*Cost-effectiveness of tailoring anticoagulant therapy by a VTE recurrence prediction model in patients with venous thrombo-embolism as compared to care-as-usual: the VISTA study.*

INTRODUCTION AND RATIONALE
Treatment of venous thrombo-embolism (VTE) * this is deep vein thrombosis (DVT)
and pulmonary embolism (PE) * with vitamin K antagonists (VKA) is highly
effective in reducing morbidity and mortality.[1,2] It reduces mortality caused
by PE from 30 to 8%.[3,4] The initial treatment of VTE consists of at least 5
days of heparin (unfractionated or low-molecular-weight) overlapped with 3-6
months VKA (acenocoumarol, phenprocoumon or warfarin). Long-term treatment
after initial therapy reduces the risk of recurrent VTE by over 90%.[5,7]
However, after VKA discontinuation there is considerable risk of VTE
recurrence, about 9% after one year.[2,5] VTE recurrence has a case-fatality
rate of 5%.[6] VTE recurrence is highest immediately after VKA therapy
discontinuation, and stabilizes in the subsequent 9 months.[7] Long term
treatment with VKA reduces the risk of VTE recurrence to 1% a year, but at a
cost of 2% increasing risk of major bleedings.[8] Given the challenge to obtain
(substantial) reduction in recurrent VTE risk versus increased risk of
bleedings, there is much discussion about the optimal duration of VKA treatment
in patients with VTE.[8,9] National and international guidelines on this topic
suggest different strategies. The guideline from the 8th ACCP advices
indefinite anticoagulant therapy for patients with a first unprovoked proximal
DVT or PE and a low risk of bleeding.[9] The Dutch CBO consensus recommends for
the same patients 6 months VKA therapy and indefinite when recurrence
occurs.[5] However, there is limited evidence to ground these strategies.
Caregivers and patients wrestle continuously with this question resulting in
much uncertainty and widely varying durations of VKA therapy in daily
practice.[10,11]
Observational studies show in patients with a first episode of unprovoked VTE,
that a positive D-dimer test after VKA treatment indicates increased VTE
recurrence risk [12-21,23]. This finding suggests that D-dimer testing can be
used to determine in whom VKA prolongation seems beneficial. Hence, D-dimer
guided VKA treatment strategy could minimize VTE recurrence (in whom longer VKA
treatment is needed) and bleeding risk (in whom VKA treatment can be stopped).
More recently, Eichinger and colleagues developed a prediction model that
accurately predicts the risk of VTE recurrence in patients with a first episode
of unprovoked VTE [22]. This prediction model was obtained by analysing several
widely accepted factors related to VTE recurrence (Age, sex, location of VTE,
BMI, factor V Leiden mutation, D-dimer levels and thrombin levels) in a cohort
consisting of 929 VTE patients with a recurrence rate of 18.9% during a
follow-up period of 43 months after VKA withdrawal. Only male sex, localisation
(proximal DVT or PE) and elevated D-dimer levers were significantly associated
with recurrence risk using a Cox proportional hazards model. Therefore, these
factors were incorporated into the prediction model. The definition of a high
recurrence risk if the risk exceeds 5% in 1 year, or 20% in 5 years, is based
on widely accepted experts* consensus [24].
If for example * based on this model * the predicted VTE recurrence after
stopping VKA treatment rate is high, patients may benefit from prolonged VKA
treatment to prevent morbidity or even mortality that is associated with these
VTE recurrences. If on the other hand the predicted VTE recurrence rate is low,
patients can safely stop VKA treatment as possible VKA treatment complications
* including (major) bleedings * no longer outweigh the benefits of long-term
VKA treatment in these patients.
Such a *VTE recurrence prediction strategy* could increase the
cost-effectiveness of VKA therapy in patients with VTE. Moreover, it gives
physicians a more objective tool to tailor optimal duration of VKA therapy.
This strategy of formal risk assessment has never been compared to
care-as-usual in a randomized clinical trial (RCT). Only one small RCT exists
on a D-dimer guided duration of VKA treatment. This RCT randomized VTE patients
with high D-dimer levels at 3 months, to either stopping (control group) or
continuing (index group) VKA treatment.[16] Patients with elevated D-dimer
level at 3 months of treatment indeed benefit from prolonged VKA therapy: VTE
recurrence over on average 18 months was 15.0% in the control group (with no
bleedings) versus 2.9% (with 1% bleedings) in the index group. As it was the
first study in this area, it only included a small group of patients with
unprovoked VTE. Moreover, it only randomized patients with an elevated D-dimer
level. Hence, it remains unclear whether an even more accurate prediction model
that also includes other patient characteristics and D-dimer testing has an
improved overall effectiveness compared to care-as-usual with duration of VKA
treatment to the physicians* discretion, let alone whether it is
cost-effective. Prompted by the promising results of this small trial, the less
straightforward clinical guidelines resulting in the large uncertainty of
physicians when to stop VKA treatment, the resulting practice variation, we
propose this study of evaluating the existing guideline with a formal risk
assessment.

OBJECTIVES
The objective is to quantify the cost-effectiveness and efficacy of formally
applying the guideline evaluating the risk-benefit ratio of the duration of
anticoagulant therapy (VKA treatment) by a previously developed VTE recurrence
prediction model (including type of VTE, gender and D-Dimer testing [22]) in
patients with a (first) episode of unprovoked VTE as compared to care-as-usual
where the duration of VKA treatment is conform current guidelines and
physicians* discretion.
The tailoring of the (dis)continuation of VKA treatment will be done after 6
months of initial VKA treatment. The effectiveness will be expressed by
comparing the cumulative incidences in recurrent VTE and bleeding complications
over 24 months of follow-up.

STUDY DESIGN
This is a pragmatic randomised multi-center trial in patients with a documented
(first) episode of unprovoked DVT or PE. After presentation, informed consent
is obtained. 5 months after the initial VTE event, patients are randomised to
the index group (prediction model guided treatment of VKA, see flowchart below)
or to the control group (care-as-usual, see flowchart below). The blocked
randomisation will be performed within strata for centre and for type of VTE
(DVT or PE).
Patients randomized to the index group that have a high risk of VTE recurrence
after an initial 6 months of VKA treatment continue VKA treatment during the
entire follow-up period of 24 months. A priori, we define a high risk of
recurrence as > 5% in the first year or > 20% in the first five years (based on
the prediction model [22] and experts* consensus [24]).
Patients randomized to the index group that (according to the prediction model)
have a low risk of VTE recurrence after stopping VKA treatment (i.e. < 5% in
the first year or < 20% in the first five years) stop long-term VKA treatment
after a standard initial treatment period of 6 months. After these 6 months,
patients are followed-up for 24 months.
Treating physicians however can overrule this treatment decision if the risk of
(major) bleeding during long-term VKA treatment is considered too high. This
(usually) will be the case in patients with either one or more of the following
characteristics: history of non reversible bleeding during VKA treatment,
history of persistently poor anticoagulant control, chronic renal or hepatic
failure, history of non cardio-embolic stroke, patient older than 75 years when
diagnosed with VTE or if an informed patient wants to stop VKA treatment.
Treating physicians are allowed to stop VKA treatment in these patients, even
if VTE recurrence risk is high or continue treatment although the risk on
recurrent VTE is low. As this is a pragmatic study aimed at improving clinical
practice, such patients remain in the study and subsequently enter the
follow-up period (after initial treatment of 6 months with VKA conform
guidelines).
Patients randomized to the control group receive care-as-usual, according to
current guidelines: usually 6 months of VKA treatment in all patients with a
first episode of unprovoked VTE, sometimes longer. The decision whether or not
to prescribe long-term VKA treatment is maily based on physicians* discretion.
For comparison reasons, a D-dimer test will be performed after 6 months of
initial VKA therapy and one month after discontinuation of VKA. To be able to
confirm a recurrent VTE during follow up, a reference compression ultrasound
will be performed after 6 months of treatment in both intervention arms.All
patients, both controls and index, are followed-up for 24 months.

Patients randomized to the index group, undergo a prediction model guided
long-term treatment (see flow chart and study design). After six months of
initial VKA therapy, a prediction model is applied on the day the patient is
instructed to stop VKA treatment (day t=0). If the prediction model
demonstrates a low risk of VTE recurrence (< 5% in the first year or < 20% at
five years) at both visits (i.e. day t=0 and t=28 days), treatment stays ended.
As soon as the result of the prediction model demonstrates a high risk of VTE
recurrence, VKA treatment is continued or resumed for another 24 months.
Treating physicians are allowed to overrule this treatment decision if patients
have a high risk of bleeding during VKA treatment based on one of the following
characteristics: history of non reversible bleeding during VKA treatment,
history of persistently poor anticoagulant control, chronic renal or hepatic
failure, history of non cardio-embolic stroke, patient older than 75 years when
diagnosed with VTE, an active peptic ulcer or if an informed patient wants to
stop VKA treatment.
As the D-dimer result (day 0) can be negative influenced by the VKA treatment,
we choose to perform a second risk assessment in all * low risk* patients, 28
days after VKA discontinuation. As soon as one risk assessment yields an
increased recurrence risk* including at day 0 (after stopping) * treatment is
immediately resumed for the follow-up period of 24 months, to rebalance the
risk of recurrent VTE.
Although a very recent meta-analysis [23] proved that the timing of D-dimer
testing after stopping VKA treatment does not affect the ability to distinguish
patients with a higher or lower risk for recurrent VTE, we are certainly aware
of the risk of recurring VTE during this period when VKA therapy is suspended.
The only previous trial on this issue * as we discussed above [16] * performed
a D-dimer assay only once at day 28, to determine the (dis)continuation of VKA
treatment. Therefore, we will perform the first D-dimer assay at the moment of
VKA withdrawal too.
After six months of VKA treatment, all patients will undergo a routine
compression ultrasound (CUS). The investigators are not aware of the results of
this CUS; only in case of suspected recurrence this 6-months-CUS will be used
to be able to confirm the diagnosis recurrent DVT. Recanalization of the veines
is assessed, resulting either in complete or non-complete recanalization.

Study procedures:
At baseline (intake visit during VKA use), several patient characteristics will
be asked to all patients (including gender, age, type of VTE event, co-morbid
diseases and medication use, and known risk factors for VTE recurrence)
Randomisation will be performed 5 months after the initial VTE event.
After 6 months of initial therapy, the intervention phase of the study will
start. Patients randomized to the index group may undergo a maximum of 2 risk
assessments, for which a maximum of 2 venapunctures for D-dimer testing will be
required. For comparison reasons, patients randomized to the control group also
undergo 2 venapunctures for D-dimer testing. The first blood sample will be
drawn after 6 months of initial treatment (T=0); the second sample will be
taken 28 days after discontinuation of VKA treatment. This test outcome has no
consequences for the treatment duration; the physician is unaware of the
outcome of this test as the investigators order the test. Only in case of
suspected recurrence of VTE, the extent of recanalisation will be performed. At
time of intake and intervention, as well as during all follow-up contacts, all
patients will be asked to fill out *Quality of Life* questionnaires (using the
SF36 and EuroQol questionnaires) & questions on health care use for the
cost-effectiveness analyses. Furthermore the investigators will approach all
patients at 3, 12 and 24 months during the follow-up phase of the study.
Patients will be asked for the occurrence of recurrent VTE events.