Among women in the general population between 20 and 40 years of age, the annual frequency of deep venous thrombosis is 1.8% [3]. In other observational studies [4-7], the frequency of venous thromboembolism has varied from 1.3% to 7% during pregnancy and from 6.1% to 23% during the postpartum period. These findings indicate that the risk for venous thromboembolism might be increased only moderately during pregnancy but that it is clearly enhanced during the postpartum period. However, the absolute frequencies during both of these periods remain low.

In contrast, in women who have an inherited deficiency of a naturally occurring anticoagulant-antithrombin, protein C, or protein S-it has been reported that pregnancy and the postpartum period are associated with a greatly increased risk for venous thromboembolism [8-11]. During pregnancy, the observed frequency of venous thromboembolism per woman per pregnancy varied from 12% to 48% in antithrombin-deficient women and from 2% to 8% in protein C-deficient women; no thrombosis was seen in protein S-deficient women [8-11]. During the postpartum period, the following frequencies were seen: 28% to 47% in antithrombin-deficient women, 11% to 20% in protein C-deficient women, and 14% in protein S-deficient women [8-11].

However, the usefulness of these data for clinical decision making is limited because the studies that produced the data lacked appropriate control groups and included many patients who were identified because they presented with venous thrombosis, either associated with or unrelated to pregnancy. In addition, most of the identified thrombotic events were diagnosed on the basis of clinical findings only, which are known to be nonspecific, particularly during pregnancy [12]. Hence, the frequencies reported previously are likely to be overestimates.

As a result, the best approach to anticoagulative prophylaxis for venous thromboembolism in anticoagulant factor-deficient women during pregnancy and the postpartum period is actively debated. Advocated regimens range from surveillance combined with noninvasive tests for deep venous thrombosis only to prophylaxis with therapeutic doses of heparin and oral anticoagulants throughout pregnancy and the postpartum period [13, 14] combined with intravenous administration of the lacking proteins.

Also debated are the maternal and fetal adverse effects associated with the use of heparin and oral anticoagulants during pregnancy. The administration of coumarins during pregnancy may cause embryopathy [15], and the use of heparin has been associated with osteoporosis, which may result in bone fractures [16, 17]. Both of these drugs may also induce hemorrhagic complications, especially during delivery [15, 18].

Because the lack of accurate clinical data on the frequency of venous thromboembolism contributes to a wide variation in clinical practice regarding this event, physicians obviously need better information about the frequency of venous thromboembolism in anticoagulant factor-deficient women. We therefore sought to determine the frequency of venous thromboembolism during pregnancy and the postpartum period among otherwise asymptomatic women with a deficiency of antithrombin, protein C, or protein S. We investigated all female family members of probands known to have a deficiency of one of these factors and assessed the frequency of pregnancy-related venous thromboembolism in this group. The deficiency status of the female family members was determined only after a careful, structured history was obtained. The female family members found to be nondeficient were used as a representative control group.

Methods

Top Methods Results Discussion References

Patients

Female members of 69 families that had a documented deficiency of antithrombin, protein C, or protein S were investigated. The study participants were identified through the family trees of unselected patients who had an objective diagnosis of venous thromboembolism and were referred to the participating study centers (Academic Medical Center, Amsterdam, the Netherlands; Institute of Medical Semeiotics, Padua, Italy). These probands were excluded from further study.

All of the women were interviewed by an investigator blinded to anticoagulant factor-deficiency status. A medical history, with attention to episodes of venous thromboembolism and to events in the obstetric history (such as pregnancy, childbirth, and postpartum periods), was obtained from each participant. An episode of venous thromboembolism was considered to have occurred only if it had been documented by objective tests (ultrasonography, impedance plethysmography, or venography for deep venous thrombosis; ventilation-perfusion lung scanning or pulmonary angiography for pulmonary embolism) or if it had been clinically diagnosed and the patient had been treated with anticoagulative drugs for at least 3 months. If a patient reported having had a venous thromboembolic event, further medical information was sought and reviewed by one of the investigators to establish the methods that had been used to diagnose the event.

A venous thromboembolic event was considered to be pregnancy related if it occurred during pregnancy or within 3 months after childbirth. Episodes of pregnancy-related venous thromboembolism were classified according to the period in which they occurred: first trimester of pregnancy, second trimester of pregnancy, third trimester of pregnancy, puerperium ( 7 days after delivery), and the remaining postpartum period. If a patient had an episode of venous thromboembolism, all subsequent pregnancies were excluded from the analysis to avoid risk enhancement and because anticoagulative prophylaxis is often given after such an episode.

After the history was recorded, blood samples were collected for the determination of anticoagulant factor status. Blood samples (20 mL) were collected by venipuncture with 21-gauge butterfly infusion sets into a plastic syringe containing 3.8% sodium citrate; the ratio of the volume of anticoagulant to the volume of blood was 0.1:0.9. Plateletpoor plasma was obtained by using centrifugation at 2000 g for 20 minutes and was stored at –80°C until it was analyzed.

Antithrombin antigen concentrations were measured using the Asseraplate Antithrombin III Kit (Boehringer Mannheim, Mannheim, Germany); antithrombin activity was measured using Berichrom ATIII (Behringwerke, Marburg, Germany).

Protein C antigen concentrations were measured with enzyme-linked immunosorbent assay (ELISA) using rabbit anti-protein C polyclonal antibody (DAKO, Glostrup, Denmark) as catching antibody. Rabbit anti-protein C polyclonal horseradish peroxidase conjugated antibody (DAKO) was used as the second antibody according to the manufacturer's instructions. Protein C activity was measured using the Protein C Reagent Kit (Behringwerke).

Concentrations of total and free protein S were measured by ELISA using rabbit anti-protein S polyclonal antibody (DAKO). The 15C4 anti-protein S monoclonal antibody (Serbio, Gennevilliers, France) was used as catching antibody, and the rabbit anti-protein S polyclonal horseradish peroxidase conjugated antibody (DAKO), diluted 1:1000, was used as the second antibody. The 15C4 anti-protein S monoclonal antibody recognized only free protein S antigen. Protein S activity was measured using the Protein S IL-Kit (Instrumentation Laboratories, Milan, Italy).

A participant was considered to be deficient if repeated tests done 1 month apart showed values that were subnormal for the protein deficiency in that participant's family. The following reference values were used: antithrombin antigen concentration, 0.80 to 1.20 U/mL; antithrombin activity, 0.80 to 1.20 U/mL; protein C antigen concentration, 0.70 to 1.30 U/mL; protein C activity, 0.70 to 1.30 U/mL; total protein S concentration, 0.70 to 1.20 U/mL; and free protein S concentration, 0.26 to 1.08 U/mL. The criteria used to classify antithrombin, protein C, and protein S deficiencies accord with those reported in the current literature [19].

Activated partial thromboplastin time and prothrombin time were determined in an effort to exclude vitamin K deficiency. Activated partial thromboplastin time was measured by using ActinFS (Dade, Miami, Florida) (normal range, 25 to 36 seconds); prothrombin time was measured by using Thromboplastin IS (Dade) (normal range, 11 to 14 seconds). During the time of the laboratory investigation, none of the study participants were pregnant or in the postpartum period [20-23].

The participants were categorized according to deficiency status into a nondeficient group and a deficient group. The deficient group was further subdivided according to specific deficiency.

Statistical Analysis

In each group, the frequency of venous thromboembolism was calculated by dividing the number of pregnancy-related venous thromboembolic episodes by the total number of recorded pregnancies and by the total number of women. The percentages of deficient and nondeficient women who had thromboembolic episodes were compared using the Fisher exact test. Using a Cox model, we calculated the hazard ratio (and its mid-p corrected 95% CI) for pregnancy-related venous thromboembolism in deficient women compared with nondeficient women (SAS, Inc., version 6.11, Cary, North Carolina). The P values were calculated by using the exact log-rank test. For this purpose, a pregnancy was regarded as the unit of time and women who had not had a venous thromboembolic event by the end of their last pregnancy were considered to be censored.

Results

Top Methods Results Discussion References

Of the female members of the 69 families, 282 women were potentially eligible for the study. Of these, 52 could not be contacted because they lived abroad (n = 12) or had lost contact with the family (n = 40). Twenty others refused to participate. Thus, 210 women were interviewed. Of these, 129 had been pregnant at least once. Sixty-nine participants (53%) were nondeficient and 60 (47%) were deficient (13 had antithrombin deficiency, 19 had protein C deficiency, and 28 had protein S deficiency). Mean age at the time of first childbirth was 25 years (range, 16 to 40 years) in the nondeficient group and 26 years (range, 17 to 43 years) in the deficient group. Similar findings were obtained separately in the groups with deficiency of antithrombin, protein C, and protein S.

The 129 women who participated in the study had a total of 367 qualifying pregnancies (pregnancies that occurred before a thrombotic event occurred). In the nondeficient group, 198 pregnancies (54%) were recorded; 169 pregnancies (46%) were recorded in the deficient group. Thirty-three pregnancies were recorded in the antithrombin-deficient group, 60 were recorded in the protein C-deficient group, and 76 were recorded in the protein S-deficient group.

In eight cases, pregnancy or the postpartum period was associated with an episode of venous thromboembolism (Table 1). In three of these eight cases, the diagnosis had been established by objective testing; in the other five cases, the diagnosis had been made on the basis of clinical signs and symptoms and had been followed by at least 3 months of anticoagulative drug therapy. One of the 198 pregnancies (0.5%) in the nondeficient group was complicated by an episode of venous thromboembolism, whereas thromboembolism occurred in 7 of 169 pregnancies (4.1%) in the deficient group (1 of 33 pregnancies in the antithrombin-deficient group [3.0%], 1 of 60 pregnancies in the protein C-deficient group [1.7%], and 5 of 76 pregnancies in the protein S-deficient group [6.6%]). Thus, overall, 1 of the 69 women in the nondeficient group (1.5% [95% CI, 0.0% to 7.8%]) and 7 of the 60 women in the deficient group (11.7% [CI, 4.8% to 22.6%]) had an episode of pregnancy-related venous thromboembolism (P = 0.02 for the difference). In the deficient group, two episodes of venous thromboembolism occurred during pregnancy, resulting in a frequency of 1.2% of pregnancies, and five episodes occurred during the postpartum period, resulting in a frequency of 3% of pregnancies. The risk for venous thromboembolism during pregnancy and the postpartum period was increased eightfold in deficient women compared with nondeficient women (hazard ratio, 8.0 [CI, 1.2 to 184]; P < 0.03). The assumptions underlying the Cox model were met, and no observations had an unduly high influence on the model. The occurrence of a first thromboembolic event in consecutive pregnancy and postpartum periods is shown in Table 2.

In the single nondeficient patient (patient 1; Table 1) with deep venous thrombosis, the event occurred during immobilization that resulted from a fracture of the pelvic bone that occurred during childbirth. This patient was treated with heparin and oral anticoagulants. She had several more subsequent episodes of venous thromboembolism.

The one antithrombin-deficient patient with deep venous thrombosis (patient 2) had this event during the third trimester of pregnancy. Three weeks later, she had a stillbirth at 38 weeks of pregnancy. Two years after that event, the patient had a recurrent period of deep venous thrombosis. In the one protein C-deficient patient who had deep venous thrombosis (patient 3), the event occurred during the third trimester of pregnancy and was followed by pulmonary embolism 3 days later. All five of the protein S-deficient patients who had venous thromboembolism (patients 4 to 8) had this event during the postpartum period; two of these episodes (those in patients 4 and 7) occurred during the puerperium. Two of the patients (patients 2 and 7, who were deficient in protein C and protein S, respectively) had several episodes of recurrent deep venous thrombosis.

Discussion

Top Methods Results Discussion References

We sought to obtain a more accurate understanding of the frequency of venous thromboembolism during pregnancy and the postpartum period in anticoagulant factor-deficient women who had not previously had an episode of venous thromboembolism.

Venous thromboembolism was seen, overall, in 4.1% of pregnancies in anticoagulant factor-deficient women; two episodes (1.2%) occurred during the third trimester, and five episodes (3.0%) occurred during the postpartum period. Overall frequencies for women with specific deficiencies were 3.0% of pregnancies for women with antithrombin deficiency, 1.7% of pregnancies for women with protein C deficiency, and 6.6% of pregnancies for women with protein S deficiency. The estimated overall hazard ratio of 8.0 [CI, 1.2 to 184] for venous thromboembolism during pregnancy among deficient women compared with nondeficient women is close to 1) the odds ratios (9.8 to 13.7) for all cases of venous thromboembolism reported in a literature review [24] that included available cross-sectional family studies and 2) the odds for venous thromboembolism outside of pregnancy and the postpartum period that have been found in other studies of families with anticoagulation factor deficiencies [25-27]. However, the absolute frequencies are substantially lower than those reported earlier, which were 12% to 48% in antithrombin-deficient women, 2% to 8% in protein C-deficient women, and 0% in protein S-deficient women during pregnancy and 28% to 47% in antithrombin-deficient women, 11% to 20% in protein C-deficient women, and 14% in protein S-deficient women during the postpartum period.

The most likely explanation for this discrepancy is the avoidance of selection bias in our study. We accomplished this by excluding the probands from the analysis and including participants in the study before deficiency status or history of previous thrombotic events was known. Similar studies done in women with factor V Leiden mutation are currently being done, but the data from these studies are not yet available.

Several important methodologic issues in our study warrant comment. In the group with anticoagulant factor deficiency, three thromboembolic episodes had been diagnosed with objective tests and the other four (as well as the single episode in the nondeficient group) had been diagnosed on the basis of clinical grounds alone. This method of diagnosis was used because, in most patients, the diagnosis had been established before objective diagnostic testing was generally available. Because it has been shown that the clinical diagnosis of venous thromboembolism is inaccurate, especially during pregnancy, the low frequencies found in our study could be overestimates. However, the relative risk can be considered accurate and realistic because we included an appropriate control group and determined the deficiency status of participants only after the medical history was taken. Moreover, because our study was retrospective, our results could have been biased by the presence of undetected excess mortality in the deficient group (that is, some women at high risk for thromboembolism may not have survived to become pregnant). However, recent reports [28, 29] have not shown excess mortality among anticoagulant factor-deficient persons.

To what extent do our results help physicians decide whether to administer or withhold anticoagulative prophylaxis during pregnancy and the postpartum period in asymptomatic women with a deficiency of antithrombin, protein C, or protein S? The observed overall frequency of pregnancy-related symptomatic venous thromboembolism-4.1%-is low relative to previously reported frequencies per pregnancy, but it is similar to the frequency of symptomatic venous thromboembolism that was seen in patients having hip surgery who had not received anticoagulative prophylaxis [30, 31].

Given the efficacy of subcutaneous low-molecular-weight heparin in the prevention of deep venous thrombosis in patients having major surgery [32], prophylaxis with this agent could be advocated for pregnant, anticoagulant factor-deficient women who have not previously had venous thromboembolism. Such an approach is likely to avoid the reported hemorrhagic complications during childbirth that are associated with high-dose unfractionated heparin [15, 18], provided that the low-molecular-weight heparin therapy is discontinued in a timely manner before delivery.

Intrinsic to the decision about administering anti-coagulative prophylaxis during pregnancy is the selection of a proper time to start this prophylaxis. In our study, venous thromboembolism primarily occurred during the postpartum period; this agrees with data reported earlier [10, 11]. Ginsberg and colleagues [33], who studied pregnant women with deep venous thrombosis whose deficiency status was not assessed, reported an equal distribution of venous thromboembolism over all trimesters of pregnancy [33]. Currently available data suggest that the highest prophylactic effectiveness would be obtained with the use of anticoagulative prophylaxis around the period of childbirth. Because the prevention of venous thromboembolism throughout the entire pregnancy would necessitate the administration of anticoagulants for a much longer period, the added beneficial effect, expressed as episodes of venous thromboembolism prevented per month of treatment, would be relatively small.

The therapeutic option of using prophylactic dosages of low-molecular-weight heparin around the time of delivery is considered valid because it is assumed that the risk for major hemorrhagic complications associated with the use of low-molecular-weight heparin during the third trimester of pregnancy and the postpartum period is similar to the 1% risk seen in patients having hip surgery [32]. However, one might argue that if this drug is administered for a period longer than the third trimester and the postpartum period combined, this risk for major hemorrhage will be higher and thus may outweigh the benefits of treatment. Given this consideration, the alternative strategy of surveillance combined with noninvasive tests for deep venous thrombosis (watchful waiting) throughout pregnancy may constitute an option to be considered.

In conclusion, the risk for venous thromboembolism during pregnancy and the postpartum period is increased in women with a deficiency of antithrombin, protein C, or protein S compared with women without these deficiencies. On the basis of the absolute risk for venous thromboembolism in the period surrounding childbirth, we suggest that anti-coagulant factor-deficient patients should receive anticoagulative prophylaxis with low-molecular-weight heparin during the third trimester and the postpartum period. In the absence of data from randomized trials, however, these recommendations must be considered tentative. Such trials are needed to convert this preliminary advice into a well-defined, evidence-based guideline.

From the Academic Medical Center, Amsterdam, the Netherlands; and the Institute of Medical Semeiotics, University of Padua, Padua, Italy.

Drs. Simioni, Zanardi, Prandoni, and Girolami: Institute of Medical Semeiotics, via Ospedale 105, 35100 Padua, Italy.

Dr. Huisman: Department of Internal Medicine, University Hospital Leiden, Rijnsburgerweg 10, PO Box 2333AA, Leiden, the Netherlands.

Dr. Buller: Center for Hemostasis, Thrombosis, Atherosclerosis, and Inflammation Research, Academic Medical Center, Meibergdreef 9, 1105 AZ Amsterdam, the Netherlands.