*A total of 9 individuals were lost during the study but later found.
Thus, these individuals did not receive all 4 questionnaires, but information
was complete during follow-up.
Patients with and without thrombophilia during the period from the end
of the initial anticoagulation period (90 days) until January 1, 2000. The
crude hazard ratio of thrombophilia compared with no thrombophilia was 1.3
(95% confidence interval, 0.8-2.0); the hazard ratio adjusted for age, sex,
and oral anticoagulation as a time-dependent covariate was 1.4 (95% confidence
Error bars indicate 95% confidence intervals.
Sverre C. Christiansen, Suzanne C. Cannegieter, Ted Koster, Jan P. Vandenbroucke, Frits R. Rosendaal. Thrombophilia, Clinical Factors, and Recurrent Venous Thrombotic Events. JAMA. 2005;293(19):2352–2361. doi:10.1001/jama.293.19.2352
Author Affiliations: Departments of Clinical
Epidemiology (Drs Christiansen, Cannegieter, Koster, Vandenbroucke, and Rosendaal)
and Hematology (Dr Rosendaal), Leiden University Medical Center, Leiden, the
Context Data on the recurrence rate of venous thrombotic events and the effect
of several risk factors, including thrombophilia, remain controversial. The
potential benefit of screening for thrombophilia with respect to prophylactic
strategies and duration of anticoagulant treatment is not yet known.
Objectives To estimate the recurrence rate of thrombotic events in patients after
a first thrombotic event and its determinants, including thrombophilic abnormalities.
Design, Setting, and Patients Prospective follow-up study of 474 consecutive patients aged 18 to 70
years without a known malignancy treated for a first objectively confirmed
thrombotic event at anticoagulation clinics in the Netherlands. The Leiden
Thrombophilia Study (LETS) was conducted from 1988 through 1992 and patients
were followed up through 2000.
Main Outcome Measures Recurrent thrombotic event based on thrombophilic risk factors, sex,
type of initial thrombotic event (idiopathic or provoked), oral contraceptive
use, elevated levels of factors VIII, IX, XI, fibrinogen, homocysteine, and
Results A total of 474 patients were followed up for mean (SD) of 7.3 (2.7)
years and complete follow-up was achieved in 447 (94%). Recurrence of thrombotic
events occurred in 90 patients during a total of 3477 patient-years. The rate
of thrombotic event recurrence was 25.9 per 1000 patient-years (95% confidence
interval [CI], 20.8-31.8 per 1000 patient-years). The incidence rate of recurrence
was highest during the first 2 years (31.9 per 1000 patient-years; 95% CI,
20.3-43.5 per 1000 patient-years). The risk of thrombotic event recurrence
was 2.7 times (95% CI, 1.8-4.2 times) higher in men than in women. Patients
whose initial thrombotic event was idiopathic had a higher risk of a thrombotic
event recurrence than patients whose initial event was provoked (hazard ratio
[HR], 1.9; 95% CI, 1.2-2.9). Women who used oral contraceptives during follow-up
had a higher thrombotic event recurrence rate (28.0 per 1000 patient-years;
95% CI, 15.9-49.4 per 1000 patient-years) than those who did not (12.9 per
1000 patient-years; 95% CI, 7.9-21.2 per 1000 patient-years). Recurrence risks
of a thrombotic event by laboratory abnormality ranged from an HR of 0.6 (95%
CI, 0.3-1.1) in patients with elevated levels of factor XI to an HR of 1.8
(95% CI, 0.9-3.7) for patients with anticoagulant deficiencies.
Conclusions Prothrombotic abnormalities do not appear to play an important role
in the risk of a recurrent thrombotic event. Testing for prothrombotic defects
has little consequence with respect to prophylactic strategies. Clinical factors
are probably more important than laboratory abnormalities in determining the
duration of anticoagulation therapy.
The incidence rate of a first venous thrombosis is 1 to 2 events per
1000 patient-years.1- 3 Venous
thrombosis manifests mainly as deep venous thrombosis (DVT) and pulmonary
embolism. Environmental risk factors include immobilization, surgery, malignancies,
pregnancy, puerperium, and exogenous female hormones.4 Genetic
abnormalities increasing the risk of a thrombotic event have been known for
several decades and include deficiencies of the natural anticoagulants antithrombin,
protein C, and protein S. Additional biochemical risk factors for a thrombotic
event are factor V Leiden, prothrombin G20210A, high levels of factors VIII,
IX, or XI, homocysteine, and fibrinogen.5- 11 Knowledge
of the risk of a thrombotic event recurrence and its determinants is relevant
for clinical policy regarding screening for thrombophilia, duration of anticoagulant
treatment, and prophylactic strategies in circumstances of increased risk.
Estimates of the 5-year cumulative incidence of recurrent thrombotic
events are around 25%.12,13 One
study14 reported a high recurrence rate of
20% (17 cases) in a total of 83 patients during a 10-month follow-up. While
estimates of the overall risk of a thrombotic event recurrence vary, reports
on contributing factors are contradictory. Several studies have been published
linking factor V Leiden, the prothrombin G20210A mutation, and the recurrence
of a thrombotic event.14- 25 Most
studies showed little effect of carriership of these mutations on the risk
of recurrence.14- 20,23 However,
in some studies a 4- to 5-fold higher risk of a thrombotic event recurrence
has been noted in carriers compared with noncarriers.21,22,24,25 A
recent critical review26 of 4 studies highlighted
hazard ratios (HRs) ranging from 1.1 to 4.1 in carriers of factor V Leiden
compared with noncarriers. In studies comparing carriers of prothrombin G20210A
with noncarriers, HRs of a thrombotic event recurrence varied from 0.9 to
A 6-fold increased risk of a thrombotic event recurrence was reported
for patients with high plasma levels of factor VIII.27 A
slightly increased risk of a thrombotic event recurrence was reported for
patients with high levels of factor IX.28 No
recurrence data are available for elevated levels of factor XI or fibrinogen,
which have been shown to increase the risk for a first event.9,11 Hyperhomocysteinemia
increases the risk of a thrombotic event.10 It
was also found to be prevalent in patients with a recurrent thrombotic event.29,30
Two reviews31,32 pointed
out that the contradictory results on contributing factors of thrombotic event
recurrence may have resulted from (1) differences in study design, (2) lack
of proper inception cohorts, (3) incomparability of anticoagulation profiles,
(4) differences in quality of documentation of events, or (5) differences
in the interpretation of clinical outcomes and laboratory tests. A particular
issue may be that selected subgroups of patients were studied, ie, those referred
to specialized centers for thrombophilia work-up, who may well have harbored
additional, yet unknown defects that could have affected risks. Most study
cohorts were small and followed up for only a short period.
We set out to determine the risk of a recurrent thrombotic event in
474 patients who had participated in a large population-based case-control
study of risk factors for a first DVT. Many risk factors for a thrombotic
event were investigated and these patients were followed up for up to 12 years.
In particular, we examined the effect of several thrombophilic risk factors
on the risk of recurrence, as well as the effect of sex, oral contraceptive
use, and whether the first event was idiopathic or provoked.
We included 474 consecutive patients with a first, objectively confirmed
episode of DVT. Patients were diagnosed between January 1, 1988, and December
30, 1992, and were participants in the Leiden Thrombophilia Study (LETS),11,33 which was a case-control study of
the etiology of DVT. In the Netherlands, patients with a thrombotic event
are treated at anticoagulation clinics, which are regionally organized. Therefore,
all patients living in a certain area are monitored by the same clinic, irrespective
of the hospital they were admitted to or the physician who started the treatment.
Patients participating in LETS were identified from the files at the anticoagulation
clinics in Leiden, Amsterdam, and Rotterdam. Ninety percent of eligible patients
were willing to participate in LETS. Patients older than 70 years and those
with malignancies were excluded. There were no major differences at baseline
in characteristics between the patient groups from the 3 clinics; 453 patients
had a DVT and 21 had a thrombosis in the arm. LETS was approved by the medical
ethics committee of the Leiden University Medical Center and has been described
Patients were initially seen at least 3 months after the discontinuation
of oral anticoagulant treatment, except in cases when this treatment could
not be stopped (n = 48). Patients were seen in person by one of
us (T.K.) between October 1990 and January 1994. The median time between a
thrombotic event and venipuncture was 19 months (range, 6-68 months). At the
examination, information on acquired risk factors was collected and a venous
blood draw was performed. Information was also collected on surgery, trauma,
immobilization, use of oral contraception shortly before the diagnosis of
a thrombotic event, family history, and reproductive history.
Blood was collected from the antecubital vein and placed in 0.106 M
of trisodium citrate. Plasma was prepared by centrifugation for 10 minutes
at 2000g at room temperature and stored at –70°C
in a 1.5-mL container. DNA was extracted by standard salting-out methods.
When a deficiency of protein C, protein S, or antithrombin was suspected,
patients were asked to have blood redrawn to confirm the diagnosis and then
were informed of their deficiency.34 In subsequent
years, we investigated resistance to activated protein C with factor V Leiden
(1994) and prothrombin G20210A (1996) in all participants and informed the
patients if they had an abnormal result.5,6,35 None
of the other test results (on levels of factors VIII, IX, or XI, homocysteine,
and fibrinogen) were communicated to the patients.7- 11
Details regarding the methods of measuring levels of the coagulation
factors were described in detail in LETS.11 The
factor VIII activity was measured by a 1-stage coagulation assay.7 Factor IX antigen levels were measured by sandwich
enzyme-linked immunosorbent assays using commercial polyclonal antibodies
(Dako A/S, Glostrup, Denmark).8 Factor XI antigen
levels were measured by using a monoclonal antifactor XI capture antibody
and polyclonal antifactor XI tagging antibody.9 The
fibrinogen concentration was measured according to the Clauss method using
a Dade reagent (Baxter, Miami, Fla). Protein C activity and antithrombin activity
were measured with Coamate (Chromogenix, Mölndal, Sweden) on an ACL 200
(Instrumentation Laboratory, Milan, Italy).34 Total
protein S was measured by a polyclonal enzyme-linked immunosorbent assay.34 All coagulation factors were expressed as units per
deciliter, in which 1 U is the amount of coagulation factor present in 1 mL
of pooled plasma. Fibrinogen levels were expressed as grams per liter. Total
homocysteine concentration was measured in a nonfasting state by a modified
method of automated high-performance liquid chromatography with reverse phase
separation and fluorescent detection with a 232-401 sample processor (Gilson
Inc, Middleton, Wis), an 8800 solvent-delivery system (Spectra-Physics, Mountain
View, Calif), and an LC 304 fluorometer (Spectra-Physics).10 Prothrombin
G20210A and factor V Leiden (FV G1691A) genotypes were assessed by standard
polymerase chain reaction.6,36
All 474 patients gave informed consent for follow-up and for the collection
of information from hospitals during a suspected thrombotic event recurrence.
Follow-up started 90 days after the date of the initial thrombotic event that
occurred in 1988-1992 (this 90-day period was defined as the period of initial
anticoagulation) and ended on January 1, 2000. Information relevant to the
follow-up after the thrombotic event was first gathered at the interview at
baseline, and subsequently by repeated mailed questionnaires. The questionnaires
were used as a screening tool for the occurrence of risk situations and recurrent
thrombotic events, and also included items on relevant clinical circumstances
such as anticoagulation (type, duration, indication), surgery, trauma, immobility,
use of oral contraception, and pregnancies during the period covered by each
questionnaire. Patients were further interviewed by telephone if they responded
positively on any item from the questionnaire or if they did not respond to
a questionnaire. Subsequently, confirmation of all relevant clinical information
pertaining to recurrent thrombotic events or risk situations was obtained
from the treating physicians.
Recurrent thrombotic events were confirmed and reports on diagnostic
methods were obtained by collecting the discharge letters from the treating
hospitals. Recurrent thrombotic events were adjudicated when they were objectively
confirmed with Doppler ultrasound, venography, or impedance plethysmography.
Recurrences of pulmonary embolism required a positive perfusion lung scan
(at least 1 segmental perfusion defect), a ventilation-perfusion lung scan
(intermediate or high probability), or a computerized tomographic scan.
Deep vein thromboses or pulmonary embolisms that occurred within the
initial anticoagulation period (90 days) were not considered thrombotic event
recurrences, but were considered a progression of the initial event (this
occurred in 2 patients).
Idiopathic was defined as an initial thrombotic event that occurred
in the absence of (1) pregnancy, (2) puerperium, (3) oral contraceptive use
within 30 days, (4) trauma, surgery, immobilization, or use of a plaster cast
within 3 months before the event. All others were classified as provoked.
End of follow-up was at the first thrombotic event recurrence, date
of death, date of emigration, or the end of the study, whichever occurred
first. Observation time was calculated as the time at risk from the end of
the anticoagulation treatment for the first thrombotic event to the end of
follow-up. Incidence rates of recurrent thrombotic events were calculated
as the number of events over the accumulated patient-time. Cumulative incidence
was calculated by Kaplan-Meier survival analysis.
The Cox-proportional hazards model was used to evaluate risks between
groups and was adjusted for age and sex. Anticoagulant therapy was entered
in the model as a time-dependent covariate. Separate analyses were performed
to assess the effect of prothrombotic abnormalities on the risk of recurrence
(factor V Leiden, prothrombin G20210A, hyperhomocysteinemia, high levels of
factors VIII, IX, or XI or of fibrinogen, and deficiencies of protein C, protein
S, or antithrombin). We assessed the risk of thrombotic event recurrence by
sex and by idiopathic or provoked classification of initial thrombotic event.
The effect of oral contraceptive use was determined by calculating thrombotic
event recurrence rates for women who used an oral contraceptive at any time
(either continued use or restarted use) during the follow-up period, stratified
by use at the time of the first event. In addition, we estimated the risk
separately for a second contralateral compared with a second ipsilateral thrombotic
During the follow-up period, some patients experienced periods with
an increased risk of a thrombotic event (trauma, immobilization, operations,
oral contraception, pregnancy) or a decreased risk (oral anticoagulation treatment).
To determine the effect of blood abnormalities on risk of thrombotic event
recurrence without the interference of these episodes, we repeated the analysis
while excluding all such periods.
For continuous phenotypes, we used the following cut-off values: 166
IU/dL for factor VIII; 129 IU/dL for factor IX; 121 IU/dL for factor XI; 4.1
g/L for fibrinogen; and 16.7, 19.8 or 20.3 μmol/L for homocysteine (3 different
cut-off levels were used as a consequence of different processing times in
the 3 clinics). Patients were considered deficient for protein C or protein
S when levels were below the lower limit of normal (67 U/dL or 33 U/dL when
using oral anticoagulation at the blood draw). Patients were considered deficient
of antithrombin when levels were repeatedly below 80 U/dL. All analyses were
performed with SPSS version 11.0 (SPSS Inc, Chicago, Ill).
A total of 474 patients were followed up for a mean (SD) of 7.3 (2.7)
years and complete follow-up was achieved for 447 patients (94%) for a total
observation time of 3477 patient-years (Figure
1). Twenty-seven patients were lost from observation (22 untraceable,
4 refusals, 1 disabled) and were included until their last observation. Of
the remaining 447 patients, 5 patients emigrated and 14 patients died during
follow-up. Follow-up was complete until death for 6 patients. The other 8
patients died before the subsequent questionnaire was due and follow-up for
a thrombotic event was considered complete up until the date of their last
questionnaire. The response rates after each questionnaire varied between
96% and 99%.
The general characteristics of the cohort are listed in Table 1. There were more women (n = 272) than men (n = 202) included
in this follow-up study. The mean (SD) age of the cohort was 45 (13.7) years.
The overall mean age was 6 years higher in men, which differed markedly according
to the type of initial thrombotic event. Idiopathic first events were more
common in men, who were on average 5 years younger than women with an idiopathic
first event. Most provoked first events occurred in women. In the patients
with provoked events, men were on average 9 years older than women, in whom
oral contraceptive use was a common determinant. Apart from elevated levels
of factor IX (more frequent in women) and hyperhomocysteinemia (more frequent
in men), prothrombotic factors were equally distributed between the sexes
During follow-up, 90 patients had a recurrent thrombotic event. Of these
patients, 73 had a DVT in the leg, 4 had a thrombosis in the arm, 12 had a
pulmonary embolism, and 1 had Budd-Chiari syndrome with an extension into
the vena cava. Two patients who initially had a DVT later had an arm thrombosis.
One patient who initially had an arm thrombosis later had a DVT. Two other
patients with an initial arm thrombosis later had a thrombosis in the opposite
arm as their recurrent thrombotic event. Of the 72 patients who had a DVT
as their recurrent event, 41 were ipsilateral and 31 were contralateral.
Follow-up started 90 days after the first event. At this time point,
195 individuals had finished their initial anticoagulant treatment. Of the
other 279 patients, 174 had a prolonged period of initial anticoagulation
treatment, which was less than 3 months for the majority (n = 106; Table 2). All others had additional periods of
oral anticoagulant use during follow-up. For 116 patients (67%), the total
duration of oral anticoagulant use was less than 12 months. The main reasons
for anticoagulant use were prophylaxis of DVT during risk situations, such
as surgery or pregnancy. Fifty-seven patients took an oral anticoagulant for
more than 12 months in total, which was for cardiac reasons or arterial prophylaxis
in 16 patients. The exact reasons for oral anticoagulant use were not known
in 53 patients.
Of the 57 patients who took an oral anticoagulant for more than 12 months
in total, 45 (79%) had 1 or more prothrombotic abnormalities. The mean (SD)
duration of oral anticoagulant use during follow-up of 45 patients was 4.7
(2.5) years per patient compared with 3.1 (2.3) years in the 12 patients without
any abnormality. This difference was due mostly to patients with anticoagulant
deficiencies (protein C, protein S, and antithrombin) who received anticoagulation
for a mean (SD) duration of 6.5 (0.9) years per patient. No major differences
in prescription of anticoagulation were observed among carriers of the other
prothrombotic risk factors, including factor V Leiden and prothrombin G20210A
carriers compared with noncarriers.
The overall incidence rate of recurrent thrombotic event was 25.9 per
1000 patient-years (95% CI, 20.8-31.8 per 1000 patient-years), corresponding
to an annual risk of 2.6%. Figure 2 shows
the cumulative incidence of recurrence over the 12-year follow-up period.
The risk of recurrence was 12.4% after 5 years of follow-up (95% CI, 9.5%-15.4%)
and 16.5% (95% CI, 13.1%-19.8%) after 7 years of follow-up. During the first
2 years after the discontinuation of the initial anticoagulant treatment,
the incidence rate was highest (31.9 per 1000 patient-years; 95% CI, 20.3
to 43.5 per 1000 patient-years). Subsequently, the incidence rate decreased
slowly with time (Figure 3).
Men had a 5-year cumulative incidence of thrombotic event recurrence
of 19.3% (95 CI, 13.9%-24.8%) compared with 7.4% (95% CI, 4.3%-10.5%) in women.
The cumulative incidence after 7 years was 25.3% (95% CI, 19.3%-31.2%) in
men compared with 9.9% (95% CI, 6.4%-13.5%) in women. The overall age-corrected
HR for risk of thrombotic event recurrence in men compared with women was
2.7 (95% CI, 1.8-4.2).
Of all 474 patients, 319 (67%) had at least 1 laboratory abnormality
at their first examination. After adjustment for age, sex, and anticoagulation,
no clear excess risk of recurrence was observedwhen we contrasted 319 patients
with 1 or more prothrombotic abnormalities to those with none (HR, 1.4; 95%
CI, 0.9-2.2) (Table 3). We did not observe
an increased risk of recurrence for any of the following prothrombotic risk
factors (using those without the specific abnormality as the reference group):
factor V Leiden, prothrombin G20210A, elevated levels of factor VIII, elevated
levels of factor IX, elevated levels of factor XI, and hyperhomocysteinemia
(Table 4). Adjustment for age, sex,
and periods of anticoagulation did not change these risk estimates (Table
4). In the patients with a deficiency of 1 of the natural anticoagulants protein
C, protein S, or antithrombin, a mildly increased risk of a recurrent thrombotic
event was observed (HR, 1.8; 95% CI, 0.9-3.7). Fibrinogen levels exceeding
4.1 g/L were also found to be associated with a slightly increased risk of
thrombotic event recurrence (HR, 1.6; 95% CI, 1.0-2.6; after adjustment for
age, sex, and anticoagulation, the HR was 1.7 [95% CI, 1.1-2.8]).
Only 1 of the 8 patients homozygous for factor V Leiden experienced
a recurrence during a mean (SD) follow-up of 8 (3.5) years. Their 5-year cumulative
incidence of 12.5% did not differ from all patients, while none of the 8 received
long-term anticoagulant treatment.
Several patients had more than 1 of the abnormalities studied. Patients
with 1 abnormality had a 1.2-fold increased risk of a thrombotic event recurrence
compared with the 155 patients without an abnormality. In those with more
than 1 abnormality, the recurrence rate increased 1.4-fold compared with those
without an abnormality. After correction for age, sex, and periods of anticoagulation
(Table 3), the HR increased to 1.6 in those with more than 1 abnormality,
while it stayed the same in those with only 1 abnormality.
Because factor V Leiden is the most common genetic abnormality, we studied
its combinations with other prothrombotic defects in more detail. Sixty-three
patients had factor V Leiden and 1 of the other prothrombotic risk factors
(prothrombin G20210A, elevated levels of factors VIII, IX, or XI, fibrinogen,
homocysteine, or deficiencies of protein C, protein S, or antithrombin), while
29 patients carried factor V Leiden without any additional abnormality. The
thrombotic event recurrence rate was 27.9 per 1000 patient-years (95% CI,
14.9-47.8 per 1000 patient-years) in those with factor V Leiden and any other
abnormality and 33.9 per 1000 patient-years (95% CI, 13.6-69.9 per 1000 patient-years)
in those with factor V Leiden without any other abnormality.
In patients with a combination of factor V Leiden and elevated levels
of fibrinogen, a high recurrence rate of 69.8 per 1000 patient-years (95%
CI, 25.6-152.1 per 1000 patient-years) was found. Other combinations of biochemical
risk factors (prothrombin G20210A, elevated levels of factors VIII, IX, or
XI, or homocysteine, deficiencies of protein C, protein S, or antithrombin)
with factor V Leiden did not show an additional increased risk.
The incidence rate of an ipsilateral second DVT (n = 41) was
12.4 per 1000 patient-years (95% CI, 8.9-16.9 per 1000 patient-years), while
the incidence rate of a contralateral second DVT (n = 31) was only
slightly lower at 9.5 per 1000 patient-years (95% CI, 6.5-13.5 per 1000 patient-years).
The incidence rate of an ipsilateral recurrent DVT in patients with
a prothrombotic risk factor was 13.0 per 1000 patient-years (95% CI, 8.6-18.7
per 1000 patient-years), while the incidence rate of a contralateral recurrent
DVT was 9.8 per 1000 patient-years (95% CI, 6.1-15.0 per 1000 patient-years).
In patients without prothrombotic risk factors, the incidence rate of an ipsilateral
recurrent DVT was 11.4 per 1000 patient-years (95% CI, 6.1-19.6 per 1000 patient-years),
while the incidence of a contralateral recurrent DVT was 9.0 per 1000 patient-years
(95% CI, 4.3-16.5 per 1000 patient-years).
The recurrence rate was highest in those with an idiopathic first thrombotic
event at 33.2 per 1000 patient-years (95% CI, 25.4-42.6 per 1000 patient-years)
compared with patients with a provoked first thrombotic event in whom the
recurrence rate was 17.7 per 1000 patient-years (95% CI, 11.9-25.4 per 1000
patient-years) (HR, 1.9; 95% CI, 1.2-2.9). In both men and women, the risk
of thrombotic event recurrence was higher in those who had had an idiopathic
first thrombotic event, but the effect was slightly higher in women (Table 5). Likewise, the effect of sex on the
risk of thrombotic event recurrence was the same irrespective of type (idiopathic
or provoked) of thrombotic event, with a higher risk in men than in women
In patients with an idiopathic first event, the recurrence rates were
equal in those with prothrombotic abnormalities (33.6 per 1000 patient-years;
95% CI, 24.3-45.2 1000 patient-years) and without an abnormality (32.4 per
1000 patient-years; 95% CI, 19.2-51.2 per 1000 patient-years) for a HR adjusted
for sex, age, and anticoagulation of 1.2 (95% CI, 0.7-2.2). In patients with
a nonidiopathic first thrombotic event, the recurrence rate was somewhat higher
among those with prothrombotic abnormalities (20.9 per 1000 patient-years;
95% CI, 12.9-31.9 per 1000 patient-years) compared with those without an abnormality
(12.6 per 1000 patient-years; 95% CI, 5.4-24.8 per 1000 patient-years) for
a HR adjusted for sex, age, and anticoagulation of 1.7 (95% CI, 0.7-3.8).
A substantial number of women (n = 128, 47%) used oral contraceptives
at the time of the first thrombotic event, most of whom discontinued use after
the event. However, 58 women continued or restarted use of an oral contraceptive
during follow-up. Eleven thrombotic event recurrences occurred during use
of oral contraception or within 1 month after cessation. The recurrence rate
in women who did not use oral contraceptives during follow-up was 12.9 per
1000 patient-years (95% CI, 7.9-21.2 per 1000 patient-years), while it was
28.0 per 1000 patient-years (95% CI, 15.9-49.4 per 1000 patient-years) in
women who used oral contraceptives at some point during the follow-up period
(either continuing or restarting).
Among women who did not use oral contraceptives during follow-up, the
recurrence rate was slightly higher in women who had never used an oral contraceptive
(16.2 per 1000 per 1000 patient-years; 95% CI, 8.7-30.2 per 1000 patient-years)
compared with women who used oral contraceptives at the time of the first
thrombotic event but discontinued use (9.7 per 1000 patient-years; 95 CI,
4.3-21.5 per 1000 patient-years) (Table 6).
Among women who used oral contraceptives during follow-up, the risk of thrombotic
recurrence was more or less equal in women who had also used an oral contraceptive
at the time of the first thrombotic event (27.3 per 1000 patient-years; 95%
CI, 14.7-50.7 per 1000 patient-years) compared with those who had not used
oral contraceptives at the time of their first event (32.5 per 1000 patient-years;
95% CI, 8.1-130.0 per 1000 patient-years). These rates were higher when only
the years that the oral contraceptives were actually used were taken into
account (Table 6).
Two of the 11 thrombotic events that arose during oral contraceptive
use occurred within 2 weeks after starting use. Two events occurred after
1 year of use, and the other events happened after a longer period of use,
varying between 3 and 9 years. Of the 58 women who used oral contraceptives
during follow-up, 15 had factor V Leiden. Only 1 thrombotic event recurrence
occurred in this group (17.2 per 1000 patient-years; 95% CI, 2.4-122.0 per
To determine the effect of prothrombotic abnormalities on recurrence
risk without the interference of episodes with an increased or decreased risk
of thrombosis, we repeated the analysis while excluding all postoperative
periods (4 weeks following surgical interventions), pregnancy and puerperium
(6 weeks following delivery), periods of oral contraceptive use, and all periods
of anticoagulation treatment. This left 66 events over 2862 patient-years,
for an incidence rate of 23.1 per 1000 patient-years (95% CI, 17.8-29.3 per
1000 patient-years). For those with prothrombotic abnormalities (factor V
Leiden, prothrombin G20210A, hyperhomocysteinemia, deficiencies of the protein
C, protein S, or antithrombin, elevated levels of the factor VIII, factor
IX, factor XI, or fibrinogen), the incidence rate was 24.8 per 1000 patient-years
(95% CI, 18.2-33.1 per 1000 patient-years). For those patients without a prothrombotic
abnormality, the incidence rate was 19.8 per 1000 patient-years (95% CI, 12.1-30.6
per 1000 patient-years) (HR, 1.2; 95% CI, 0.7-2.1). The effects of each of
the prothrombotic defects separately on the risk of spontaneous recurrence
were again equal to those found in the overall analysis (Table 4).
In a large cohort of patients followed up for a prolonged time after
a first venous thrombotic event, we found an annual risk of thrombotic event
recurrence of 2.6%. The cumulative risk of recurrence was 12.4% after 5 years
and 16.5% after 7 years of follow-up. Although the incidence rate was slightly
higher in the first 2 years, at an annual rate of 3.2%, the risk of thrombotic
event recurrence persisted at a high level of more than 2% during the following
years. Others have reported higher recurrence rates of approximately 25% after
5 years of follow-up.12,13 An
important difference of these studies was their inclusion of elderly and cancer
patients—groups likely to have recurrent thrombotic events. Therefore,
our results apply to patients who are younger than 70 years at their first
thrombotic event, who do not have malignancy, but who are otherwise unselected.
We found a few clinical factors that affected the risk of recurrence
(ie, male sex, an idiopathic first thrombotic event, and oral contraceptive
use). Sex and the type of event were related: idiopathic first thrombotic
events were more common in men, while provoked first thrombotic events were
seen almost 5 times more frequently in women because of oral contraceptive
use. Men had a 2.5-fold higher risk of thrombotic event recurrence than women.
This effect of sex was the same in patients with a first idiopathic thrombotic
event as in those with a first provoked thrombotic event. Similarly, the risk
of thrombotic event recurrence in both men and women was higher in those who
had had an idiopathic event. Although sex and type of first thrombotic event
were strongly related, their effects on recurrence were not connected. The
effect of prothrombotic abnormalities was small both in men and women and
could therefore not explain the difference between the sexes. These findings
confirm recent results from a British and an Austrian study.37- 39
Use of oral contraception increased the risk of thrombotic event recurrence.
Advice to refrain from further oral contraceptive use would be a simple and
effective way to reduce the risk of a second thrombotic event in women. We
did not see a major effect of postthrombotic damage on the risk of recurrence.
We found similar rates of thrombotic event recurrence in the ipsilateral and
contralateral leg. This makes it tempting to hypothesize that a systemic effect
is as likely to contribute to the overall recurrence risk as persisting remnants
of the initial clot.40
Sixty-seven percent of the patients had at least 1 prothrombotic abnormality.
In these patients the recurrence risk was only slightly increased (1.4-fold)
compared with those without such abnormalities. In patients with more than
1 abnormality, the recurrence rate was higher than in those with only 1 abnormality
(1.6-fold vs 1.2-fold). The effect of the prothrombotic risk factors separately
varied somewhat but on the whole, they seemed to be a weak determinant of
recurrences. We found no evidence of an increased risk of recurrence for carriers
of factor V Leiden or the prothrombin G20210A mutation. Similarly, we could
not find an excess recurrence risk for individuals with high levels of factors
VIII, IX, or XI. Hyperhomocysteinemia did not show any effect on recurrence
risk either. It should be noted that in the Netherlands vitamin supplementation
is not currently advised to patients with hyperhomocysteinemia. Therefore,
these results represent the natural course of this condition. A mildly increased
risk (1.8-fold) was observed in those with the strongest risk factors for
first thrombotic events, deficiencies of protein C, protein S, and antithrombin.
High fibrinogen levels also conferred a slightly increased risk of recurrence
(1.7-fold). These results are at variance with some other studies in which
increased recurrence risks were found for protein C, protein S, and antithrombin,13 for hyperhomocysteinemia,29,30 for
increased levels of factor VIII and factor IX,27,28 and
for factor V Leiden and prothrombin G20210A.21,22,24,25 These
studies differed considerably with respect to design and study population,
methods, sample size, and duration of follow-up, which could explain the discrepancies.31,32 In a prospective cohort study of
unselected patients with a similar design as our study, no effect of thrombophila
was found either.37 In summary, we saw no major
effect for any of these factors, which is internally consistent because it
is difficult to understand why some prothrombotic abnormalities would increase
the risk of recurrence and others would not.
Venous thrombosis is a multicausal disease.4 Individuals
need a certain combination of risk factors, each adding to the thrombotic
event potential, which exceeds the thrombosis threshold. When patients have
similar thrombotic event potentials, recurrence risks may be similar, too.
This explains the equal risks we found for all thrombophilic defects. In patients
whose first thrombotic eventwas idiopathic, the recurrence rate was equal
in those with and without a prothrombotic abnormality. This can be explained
from the existence of a not yet identified prothrombotic abnormalities, which
hold the same thrombotic event potential as the known prothrombotic abnormalitites.
The recurrence risk was actually only increased in those patients who had
2 or more abnormalities, or in other words, only those with a somewhat higher
thrombotic event potential stood out.
Among patients with thrombophilia, those who had a provoked first thrombotic
event had a recurrence risk that was still lower than patients who had an
idiopathic first thrombotic event. This is remarkable because one would expect
these rates to be equal after the environmental factor (surgery, puerperium)
that contributed to the initial thrombotic event has been removed. This can
only be explained when patients with thrombophilia and an idiopathic first
thrombotic event have a higher thrombotic event potential than patients with
thrombophilia and a provoked first thrombotic event—this could be an
extra (a not yet identified) laboratory risk factor or a local factor such
as an anatomical abnormality.
These findings have important implications for clinical strategies.
Patients with an idiopathic first thrombotic event are often extensively tested
for prothrombotic defects. However, a positive result of a defect does not
predict the risk of thrombotic event recurrence and therefore has no clinical
Our study may be limited with respect to the generalizability of the
findings because we excluded patients older than 70 years and cancer patients.
Our main findings may therefore not be applicable to these groups. Also, in
our study population the use of prophylactic anticoagulation for short periods
was quite high. This may have affected the overall relatively low rate of
recurrences. However, we do not expect that this has had an influence on the
lack of effect we found for the separate prothrombotic abnormalities. In all
analyses, the use of anticoagulation was adjusted for and even when we excluded
all risk-enhancing (surgery, oral contraception use, pregnancy, puerperium)
and risk-decreasing (oral anticoagulation) situations, we did not find an
effect of thrombophilia. Another issue is the diagnosis of thrombotic event
recurrence because a new event is often difficult to distinguish from postthrombotic
syndrome.41 An incorrect classification of
a thrombotic event recurrence could have affected the total rate of recurrent
events. However, the incidence rate of a contralateral second DVT was only
slightly lower than that of an ipsilateral DVT. It is therefore unlikely that
the low rate of recurrent events was overestimated. The distribution of the
prothrombotic risk factors was also equal between ipsilateral and contralateral
DVTs. Therefore, our conclusion with respect to the effect of thrombophilia
would remain unchanged.
In conclusion, patients who had a first thrombotic event had a high
risk of recurrence. This risk is higher in men, in patients whose first thrombotic
event was idiopathic, in women who use oral contraceptives, and in patients
with 2 or more prothrombotic risk factors. Solitary laboratory abnormalities
appear not to predict the risk of recurrence. Therefore, extensive, if any,
thrombophilic work-up after a first thrombotic event is not likely to confer
a clinical benefit to the patient. Similarly, a differential treatment with
regard to duration of oral anticoagulation in patients with prothrombotic
abnormalities does not seem to be rational based on these data. Adequate prophylactic
anticoagulation during risk situations for all patients with a history of
a thrombotic event may be the most important measure to reduce the risk of
a recurrent event. Women using oral contraceptives should be advised to refrain
from further use. The decision on optimal duration of anticoagulation therapy
after a first thrombotic event will probably need to be based on clinical
factors (male sex, oral contraceptive use, and idiopathic first thrombotic
event) rather than laboratory abnormalities.
Corresponding Author: F. R. Rosendaal, Leiden
University Medical Center, PO Box 9600, NL-2300 RC Leiden, the Netherlands
Author Contributions: Dr Rosendaal had full
access to all of the data in the study and takes responsibility for the integrity
of the data and the accuracy of the data analysis.
Study concept and design: Koster, Vandenbroucke,
Acquisition of data: Christiansen, Koster.
Analysis and interpretation of data: Christiansen,
Cannegieter, Koster, Rosendaal.
Drafting of the manuscript: Christiansen, Cannegieter,
Critical revision of the manuscript for important
intellectual content: Christiansen, Cannegieter, Koster, Vandenbroucke,
Statistical analysis: Christiansen, Cannegieter,
Obtained funding: Rosendaal.
Administrative, technical, or material support:
Study supervision: Cannegieter, Vandenbroucke,
Financial Disclosures: None reported.
Funding/Support: The LETS study was funded
by grant 89.063 from the Netherlands Heart Foundation and the follow-up study
was funded by grant 2827170 from the Prevention Fund/ZonMW.
Role of the Sponsors: The funding organizations
are public institutions and had no role in the design and conduct of the study;
collection, management, analysis, and interpretation of the data; and preparation,
review, or approval of the manuscript.
Acknowledgment: We are grateful to the personnel
of the Anticoagulation Clinics of Leiden, Rotterdam, and Amsterdam who facilitated
the inclusion of the patients. We thank Ank Schreijer, Ingeborg de Jonge,
and Inge Noordermeer for data management and all participating patients for