Extended Anticoagulation in Unprovoked VTE

Bottom line first

After 3 to 6 months of standard anticoagulation for a first unprovoked VTE, the question is no longer "DOAC or warfarin?". It is "continue or stop?". Two large RCTs answered the first half of the question:

Apixaban 2.5 mg BD (AMPLIFY-EXT) reduced recurrence by about 7 percentage points versus placebo over 12 months with no increase in major bleeding.
Rivaroxaban 10 mg OD (EINSTEIN CHOICE) reduced recurrence by about 3 percentage points versus aspirin 100 mg with no increase in major bleeding.

For the second half, who might reasonably stop, clinical prediction tools can support the discussion, but they should not replace clinical judgement, bleeding-risk assessment, patient preference, or local thrombosis policy. HERDOO2 has the strongest prospective management validation for identifying low-risk women; DASH and Vienna provide additional recurrence-risk estimates but should be used cautiously. Each is shown below with an interactive calculator.

How to read this page

The two RCTs are presented first, each in the same structure: PICO, headline numbers, critical appraisal, bottom line. The risk stratification section follows, with a working calculator for HERDOO2 and DASH. The UK practice synthesis, teaching points, and references sit at the bottom.

1. AMPLIFY-EXT (2013) — apixaban vs placebo

Agnelli G et al. N Engl J Med 2013;368(8):699–708 · After 6–12 months of standard anticoagulation

PICO

P · 2,486 adults with confirmed VTE who had completed 6–12 months of anticoagulation and for whom the decision to continue was in equipoise

I · Apixaban 2.5 mg twice daily, or 5 mg twice daily, for 12 months

C · Placebo for 12 months

O · Symptomatic recurrent VTE or VTE-related death

Headline results

Outcome	Placebo (n=829)	Apixaban 2.5 mg BD (n=840)	Apixaban 5 mg BD (n=813)
Recurrent VTE or VTE death	73 (8.8%)	14 (1.7%)	14 (1.7%)
Absolute risk reduction vs placebo	—	7.2% (95% CI 5.0–9.3)	7.0% (95% CI 4.9–9.1)
P value vs placebo	—	< 0.001	< 0.001
Major bleeding	0.5%	0.2%	0.1%
Clinically relevant non-major bleeding	2.3%	3.0%	4.2%
All-cause death	1.7%	0.8%	0.5%

Critical appraisal

Strengths. Randomised, double-blind, phase III, 12-month treatment, hard outcome. Two apixaban doses were tested against placebo. Recurrence rates were very similar with the lower dose (2.5 mg BD, the thromboprophylactic dose) and the treatment dose (5 mg BD), with numerically less bleeding at the lower dose. Number needed to treat to prevent one recurrent VTE is about 14 over 12 months.

Limitations. The trial was not designed to compare the two apixaban doses against each other, so it does not establish that 2.5 mg BD and 5 mg BD are equivalent. "Clinical equipoise" was an investigator judgement, so the trial enrolled patients who, by definition, were not the highest-risk recurrence cases. The enrolled population included both provoked and unprovoked VTE; when applying the data to a first unprovoked VTE, baseline recurrence risk must still be judged individually. Generalisability to patients with very high recurrence risk (for example active cancer or severe thrombophilia) is limited.

Bottom line

For VTE patients in clinical equipoise after 6–12 months of anticoagulation, apixaban 2.5 mg BD for a further 12 months reduces recurrence substantially (from 8.8% to 1.7% in the trial) without increasing major bleeding. Both apixaban doses produced very similar recurrence rates. The 2.5 mg BD dose is the licensed extended-phase dose after completion of initial treatment and is generally preferred for extended secondary prevention when reduced-intensity therapy is appropriate.

1A · Phase III RCT

2. EINSTEIN CHOICE (2017) — rivaroxaban vs aspirin

Weitz JI et al. N Engl J Med 2017;376(13):1211–1222 · After 6–12 months of standard anticoagulation

PICO

P · 3,396 adults with VTE who had completed 6–12 months of anticoagulation, in equipoise for continued treatment

I · Rivaroxaban 20 mg once daily, or 10 mg once daily, for 12 months

C · Aspirin 100 mg once daily

O · Symptomatic recurrent fatal or non-fatal VTE

Headline results

Outcome	Aspirin (n=1131)	Rivaroxaban 20 mg (n=1107)	Rivaroxaban 10 mg (n=1127)
Recurrent VTE	50 (4.4%)	17 (1.5%)	13 (1.2%)
HR vs aspirin	—	0.34 (95% CI 0.20–0.59)	0.26 (95% CI 0.14–0.47)
P value vs aspirin	—	< 0.001	< 0.001
Major bleeding	0.3%	0.5%	0.4%
Clinically relevant non-major bleeding	1.8%	2.7%	2.0%

Critical appraisal

Strengths. Modern three-arm phase III design, tested head-to-head against aspirin (the historical default in some health systems). Both rivaroxaban 10 mg OD and 20 mg OD were superior to aspirin. The trial was not designed to establish superiority of 10 mg over 20 mg; the similar efficacy and low bleeding rate support 10 mg OD as an extended-phase option in suitable patients. Number needed to treat is about 33 over 12 months versus aspirin.

Limitations. No placebo arm. Aspirin is an active comparator with its own modest antithrombotic effect. The "equipoise" entry criterion again excludes the highest-risk patients. The trial population included patients with provoked and unprovoked VTE, so when applying the data to a first unprovoked VTE, baseline recurrence risk must still be judged individually. Follow-up was 12 months only; longer-term data come from registries and post-hoc analyses.

Bottom line

Rivaroxaban 10 mg OD reduces recurrent VTE around 3-fold compared with aspirin 100 mg, with bleeding rates only marginally higher. Aspirin is no longer the default for extended VTE prophylaxis in patients who can tolerate a DOAC.

1A · Phase III RCT

3. Risk stratification — three tools that support the decision

The trials above show that treatment works. They do not identify who needs it. Clinical prediction tools can support that judgement, alongside bleeding-risk assessment and patient preference. Three are in routine use:

HERDOO2: applies to women only. Identifies low-recurrence-risk women who can be considered for stopping.
DASH: applies to both sexes. Pools four predictors (D-dimer, Age, Sex, Hormones) into a score.
Vienna prediction model: uses sex, location of index VTE, and D-dimer to estimate a continuous recurrence probability.

A practical approach: complete at least 3 months of anticoagulation, then reassess recurrence risk, bleeding risk, treatment tolerance and patient preference. If using a prediction tool, follow the D-dimer timing used by that tool. HERDOO2 uses on-treatment D-dimer; DASH and Vienna use post-treatment D-dimer measurements.

HERDOO2 — women only

HERDOO2 (Rodger 2008, CMAJ) identifies women at low risk of VTE recurrence after a first unprovoked VTE. The rule does not apply to men, as no low-risk male subgroup could be defined.

Score 1 point for each of the following:

Hyperpigmentation, oedema, or redness of either leg
D-dimer ≥ 250 µg/L while still on anticoagulation
Obesity (BMI ≥ 30 kg/m²)
Older age (≥ 65 years)

The original derivation study measured D-dimer while patients were taking a vitamin K antagonist. The threshold has not been separately derived in patients treated with a DOAC.

Score 0 or 1: low risk. Annual recurrence was about 1.6% in the derivation study (95% CI 0.3 to 4.6).
Score 2 or more: high risk. Annual recurrence was about 14.1% (95% CI 10.9 to 17.3).

HERDOO2 was prospectively validated in a multinational cohort management study (Rodger 2017, BMJ). In that study, low-risk women who discontinued anticoagulation had recurrent VTE at about 3.0% per patient-year (95% CI 1.8 to 4.8). This is still generally considered low enough to support stopping anticoagulation in appropriately selected women after shared decision-making. Derivation figures such as the 1.6% tend to be more optimistic than prospective validation figures.

HERDOO2 — interactive score

For female patients with a first unprovoked VTE who have completed initial anticoagulation.

Select the criteria that apply, then calculate. A score of 0 is a valid result.

DASH — both sexes

DASH (Tosetto 2012, JTH) pools four predictors into a single score, derived from 1,818 patients with unprovoked VTE.

D-dimer abnormal after stopping anticoagulation → +2
Age < 50 years → +1
Sex (male) → +1
Hormone-associated VTE (in women) → −2

Score ≤ 1 → annual recurrence about 3.1% (95% CI 2.3–3.9)
Score 2 → annual recurrence about 6.4% (95% CI 4.8–7.9)
Score ≥ 3 → annual recurrence about 12.3% (95% CI 9.9–14.7)

ROC AUC 0.71. Applies to both sexes. About half of patients with a first unprovoked VTE fall into the score ≤ 1 group.

DASH — interactive score

Select the criteria that apply, then calculate. A score of 0 is a valid result.

Vienna prediction model — continuous risk estimate

Vienna prediction model (Eichinger 2010, Circulation) was derived from 929 patients followed for a median of 43 months after stopping anticoagulation. The model excludes patients with a strong thrombophilic defect (inhibitor deficiency, lupus anticoagulant, homozygous or combined defects).

Independent predictors of recurrence:

Male sex: HR 1.90 (95% CI 1.31–2.75) vs female
Proximal DVT: HR 2.08 (95% CI 1.16–3.74) vs distal
Pulmonary embolism: HR 2.60 (95% CI 1.49–4.53) vs distal DVT
D-dimer (per doubling): HR 1.27 (95% CI 1.08–1.51)

The original publication provides a nomogram. A web-based calculator implementing a later recalibration of the model is available at clinicalbiometrics.shinyapps.io/VPM_lowrisk. Because the online tool reflects an updated model, results may differ slightly from the 2010 nomogram. Unlike HERDOO2 (binary) and DASH (categorical), the Vienna model returns a continuous probability, which can help shared-decision conversations where the patient wants a specific number.

Comparison of the three risk tools

Tool	Population	Inputs	Output	Best used for
HERDOO2	Women only	4 binary items, includes D-dimer on AC	Low vs high risk (binary)	Quickest decision rule; strongest prospective management validation
DASH	Both sexes	4 items, includes D-dimer off AC	3 risk strata	Mixed-sex outpatient clinic
Vienna	Both sexes (no strong thrombophilia)	Sex, location, D-dimer	Continuous probability	Patients wanting a specific number, shared decision-making

What this means for UK practice

NICE NG158 recommends anticoagulation for at least 3 months for confirmed proximal DVT or PE. After 3 months, reassess the benefits and risks of continuing, stopping, or changing anticoagulation. For unprovoked proximal DVT or PE, continuing anticoagulation beyond 3 months should be considered, balancing recurrence risk, bleeding risk, treatment tolerance, renal function, comorbidity, patient preference, and local thrombosis policy.

A practical approach:

Confirm whether the event was provoked or unprovoked, and whether there are persistent risk factors.
Review bleeding risk, renal function, weight, drug interactions, adherence, and patient preference.
Bleeding risk: assess clinically and consider HAS-BLED, as recommended by NICE NG158, for people having anticoagulation for unprovoked proximal DVT or PE. NICE advises discussing stopping anticoagulation if the HAS-BLED score is 4 or more and cannot be modified. RIETE may be mentioned as an alternative specialist bleeding-risk model, but it should not be presented as the NICE-recommended tool.
If considering stopping after unprovoked VTE, do not rely solely on prediction tools. Use HERDOO2, DASH, or Vienna only as adjuncts, and follow the D-dimer timing used in the original model.
If extended anticoagulation is chosen and reduced-intensity DOAC is appropriate, options include apixaban 2.5 mg BD after 6 months of treatment, or rivaroxaban 10 mg OD after at least 6 months of treatment.
Review patients on long-term anticoagulation or aspirin at least annually.

A few practical notes:

Aspirin is not the default fallback. EINSTEIN CHOICE showed rivaroxaban 10 mg OD is roughly three times more effective than aspirin 100 mg with only marginally more bleeding.
Cancer-associated VTE is a different track. These patients usually continue full-dose anticoagulation while cancer is active (Caravaggio, Hokusai-VTE Cancer, SELECT-D). NICE NG158 gives specific guidance.
Provoked VTE (surgery, immobilisation, oestrogen) has a much lower recurrence risk and usually stops at 3 months. The tools above are for unprovoked events.
Annual review. Patients on indefinite anticoagulation should have an annual review of bleeding risk, renal function, weight, and adherence.
Document the conversation. Indefinite anticoagulation is a long-term commitment; the rationale, alternatives, and the patient's preference should be in the notes.

Teaching points

"Extended" anticoagulation does not mean "lifelong". It means "indefinite, reviewed annually." Patients who can stop one year often cannot another year if bleeding risk rises.
Lower-intensity DOAC dosing was a deliberate trial design, and it works. In the extended-phase trials, apixaban 2.5 mg BD and rivaroxaban 10 mg OD produced recurrence rates similar to the full-treatment doses, with numerically less bleeding. Neither trial was designed to compare the two doses head-to-head, so equivalence is not formally established. The reduced-intensity dose is the licensed extended-phase option after completion of initial treatment and is generally preferred when extended therapy is appropriate.
D-dimer is central to risk stratification. Both HERDOO2 (on treatment) and DASH/Vienna (off treatment) depend on it. A negative D-dimer 3–4 weeks after stopping is reassuring; a positive D-dimer should prompt reassessment.
Men with a first unprovoked VTE have a notably higher recurrence risk than women. This is a robust finding across HERDOO2, DASH, and Vienna. The decision to stop in a man with a first unprovoked PE should be made carefully; many will stay on indefinite anticoagulation.
Aspirin still has a role, but it is a small one. Some patients refuse or cannot tolerate a DOAC; aspirin reduces recurrence by about a third (WARFASA, ASPIRE) and is reasonable in that situation. It is no longer the first-line extended therapy.

Discussion questions

A 42-year-old man presents 6 months after a first unprovoked PE. He is on rivaroxaban 20 mg OD and asks whether he can stop. How would you frame the conversation? Which tool would you use?
A 67-year-old woman with a BMI of 32 and a recent left-leg unprovoked DVT has completed 6 months of apixaban. Her D-dimer on treatment is 180 µg/L and her legs are clinically normal. Walk through HERDOO2 step by step: which criteria does she meet, what is her score, and what would you recommend? (Hint: it is not as low-risk as the low D-dimer might suggest.)
Why does EINSTEIN CHOICE compare rivaroxaban with aspirin rather than placebo? What ethical and practical reasoning sits behind that design choice?

References

Bibliographic details verified against PubMed, May 2026.

Agnelli G, Buller HR, Cohen A, et al. Apixaban for extended treatment of venous thromboembolism. N Engl J Med. 2013;368(8):699–708. PMID: 23216615 · DOI: 10.1056/NEJMoa1207541
Weitz JI, Lensing AWA, Prins MH, et al. Rivaroxaban or aspirin for extended treatment of venous thromboembolism. N Engl J Med. 2017;376(13):1211–1222. PMID: 28316279 · DOI: 10.1056/NEJMoa1700518
Rodger MA, Kahn SR, Wells PS, et al. Identifying unprovoked thromboembolism patients at low risk for recurrence who can discontinue anticoagulant therapy. CMAJ. 2008;179(5):417–426. PMID: 18725614 · DOI: 10.1503/cmaj.080493
Rodger MA, Le Gal G, Anderson DR, et al. Validating the HERDOO2 rule to guide treatment duration for women with unprovoked venous thrombosis: multinational prospective cohort management study. BMJ. 2017;356:j1065. PMID: 28314711 · DOI: 10.1136/bmj.j1065
Tosetto A, Iorio A, Marcucci M, et al. Predicting disease recurrence in patients with previous unprovoked venous thromboembolism: a proposed prediction score (DASH). J Thromb Haemost. 2012;10(6):1019–1025. PMID: 22489957 · DOI: 10.1111/j.1538-7836.2012.04735.x
Eichinger S, Heinze G, Jandeck LM, Kyrle PA. Risk assessment of recurrence in patients with unprovoked deep vein thrombosis or pulmonary embolism: the Vienna prediction model. Circulation. 2010;121(14):1630–1636. PMID: 20351233 · DOI: 10.1161/CIRCULATIONAHA.109.925214