The duration of oral anticoagulant therapy after a second episode of venous thromboembolism. The Duration of Anticoagulation Trial Study Group.

Abstract

A consensus has not been reached about the optimal duration of oral anticoagulant therapy after a second episode of venous thromboembolism.
In a multicenter trial, we compared six months of oral anticoagulant therapy with anticoagulant therapy continued indefinitely in patients who had had a second episode of venous thromboembolism. Of 227 patients enrolled, 111 were randomly assigned to six months of anticoagulation and 116 were assigned to receive anticoagulant therapy indefinitely; for both groups, the target international normalized ratio was 2.0 to 2.85. The initial episodes of deep-vein thrombosis (n = 193) and pulmonary embolism (n = 34), as well as recurrent episodes, were all objectively confirmed.
After four years of follow-up, there were 26 recurrences of venous thromboembolism that fulfilled the diagnostic criteria, 23 in the group assigned to six months of therapy (20.7 percent) and 3 in the group assigned to continuing therapy (2.6 percent). The relative risk of recurrence in the group assigned to six months of therapy, as compared with the group assigned to therapy of indefinite duration, was 8.0 (95 percent confidence interval, 2.5 to 25.9). There were 13 major hemorrhages, 3 in the six-month group, (2.7 percent) and 10 in the infinite-treatment group (8.6 percent). The relative risk of major hemorrhage in the six-month group, as compared with the infinite-treatment group was 0.3 (95 percent confidence interval, 0.1 to 1.1). There was no difference in mortality between the two groups.
Prophylactic oral anticoagulation that was continued for an indefinite period after a second episode of venous thromboembolism was associated with a much lower rate of recurrence during four years of follow-up than treatment for six months. However, there was a trend toward a higher risk of major hemorrhage when anticoagulation was continued indefinitely.

Full text

Page 1

Volume 336Number 6

?

393

The New England
Journal
of



Medicine

© Copyright, 1997, by the Massachusetts Medical Society

VOLUME 336

F

EBRUARY

6, 1997

NUMBER 6

THE DURATION OF ORAL ANTICOAGULANT THERAPY AFTER A SECOND
EPISODE OF VENOUS THROMBOEMBOLISM

S

AM

S
P

CHULMAN
ER
L
INDMARKER
G

, M.D., S

TAFFAN

G

RANQVIST
N
ICOL
ARBRO
URATION

, M.D., M
, M.D., S
L
EIJD

OF
A

ARGARETA
VEN
-G
UNNAR
, M.D., O
NTICOAGULATION

H
E
INDER
T

OLMSTRÖM
KLUND
, M.D., E
RIAL
S

, M.D., A
, M.D., S
NNO
TUDY
G

NDERS
N
ORDLANDER
OOGNA
, M.D.,
ROUP
*

C

ARLSSON
, M.D.,

, M.D.,




, M.D., P
ÄRFARS
AND

ETER










UNE
L




ERD

L


, M.D., B

THE





LLE
O

L










D













A

BSTRACT
Background
about the optimal duration of oral anticoagulant
therapy after a second episode of venous thrombo-
embolism.
Methods
In a multicenter trial, we compared six
months of oral anticoagulant therapy with anticoag-
ulant therapy continued indefinitely in patients who
had had a second episode of venous thromboembo-
lism. Of 227 patients enrolled, 111 were randomly
assigned to six months of anticoagulation and 116
were assigned to receive anticoagulant therapy in-
definitely; for both groups, the target international
normalized ratio was 2.0 to 2.85. The initial episodes
of deep-vein thrombosis (n
embolism (n
?
34), as well as recurrent episodes, were
all objectively confirmed.
Results
After four years of follow-up, there were 26
recurrences of venous thromboembolism that ful-
filled the diagnostic criteria, 23 in the group assigned
to six months of therapy (20.7 percent) and 3 in the
group assigned to continuing therapy (2.6 percent).
The relative risk of recurrence in the group assigned
to six months of therapy, as compared with the group
assigned to therapy of indefinite duration, was 8.0 (95
percent confidence interval, 2.5 to 25.9). There were
13 major hemorrhages, 3 in the six-month group (2.7
percent) and 10 in the indefinite-treatment group (8.6
percent). The relative risk of major hemorrhage in the
six-month group, as compared with the indefinite-
treatment group, was 0.3 (95 percent confidence in-
terval, 0.1 to 1.1). There was no difference in mortality
between the two groups.
Conclusions
Prophylactic oral anticoagulation that
was continued for an indefinite period after a second
episode of venous thromboembolism was associat-
ed with a much lower rate of recurrence during four
years of follow-up than treatment for six months.
However, there was a trend toward a higher risk of
major hemorrhage when anticoagulation was contin-
ued indefinitely. (N Engl J Med 1997;336:393-8.)
©1997, Massachusetts Medical Society.


A consensus has not been reached



?

193) and pulmonary








From the Departments of Internal Medicine at Karolinska Hospital,
Stockholm (S.S., P .L.), Huddinge Hospital, Huddinge (M.H.), Danderyd
Hospital, Danderyd (A.C.), Köping Hospital, Köping (P .N.), Södertälje
Hospital, Södertälje (S.-G.E.), Västerås Central Hospital, Västerås (S.N.),
Södersjukhuset, Stockholm (G.L.), St. Göran Hospital, Stockholm (B.L.),
Örebro Regional Hospital, Örebro (O.L.), and Nacka Hospital, Stockholm
(E.L.); and the Department of Radiology, Ersta Hospital, Stockholm
(S.G.) — all in Sweden. Address reprint requests to Dr. Schulman at the
Department of Internal Medicine, Karolinska Hospital, S-171 76 Stockholm,
Sweden.
Other authors were Hans Walter, M.D., Sabbatsberg Hospital, Stock-
holm; Stanka Viering, M.D., Norrtälje Hospital, Norrtälje; Martin Hjorth,
M.D., Lidköping Hospital, Lidköping; and Jonas Boberg, M.D., Uppsala
Academic Hospital, Uppsala.
*The investigators and institutions participating in the Duration of An-
ticoagulation (DURAC II) Trial Study Group are listed in the Appendix.

RAL anticoagulant therapy is routinely
given to most patients who have had ep-
isodes of venous thromboembolism. Two
recent multicenter trials have demonstrat-
ed that if the duration of treatment after a first epi-
sode of thromboembolism is extended to three to
six months, instead of four to six weeks, the rate of
recurrence can be reduced, especially among pa-
tients with permanent risk factors such as throm-
boembolism that was idiopathic in nature and venous
insufficiency.
The optimal duration of secondary
prophylaxis after a second episode of venous throm-
boembolism is unknown. In one study, patients
were stratified according to whether the episode of
venous thromboembolism was their first or second,
but the number of patients with second episodes
was small and no conclusions could be drawn.
the absence of data, it has been assumed that the
risk of recurrence is greater after a second episode
than after a first. Sixty percent of Swedish hospitals
recommend that patients with second episodes of
venous thromboembolism receive oral anticoagulant

1,2


3

In
Downloaded from www.nejm.org by SAM SCHULMAN on June 03, 2004.
Copyright © 1997 Massachusetts Medical Society. All rights reserved.

Page 2

394

?

February 6, 1997

The New England Journal of Medicine

therapy for three to six months or even longer; some
centers recommend continuing therapy for more
than two years.
It has also been suggested that oral
anticoagulation be continued indefinitely after re-
current venous thromboembolism.
We undertook this trial to compare six months of
oral anticoagulant therapy after a second occurrence
of deep-vein thrombosis or pulmonary embolism
with the same treatment continued indefinitely. The
end points were recurrent thromboembolism, major
hemorrhagic complications, and death during four
years of follow-up.

4


5



METHODS

Study Design

The portion of the Duration of Anticoagulation (DURAC) tri-
al that we describe here was a randomized, open trial of anti-
coagulant therapy in patients with second episodes of venous
thromboembolism, in which 16 medical centers in central Swe-
den participated. It was conducted in parallel with a study of an-
ticoagulation after a first episode of venous thromboembolism,
and followed the same protocol except for the duration of treat-
ment. Consecutive patients at least 15 years of age who had acute
pulmonary embolism or deep-vein thrombosis in the leg, the iliac
veins, or both were included.
The diagnostic procedures have been described previously.
Briefly, objective diagnosis based on the results of venography in
the case of deep-vein thrombosis and on the results of angiogra-
phy or the combination of chest radiography and ventilation–per-
fusion lung scanning in the case of pulmonary embolism was re-
quired. The exclusion criteria were identical to those
of patients with the first episode of venous thromboembolism.
Oral informed consent was obtained from all patients before en-
rollment.
Randomization, which was stratified only according to medical
center, took place at the end of hospitalization and was performed
centrally. A computer-generated allocation schedule was used to
assign patients, in blocks of 10, to receive oral anticoagulant ther-
apy either for six months or indefinitely; the duration of therapy
was counted from the date when stable prothrombin times within
the target range were achieved.

2


2




in the study

2


Anticoagulant Therapy

The initial treatment of the venous thromboembolism consist-
ed of unfractionated or low-molecular-weight heparin adminis-
tered intravenously or subcutaneously for at least five days (until
the prothrombin time had been within the target range for two
days). If the treating physician thought it was indicated, throm-
bolytic therapy could be given at the start of the study. Patients
with deep-vein thrombosis were provided with a graduated-com-
pression stocking and instructed to wear it during the day for at
least one year.
Oral anticoagulation with warfarin sodium or dicumarol was
usually begun at the same time as heparin therapy; the target cho-
sen for the international normalized ratio (INR) was 2.0 to 2.85,
partly because a pilot study showed an excess of hemorrhagic
complications when the INR exceeded that range.
of prothrombin times was performed as previously described,
with use of thromboplastin reagents with international sensitivity
indexes of 0.86 to 1.00. When a stable prothrombin time within
the target range had been achieved, the test was repeated weekly
for the first three weeks and then at least once every four weeks.
Oral anticoagulant therapy was discontinued after six months,
without tapering, in the patients randomly assigned to six months
of therapy, usually at the time of the six-month visit.
Before anticoagulant therapy was initiated, we obtained plasma

6

The analysis

2

samples from patients who were less than 50 years old and those
with a family history of venous thromboembolism for measure-
ment of antithrombin, protein C, and protein S, as previously de-
scribed.
The patients were instructed to abstain from taking analgesic
agents containing aspirin and, if antiinflammatory treatment was
required, to use only ibuprofen. All the patients were informed
about the symptoms of deep-vein thrombosis and pulmonary em-
bolism. They were told to report immediately to the emergency
room if any such symptoms occurred, and patients receiving oral
anticoagulant therapy were also asked to report all hemorrhagic
complications. Follow-up evaluations by one of the investigators
or by a specially trained nurse or physiotherapist were scheduled
for 1.5, 3, 6, 9, 12, 24, 36, and 48 months after randomization.
At each visit, the patients were asked about new symptoms of
venous thromboembolism and, if they were still receiving an an-
ticoagulant drug, about possible hemorrhage. They were also re-
minded of the symptoms of deep-vein thrombosis and pulmonary
embolism and reminded to come to the emergency room if such
symptoms occurred and to report bleeding episodes.

2


End Points
The principal end points of the study were major hemorrhage,
recurrent venous thromboembolism, and death during the four-
year period. Major hemorrhages were defined as episodes of bleed-
ing that resulted in death or required hospitalization, treatment
with blood products or vitamin K, or any combination of these
outcomes.
Recurrent thromboembolic events were objectively verified by
the same methods as the index events. In addition, to be consid-
ered confirmed, a recurrent deep-vein thrombosis had to be char-
acterized by one of the following: thrombus in the contralateral
leg; thrombus in another deep vein of the same leg as the original
thrombus; or thrombus in the same vein as the original event,
with proximal extension of at least 5 cm if the upper limit of the
original thrombus had been visualized or, if the upper limit of the
original thrombus had not been determined, the presence of a
constant filling defect surrounded by contrast medium. Recurrent
pulmonary embolism had to be confirmed by defects in previous-
ly perfused areas, unless another scan during the intervening
period had shown resolution of the original defects. Fatal pul-
monary embolism had to be confirmed by autopsy. Initial and
subsequent venograms in patients with confirmed and uncon-
firmed recurrences of deep-vein thrombosis and lung scans in
patients with pulmonary embolism were reviewed by an inde-
pendent radiologist who was blinded to the patients’ treatment
assignments and the order of the examinations.
Patients who were lost to follow-up were repeatedly cross-
checked against data in the national Death Registry; no deaths have
been missed. Names were also checked against the registry of hos-
pitalizations. The ascertainment of recurrences or major hemor-
rhages is almost certainly complete.
An independent safety committee reviewed the number of pa-
tients included and the number of major end points twice during
the study. The committee was instructed to stop the study in case
of a significant difference between the treatment groups in the
rate of major hemorrhages.


Statistical Analysis
For the calculation of the required number of patients in each
treatment group, we assumed an annual rate of recurrence of
1 percent among patients receiving oral anticoagulation and 5 per-
cent among those not receiving such treatment, or a cumulative in-
cidence after four years of 4 percent and 18 percent, respectively.
With an alpha error of 5 percent and a beta error of 20 percent
(two-tailed), we needed 88 patients per group; since we estimat-
ed that 20 percent would be lost to follow-up, recruitment of 110
patients per group was required.
All statistical analyses were performed on an intention-to-treat
basis, although some patients in both groups received oral anti-

Downloaded from www.nejm.org by SAM SCHULMAN on June 03, 2004.
Copyright © 1997 Massachusetts Medical Society. All rights reserved.

Page 3

DURATION OF ORAL ANTICOAGULANT THERAPY AFTER A SECOND EPISODE OF VENOUS THROMBOEMBOLISM

Volume 336 Number 6

?

395

coagulation for shorter or longer periods than called for in the
protocol, and some were discovered after randomization to have
cancer. For statistical analysis, we used Wilcoxon’s rank-sum test
and the log-rank test (the Lifetest procedure in the SAS software
system) and Fisher’s exact test for two groups. Patients were lost
to follow-up for a total of 442 months (4 percent of the total);
these person-months were accounted for in the log-rank test.
The group in this study that was assigned to six months of ther-
apy has also been compared with a group of patients in another
portion of the DURAC trial who had a first episode of venous
thromboembolism and were treated with oral anticoagulant
drugs for six months.
The latter group has also completed four
years of follow-up and was recruited during the same period.
The protocols for these two groups have in all respects been
identical. Ninety-five percent confidence intervals are shown for
all results. The study was approved by the regional and local eth-
ics committees.

2


RESULTS

Enrollment took place from April 12, 1988,
through April 18, 1991; during this period, 227 pa-
tients were randomly assigned to treatment groups.
Two patients assigned to indefinite anticoagulation
had suspected congenital protein C deficiency, but
because of the difficulties of confirming this diagno-
sis during treatment they were not withdrawn from
the study. No congenital protein S or antithrombin
deficiency was detected.
According to the logbooks, which were available
from 12 of the 16 medical centers, corresponding
to 84 percent of all patients (191 of 227), 63 per-
cent of patients evaluated were enrolled (191 of
301). The patients from the centers with missing
logbooks were equally divided between the two
treatment groups. Five eligible patients were not en-
rolled because the investigators did not have time to
enroll them.
Of the 227 patients recruited for the study, 111
were randomly assigned to six months of treatment
and 116 to indefinite treatment. The treatment
groups were similar at entry (Table 1). In the six-
month group, 10 patients received treatment for a
longer period than intended (1 to 42 months long-
er); as a result, the mean duration of anticoagulation
was actually 7.7 months. In the group assigned to in-
definite treatment, 26 patients had shorter periods
of treatment (1 to 43 months shorter), resulting in
a mean duration of treatment of 42.7 months dur-
ing the 4 years of follow-up. The main cause for
these deviations was a refusal by the patient to ad-
here to the protocol. The percentages of patients
who complied with the instructions for the use of
compression stockings were 95 percent in the six-
month group and 94 percent in the indefinite-treat-
ment group at 3 months; 82 percent and 77 per-
cent, respectively, at 12 months; 55 percent and 43
percent at 24 months; and 38 percent and 37 per-
cent at 48 months (there were no significant differ-
ences in these rates between the groups).
During the four years of follow-up, 26 patients
died and 14 dropped out. The principal end points

*Of the patients with deep-vein thrombosis as the index event, three in
the six-month group and four in the indefinite-treatment group had pre-
viously had pulmonary embolism.
†Temporary risk factors were surgery, trauma, temporary immobiliza-
tion, travel, the receipt of estrogen, infection, Baker’s cyst, and pregnancy;
permanent risk factors were idiopathic venous thromboembolism and
venous insufficiency.

T

ABLE
A

1.

C

HARACTERISTICS
CCORDING

TO



OF



THE

P
OF

ATIENTS
A
SSIGNED



AT

E
T

NROLLMENT
REATMENT

,






THE

D

URATION








.

C

HARACTERISTIC

6 M
(N
?

O




111)
I
(N

NDEFINITE
?
116)





mean

?

SD

Age (yr)
Years since previous thromboembolic event
65.0
8.1

?
?

12.4
11.8
64.0
6.4

?
?

12.5
7.2





no. (%)

Male sex
Type of index event
Pulmonary embolism
Deep-vein thrombosis
Proximal thrombosis
Site of previous deep-vein thrombosis*
Ipsilateral leg
Contralateral leg
Temporary risk factor†
Subsequent cancer
Family history of venous thromboembolism
Thrombolytic therapy
Therapy with low-molecular-weight heparin
70 (63) 68 (59)
17 (15)
94 (85)
68 (72)
17 (15)
99 (85)
65 (66)
54 (59)
37 (41)
22 (20)
8 (7)
24 (22)
4 (4)
3 (3)
45 (47)
50 (53)
21 (18)
8 (7)
22 (19)
4 (3)
8 (7)

are shown in Table 2. There was no statistically sig-
nificant difference in mortality between the two treat-
ment groups. No cases of fatal pulmonary embolism
could be confirmed, although it was suspected in a
patient in the six-month group who died suddenly
at 27 months.
There was a trend toward more major hemorrhag-
es in the group assigned to indefinite anticoagula-
tion. In the six-month group, only one of the three
major hemorrhages occurred during anticoagulant
therapy (an episode of vaginal bleeding, triggered by
an occult cancer). The remaining two episodes, both
of which were cerebral hemorrhages, occurred 14.5
and 18 months after the discontinuation of therapy;
one was fatal. In the group receiving anticoagulation
for an indefinite period, there were a fatal subarach-
noid hemorrhage, a case of fatal hemorrhagic pan-
creatitis, an episode of severe epistaxis requiring hos-
pitalization, three gastrointestinal hemorrhages (two
of which required transfusions), two episodes of
hematuria requiring hospitalization (one of which
occurred after cystoscopy), a post-traumatic sub-
cutaneous hematoma requiring hospitalization, and
an episode of intraabdominal bleeding treated with
transfusions. In five of the patients with hemorrhag-
ic complications, the intensity of anticoagulation
was greater than the target range at the time of ad-
mission (INR, 3.7 to 6.7). None of the patients with
Downloaded from www.nejm.org by SAM SCHULMAN on June 03, 2004.
Copyright © 1997 Massachusetts Medical Society. All rights reserved.

Page 4

396

?

February 6, 1997

The New England Journal of Medicine

hemorrhage received vitamin K alone without hos-
pitalization.
The difference in the rate of recurrent venous
thromboembolic events between the six-month group
(20.7 percent; 95 percent confidence interval, 13.1 to
28.3 percent) and the group receiving therapy indef-
initely (2.6 percent; 95 percent confidence interval,
1.1 to 4.1 percent) was significant (P
four years of follow-up.
The cumulative probability of a recurrent event is
shown in Figure 1. In the six-month group there was
a progressive accumulation of recurrent events, dis-

?

0.001) after
tributed over the three and a half years after the dis-
continuation of anticoagulation. In the group as-
signed to indefinite anticoagulation, there were only
three recurrent thromboembolic events (at months
26, 29, and 42), all of which occurred 1 to 10
months after the premature discontinuation of anti-
coagulant therapy. There was thus no recurrence
during anticoagulant therapy in any group. One of
the three events was fatal: a mesenteric-vein throm-
bosis, verified by laparotomy, in a patient with dia-
betes mellitus and cirrhosis of the liver, in whom the
anticoagulant therapy was discontinued prematurely
after 28 months, 1 month before the event occurred.
The remaining recurrences consisted of eight cases
of pulmonary embolism and eight cases of deep-vein
thrombosis in the same leg as the index event and
nine in the contralateral leg. None of the recurrenc-
es occurred in a high-risk situation (e.g., after sur-
gery or while the patient was immobilized).
Three of the recurrences (all in the six-month
group) were detected at a follow-up visit and the re-
mainder when the patients came to the emergency
room because they had new symptoms. In six pa-
tients in the six-month group and two in the group
assigned to therapy of indefinite duration who came
to the emergency room because of new symptoms,
venograms or lung scans did not demonstrate a re-
currence. Four additional patients (all in the six-
month group) were hospitalized because of new
symptoms and filling defects on the venogram (two
patients) or perfusion defects on the lung scan (two
patients). On subsequent review, these abnormali-
ties did not meet the criteria for a recurrence, how-
ever, and they are not included in the statistical anal-
ysis. Sixty-two percent of the prothrombin times
were within the target range for oral anticoagulant
therapy (INR, 2.0 to 2.85).
The group assigned to six months of therapy in
this study was also compared with a group of pa-
tients treated for the same length of time and ac-
cording to the same protocol who had had only one
episode of venous thromboembolism at the time of
enrollment. The difference after three and four years
of follow-up did not reach statistical significance, ei-
ther by the chi-square test (P
points) or by the log-rank test (P
more weight to late events.
The patients with recurrences did not differ signif-
icantly from the rest with respect to age, sex, wheth-
er the index event was a pulmonary embolism or
deep-vein thrombosis, whether the index thrombus
was proximal or distal, whether the factor triggering
the index thrombus was temporary or permanent,
the length of time between the first thromboembol-
ic event and the index episode, and the presence or
absence of a family history of venous thromboem-
bolism. The numbers in these subgroups were too
small, however, for reliable analyses.

?

0.2 for both time
?
0.2), which gives



Figure 1.
boembolism in Patients with a Second Episode, According to
the Duration of Assigned Anticoagulant Therapy.

Cumulative Probability of Recurrent Venous Throm-
0
30
20
10
06 121824 30 364248
Months
Cumulative Probability
of Recurrence (%)
Six-month group
Indefinite-treatment group

*Relative risks are expressed as the ratio of the number of patients with
the specified end point to the total number of patients in the six-month
group, divided by the corresponding ratio in the indefinite-treatment
group. CI denotes confidence interval.
†Two of the hemorrhages occurred after the discontinuation of oral an-
ticoagulation.
‡All recurrences in this group occurred after the premature discontinu-
ation of anticoagulation.
§There were 95 patients in the six-month group and 96 in the indefinite-
treatment group at these 12 hospitals.

T
EARS

ABLE

2.
CCORDING

F

REQUENCY


THE

OF

P
D

RINCIPAL
URATION

E

ND
OF

P
A

OINTS
SSIGNED



AFTER
T

F

OUR


Y


, A




TO












REATMENT

.

E

ND

P

OINT

6 M
(N
?

O



111)
I
(N?116)

NDEFINITE

RELATIVE RISK
(95% CI)* P VALUE
no. (%)
Major hemorrhage
Recurrence
In hospitals with
logbooks§
Death
3 (2.7)†
23 (20.7)
21 (22.1)
10 (8.6)
3 (2.6)‡
3 (3.1)
0.3 (0.1–1.1)
8.0 (2.5–25.9)
7.1 (2.2–22.9)
0.084
?0.001
?0.001
16 (14.4)10 (8.6)1.7 (0.8–3.5) 0.21
Downloaded from www.nejm.org by SAM SCHULMAN on June 03, 2004.
Copyright © 1997 Massachusetts Medical Society. All rights reserved.

Page 5

DURATION OF ORAL ANTICOAGULANT THERAPY AFTER A SECOND EPISODE OF VENOUS THROMBOEMBOLISM
Volume 336Number 6
?
397
DISCUSSION
Because little is known about anticoagulant ther-
apy in patients with second episodes of venous
thromboembolism, we studied such patients in a
separate trial. The demonstration of a difference or
equivalence in outcome between two groups of such
patients after random assignment to different, limit-
ed periods of anticoagulation — for example, six
months as compared with one year — would require
many more patients than were included in this trial.
Although long-term anticoagulation has been rec-
ommended after recurrent venous thromboembo-
lism,5 there is a justified fear that major hemorrhagic
complications will result. With an intensity of anti-
coagulation corresponding to an INR of 2.0 to 3.0,
the incidence of major or fatal bleeding is 0.6 to 0.7
percent per month.1,7 With the target intensity used
in our trial (INR, 2.0 to 2.85), six major hemor-
rhages occurred in patients who had had a first epi-
sode of venous thromboembolism during 3399 per-
son-months (incidence, 0.18 percent per month).2
In the present study the actual total duration of an-
ticoagulation during four years of follow-up was
5561 person-months; there were 11 major hemor-
rhages during treatment, for a monthly incidence of
0.20 percent. Any comparison with the results of
previous studies must be made with great caution,
because of slight differences in the definitions of
major hemorrhage. When we took into account the
serious hemorrhages, which occurred even without
anticoagulation, the difference in the incidence of
major hemorrhages between the two study groups
amounted only to a statistical trend. There is also a
possibility of bias resulting from underestimation of
the incidence of hemorrhage in the six-month group,
since these patients were not actively questioned
about hemorrhages after the discontinuation of an-
ticoagulant therapy.
The risk of recurrent thromboembolism was, on
the other hand, significantly reduced when the oral
anticoagulant therapy was continued indefinitely. We
tried to minimize the risk of bias due to the unblind-
ed study design by having the venograms and lung
scans reviewed by an independent, blinded radiolo-
gist. Furthermore, all but three of the recurrences
were detected by physicians who were not involved in
the study, and the number of negative examinations
of possible recurrences was small in both groups, in-
dicating that there was probably no tendency to over-
diagnose recurrences in the six-month group. Except
for the effects of any remaining bias, the intensity of
anticoagulation we used (target INR, 2.0 to 2.85)
proved effective, since no patient actually receiving
anticoagulant therapy had a confirmed recurrence.
Our results therefore suggest that long-term second-
ary prophylaxis is effective in patients with recurrent
venous thromboembolism and that it does not entail
a high risk of hemorrhagic complications.
Nonetheless, the occurrence of hemorrhages in
patients receiving anticoagulant therapy cannot be
disregarded. Our results indicate that if 100 patients
were treated indefinitely instead of for only six
months, 0.43 episode of recurrent thromboembo-
lism would be averted per month, at a cost of 0.20
major hemorrhage per month (albeit not caused ex-
clusively by anticoagulant therapy). It would thus be
of value to investigate whether a reduction in the in-
tensity of anticoagulation — to a target INR of 1.5
to 2.0 after, for example, one year — could eliminate
the risk of bleeding while still offering the same pro-
tective effect.
Our results do not indicate which subgroups of
patients may have a special need for prolonged anti-
coagulation, partly because of the limited numbers
of patients in the study. It is possible that most pa-
tients who have second thromboembolic episodes
are at such a high risk for further recurrences that a
division into subgroups would yield little informa-
tion. We were unable to demonstrate an increased
risk of recurrence after a second episode, as compared
with the risk after a first episode, among patients
treated for six months according to identical proto-
cols. This failure may reflect an insufficient number
of patients with second thromboembolic episodes,
or it may mean that no difference exists.
Supported by grants from the Swedish Heart Lung Foundation, the
Swedish Society of Medicine, the Karolinska Institute, Skandia, Trygg-
Hansa, Triolab, Nycomed, and Stago.
We are indebted to Professor Jack Hirsh of the Hamilton Civic Hos-
pitals Research Centre, McMaster University, for his constructive crit-
icism of the study design, and to Professor Leon Poller and Dr. Jean
Thomson of the United Kingdom Reference Laboratory for Anticoag-
ulant Reagents and Control and Dr. A.M.H.P. van den Besselaar of
the Stichting Referentie-Instituut Laboratorium Onderzoek Anti-
stollingscontrole for their assistance with quality control.
APPENDIX
The DURAC II Trial Study Group consisted of the following
investigators: Danderyd Hospital, Danderyd — A. Carlsson,
C. Gustafsson, and A. Gröndahl; Huddinge Hospital, Huddinge
— A.-S. Rhedin, E. Törnebohm, M. Holmström, and D. Lock-
ner; Karolinska Hospital, Stockholm — S. Schulman, P . Lindmark-
er, and H. Johnsson; Köping Hospital, Köping — P . Nicol, J. Ko-
bosko, B. Malmros, N. Arcini, and J. Saaw; Nacka Hospital,
Stockholm — E. Loogna and R. Stig; Norrtälje Hospital, Norrtälje
— S. Viering; Nyköping Hospital, Nyköping — B. Ljungberg,
S. Wilhelmsson, and Å. Ohlsson; Sabbatsberg Hospital, Stockholm
— H. Walter, K. Malmqvist, and F. Al-Khalili; St. Göran Hospital,
Stockholm — B. Leijd and A. Petrescu; Södersjukhuset, Stockholm
— J. Brohult, G. Lärfars, and J. Hulting; Södertälje Hospital,
Södertälje — S.-G. Eklund, E. Svensson, and L. Dahlin; Uppsala
Academic Hospital, Uppsala — J. Boberg; Västerås Central Hospi-
tal, Västerås — S. Nordlander and B. Marjanovics; Örebro Region-
al Hospital, Örebro — O. Linder; Linköping Regional Hospital,
Linköping — K.-Å. Jönsson and C. Malm; Lidköping Hospital,
Lidköping — M. Hjorth and A. Lindgren: Safety Committee —
B. Fagrell, Karolinska Hospital, and M. Kallner, Löwenströmska
Hospital; Radiologic Assessment — S. Granqvist, Ersta Hospital;
Steering Committee — J. Boberg, J. Brohult, S.-G. Eklund,
B. Fagrell, H. Johnsson, B. Ljungberg, D. Lockner, P . Nicol,
Downloaded from www.nejm.org by SAM SCHULMAN on June 03, 2004.
Copyright © 1997 Massachusetts Medical Society. All rights reserved.

Page 6

398
?
February 6, 1997
The New England Journal of Medicine
S. Schulman (chair and coordinator), S. Wilhelmsson, and
B. Wadman, Örebro Regional Hospital; Statistical Analysis —
M. Snyder, Tadiran Information Systems, Givat Shmuel, Israel.
REFERENCES
1. Research Committee of the British Thoracic Society. Optimum dura-
tion of anticoagulation for deep-vein thrombosis and pulmonary embo-
lism. Lancet 1992;340:873-6.
2. Schulman S, Rhedin A-S, Lindmarker P, et al. A comparison of six weeks
with six months of oral anticoagulant therapy after a first episode of venous
thromboembolism. N Engl J Med 1995;332:1661-5.
3. Schulman S, Lockner D, Juhlin-Dannfelt A. The duration of oral anti-
coagulation after deep vein thrombosis: a randomized study. Acta Med
Scand 1985;217:547-52.
4. Johansson J, Johnsson H, Wilhelmsson S. Diagnos och behandling
av venös tromboembolism vid svenska internmedicinska kliniker — en
översikt. Lakartidningen 1987;84:22-5.
5. Hyers TM, Hull RD, Weg JG. Antithrombotic therapy for venous
thromboembolic disease. Chest 1993;102:Suppl:408S-425S. [Erratum,
Chest 1993;103:1636.]
6. Schulman S, Stigendal L, Jansson J-H, Brohult J. Haemorrhagic and
thromboembolic complications versus intensity of treatment of venous throm-
boembolism with oral anticoagulants. Acta Med Scand 1988;224:425-30.
7. Hull RD, Raskob GE, Rosenbloom D, et al. Heparin for 5 days as com-
pared with 10 days in the initial treatment of proximal venous thrombosis.
N Engl J Med 1990;322:1260-4.
Downloaded from www.nejm.org by SAM SCHULMAN on June 03, 2004.
Copyright © 1997 Massachusetts Medical Society. All rights reserved.