Choosing an anticoagulant comes down to a few axes: what it inhibits (thrombin vs factor Xa vs vitamin-K-dependent synthesis), how fast it turns on and off (half-life), how much depends on the kidney, whether it needs monitoring, and whether it can be reversed. The renal-clearance column is the one most likely to change a decision in CKD — dabigatran is the most renally cleared (80%); apixaban the least (27%).

First branch point is always life-threatening vs not. For non-major bleeding: hold the drug, apply local hemostasis, support — most anticoagulants clear on their own given their half-lives. Reverse only for life-threatening or critical-site bleeding (intracranial, intraspinal, pericardial, airway) or an emergent procedure. Always loop in pharmacy and the relevant specialist.

Warfarin — Vitamin K (IV for urgent reversal) plus 4F-PCC (KCentra) for severe bleeding or emergent procedure; vitamin K is always co-given because PCC factors are short-lived while warfarin's effect persists.

Factor Xa inhibitors (apixaban / rivaroxaban / edoxaban) — for severe/life-threatening bleeding with clinically significant anticoagulant effect, 4F-PCC 2,000 units IV once is an off-label U.S. guideline option. FDA concluded Andexxa's risks outweigh its benefits; U.S. commercial sales ended December 22, 2025.

Dabigatran — Idarucizumab (Praxbind) 5 g IV as two consecutive 2.5-g vials is the specific labeled antidote. If unavailable and urgent reversal is required: PCC or activated PCC 50 units/kg IV once (max 4,000 units) is an off-label U.S. guideline fallback. Limited labeling data show about 49% dialysis removal over 4 hours at 200 mL/min blood flow, followed by redistribution.

UFH / LMWH — Protamine is given slowly, with no more than 50 mg over any 10-minute period. It neutralizes UFH, but enoxaparin anti-factor-Xa activity is never completely neutralized (maximum about 60%). Fondaparinux has no specific reversal agent — use specialist-directed supportive and hemostatic management for life-threatening bleeding.

KCentra contains heparin (added to blunt its own clotting risk), so a HIT history is a labeled contraindication — along with DIC and anaphylaxis to its components. About the timing: HIT antibodies are transient and usually undetectable by ~40–100 days, and HIT has no true memory response — but the calendar alone doesn't clear a patient. "Remote HIT" that tolerates heparin means recovered platelets plus a negative PF4 antibody (ideally a negative functional assay too), and that re-exposure evidence comes from planned cardiac/vascular surgery — not an emergent bleed where there's no time to test. So if there's a HIT history and you can't confirm antibodies have cleared, treat KCentra as contraindicated and reach for a heparin-free option:

FEIBA (activated PCC) is the usual fallback in HIT because it's heparin-free; it runs a higher thrombotic risk at higher doses, and you should not dose it to normalize the aPTT (that can drive DIC). 3F-PCC (e.g. Profilnine, heparin-free) works but is light on factor VII, so it's paired with a VII source (low-dose rFVIIa or a little FFP) for warfarin. rFVIIa (NovoSeven) is a refractory/last-resort adjunct — high thrombotic risk and it doesn't replace the other factors. FFP reverses too but needs large volumes and acts slowly (volume overload, slow INR drop) — reserve it for when no PCC is available. Vitamin K is still given with any warfarin reversal.

Platelets: give them for genuine thrombocytopenia or before neurosurgery — but don't reflexively transfuse for spontaneous antiplatelet-associated ICH. The PATCH trial found platelet transfusion worsened death/dependence (adjusted OR ~2). Calcium: blood products are citrated, and citrate chelates calcium (and magnesium). Large-volume FFP/platelet/PRBC transfusion drops ionized calcium — and calcium (factor IV) is required for the clotting cascade and platelet function, so hypocalcemia is itself a coagulopathy. Check ionized calcium and replace it (calcium gluconate or chloride) during massive transfusion.

DOACs are designed not to need routine monitoring — fixed dosing, predictable PK. Levels matter only in edge cases (bleeding, urgent surgery, extremes of weight, renal failure, suspected non-adherence) and require drug-specific, calibrated assays, not PT/aPTT.

The anti-Xa assay is chromogenic — the lab adds a fixed amount of factor Xa plus a substrate that turns the sample a color when Xa cleaves it. Heparin (through antithrombin) inhibits some of that Xa, so less color = more heparin, read off a curve calibrated to heparin. The catch: a direct Xa inhibitor (apixaban, rivaroxaban, edoxaban) also inhibits that Xa and dims the color — but the analyzer can't tell which drug did it, so it attributes the whole effect to heparin → a falsely high heparin anti-Xa. The error can be several-fold, and once a DOAC is on board the heparin level is essentially meaningless — you can't titrate to it. Interference usually lasts 48–72 h, sometimes ≥ 96 h with interacting drugs or renal impairment.

Practical move: if the patient was recently on a Xa-inhibitor DOAC, draw a baseline anti-Xa before starting heparin. If it's already elevated, the assay is contaminated — switch to aPTT monitoring until the DOAC clears.

Second trap: many anti-Xa reagents add exogenous antithrombin, which can mask antithrombin-deficiency heparin resistance — the level reads therapeutic while the patient is actually under-anticoagulated. Suspect it when infusion rates run very high (> 25–30 units/kg/hr) against a flat aPTT.

What it is: heparin-induced thrombocytopenia is an immune reaction — antibodies against the heparin–platelet factor 4 (PF4) complex bind and activate platelets via their Fc receptors, driving a prothrombotic state. Counterintuitively, the danger is thrombosis (arterial or venous), not bleeding, despite the falling platelet count.

Timing: platelets typically fall 5–10 days after heparin starts, or within ~1 day if there was heparin exposure in the prior ~100 days (anamnestic). Far more common with UFH than LMWH.

4T logic: the score weighs Thrombocytopenia magnitude, Timing, Thrombosis/sequelae, and whether oTher causes exist. Low score (≤3) has a high negative predictive value — pursue other causes before committing to a HIT workup. Use the 4T calculator in the Heparin/DTI tab.

What to do: stop all heparin sources (flushes, line locks, LMWH) and start a non-heparin anticoagulant — a direct thrombin inhibitor (argatroban, bivalirudin), fondaparinux, or a DOAC (off-label). Avoid prophylactic platelet transfusions. Do not start warfarin until platelets recover (≥ 150); risk of venous limb gangrene; overlap when transitioning. Send a PF4 ELISA, confirm with a functional assay (SRA).

Sick / unstable, bleeding risk, or a procedure may be needed → DTI infusion. Argatroban or bivalirudin are IV, titratable, and clear fast, so you can turn them off. Argatroban for renal failure (hepatic clearance); bivalirudin if hepatic impairment or short offset is the priority.

Clinically stable, average bleed risk, no limb/life-threatening clot → fondaparinux or a DOAC. Once-daily SC (fonda) or oral, no routine monitoring, outpatient-friendly. The trade-off is a long offset and no reversal — fondaparinux is off-label for HIT and contraindicated in severe renal failure (renally cleared). Avoid it in unstable patients or weight < 50 kg.

Bridging is mostly a warfarin question. DOACs have short half-lives, so the drug clears on its own during a brief hold — bridging is not needed. For warfarin, bridge with LMWH/UFH only when thromboembolic risk is high (recent stroke/TIA < 3 months, mechanical mitral valve, CHA₂DS₂-VASc very high). The BRIDGE trial showed no-bridge was non-inferior for stroke with less bleeding in most AF patients.

DOAC hold by bleed risk + renal function: minimal-bleed-risk procedures generally don't require interruption; high-bleed-risk procedures hold ≥ 2 days (4–5 half-lives), longer for dabigatran as CrCl falls. Use the Periop tab's hold calendar for exact dates.

A measured drug level under ~30 ng/mL can permit earlier neuraxial placement when timing is unavoidable. Coordinate with the proceduralist and anesthesiology.

The governing idea: never leave a gap in coverage and never stack two anticoagulants longer than necessary. Warfarin's slow on/off makes its transitions the fiddly ones; DOACs and parenterals switch cleanly on timing.

Why warfarin needs a bridge (sometimes). Warfarin doesn't work for days — it lowers factors II/VII/IX/X only as the existing ones decay, and it transiently lowers protein C/S first, so the first ~5 days can be net pro-thrombotic. When you stop a fast anticoagulant and start warfarin, there's a coverage gap until the INR is therapeutic. A parenteral "bridge" (LMWH/UFH) fills that gap — but only worth the bleeding cost when clot risk is high (mechanical mitral valve, recent VTE/stroke). For most AF patients, no-bridge won out (BRIDGE trial: similar stroke, less bleeding).

Why dabigatran (and the other DOACs) need overlap going to warfarin. Warfarin takes days to become effective, so you must keep the DOAC running while warfarin loads — but because dabigatran is renally cleared, the overlap window is set by kidney function: start warfarin 3 days before stopping if CrCl ≥ 50, 2 days if 30–49, 1 day if 15–30. Note dabigatran itself prolongs the INR, so the INR isn't reliable until ~2 days after it's stopped.

The clean direction. Going from warfarin to a DOAC, or between parenteral and DOAC, is just timing — no bridge — because the DOAC turns on within hours. Stop warfarin, start the DOAC once the INR has drifted down into each agent's threshold (above).

Vitamin K only fixes a problem caused by lack of vitamin K. Give it for warfarin reversal, or for a true deficiency — not reflexively for every high INR. The trap in the hospital is giving it for an elevated INR that is actually liver synthetic failure, where it does nothing.

Suspect vitamin K deficiency when the INR is up and the patient has a reason to be depleted: NPO / poor intake for several days, malnutrition (common in long-stay ICU), prolonged antibiotics (wipe out gut flora that make vitamin K), or fat malabsorption / cholestasis (celiac, short gut, CF, biliary obstruction, bariatric). Vitamin K is fat-soluble, so anything blocking fat absorption blocks it. You usually don't need a level — the risk-factor history plus a vitamin-K-responsive INR is the practical diagnosis.

Factor V. It's made by the liver but is not vitamin-K-dependent. So: normal factor V + low factor VII (or just a vitamin-K-responsive INR) → vitamin K deficiency (give it, it'll work). Low factor V + low VII → hepatic synthetic failure (vitamin K won't correct it; the substrate isn't the problem). Low albumin and high bilirubin point the same way — toward liver.

The cheapest test is a trial. Give vitamin K and recheck the INR at 12–24 h — meaningful correction confirms deficiency (or warfarin effect); no movement means it's hepatic. Especially handy in a sick patient who has both malnutrition and liver dysfunction.

In cirrhosis, most elevated INRs are not vitamin K deficiency — so don't chase the number. The diseased liver runs a "rebalanced" hemostasis: it makes less of both the pro-coagulant and the anticoagulant factors, so a high INR mostly reflects disease severity (it's a MELD input) rather than net bleeding risk. Routine or repeated high-dose phytonadione barely moves the INR and doesn't reduce bleeding. Reserve it for a genuine deficiency reason (cholestasis, malnutrition, prolonged antibiotics) or actual warfarin reversal.

2.5–10 mg covers almost everything — higher doses rarely reverse any faster, and > 10 mg saturates hepatic stores → warfarin resistance for days to ~a week (you'll struggle to re-anticoagulate). Match route to urgency: oral is reliable when you can wait; IV is fastest (measurable INR drop within a few hours) but carries a small anaphylactoid risk, so give it diluted and slow as an IVPB, never a push. Skip SC and IM — subcutaneous absorption is erratic and IM risks a hematoma in an anticoagulated patient. All routes reach full effect by ~24 h.

Most medical inpatients with reduced mobility get pharmacologic prophylaxis unless bleeding risk is high. Default is enoxaparin 40 mg SC daily (or 30 mg q12h in trauma/ortho), UFH 5,000 units SC q8–12h when renal function is poor (CrCl < 30) or rapid reversibility matters, or fondaparinux if HIT history. Mechanical (SCDs) when anticoagulation is contraindicated. Prophylaxis doses are fixed and low — they are not treatment doses and don't need monitoring.

Different problem: antiplatelets target arterial, platelet-rich clot (coronary, cerebrovascular, peripheral), anticoagulants target venous/stasis, fibrin-rich clot (VTE, AF cardioembolism). They are not interchangeable, and combining them multiplies bleeding risk — only do it with a clear indication and a planned duration.

Aspirin + a P2Y12 inhibitor after ACS or stent — typically ~12 months, shortened when bleeding risk is high, extended when ischemic risk dominates. Ticagrelor or prasugrel are preferred over clopidogrel in ACS; clopidogrel is preferred for stable/elective PCI. If the patient also needs an OAC (e.g. AF + stent): drop aspirin early (often by discharge) and run OAC + clopidogrel — triple therapy only briefly, because stacking three agents bleeds.

These are therapeutic (treatment) doses — distinct from the fixed low prophylaxis doses. In the US the two agents you'll actually reach for are enoxaparin (the default) and dalteparin (cancer VTE); tinzaparin isn't marketed here.

Enoxaparin is renally cleared, so it accumulates and bleeds when kidneys fail. At CrCl < 30, reduce treatment dosing to 1 mg/kg once daily and consider anti-Xa monitoring — or, often better, switch to a UFH infusion (not renally cleared, fully reversible; one QI series cut major bleeding from ~13% to ~4% doing exactly this). Dalteparin bioaccumulates less and is the more defensible LMWH if you must use one in renal impairment or the elderly. Fondaparinux is a hard stop below CrCl 30.

Enoxaparin — default for VTE, ACS, and warfarin/DOAC bridging: familiar, stocked everywhere, flexible (q12h or daily). Dalteparin — cancer-associated VTE (the CLOT weight-based regimen) and a reasonable renal/elderly choice. Fondaparinux — when you need a HIT-safe, no-monitoring, once-daily agent and renal function is fine. UFH infusion — severe renal failure, high bleeding risk, or when a procedure may be imminent (rapid off, reversible).

DOACs (apixaban, edoxaban, rivaroxaban) are now first-line for most cancer-associated VTE. Keep LMWH (dalteparin) when there's a luminal GI or GU cancer (DOACs — especially edoxaban/rivaroxaban — bleed more there), significant thrombocytopenia, vomiting/absorption problems, or major drug interactions with anticancer therapy.

Most patients need none. Check a peak anti-Xa 4 h after the 3rd–4th dose in the edge cases: severe renal impairment on prolonged therapy, extremes of weight, and pregnancy. Rough peaks — q12h dosing ~0.6–1.0 IU/mL, once-daily >1.0. (ASH 2018 actually leans toward dose-adjust or switch over routine monitoring, since levels are often slow to come back.)