Professor Ingrid Pabinger MD, Department of Internal Medicine I, Division of Haematology and Blood Coagulation, Waehringer Guertel 18–20, A 1090 Vienna, Austria. Tel.:+43 140400 4409; fax:+43 14026930; e-mail: firstname.lastname@example.org
Summary. Background: Limited data exist on the risk of pregnancy-associated venous thromboembolism (VTE) in women with a history of VTE. Objective: To evaluate the risk of recurrent pregnancy-associated thrombosis in women with previous VTE in a large retrospective cohort study. Patients and methods: One hundred and fifty-nine women with at least one pregnancy (293 pregnancies in total) after a VTE were included into the study. The patients underwent a standardized interview on their history of thrombosis and pregnancy-associated complications. Results: Eight recurrent events occurred during 197 pregnancies without thrombosis prophylaxis. The probability of VTE during pregnancy without thrombosis prophylaxis was 6.2% (95% confidence interval 1.6–10.9%). The risk was constant over the whole period of pregnancy. Of the eight women with VTE during pregnancy four had heterozygous FV:R506Q, two in combination with hyperhomocysteinemia. No VTE occurred during 87 pregnancies with thrombosis prophylaxis. In the postpartum period 15 VTEs occurred, two of 83 (2.4%) after pregnancy termination, one of 53 (1.9%) after miscarriage, three of 10 (30%) after stillbirth and nine of 138 (6.5%) after live birth. Conclusions: Without thrombosis prophylaxis the risk for recurrent symptomatic VTE is substantial during the whole period of pregnancy in women with previous VTE. The majority of events occurred after delivery, reflecting the very high risk during the postpartum period. Prospective and comparative trials to ascertain efficacy and safety of prophylactic heparin are urgently needed.
Venous thromboembolism (VTE) during pregnancy is a serious complication and a diagnostic and therapeutic challenge. In the general population the incidence of pregnancy-associated VTE has been estimated to vary from one in 1000 to one in 2000 deliveries [1,2]. The risk of VTE is five times higher in a pregnant woman than in a non-pregnant woman of similar age . In individuals with well-defined hereditary thrombosis risk factors, such as the factor V:R506Q mutation, the factor II:G20210A variation, antithrombin deficiency or protein C deficiency, a relative risk of pregnancy-associated VTE between 3.4 and 15.2 has been found . For women with heterozygous factor V:R506Q an incidence of thrombosis of one in 400–500  for those with homozygous factor V:R506Q of approximately one in 25  was estimated.
Limited data exist on the pregnancy-associated VTE risk in women with a history of VTE. In a retrospective evaluation including 72 patients a recurrence rate of 12% was found. An antepartum recurrence rate of only 2.4% was found in a recently published prospective study on 125 women, who had a pregnancy after a venous thromboembolic event and did not receive prophylaxis . Patients with known thrombophilia were excluded from enrollment in this prospective study and observation started at a mean duration of pregnancy of 15 ± 6 SD weeks.
To evaluate the incidence of recurrent VTE over the whole period of pregnancy we conducted a retrospective study in a large cohort of women with previous thrombosis, who had had at least one pregnancy after VTE.
Patients and methods
Nine hundred and seventy-three consecutive women with a history of VTE that had occurred between 13 and 45 years of age were investigated for risk factors of thrombosis between January 1985 and December 1998 in our department. In all patients VTE was documented by objective methods (phlebography, duplex ultrasonography, perfusion ventilation lung scan or computed tomography). These women were invited for reinvestigation. Five hundred and seven women accepted our invitation, had a blood sample drawn and underwent a standardized interview on their history of thrombosis and pregnancy-associated complications. Recurrent thrombotic events were categorized as definite when they were objectively confirmed, and as probable when the clinical feature was typical and had led to anticoagulant treatment. A VTE event was defined as pregnancy-associated when it occurred during pregnancy or within 6 weeks after delivery, or after the termination of pregnancy due to other reasons (for example, miscarriage). Of 507 women, 112 had never been pregnant; 395 reinvestigated women had been pregnant at least once and 159 women had had at least one pregnancy after VTE and were included in the present study. Twenty-five of 159 patients had been enrolled in the prospective AUREC study . The interval between current investigation and last VTE was at least 4 months.
There were no uniform decisions as to whether or not a woman should receive heparin prophylaxis during pregnancy. Women who contacted our department when they were pregnant were advised to use thrombosis prophylaxis during pregnancy and were prescribed standard heparin (2–3 times 5000 Units/day) or, as soon as licensed for pregnant women, low molecular weight heparin (LMWH) (4000–5000 anti Xa Units/day). Those with antithrombin deficiency or a lupus anticoagulant were specifically informed to contact our department in case of pregnancy. Women who were cared for in other hospitals mainly did not receive thrombosis prophylaxis during pregnancy.
Plasma samples were obtained from patients after overnight fasting and centrifuged at 2000 g for 20 min. Coagulation tests (lupus anticoagulant, factor VIII) were performed within 3 h of blood sampling. For determination of natural coagulation inhibitors plasma was frozen at −20 °C until analysis. For determination of homocysteine, samples were cooled immediately at 4 °C and centrifuged within 30 min of sampling, snap-frozen and stored at – 70 °C. Diagnosis of the lupus anticoagulant was made according to the criteria of the International Society of Thrombosis and Haemostasis  using two different screening tests (activated partial thromboplastin time and diluted Russels viper venom time) and confirmatory tests, as described previously . Antithrombin and protein C activity were determined on the analyzer STA. Free Protein S antigen was determined by enzyme-linked immunosorbent assay (ELISA). Factor VIII clotting activity (95th percentile of 307 healthy individuals 248%) was determined by one-step clotting assay on a KC 10 coagulometer (Amelung, Lieme, Germany). Total homocysteine (normal range for women < 13.6 μmol/L, for men < 16.3 μmol/L) was determined using a high performance liquid chromatography (HPLC) kit by Immunodiagnostic, Bensheim, Germany, as described previously . Analyses of the factor V:R506Q and the factor II:A20210G genotypes were performed by multiplex polymerase chain reaction (PCR) following the general principle of mutagenically separated PCR. An individual heterozygous for both mutations was included in each experiment as positive control.
Variables of interest are described by mean ± standard deviation (SD) if not mentioned otherwise. Kaplan–Meier's  method was used to estimate the probability of VTE during pregnancy. Pregnancies without VTE ending by termination, miscarriage or stillbirth were censored at the respective termination times. Pregnancies without VTE ending with live birth were considered as completely observed and censored at time of delivery. The estimated probability of VTE at the end of pregnancy is mentioned together with the 95% confidence interval. Fisher's exact test was used to compare the frequency of recurrent VTE in patients with and without a thrombosis risk factor and to evaluate the importance of precipitating conditions at first event.
One hundred and fifty-nine women were pregnant at least once after a VTE and were included into the study. The actual mean age of the study patients was 39.5 ± 11.3 years; the mean age at first VTE was 25 ± 6.3 years. One hundred and nine women had at least one pregnancy without thrombosis prophylaxis. At the initial episode only 18 women had VTE without any temporary risk factor, 141 had a temporary risk factor; corresponding data for the 109 women who did not receive thrombosis prophylaxis during pregnancy were 15 and 94 women. In the following, numbers applying to the group of women without thrombosis prophylaxis in one of their pregnancies (n = 109) are additionally given. The most common temporary risk factor was oral contraceptive (OC) use, including patients without, and those with other additional precipitating conditions (n = 116 and 72, respectively). Other temporary risk factors were surgery in 19 and 14 women (eight and five, respectively, had additional OC), trauma in 16 and 11 (11 and six, respectively, with OC), immobilization in seven each (six and five, respectively, with OC), prolonged bed rest due to acute bacterial or viral infection in six of the total group (four with OC), and other conditions in two each. The first VTE occurred during pregnancy in 16 and five patients, after abortion in two and during the postpartum period in 24 and 15 women. Laboratory thrombosis risk factors are listed in Table 1.
Table 1. Thrombosis risk factors in all 159 women and in those 109 women without thrombosis prophylaxis during pregnancy
All women n = 159
Women without prophylaxis n = 109
In 141 patients all thrombosis risk factors were determined,
**in 101 women all thrombosis risk factors were determined.
The 159 women had in total 293 pregnancies after a VTE. Nine pregnancies (all without VTE) were excluded from evaluation as information on thrombosis prophylaxis was not available. During 197 pregnancies in 109 women, no thrombosis prophylaxis was given; of those, eight pregnancies (five deep leg, one deep arm VT, two pulmonary embolism, all definite) were complicated by VTE during pregnancy (Fig. 1a). All women had only a single thrombotic event prior to the recurrent event during pregnancy.
Five events occurred during the first, two during the second and one during the third trimester. The probability for development of thrombosis during pregnancy in patients without prophylaxis (Fig. 2) was 6.2% (95% confidence interval (CI) 1.6–10.6%) derived from the Kaplan–Meier analysis. Of the eight women with VTE during pregnancy, four had heterozygous factor V:R506Q, two in combination with hyperhomocysteinemia (Table 2). For evaluation of the impact of laboratory risk factors on the recurrence rate we have restricted analysis to the 101 women without heparin prophylaxis and full laboratory evaluation. The frequency of pregnancy-associated recurrent thrombosis was four of 62 (6.6%) with and four of 39 (10.2%) without laboratory risk factors (P = 0.7). With regard to a temporary risk factor at first thrombosis, none of the 15 without any temporary risk factor and eight of 94 (8.5%) with a temporary risk factor had a recurrent event (P = 0.6). When the analysis was restricted to pill intake as temporary risk factor at first thrombosis, a higher number of women with pill intake at first thrombosis (seven of 72, 10%) in comparison to those who developed their first thrombosis without pill intake (one of 37, 2.7%) had recurrent pregnancy associated thrombosis; however, the difference was statistically not significant (P = 0.3).
Table 2. Women with VTE during pregnancy (all without prophylaxis)
The median time period between the first VTE and recurrent pregnancy associated VTE in the eight women was 5.8 years (range 0.5–13.5 years). In comparison, in the 101 women without thrombosis prophylaxis and without VTE the median time interval between VTE and first pregnancy after VTE was 2.6 years (range 0.4–15 years).
During 87 pregnancies, thrombosis prophylaxis with either 5000 Units standard heparin twice daily or prophylactic doses of low molecular weight heparin (Dalteparin or Enoxaparin) were given. In none of these pregnancies had a recurrent VTE occurred.
Outcome of pregnancy and the frequency of thrombotic events after end of pregnancy are summarized in Fig. 1b; there were 138 life births, 10 stillbirths, 53 miscarriages and 83 terminations. One surgical removal of extrauterine pregnancy and one induced abortion were complicated by VTE. The latter patient had received prophylaxis with 5000 U Dalteparin once daily. Of 53 pregnancies ending prematurely with miscarriage, one was complicated by VTE (without prophylaxis). Stillbirth occurred in 10 pregnancies, three were complicated by VTE (all without prophylaxis). One hundred and thirty-eight pregnancies ended with delivery of a viable infant; in nine cases a venous thromboembolic event (eight definite, one probable event) occurred within 6 weeks after delivery. Of 65 women without thrombosis prophylaxis, four (6.1%) experienced VTE and five of 73 (6.9%) women had thrombosis despite heparin prophylaxis (Table 3).
Table 3. Prevalence of VTE after delivery/end of pregnancy in women with and without thrombosis prophylaxis
In women with a history of VTE we found a probability of 6.2% (95% CI 1.6–10.6%) of antepartum recurrent symptomatic VTE without thrombosis prophylaxis over the full length of pregnancy estimated by Kaplan–Meier analysis. Due to the complete data collection we were able to estimate the risk of thrombosis throughout the full length of pregnancy. The risk appears to be constant over the whole period of pregnancy including the first trimester. It is therefore very probable that not mechanical factors but hormonal changes occurring already very early in pregnancy are mainly responsible for the increased risk. This is also compatible with the fact that women with antithrombin deficiency have a high risk of thrombosis from the beginning of pregnancy [14,15].
We found a somewhat higher risk of thrombosis during pregnancy compared to the very low recurrence rate described in a Canadian study published by Brill-Edwards et al. , who reported an antepartum recurrence of VTE in three of 125 women (2.4%). However, in both studies the risk for recurrent VTE was clearly below 10% and the confidence intervals are overlapping. The lower risk in the prospective study of Brill-Edwards et al. may be due to the fact that the women were included at a median duration of pregnancy of 15 ± 6 weeks, whereas our retrospective evaluation covered the whole pregnancy period. It cannot be excluded completely that the slightly higher risk in our study is due to a selection bias, preferentially including women with recurrence. Because the overall recurrence rate (16) is low in our patient population, a considerable selection towards women with a higher recurrence rate is unlikely. In the retrospective analysis by Tengborn et al. , a recurrence rate of 12% (eight during 67 pregnancies) in two small prospective studies, a recurrence rate of one of 20 (5%)  and one of 26 (4%)  women without antenatal prophylaxis was found. In another retrospective analysis including pregnancy and the puerperium a recurrence rate of 12% was described . As has been published previously by our group, the risk of thrombosis is definitely higher during pregnancy compared to non-pregnant periods .
The proportion of women with congenital abnormalities was not higher than in the overall population of women with thrombosis. The presence of a risk factor for thrombosis did not help to differentiate clearly between women at high or low risk for pregnancy associated recurrence. Each woman with a recurrence during pregnancy had a temporary risk factor at first thrombotic event. Because only a small number of women had no temporary risk factor at first thrombosis (11.3%), we cannot draw any conclusions from these data. There was a long time period between VTE and pregnancy-associated thrombosis. Therefore, we suggest that not only women who become pregnant shortly after a VTE are at increased risk.
None of the 87 pregnancies with heparin prophylaxis was complicated by a thromboembolic event. As this is a retrospective analysis we have not performed statistical comparisons between pregnancies with and without heparin. Although this is not a prospective and controlled trial, our results suggest that heparin is able to reduce the risk of recurrence during pregnancy.
There is no consensus as to whether all women with previous thrombosis should receive prophylaxis during pregnancy. In the Seventh ACCP Consensus Conference on Antithrombotic Therapy  administration of prophylactic heparin is not recommended routinely, but is advised in those with thrombophilia and spontaneous thrombosis. It has been shown convincingly that low molecular weight heparin is safe during pregnancy [21,22].
In the postpartum period 15 women from our study suffered from VTE. These data indicate and confirm that the risk of VTE is higher in the postpartum period. The rate of thrombosis in the postpartum period was highest after stillbirth (three of 10). Five of 15 thrombotic events occurred despite heparin prophylaxis at usual dosages. This illustrates the high risk shortly after delivery or termination of pregnancy due to other reasons. Also in the prospective study by Brill-Edwards et al. postpartum thrombosis occurred despite patients receiving thrombosis prophylaxis. It might be concluded that usual prophylactic dosage regimens are not fully effective during this period of very high risk; it is possible that higher doses or other substances are required to protect women from postpartum recurrence.
The rate of induced pregnancy termination was high in our patient group. It can be speculated that at least some of these terminations were triggered by an anticipated increased risk for pregnancy-induced VTE. However, it is not possible to make any comparisons with the normal population as no systematic registry on pregnancy termination exists in Austria.
There are certain limitations due to the retrospective design of the study. A recall bias cannot be excluded. However, it is very unlikely that symptomatic recurrent thrombosis or pulmonary embolism are not remembered and reported by the patients. The retrospective nature of the study has important advantages because we could evaluate the whole period of pregnancy including the first trimester, whereas the very early phase of pregnancy is probably missed in a prospective study and may therefore lead to an underestimation of the risk.
We conclude that the risk for recurrent symptomatic VTE is substantial during the whole period of pregnancy without thrombosis prophylaxis in women with a history of VTE. The risk of VTE is highest after delivery and may occur despite heparin prophylaxis, reflecting the very high risk during the postpartum period. Prospective and comparative trials are urgently needed to improve patient care in this specific situation.
The skillful technical assistance of Laura Ovissi and Silvia Koder is acknowledged.