Call Poison Control: 1-800-222-1222 (24/7, free, connects to your regional center and a medical toxicologist). Every dose below is a starting point for an adult and assumes concurrent airway, breathing, and circulation management. Confirm against your institutional protocol and the current package insert before administering. Antidote dosing is one of the highest-consequence decisions in pharmacy practice; this card is a memory aid, not an order set.
Rarely appropriate in overdose. See the Danger Zone below before considering it.
Can convert a supportive-care problem into a seizure you have just made harder to treat.
Underlined entries jump to a detailed section further down the card. Doses are adult unless stated. Pediatric dosing differs materially for several agents and is outside the scope of this card. Anticoagulant reversal beyond protamine is covered separately; see the note at the bottom.
The 2023 US/Canada consensus statement (Dart et al., JAMA Netw Open) reframed NAC as a total dose delivered over a duration set by stopping criteria, rather than a branded protocol you run to completion. It also retired the term “massive ingestion” in favor of “high-risk ingestion.”
Regimen
Dosing
Notes
IV 3-bag (21 h)
150 mg/kg over 60 min → 50 mg/kg over 4 h → 100 mg/kg over 16 h
Total 300 mg/kg. Most widely stocked protocol.
IV 2-bag (20 h)
200 mg/kg over 4 h → 100 mg/kg over 16 h
Same total. Markedly fewer anaphylactoid reactions. FDA-approved 2024 for patients ≥41 kg. Observational data show no difference in hepatotoxicity, but the label states comparative efficacy for preventing hepatotoxicity is not established.
Oral (72 h)
140 mg/kg load, then 70 mg/kg q4h × 17 doses
Total 1330 mg/kg. Reasonable when IV is unavailable or the patient reacted to IV.
When to treat
Scenario
Threshold
Acute single ingestion, level drawn 4-24 h
Above the Rumack-Matthew treatment line (150 mcg/mL at 4 h, declining with a 4-hour half-life).
High-risk ingestion
≥30 g ingested, or level above the 300 mcg/mL at 4 h line. Many poison centers scale the third-bag rate 2-4× by nomogram line; note that this is common practice rather than a formal 2023 consensus recommendation, and the benefit in patients treated early remains unproven.
Repeated supratherapeutic ingestion
APAP >20 mcg/mL or elevated transaminases. The nomogram does not apply.
Co-ingested opioid or anticholinergic
If the first level at 4-24 h is ≤10 mcg/mL, NAC is not needed. Otherwise absorption may be delayed; repeat the level.
During hemodialysis
NAC is dialyzable. Increase the IV infusion rate to at least 12.5 mg/kg/h during dialysis (double the standard third-bag rate). Oral dosing needs no adjustment.
Stopping criteria (all four must be met)
1
APAP <10 mcg/mL
2
INR <2.0
3
AST/ALT normal, at baseline, or down 25-50% from peak
4
Patient clinically well and improving
⚠️
The named error: the consensus statement specifically calls out stopping NAC at 20-21 hours without reassessment as a common clinical mistake. The infusion ends when the criteria are met, not when the bag does.
💡
Anaphylactoid reactions are not IgE-mediated and are inversely related to APAP concentration: roughly 25% incidence when APAP is <150 mcg/mL versus ~3% when >300 mcg/mL. Flushing alone needs nothing. Urticaria gets diphenhydramine with the infusion running. Only bronchospasm or hypotension requires holding the drug. A prior reaction is not an absolute contraindication to re-treatment.
💡
Fomepizole is an emerging adjunct in high-risk acetaminophen overdose (CYP2E1 and JNK inhibition), typically 15 mg/kg IV then 10 mg/kg q12h. The evidence is case-series level, it is not in the 2023 consensus, and it is never a substitute for NAC.
The 2024 expert consensus (Am J Med) shifted several long-taught thresholds. Most notably, the potassium trigger moved. Note that the panel took a neutral position on Fab dosing itself and deferred to the product labeling, so the vial counts below come from the DigiFab prescribing information rather than from the consensus.
Question
Answer
Empiric dose, acute, unknown amount
20 vials (may give 10, reassess, then 10 more). Cardiac arrest: 20 vials by bolus.
Empiric dose, chronic toxicity
6 vials in adults and children ≥20 kg. Titrated low-dose strategies (1-2 vials, reassess at 60 min) are used in stable chronic toxicity in some centers.
Calculated from level
vials = (serum digoxin ng/mL × weight kg) ÷ 100. Valid only on a post-distribution level (≥6-8 h post-ingestion); an early level wildly overestimates the vial count.
Calculated from amount ingested
vials = mg ingested ÷ 0.5 (current DigiFab labeling). The older Digibind formula multiplied by 0.8 for tablet bioavailability; capsules and IV digoxin are ~100% bioavailable and take no such factor.
Potassium trigger
K+ ≥6 mEq/L is the strongly endorsed 2024 threshold. The panel reached no consensus at 5.5 mEq/L, which is the number most of us were taught.
Level as a sole trigger
Weakly endorsed only above 4 ng/mL. The panel explicitly stated there is no consistent relationship between serum concentration and clinical effect.
Bradyarrhythmia
Treat antidotally with Fab rather than placing a temporary transvenous pacemaker (strong recommendation).
Expected response
Rhythm and conduction improve in 30-45 min; 50-90% improve within an hour. Potassium normalizes over 2-6 h; watch for rebound hypokalemia.
⚠️
Post-Fab digoxin levels are a trap. Fab pulls digoxin out of tissue into plasma as an inactive complex, so total digoxin (what most assays measure) rises many-fold within minutes while free (active) digoxin plummets. The assay stays uninterpretable for several days with normal kidneys and a week or more in renal impairment. Titrate to rhythm, symptoms, and potassium instead.
⚠️
Rebound. (See also calcium in digoxin-associated hyperkalemia in the Danger Zone.) Free digoxin can rebound at 12-24 h with normal renal function, and 24-96 h in renal dysfunction. Dialysis removes neither digoxin nor the Fab complex. Monitor well beyond the infusion in renal failure.
15 mg/kg IV load, then 10 mg/kg q12h × 4 doses, then 15 mg/kg q12h thereafter. Dilute in ≥100 mL, infuse over 30 min.
Why the dose increases at 48 h
Fomepizole induces its own metabolism via CYP2E1; clearance rises after roughly 30-40 h of therapy and the 50% bump keeps levels above the ADH-inhibitory threshold.
During hemodialysis
Dose q4h, or run 1-1.5 mg/kg/h. Fomepizole is dialyzable. At the end of HD, give 50% of the next dose if 1-3 h have passed, a full dose if >3 h.
When to start
Do not wait for a confirmatory level. Treat on a documented concentration >20 mg/dL, a credible ingestion history, an unexplained osmol gap, or an unexplained anion-gap acidosis (pH <7.3 or bicarbonate <20).
Cofactors
Methanol: folinic acid (leucovorin) 50-100 mg IV q4-6h (folic acid if unavailable). Ethylene glycol: thiamine 100 mg and pyridoxine 100 mg IV. Cheap, safe, mechanistically sound; dosing intervals vary by source.
When to stop
Concentration undetectable or <20 mg/dL, patient asymptomatic, and arterial pH normal. With ADH blocked and no dialysis, methanol’s half-life is ~45-54 h, so this can take days.
💡
The biggest practice change: for ethylene glycol in a patient who presents early with normal renal function, no significant acidosis, and no end-organ injury, fomepizole alone is sufficient and dialysis is not required, even at concentrations well above the old 50 mg/dL trigger. EXTRIP 2023 suggests dialysis above roughly 310 mg/dL when fomepizole is running. Concentration is not the only trigger, though: EXTRIP recommends dialysis regardless of antidote for anion gap >27 mmol/L, glycolate >12 mmol/L, coma, or seizures, and suggests it for anion gap 23-27 or glycolate 8-12. Many institutional order sets still carry the older, far more aggressive threshold. Methanol is different: EXTRIP recommends dialysis above 70 mg/dL on fomepizole, or for coma, seizures, new visual deficits, pH ≤7.15, or anion gap >24.
💡
Vial solidifies below 25°C. That is expected, not a defect. Warm it in your hand or under running water. Pharmacists get called about this.
Respiratory depression, known or suspected opioid dependence
0.04-0.1 mg IV, then double every 2-3 min (0.04 → 0.08 → 0.16 → 0.4 → 2 → 4 → 10 mg)
Respiratory depression, not dependent or unknown
0.4 mg IV, escalate q2-3 min
Apnea or agonal respirations
0.4-2 mg IV. Bag-mask ventilate first and throughout.
Community / take-home
4 mg intranasal (or 8 mg, or 3 mg product), or 0.4-2 mg IM, repeat q2-3 min with rescue breathing between doses
No response at 10 mg cumulative
Reconsider the diagnosis. Think inadequate ventilation, a co-ingested sedative, xylazine or medetomidine, hypoxic brain injury, or a non-opioid cause.
Continuous infusion: the two-thirds rule
1
Identify the cumulative bolus dose that restored adequate ventilation (the waking dose).
2
Run the infusion at two thirds of that dose per hour. Waking dose 0.8 mg total → run 0.5 mg/h.
3
Because steady state takes 30-60 min, many toxicologists give a bolus of half the waking dose about 15 min after starting the drip to cover the trough.
4
Titrate up by 0.1-0.2 mg/h and rebolus if depression recurs. Wean by 0.1-0.2 mg/h every 2 h.
5
Observe at least 12 h after the infusion is stopped. Indicated for anyone who needed repeat boluses, or who took methadone, an ER/LA opioid, or buprenorphine.
💡
Higher first doses for fentanyl are not supported. Pooled data show a cumulative 4 mg was sufficient in 97% of presumed fentanyl overdoses, and multiple poison centers and ACMT have concluded that standard dosing reverses fentanyl analogs. What the literature actually shows is that clinicians have used more, not that more works better. A higher starting dose buys precipitated withdrawal, which is not benign: noncardiogenic pulmonary edema is a recognized complication.
💡
Xylazine does not respond to naloxone and there is no human reversal agent. Give naloxone anyway; the dangerous respiratory depression in a fentanyl-xylazine overdose is the fentanyl. Expect persistent sedation afterward and treat it supportively. Do not read it as naloxone failure and keep escalating.
💡
Buprenorphine is the exception that needs more. Its receptor dissociation half-life is roughly 40 minutes, so reversal requires a bolus followed immediately by an infusion and full effect may take 30 minutes. Ventilate and be patient rather than declaring failure early.
Total units infused over the preceding 2-2.5 h, then 1 mg per 100 units. Example: 1200 units/h × 2 h = 2400 units → 24 mg. The rate drives the dose, not the 24-hour total.
Enoxaparin ≤8 h
1 mg protamine per 1 mg enoxaparin
Enoxaparin >8 h
0.5 mg per 1 mg. At ≥12 h protamine may not be required.
Enoxaparin, still bleeding at 2-4 h
Additional 0.5 mg per 1 mg
Dalteparin
1 mg per 100 anti-Xa units if ≤8 h; 0.5 mg per 100 units at 8-12 h
Fondaparinux
Protamine does not work. No approved reversal agent. Off-label options are aPCC (FEIBA) ~20 units/kg or rFVIIa 90 mcg/kg, both on low-quality evidence.
Hard limits
Maximum single dose 50 mg. Maximum rate 5 mg/min (50 mg over at least 10 min). Rate is the single biggest modifiable driver of hypotension and anaphylactoid reaction.
⚠️
Never round up. Protamine given in excess is itself an anticoagulant: free protamine impairs platelet function, thrombin generation, and fibrin polymerization. In a bleeding patient, overdosing looks like “more reversal” and produces more bleeding. Under-dose, recheck aPTT or anti-Xa at 10-15 min, and redose.
💡
Sort the allergy risk factors by evidence.NPH insulin use and prior protamine exposure are the best supported: a substantial minority of NPH-treated patients develop anti-protamine antibodies, and antibody-positive patients have a markedly higher reaction rate. Fish allergy is the weak one; protamine is a fish sperm nuclear protein, not the muscle parvalbumin that causes food allergy, and cross-reactivity has not been established, though isolated case reports exist. All four risk factors are label-listed and worth documenting, but fish allergy alone should not withhold a life-saving dose. Slow the infusion, have resuscitation ready, consider a 5-10 mg test dose.
💡
Watch for rebound. Protamine’s half-life is about 10 minutes, far shorter than depot heparin absorbed from a subcutaneous site. Recheck at 10-15 min and then q4h for up to 24 h, and expect to redose after subcutaneous UFH or LMWH.
Flumazenil in undifferentiated overdose. Real-world and trial data put the seizure rate around 1.3-1.4%, and most of those seizures occurred in patients with a proconvulsant co-ingestion. Meta-analysis of 13 randomized trials found serious adverse events in 2.4% versus 0.4% with placebo (RR 3.81, 95% CI 1.28-11.39, number needed to harm 50) — note the wide interval; this rests on 14 events across 994 patients. Two mechanisms: precipitated withdrawal in a dependent patient, and unmasking a co-ingested proconvulsant whose seizures are now harder to treat because you have just blocked the GABA-A site. Isolated oral benzodiazepine overdose has very low mortality with airway support alone, so the antidote converts a low-risk problem into an iatrogenic one.
Absolute contraindications: any sign of cyclic antidepressant overdose (anticholinergic toxidrome, wide QRS, terminal R in aVR), chronic benzodiazepine dependence, known seizure disorder, benzodiazepines being given for a life-threatening indication such as status epilepticus or raised ICP.
Where it is defensible: reversal of iatrogenic procedural sedation, and accidental single-agent ingestion in a benzodiazepine-naive patient with respiratory insufficiency. The two settings have different labeled ceilings and are easy to conflate: • Procedural sedation reversal: 0.2 mg IV over 15 s, then 0.2 mg at 1-minute intervals, cumulative maximum 1 mg (3 mg in any hour). • Suspected benzodiazepine overdose: 0.2 mg IV over 30 s, then 0.3 mg, then 0.5 mg at 1-minute intervals, usual cumulative maximum 3 mg; no response by 5 mg means the sedation is not benzodiazepine-mediated. Call poison control before either.
And never discharge on a flumazenil response. Its clinical duration is 20-50 minutes against diazepam’s multi-day active metabolites, and resedation typically appears at roughly 60-120 minutes.
In fairness: this is not entirely settled. A 2025 retrospective poison-center analysis argues the real-world adverse event rate is low, and drew a published rebuttal. The restrictive position above remains the mainstream one and is what the randomized data support.
☠️
Calcium in digoxin-associated hyperkalemia. The 2024 consensus strongly states that IV calcium is not helpful here, and weakly that it may be harmful. Worth knowing that the classic “stone heart” teaching is not well supported: the animal data required severe pre-existing hypercalcemia, and a 159-patient review found no difference in mortality and no dysrhythmia within four hours of calcium. The practical conclusion is the same either way. Fab is the treatment for digoxin-associated hyperkalemia, not calcium, and reaching for calcium wastes the minutes that matter.
☠️
Methylene blue in G6PD deficiency and in serotonergic patients. It requires NADPH from G6PD to work, so it may not work in G6PD deficiency and can precipitate hemolysis; G6PD deficiency is a labeled contraindication, and giving it anyway in a life-threatening case is a deliberate off-label decision. It also carries a boxed warning for serotonin syndrome as a potent MAO-A inhibitor. In life-threatening methemoglobinemia do not withhold it, but expect failure in G6PD deficiency and plan for exchange transfusion.
☠️
Lipid emulsion outside LAST. The AACT-convened Lipid Emulsion Workgroup (2016) suggests against lipid emulsion as first-line therapy for most non-local-anesthetic toxins, and was neutral rather than supportive for most of the rest; bupivacaine cardiac arrest is the one indication carrying a clear recommendation for it. ACMT’s separate 2016 position statement is somewhat more permissive in cardiotoxic lipophilic overdose, so do not present the two societies as speaking with one voice. Practical caution: lipemia renders chemistry and coagulation samples unusable for hours and clogs ECMO and CRRT circuits. Reserve it as true rescue after standard therapy has failed, and warn the lab before you give it.
Anticoagulant Reversal Beyond Protamine
Warfarin (vitamin K plus 4-factor PCC), dabigatran (idarucizumab), and the factor Xa inhibitors each carry enough nuance to deserve their own treatment. Note that andexanet alfa was voluntarily withdrawn from the US market effective December 22, 2025, on the basis of postmarketing thromboembolic event data; ANNEXA-I met its hemostatic efficacy endpoint but showed more thrombosis than usual care. That leaves 4-factor PCC as the practical option for factor Xa inhibitor reversal in the US. It remains approved outside the US as Ondexxya. See our Direct Oral Anticoagulants reference and the DOAC Quick Card.