GHERMAN, ROBERT B. MD; GOODWIN, T. MURPHY MD; LEUNG, BELINDA MD; BYRNE, JAMES D. MD; HETHUMUMI, REHENA; MONTORO, MARTIN MD
Objective: To estimate the incidence, timing, and associated clinical characteristics of objectively diagnosed pregnancy-associated venous thromboembolism.
Methods: We retrospectively reviewed venous thromboembolism cases (deep venous thrombosis and pulmonary embolism) that occurred between 1978 and 1996. Study inclusion criteria required the objective diagnosis with either Doppler ultrasound, venography, impedance plethysmography, pulmonary angiography, ventilation-perfusion scanning, or computed tomography or magnetic resonance imaging.
Results: Among 268,525 deliveries there were 165 (0.06%) episodes of venous thromboembolism (one per 1627 births). There were 127 cases of deep venous thrombosis and 38 cases of pulmonary embolism. Only 14% (23 of 165 patients) had a history of venous thromboembolism. Most cases of deep venous thrombosis were in the left leg (104 of 127, 81.9%), with nearly three quarters of them (94 of 127, 74.8%) occurring during the antepartum period. Among cases of antepartum deep venous thrombosis, half were detected before 15 weeks' gestation (47 of 95, 49.5%), and only 28 cases occurred after 20 weeks (P < .001). Most of the pulmonary embolisms occurred in the postpartum period (23 of 38, 60.5%) and were strongly associated with cesarean delivery (19 of 36,470 compared with four of 232,032, P < .001).
Conclusion: The incidence of venous thromboembolism during pregnancy is lower than has been previously described. Most cases occurred in the antepartum period, with the risk of deep venous thrombosis appearing to begin even before the second trimester.
Despite recent advances in diagnosis and treatment, venous thromboembolism remains a significant cause of obstetric morbidity and mortality. Pulmonary embolism continues to be the leading cause of maternal death after a live birth.1,2 Each component of Virchow's triad (venous stasis, increases in coagulation factors, and tissue trauma) is present at some point during pregnancy; therefore, venous thromboembolism is five times more common in pregnant women.3,4 Venous stasis occurs as capacitance vessels increase in diameter while venous return is diminished by the pressure from the gravid uterus on the iliac veins and vena cava. Increases in coagulation factors lead to a hypercoagulable state. Both vaginal and cesarean delivery lead to tissue trauma and subsequent disruption of the vascular endothelium.
Because deep venous thrombosis and pulmonary embolism occur infrequently, important issues concerning their natural history and therapy remain unresolved. In addition, most previous studies evaluating pregnancy-associated thromboembolic events have used clinical criteria only for diagnosis. Two recent review articles5,6 noted ongoing debate with respect to the overall frequency of thromboembolism and timing of thromboembolic events during pregnancy. We therefore sought to estimate the incidence, timing, and clinical characteristics of objectively diagnosed pregnancy-associated thromboembolism.
Materials and Methods
We retrospectively reviewed consecutive cases of deep venous thrombosis and pulmonary embolism at Los Angeles County and University of Southern California Women's Hospital between January 1, 1978, and December 31, 1996. This study includes some cases of venous thromboembolism previously reported in abstract form (Rutherford S, Montoro M, McGehee W, Strong T. Thromboembolic disease associated with pregnancy: An 11-year review [abstract]. Am J Obstet Gynecol 1991;164:286). Our hospital is a large tertiary care referral center, with a primarily indigent Hispanic patient population. We initially identified cases by International Classification of Diseases, ninth revision discharge diagnosis code and by review of obstetric, radiologic, and coagulation laboratory databases. Only cases with objective evidence of thromboembolism were eligible for inclusion. Patients in whom heparin therapy was started because of clinical suspicion of thromboembolic disease were excluded from analysis.
Objective modalities used to diagnose deep venous thrombosis included Doppler ultrasound, impedance plethysmography, venography, and computed tomography/magnetic resonance imaging (CT/MRI) scanning. Duplex Doppler ultrasound was generally used as the initial test to confirm thrombosis among patients presenting with symptoms or signs of deep venous thrombosis. Before 1988, venography was used to confirm abnormal results of Doppler ultrasound or impedance plethysmography. High probability ventilation and perfusion scans were considered confirmatory of pulmonary embolism. Patients with low or intermediate probability scans had pulmonary angiography for confirmation.
Hospital charts were reviewed for maternal characteristics of age, parity, height, weight, time during pregnancy or postpartum when thromboembolism occurred, medications at time of diagnosis, site and side of thromboembolism, and presenting clinical symptoms and signs. The postpartum period lasted until 42 days after delivery of the fetus or termination of the pregnancy. We defined risk factors for thromboembolism as age over 35 years, parity of three or more, weight over 165 pounds, and a personal or family history of deep venous thrombosis or pulmonary embolism.7 Other potential risk factors included a recent history of trauma or surgery, bedrest, malignancy, or autoimmune disorders. Homan's signs were noted to be positive if discomfort was elicited in the upper calf during forced dorsiflexion of the ankle joint. Leg swelling, obtained by measuring the widest circumference of the calf or the diameter of the ankle, was defined as significant if there was a difference of 2 cm between legs. Although not routinely determined in all patients during the study period, the individuals who presented with a thrombus in an atypical location, previous history of thrombosis, or history of fetal death were evaluated for deficiencies of protein S, protein C, and antithrombin III, as well as presence of activated protein C resistance and antiphospholipid antibodies.
Unpaired t tests and χ2 analysis were done with Statview 4.1 (Abacus Concepts, Inc., Berkeley, CA). A P value < .05 was considered statistically significant.
During the 19-year period, there were 268,525 births at our hospital, with 36,489 (13.6%) cesarean and 232,036 vaginal deliveries. There were 165 (0.06%) pregnancies complicated by venous thromboembolism, for a total incidence of one per 1627 births. We noted 127 cases of deep venous thrombosis and 38 cases of pulmonary embolism, for incidences of one per 2114 (0.5%) and one per 7066 (0.014%) births, respectively.
Table 1 shows maternal demographics. As evidenced in Table 2, most (95 of 127, 74.8%) deep venous thromboses were diagnosed by duplex Doppler ultrasonography. Real-time ultrasound detected one case of right ovarian vein thrombosis in a patient 7 days after an uncomplicated spontaneous vaginal delivery. In one patient presenting at 32 + 4 weeks' gestation, MRI revealed multiple filling defects in the left external iliac vein. Among patients with Doppler-confirmed thrombus, impedance plethysmography also detected a clot in 79.4% (23 of 29). In most (33 of 38, 87%) patients with pulmonary embolism, it was diagnosed by ventilation and perfusion scan, with only a few patients requiring pulmonary angiography. Slightly less than half the patients with pulmonary embolism who had Doppler evaluation for lower extremity deep venous thrombosis had evidence of thrombus (16 of 35, 45.7%).
Slightly less than one third of the patients (53 of 165, 32.1%) had any known risk factor for thromboembolism, as noted in Table 3. Among those with a prior episode of thromboembolism, antepartum deep venous thrombosis was not significantly more frequent than postpartum deep venous thrombosis (16 of 94 compared with seven of 33, P = .78). Patients with a history of venous thromboembolism who had antepartum deep venous thrombosis did not have an earlier onset compared with those without history of thrombosis (16.9 ± 3.7 compared with 17.4 ± 1.4 weeks' gestation, P = .8).
The presenting signs and symptoms of the study cases are listed in Table 4. At the time of initial presentation, the triad of pain, tenderness, and unilateral leg swelling was present in 85.7% of the patients with deep venous thrombosis. Although an enlarged leg was measured in 75% of the cases, slightly fewer than half had Homan's sign.
Nearly all deep venous thromboses identified occurred in the lower extremity (125 of 127, 98.4%). The left leg was more commonly affected than the right leg (104 of 127 compared with 22 of 127, P < .001), with one case of bilateral lower extremity deep venous thrombosis. Among antepartum deep venous thromboses, those in the right leg did not have an earlier onset (18.7 ± 1.9 compared with 15.4 ± 2.9 weeks' gestation, P = .38), nor were they more extensive in nature than those in the left leg. Overall, deep venous thrombosis was more common antepartum than postpartum (94 of 127 compared with 33 of 127, P < .001), with a mean gestational age at time of diagnosis of 16.8 ± 2.4 weeks. As seen in Figure 1, approximately half of the cases were detected before 15 weeks' gestation (47 of 95, 49.5%), with only 28 cases occurring after 20 weeks (P < .001). The mean time of postpartum deep venous thrombosis was 16.7 ± 3.9 days (range 1–42 days). Among the 33 cases of postpartum deep venous thrombosis, there was no increased incidence with surgical procedures, including cesarean delivery (n = 12), second trimester dilatation and evacuation (n = 4), and ectopic gestation (n = 1) compared with vaginal delivery (n = 16) (P = .9). Postpartum deep venous thromboses that occurred after spontaneous vaginal delivery, however, did occur earlier than those associated with cesarean (2.3 ± 0.7 compared with 6.3 ± 1.1 days, P = .004).
Although most pulmonary embolisms occurred postpartum (23 of 38 compared with 15 of 38), this was not statistically significant (P = .1). Those that did occur postpartum, however, were strongly associated with cesarean delivery (19 of 36,470 cesareans compared with four of 232,032 vaginal deliveries, P < .001). Only two patients (1.6%) had pulmonary embolism while on heparin therapy for deep venous thrombosis. All three maternal deaths were caused by surgically associated pulmonary embolism, with two occurring intraoperatively and the other taking place 1 day after cesarean. There were four (3.2%) recurrences of deep venous thrombosis, three of which were related to subtherapeutic heparin levels and one to patient noncompliance. Other recorded complications (n = 11) related to antepartum or postpartum heparin anticoagulation included thrombocytopenia, allergic reaction, and wound hematoma. We noted one case of protein S deficiency, four cases of protein C deficiency, two cases of anti-thrombin III deficiency, and four cases of activated protein C resistance. These patients did not have an earlier onset of antepartum deep venous thrombosis compared with those without hereditary thrombophilia. There were no fetal or neonatal complications resulting from maternal anticoagulation.
This study represents a large compilation of cases of objectively diagnosed thromboembolism during pregnancy. We found that, although thromboembolism rarely occurs in pregnancy, the incidence might be overestimated on the basis of previous reports. Most deep venous thrombosis cases occurred in the antepartum period, especially during the early second trimester. Only a few patients had the classic risk factors for thromboembolism.
In most previous reports, thromboembolism in pregnancy was diagnosed exclusively based on clinical signs and symptoms. During the 1950s, the frequency of thrombophlebitis was estimated to be 0.13–0.2%.8 A rate of 1.8% was reported by Bergqvist in 1979.9 Among 32,337 women delivering between 1951 and 1969, Aaro and Juergens10 noted 64 cases of deep thrombophlebitis and 13 cases of pulmonary embolism (0.24%). In the few recent studies requiring objective testing for diagnosis, the incidence of thromboembolism appears to be much lower. Kierkegaard11 found only 11 cases of deep venous thrombosis among 14,869 patients (0.07%). In the study by Bergvist and Hedner12 encompassing the years 1974–1980, the incidence of deep venous thrombosis was 0.07%. The present study likewise ound that the risk of thromboembolism during pregnancy was very low, 0.06% (one per 1627 births).
In the older studies, the risk of deep venous thrombosis traditionally was considered to be greatest during the late third trimester and postpartum period.9–11 Postpartum thrombosis was said to be three to five times as frequent as antepartum events and three to 16 times more common after cesarean compared with vaginal delivery.10,11 In accord with several recent reports3,12–15 this study confirmed that the antepartum period is the period of highest risk of deep venous thrombosis. Earlier postpartum ambulation, the shift from prolonged bedrest after uncomplicated vaginal and cesarean deliveries, avoidance of estrogenic suppression of lactation, and increased use of elastic compression stockings with graded pressure have contributed to this shift in frequency. The immediate postpartum period after cesarean continues to be the period of greatest risk for pulmonary embolism.13
In accord with the findings of several other objective studies, we found that the risk of venous thromboembolism appears to begin early in pregnancy.3,10,12,14–17 Progesterone-mediated increases in venous distensibility and capacity are evident early in gestation.5,18 Under the influence of estrogen, levels of coagulation factors I, II, VII, VIII, IX, XII, and fibrinogen are elevated by the end of the first trimester. Thrombin production, as measured by either fibrinopeptide A or thrombin-antithrombin III complex concentrations, is also higher by the end of the first trimester.19 The fibrinolytic system is inhibited, mostly in the third trimester.4,5 The relatively low incidence of venous thrombosis during the third trimester could be due to the presence of a placentally produced dermatan sulfate antithrombotic glycosamnioglycan.20
There was a clear left-sided predominance of antepartum deep venous thrombosis in this study. In the study by Ginsberg et al,16 58 of 60 women (96.7%) had venous thrombosis in the left leg alone and two women had bilateral venous thrombosis. Among the 48 cases of proximal thrombi described by Tengborn et al,3 81% occurred in the left lower extremity. Two other studies10,14 found that 75–82% of antepartum cases involved the left deep venous system. Although the exact explanation for this is unknown, increased venous stasis in the left deep venous system could be the cause. During pregnancy, the right iliac artery has an overly exaggerated compressive effect on the left common iliac vein.21 The enlarging gravid uterus could also selectively induce a compressive effect on the common iliac vein.
In this large retrospective study spanning nearly 20 years, we acknowledge that ascertainment and recall bias might be present. As most deep venous thromboses are initially silent and fewer than one third show classic symptoms, we might have underestimated the true incidence of the disease process. Using fibrogen scanning, Friend and Kakkar7 noted that three of 100 women had asymptomatic calf thrombi in the postpartum period. Because of the changing diagnostic modalities used during the years of our study, there could have been a higher incidence in the early years when ultrasound and CT and MRI were not used primarily. For example, the pregnancy-related changes of increased intra-abdominal pressure, obesity, and decreased venous return associated with lower extremity edema can be associated with false-positive diagnosis by impedance plethysmography. Although duplex Doppler ultrasound has been reported to have a 90% sensitivity and specificity for proximal vein thrombosis, it will not detect approximately 50% of small calf thrombi because of collateral venous channels.13 It is likewise possible that some of the individuals whose diagnosis of deep venous thrombosis or pulmonary embolism was clinical actually had thromboembolic phenomena. Our reported incidence of thromboembolism might be falsely low, as some of the women might have presented to other hospitals for postpartum evaluation. Finally, our reported incidence of pregnancy-associated venous thromboembolism might not reflect the actual rates in other patient populations, because of the different frequencies of heritable thrombophilias in certain ethnic groups.22
If venous thrombosis remains undiagnosed and untreated, in 15–24% of patients pulmonary embolism will develop.4,13 The embolism is fatal in nearly 15% of these individuals, with two thirds of the deaths occurring within 30 minutes of the embolic event.13 Only through early detection of venous thromboembolism and subsequent heparinization can this high maternal mortality rate be diminished. Although venous thromboembolism complicating pregnancy is a rare event, all pregnant individuals should be considered at risk. Although the patient might not have the traditional risk factors or might present early in gestation, diagnostic testing should be done immediately.
1. Atrash HK, Koonin LM, Lawson HW, Franks AL, Smith JC. Maternal mortality in the United States, 1979–1986. Obstet Gynecol 1990;76:1055–60.
2. Rochat RW, Koonin LM, Atrash HK, Jewett JF, and the Maternal Mortality collaborative. Maternal mortality in the United States: Report from the Maternal Mortality Collaborative. Obstet Gynecol 1988;72:91–7.
3. Tengborn L, Bergqvist D, Matzsch T, Bergqvist A, Hedner U. Recurrent thromboembolism in pregnancy and puerperium: Is there a need for thromboprophylaxis? Am J Obstet Gynecol 1989;160:90–4.
4. Sipes SL, Weiner CP. Venous thromboembolic disease in pregnancy. Semin Perinatol 1990;14:103–18.
5. Toglia MR, Weg JG. Venous thromboembolism during pregnancy. N Engl J Med 1996;335:108–14.
6. Barbour LA, Pickard J. Controversies in thromboembolic disease during pregnancy: A critical review. Obstet Gynecol 1995;86:621–33.
7. Friend JR, Kakkar VV. The diagnosis of deep venous thrombosis in the puerperium. J Obstet Gynaecol Br Commonwealth 1970;77:820–3.
8. Villasanta U. Thromboembolic disease in pregnancy. Am J Obstet Gynecol 1965;93:142–60.
9. Bergqvist A, Bergqvist D, Hallbook T. Acute deep vein thrombosis (DVT) after cesarean section. Acta Obstet Gynecol Scand 1979;58:473–6.
10. Aaro LA, Juergens JL. Thrombophlebitis associated with pregnancy. Am J Obstet Gynecol 1971;109:1128–33.
11. Kierkegaard A. Incidence and diagnosis of deep venous thrombosis associated with pregnancy. Acta Obstet Gynecol Scand 1983;62:239–43.
12. Bergqvist D, Hedner U. Pregnancy and venous thromboembolism. Acta Obstet Gynecol Scand 1983;62:449–53.
13. Rutherford SE, Phelan JP. Deep venous thrombosis and pulmonary embolism in pregnancy. Obstet Gynecol Clin North Am 1991;18:345–69.
14. Bergqvist A, Bergqvist D, Hallbook T. Deep venous thrombosis during pregnancy: A prospective study. Acta Obstet Gynecol Scand 1983;62:443–8.
15. Hellgren M, Nygards EB. Long-term therapy with subcutaneous heparin during pregnancy. Gynecol Obstet Invest 1982;13:76–89.
16. Ginsberg JS, Brill-Edwards P, Burrows RF, Bona R, Prandoni P, Buller HR, et al. Venous thrombosis during pregnancy: Leg and trimester of presentation. Thromb Haemost 1992;67:519–20.
17. Hull RD, Raskob GE, Carter CJ. Serial impedance plethysmography in pregnant patients with clinically suspected deep vein thrombosis: Clinical validity of negative findings. Ann Intern Med 1990;112:663–7.
18. Goodrich SM, Wood JE. Peripheral venous distensibility and velocity of venous blood flow during pregnancy or during oral contraceptive therapy. Am J Obstet Gynecol 1964;90:740–4.
19. de Boer K, ten Cate JW, Sturk A, Borm JJ, Treffers PE. Enhanced thrombin generation in normal and hypertensive pregnancies. Am J Obstet Gynecol 1989;160:95–100.
20. Delorme MA, Tam AS, Xu L. The effects of dermatan sulfate at submicrogram/ml concentrations on in vitro thrombin generation. Thromb Haemost 1996;75:747–51.
21. Cockett FB, Thomas ML, Negus D. Iliac vein compression: Its relation to iliofemoral thrombosis and the post-thrombotic syndrome. Br Med J 1967;2:14–9.
22. Rouse DJ, Goldenberg RL, Wenstrom KD. Antenatal screening for Factor Leiden mutation: A critical reappraisal. Obstet Gynecol 1997;90:848–51.