions). Allows reductions of 0.1 mg with reasonable accuracy via 1 mL syringe.
2. Compounded liquid at the pharmacy, typically at 1 mg/mL or lower for sub-mg dosing.
3. Tablet titration in water ("the Ashton water titration"): a 2 mg or 5 mg tablet dispersed in a fixed volume of water, with the appropriate volume drawn off and discarded daily. Diazepam is poorly water-soluble; the suspension must be shaken vigorously each time and used immediately. This is acceptable for short-term use but pharmacist-compounded liquid is preferable for the multi-month sub-mg phase.

Common substitution failures

Over-substituting and producing sedation

If the patient becomes sedated, ataxic, or cognitively dulled after substitution, the diazepam equivalence has been overshot. Common reasons: failure to account for very long t½ of desmethyldiazepam in elderly patients (steady-state may take 2–3 weeks to reach), or use of equivalence tables that over-estimate diazepam at very high alprazolam or clonazepam doses. Reduce diazepam by 10–20% rather than reverting to the original agent.

Under-substituting and producing breakthrough withdrawal

If the patient experiences anxiety, tremor, sweating, or insomnia after substitution, occupancy is insufficient. The temptation is to add PRN lorazepam or alprazolam — this defeats the purpose of substitution by re-introducing inter-dose oscillation. Instead, increase diazepam by 2–5 mg and re-stabilise.

Substituting when substitution is unnecessary

Patients on clonazepam who do not experience inter-dose withdrawal, who tolerate their dose schedule, and whose dose can be cut accurately with available formulations (clonazepam 0.5 mg tablets, or compounded liquid) do not require substitution. Direct hyperbolic taper of the original agent is reasonable. Substitution is a tool for short-half-life agents and granularity problems, not a universal first step.

Substituting in patients with significant hepatic disease

Diazepam is oxidatively metabolised via CYP2C19 and CYP3A4 to active metabolites with very long half-lives. In significant hepatic impairment, accumulation can be marked. Ashton herself recommended oxazepam or lorazepam (both glucuronidated, no active metabolites) as the carrier in cirrhosis. The same hyperbolic taper principles apply.

Adjuncts and what to avoid

The Ashton Manual is explicit that no adjunctive medication reliably accelerates or smooths benzodiazepine withdrawal in a way that justifies routine use. Specific points:

• Flumazenil. Subanaesthetic flumazenil infusions have been promoted for "rapid detox." Evidence is weak, and the procedure carries seizure risk in dependent patients. Not recommended outside research settings.
• Pregabalin and gabapentin. May provide symptom relief during withdrawal but carry their own dependence liability. Use with awareness that you may be substituting one GABAergic dependence for another.
• Antidepressants. Only indicated if there is a comorbid depressive or anxiety disorder requiring treatment in its own right; do not start an SSRI simply to "cover" a benzodiazepine taper.
• Beta-blockers. Propranolol 10–40 mg may attenuate autonomic symptoms (tremor, tachycardia) without affecting CNS withdrawal. Reasonable as a short-term adjunct.
• Z-drugs as substitutes. Do not substitute zolpidem or zopiclone for a benzodiazepine taper. Cross-tolerance is incomplete in the wrong direction.

Protracted withdrawal

A subset of patients — estimates from Ashton's clinic and from the Maudsley Deprescribing Guidelines suggest 10–15% of long-term users — experience symptoms persisting more than 6–12 months after the last dose: cognitive impairment, perceptual disturbances, paraesthesiae, dysautonomia, and persistent anxiety. The mechanism is not established but is presumed to involve durable changes in GABA-A subunit composition and downstream glutamatergic upregulation. There is no validated treatment. Reinstatement and a slower re-taper sometimes helps if the patient is within weeks of cessation; reinstatement years later is rarely effective. Counsel patients on long-term high-dose benzodiazepines about this risk before initiating taper, not as a deterrent but so the eventual schedule is appropriately conservative.

Clinical pearls

• Substitute for short-half-life agents (alprazolam, lorazepam, oxazepam, temazepam, triazolam) and for any agent producing inter-dose withdrawal. Direct taper is acceptable for stable clonazepam and chlordiazepoxide.
• Use Ashton equivalences (e.g., alprazolam 0.5 mg ≈ diazepam 10 mg) and expect total daily diazepam doses that look high — this is correct, not over-prescription.
• Substitute one dose at a time over 2–6 weeks, beginning with the night-time dose; do not begin tapering until substitution is complete and the patient has stabilised.
• Adopt a hyperbolic taper: reduce by approximately 10% of current dose every 2–4 weeks at high doses, dropping to 2.5–5% below 10 mg diazepam-equivalent.
• For the sub-mg phase, use 1 mg/mL diazepam oral solution or compounded liquid; tablet-splitting is not accurate below 2 mg.
• In hepatic impairment, substitute to oxazepam or lorazepam rather than diazepam to avoid active-metabolite accumulation.
• Do not add PRN short-acting benzodiazepines to manage breakthrough withdrawal during a diazepam taper — increase the scheduled diazepam dose instead.

---

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