Population Pregnant women with mechanical heart valves attending high-risk antenatal clinics, treated with enoxaparin (1 mg/kg twice daily) during pregnancy.
Methods Women with mechanical heart valves treated with enoxaparin at any stage during pregnancy (1997–2008) identified using a database of women with mechanical heart valves attending the high-risk clinics and a prospective database of women prescribed enoxaparin for any indication during pregnancy.
Main outcome measures Maternal outcomes included thromboembolic and haemorrhagic complications. Pregnancy and fetal outcomes included miscarriage, stillbirth, baby death and live birth, small-for-gestational-age infants, warfarin embryopathy and warfarin-related fetal loss.
Results Thirty-one women underwent 47 pregnancies. In 34 pregnancies (72.3%), anticoagulation was with predominantly enoxaparin and 13 (27.7%) pregnancies women received mainly warfarin, with enoxaparin given in the first trimester and/or peri-delivery. Seven (14.9%) thrombotic complications occurred, of which five (10.6%) were associated with enoxaparin treatment. Non-compliance or sub-therapeutic anti-Xa levels contributed in each case. Antenatal and postpartum haemorrhagic complications occurred in eight (17%) and 15 (32%) pregnancies respectively. Of 35 pregnancies continuing after 20 weeks’ gestation, 96% (22/23) of women taking predominantly enoxaparin had a surviving infant compared with 75% (9/12) in women taking primarily warfarin. Four perinatal deaths occurred, three attributable to warfarin.
Conclusions Compliance with therapeutic dose enoxaparin and aspirin during pregnancy in women with mechanical heart valves is associated with a low risk of valve thrombosis and good fetal outcomes, but close monitoring is essential.
Management of women with mechanical heart valve replacements during pregnancy is one of the most high-risk and difficult challenges that clinicians caring for pregnant women face. Continuation of therapeutic anticoagulation throughout pregnancy to reduce the risk of thromboembolic complications is required. Warfarin, the safest treatment option for women crosses the placenta, is teratogenic and is associated with increased rates of fetal loss. To avoid the adverse fetal effects of warfarin, recent consensus guidelines have included recommendations for treatment with therapeutic dose unfractionated or low molecular weight heparin (LMWH) regimens.1,2 However, concerns remain about the efficacy of heparins to prevent mechanical valve thrombosis, especially in women considered at high risk of valve thrombus formation.
The largest literature review of pregnancy outcomes women with mechanical heart valves anticoagulated during pregnancy reported thromboembolic complications in 3.9% of pregnancies in women taking warfarin alone, 9.2% in pregnancies to women who received unfractionated heparin (UFH) in the first trimester followed by warfarin and 33% in pregnancies treated with UFH throughout pregnancy.3 A review of 16 studies (1996–2003) of pregnant women with mechanical heart valves managed with LMWH reported complications related to valve thrombus in ten of 81 pregnancies (12.3%).4 By contrast, another review of 23 studies (1996–2005) reported thrombotic events associated with LMWH treatment occurred in 22% of pregnant women (n = 76) with mechanical valves.5 The increased thrombotic rate in the second review was because of inclusion of more single case reports of valve thrombosis in women on LMWH and unpublished data from the HIP-CAT study.6 Limiting analysis to studies of five or more pregnancies, revealed a 6.9% rate of valve thrombosis (5/73 pregnancies).7–13 The largest single centre study to date, reported the outcome of 14 pregnancies.12 Larger studies are required to better understand the risks associated with LMWH treatment for pregnant women with mechanical heart valves.
Since 1997, our service has offered pregnant women with mechanical heart valves anticoagulation with therapeutic dose LMWH, enoxaparin, as an alternative to warfarin. The aim of this current study was to assess in a single centre (1) the rate of maternal thrombotic and haemorrhagic complications and (2) pregnancy outcomes and fetal complications in women with mechanical heart valves who received treatment with enoxaparin during pregnancy.
We conducted a retrospective audit of pregnant women with mechanical heart valves treated with enoxaparin at any stage during pregnancy and who delivered between January 1997 and July 2008 at two tertiary referral hospitals in Auckland, New Zealand. The study received ethical approval from the Northern Regional Ethics Committee. Eligible women with mechanical heart valves who took enoxaparin during pregnancy were identified through (1) a database of women with mechanical heart valves attending the high-risk clinic and (2) a prospective database of women prescribed enoxaparin for any indication during pregnancy that was then cross-referenced with the following two databases; a hospital database recording discharge diagnoses and a cardiology database of patients who underwent heart valve replacements. This study includes 14 pregnancies in 11 women that were previously reported.12
Pre-pregnancy counselling is offered to women of childbearing age with mechanical heart valves in Auckland. Information is given about the risks of maternal and perinatal morbidity and mortality, the risks and benefits of each anticoagulant treatment option including advice that the safest option is to avoid pregnancy. Women are advised to contact the clinic as soon as they miss a period to enable switching from warfarin to enoxaparin before 6 weeks’ gestation. Women not seen prior to pregnancy are reviewed urgently upon confirmation of pregnancy to discuss treatment options including the risks of continuing the pregnancy. Each pregnant woman completes written informed consent for her choice of one of the following treatment options.
1 Substitution of warfarin with therapeutic dose enoxaparin (1 mg/kg bd) and aspirin 100–150 mg before 6 weeks’ gestation, continued until planned delivery.
2 Substitution of warfarin with therapeutic dose enoxaparin from 6 until 12 weeks’ gestation,6 then reverting to warfarin. Enoxaparin is re-introduced at 34–36 weeks of gestation until planned delivery. Aspirin 100–150 mg throughout pregnancy.
3 Warfarin and aspirin 100–150 mg throughout pregnancy, switching to therapeutic enoxaparin and aspirin at 34–36 weeks’ gestation until planned delivery.
Women at very high risk of valve thrombosis are advised that options 2 or 3 are preferred for maternal safety. For women treated with enoxaparin, monthly monitoring of anti-Xa levels was recommended, aiming for target levels of 0.4–0.7 IU/ml pre-dose and 0.7–1.2 IU/ml 4 hours post dose. Anti-Xa levels were initially checked 3–7 days after starting treatment or following dose modification. The enoxaparin dose was increased or decreased by 10 mg/kg bd if anti-Xa levels were levels were low or high respectively. During warfarin therapy, monthly INRs were measured with a target INR of 3–4.5 for women with a Starr-Edwards valve(s) and 2.5–3.5 for women with bi-leaflet or tilting disc valves. All adjustments to anticoagulant doses were made at the discretion of the treating clinician. The approach to management of anticoagulation peri-delivery and postpartum is outlined in Table 1. Warfarin was commenced at 10 mg on the first day of treatment, 5–10 mg on the second day and then the dose was adjusted according to the INR.
Table 1. Recommended management of anticoagulation peri-delivery
Management of delivery
Planned delivery for all women. Vaginal delivery unless obstetric indications for a caesarean section.
Women on warfarin switching on to enoxaparin at 34–36 weeks gestation.
Induction of labour or elective caesarean section at 38 ± 1 weeks’ gestation.
Last enoxaparin dose given 36 hours before induction of labour or caesarean section.
Start intravenous (iv) unfractionated heparin (1250 IU/hour, 6 hourly aPTT, target aPTT 2–3 × baseline) 24 hours before induction of labour or caesarean section. Since 2006, 5000 IU bolus unfractionated heparin is given as a loading dose.
Stop iv unfractionated heparin once in established labour or 6 hours prior to regional anaesthesia. Confirm aPTT returns to baseline before epidural catheter placement.
Postpartum anticoagulant management
4–6 hours post vaginal delivery or 6–12 hours post caesarean section, restart iv unfractionated heparin (500 IU/hour, no bolus dose, gradual increase over 24–48 hours to target aPTT) and continue until INR > 2.
Restart warfarin on day 1 postpartum if an uncomplicated vaginal delivery or on days 2–3 if caesarean section or other bleeding complications.
Information retrieved from medical records included demographic data, smoking at first antenatal visit, indication for and date of valve replacement, mechanical valve type, valve number and site, evidence of cardiac failure on X-ray or echocardiography and the presence of risk factors for thromboembolic events. These included thrombogenic mechanical valves such as Starr-Edwards, two mechanical valve replacements, previous thromboembolism (valve thrombus, transient ischaemic attack [TIA] and cerebrovascular accident [CVA], as defined below) and atrial fibrillation. Data collected on warfarin, enoxaparin and UFH therapy included the dose, gestational period during which treatment was prescribed and detailed information about peri-delivery anticoagulation. Peak (4 hour post dose) and trough (pre-injection) anti-Xa levels were recorded, as well as the gestation at which tests were carried out.
Information about the primary endpoints, thromboembolism and haemorrhagic complications was collected. Diagnosis of valve thrombus required confirmation on echocardiogram. TIA was defined as an acute neurological deficit with complete resolution by 24 hours and when the acute neurological deficit persisted for more than 24 hours, a CVA was diagnosed. A thrombotic CVA was confirmed by exclusion of cerebral haemorrhage on CT scan or MRI. Antenatal obstetric bleeding included antepartum haemorrhage (APH) or placental abruption. Other antenatal bleeding was defined as haemorrhagic events in the antenatal period after exclusion of obstetric bleeding. Postpartum haemorrhage (PPH) was classified as either a primary PPH (loss > 500 ml after vaginal delivery or >750 ml after caesarean in the first 24 hours after delivery) or a secondary PPH if the vaginal bleeding occurred more than after 24 hours postpartum. Other major postpartum bleeding included surgical site bleeding and ischiorectal or vulval haematoma that resulted in blood transfusion, surgery or re-admission. Anticoagulant status and anti-Xa levels, if available, were recorded at the time of an event.
Fetal outcome was classified as either a termination of pregnancy before 20 weeks’ gestation, miscarriage (spontaneous fetal loss < 20 weeks’ gestation), stillbirth (fetal death ≥20 weeks’ gestation), baby death (neonatal or infant death) or a live birth. Preterm birth was delivery before 37 weeks’ gestation and included both spontaneous and iatrogenic preterm births. Small for gestational age was defined as a birthweight <10th customised centile.14 Diagnosis of warfarin embryopathy required at least midline hypoplasia and epiphyseal stippling following exposure to warfarin in the first trimester.15 Warfarin-attributed baby losses included either termination of pregnancy, stillbirth or baby death due to warfarin embryopathy or fetal intracranial haemorrhage.
Statistical tests included t-test, for continuous data and Fisher’s Exact test, for binary data. A P-value < 0.05 was considered statistically significant. Mean (SE) pre-dose and peak anti-Xa levels were calculated for each 4-week interval throughout pregnancy.
Between January 1997 and July 2008, 31 women with mechanical valve replacements received enoxaparin during 47 pregnancies. Twenty-five women (80%) were Pacific Island or Maori, three (10%) Caucasian and three (10%) were Asian. Twenty-seven women (87%) required valve replacement for rheumatic valvular disease, three (10%) for congenital heart disease and one woman (3%) had a valve replacement following bacterial endocarditis. Twelve (39%) women had Starr-Edwards valves including eight with single mitral valve replacements and four with both mitral and aortic valve replacements (three with mitral and aortic Starr-Edwards valves and one with mitral Starr-Edwards valve and aortic Medtronic Hall valve replacements). Nineteen (61%) women had tilting-disc or bileaflet valve replacements. Of these, ten women had single valve replacements, including five bileaflet mitral valves, one tilting-disc mitral valve, four bileaflet aortic valves and nine women had both mitral and aortic bileaflet or tilting-disc valves.
Thirty-nine of the 47 pregnancies (83%) continued after 12 weeks’ gestation and 35 (75%) beyond 20 weeks’ gestation, Figure 1. Maternal characteristics, risk factors for thromboembolic events and anticoagulant regimens during the pregnancies are shown in Table 2. In 34 (72.3%) pregnancies, women chose to take enoxaparin throughout pregnancy from the time of booking. Twenty women started enoxaparin before six completed weeks’ gestation (median 5.6 [range 3.7–6.7] weeks) and 14 started after 6 weeks’ gestation (median 12.6 [range 7–26] weeks). Of the ‘late starting’ enoxaparin group, there were 12 pregnancies where women booked late to clinic, one in which enoxaparin was started at 24 weeks after a mechanical valve replacement and one where anticoagulation was switched from UFH to enoxaparin at 25 weeks’ gestation following a valve thrombosis. Low-dose aspirin was prescribed in 40 (85%) pregnancies. Two women developed cardiac failure, one at 26 weeks gestation and the other on the third day postpartum. Both women responded to treatment with diuretics with no permanent change in their functional status by NYHA criteria. Three of 19 (15%) nulliparous women developed pre-eclampsia.
Table 2. Maternal characteristics and thromobembolic risk factors
Maternal characteristics at booking
Pregnancies (N = 47)
Data shown as N (%) or median (range).
*Single or double valve
Smoker at booking
Thromboembolic risk factors
Two mechanical valves
≥1 Thromboembolic risk factor
Anticoagulation during pregnancy
started ≤ 6 weeks gestation
started > 6 weeks gestation
Enoxaparin 6–12 weeks + pre-delivery
Enoxaparin pre-delivery only
Overall, thromboembolic complications developed in seven (14.9%) pregnancies, five in the antenatal period and two postpartum. Thromboembolic risk factors were not more common in these pregnancies (6/7, 85.7%) compared to pregnancies with no thromboembolism (30/40, 75%), P = 1.0. In five (10.6%) cases, the thromboembolic events (three antepartum and two postpartum) were associated with enoxaparin therapy, Table 3. Compliance was an issue in three of these events and anti-Xa levels were sub-therapeutic in the other two. One woman, previously reported,12 presented with a valve thrombus at 7 weeks of gestation having stopped warfarin several months preconception. She started enoxaparin (1 mg/kg bd) and aspirin, with trough and peak anti-Xa levels were 0.22 IU/ml and 0.68 IU/ml respectively and a repeat transoesophageal echocardiogram showed probable resolution of the valve thrombus. At 22 weeks’ gestation, she presented with a TIA due to valve thrombus despite being compliant with enoxaparin. Her anti-Xa level was 0.62 IU/ml and she required urgent mosaic valve replacement. The second woman ceased warfarin at 8 weeks but only commenced enoxaparin when she presented at 12 weeks’ gestation. She was non-compliant with anti-Xa testing and clinic appointments. After a TIA at 17 weeks gestation, warfarin was recommended, but she elected to continue enoxaparin, remained non-compliant and presented with a CVA at 32 weeks. A third woman was prescribed therapeutic enoxaparin without aspirin from 7 weeks’ gestation. She had anti-Xa levels at 8 weeks (trough 0.11 IU/ml, peak 0.59 IU/ml), was non-compliant and then presented at 13 weeks with a dense left CVA. She switched over to warfarin and aspirin and had a stillbirth at 21 weeks as a result of fetal intracranial haemorrhage. The first postpartum thromboembolic event associated with enoxaparin was an asymptomatic valve thrombus within 24 hours of delivery while on intravenous UFH. This woman had been non-compliant with sub-therapeutic anti-Xa levels (trough < 0.2 IU/ml, peak 0.2 IU/ml) in the third trimester of pregnancy. The second woman, with two Starr-Edwards valves, also had sub-therapeutic anti-Xa levels (trough 0.23 IU/ml, peak 0.56 IU/ml) in late pregnancy and received the standard UFH infusion peri-delivery. On day 5 postpartum, she had a TIA while on subcutaneous therapeutic dose enoxaparin and warfarin (INR 2.4).
Table 3. Maternal thromboembolic and haemorrhagic complications
Number of events (n = 47 pregnancies)
*Includes secondary PPH and other major bleeding; primary PPH were excluded.
Unrelated to enoxaparin
Unrelated to enoxaparin
Unrelated to enoxaparin
Two women developed thromboembolic complications prior to commencing enoxaparin. One woman stopped taking warfarin before conception and presented at 24 weeks’ gestation with valve thrombus. Following commencement of enoxaparin and aspirin, the valve thrombus resolved. The second woman, transferred from a district hospital, with a St Jude’s aortic valve developed a TIA at 25 weeks’ gestation while treated with 5000 units UFH twice daily. She had no further thrombotic events once switched to therapeutic dose enoxaparin.
Antenatal bleeding complicated eight (17%) pregnancies, five of which were associated with enoxaparin and three with intravenous UFH, Table 3. Of the five bleeds on enoxaparin, one was a minor haematemesis and four were antenatal obstetric bleeds (two placental abruptions, two antepartum haemorrhages of undetermined origin). The first abruption occurred in a woman who had a subchorionic haematoma at 18 weeks. At 33 weeks, her trough anti-Xa level was 0.55 IU/ml and peak level 0.66 IU/ml. She presented at 34 weeks with an estimated 1300 ml vaginal blood loss and required massive transfusion despite attempted reversal of enoxaparin with protamine prior to caesarean section. Her baby survived. The second abruption occurred in association with pre-eclampsia in the woman, described above, who had a mosaic valve replacement during pregnancy and was on prophylactic dose enoxaparin. One APH occurred with over-anticoagulation (anti-Xa trough 0.93 IU/ml, peak 1.43 IU/ml) and the other was a minor APH at 35 weeks’ gestation.
The three antenatal bleeds unrelated to enoxaparin occurred in women on intravenous UFH. These included a concealed abruption during induction of labour, a rectus sheath haematoma and epistaxis 4 days after starting intravenous UFH because of preterm labour and an intra-abdominal bleed 6 days after a laparotomy for removal of a dermoid cyst at 16 weeks that resulted in a miscarriage.
Primary postpartum haemorrhage (PPH) occurred in six (12.8%) pregnancies, three in women with placental abruptions. There were nine (19.1%) cases of secondary PPH or other major bleeding that required blood transfusion, readmission or surgery, Table 3. Six of the nine events were related to enoxaparin (five occurred in women while on enoxaparin and warfarin, one woman on enoxaparin only) and three unrelated to enoxaparin (one on intravenous UFH and warfarin, one UFH, one warfarin). Of the six events related to enoxaparin, four were secondary PPHs after vaginal deliveries and two surgical site bleeds (one post caesarean section and one after puerperal tubal ligation). There were three secondary PPHs unrelated to enoxaparin. In addition to the secondary PPH, one woman had a caesarean section wound haematoma and another woman had a large ischiorectal haematoma after a vaginal delivery.
Anti-Xa monitoring and enoxaparin dose
The median enoxaparin dose was 1.0 mg/kg/12 hours, with interquartile dose ranges as follows; 1st interquartile range (IQR) 0.7–1.0 mg/kg/12 hours, 2nd IQR 1.0–1.0 mg/kg/12 hours, 3rd IQR 1.0–1.1 mg/kg/12 hours, 4th IQR 1.1–1.5 mg/kg/12 hours. Mean peak and trough anti-Xa levels and mean dose of enoxaparin (mg/kg/12 hours) during pregnancy are shown in Figure 2. The overall median trough anti-Xa level was 0.4 IU/ml (5th centile 0.12 IU/ml, 25th centile 0.27 IU/ml, 75th centile 0.54 IU/ml, 95th centile 0.81 IU/ml) and the overall median peak anti-Xa level was 0.8 IU/ml (5th centile 0.36 IU/ml, 25th centile 0.62 IU/ml, 75th centile 1.00 IU/ml, 95th centile 1.24 IU/ml).
There were four terminations of pregnancy (Figure 1), including one at 17 weeks for fetal hydrocephalus occurring while on 5 mg warfarin. Of the eight miscarriages, seven were in the first trimester and the second trimester loss occurred following surgery for the dermoid cyst. Delivery details and pregnancy outcome for the 35 babies born after 20 weeks are summarised in Table 4. Of pregnancies continuing beyond 20 weeks, 96% (22 of 23) of women treated predominantly with enoxaparin during pregnancy had a surviving infant compared to 75% (9 of 12) of those treated mainly with warfarin, P = 0.11. There were four (11.4%) stillbirths, neonatal or infant deaths. A woman taking enoxaparin had an early neonatal death as a result of spontaneous preterm labour at 20 weeks thought to be unrelated to anticoagulant therapy. The other three perinatal deaths occurred as a direct result of warfarin, two stillbirths as a result of fetal intracerebral haemorrhage (warfarin dose 4 mg and 5 mg, maternal INR 3.3 and 3.2) and one infant death at 2 months due to warfarin embryopathy in a woman taking 6 mg warfarin until 34 weeks’ gestation. A second infant with congenital cardiac anomalies was born to a woman with rheumatic valvular fever disease who stopped warfarin at 5.1 weeks’ gestation and took enoxaparin until 12 weeks gestation before switching back to warfarin. Small-for-gestational-age infants were more common in women who took mainly warfarin during pregnancy, with 58.3% (7 of 12) pregnancies complicated by SGA compared to 8.7% (2 of 23) in women who took enoxaparin, P = 0.003. Rates of smoking during pregnancy were high. Of 13 (27.7%) pregnancies where smoking was recorded at booking, two were in pregnancies to women treated with predominantly warfarin and 11 where women took predominantly enoxaparin. Of the nine preterm births, six occurred in pregnancies where women predominantly received enoxaparin. Two of the six were spontaneous preterm births at 20 weeks’ and 35 weeks’ gestation unrelated to anticoagulant therapy. Of the four iatrogenic preterm births, enoxaparin therapy may have contributed in two cases where placental abruption led to preterm delivery. The remaining three preterm births were associated with warfarin usage; two inductions after fetal deaths in-utero and one an induction for severe SGA.
Table 4. Pregnancy outcome in pregnancies continuing after 20 weeks of gestation
Pregnancies (n = 35)
Data expressed as mean (SD) or n (%) where n = pregnancies.
Induction of labour
Customised birthweight centile
Neonatal intensive care admission
Neonatal or infant death
This is the largest study of consecutive pregnancies in women from a single centre with mechanical heart valves who have been treated with therapeutic dose enoxaparin. Enoxaparin was the predominant anticoagulant used in 34 of the 47 pregnancies and in the remaining 13 pregnancies, warfarin was the main anticoagulant, with enoxaparin used between 6 and 12 weeks gestation and/or peri-delivery. Thromboembolic complications in our cohort occurred in seven (14.9%) pregnancies, of which five (10.6%) were directly related to enoxaparin therapy. Poor compliance with therapy or sub-therapeutic anti-Xa levels was an issue in all cases. No thromboembolic complications occurred in the 20 pregnancies where enoxaparin was commenced at ≤6 weeks’ gestation, a group that was compliant with medication and testing of anti-Xa levels. In the thirteen pregnancies where warfarin was the predominant anticoagulant, there were also no thromboembolic complications; however, there were four warfarin-attributed fetal or baby losses, including a termination for fetal hydrocephalus.
Our rate of thromboembolic complications associated with enoxaparin therapy in women with mechanical valves is similar to that reported in a recent literature review and recent Norwegian case series.17 Risk factors for valve thrombosis were present in three-quarters of the pregnancies in our study, but these were not more common in women who had thromboembolic complications. Non-compliance with enoxaparin and failure to achieve therapeutic target anti-Xa levels were the main contributing factors to the development of thromboembolic complications in our centres.
There are no data from prospective clinical trials to inform clinicians on the optimal frequency of measuring anti-Xa or on appropriate target levels for pregnant women with mechanical heart valves taking LMWH. Recent publications have differed in their recommendations with regards anti-Xa levels and with target post-dose levels of 0.7–1.2 IU/ml1 or peak anti-Xa levels of 1.0 IU/ml.3 By contrast, Elkayam and Bitar recommended target pre-dose trough anti-Xa levels of 0.6 IU/ml and 0.7 IU/ml.16 In the Norwegian study, therapeutic dose dalteparin was used, aiming at trough and peak anti-Xa targets of 0.4–0.8 IU/ml and 0.7–1.2 IU/ml respectively.17 They reported two (16.7%) thromboembolic complications in women who were initially prescribed suboptimal doses of LMWH. Our data and the Norwegian study highlight the importance of clinicians supervising care being aware of LMWH dosage regimens, target anti-Xa levels and the need for intensive follow up, especially if non-compliance is a concern.
Bleeding complications occurred in a third of our pregnancies, 17% antepartum and 19.1% postpartum. Most were related to enoxaparin treatment and likely to be, in part, a consequence of the intensity of anticoagulation required to prevent valve thrombosis during pregnancy. While low-dose aspirin may have contributed to the high rate of bleeding, as has been shown in non-pregnant populations, it also reduces the risk of thrombosis18 and is recommended for women with mechanical valves during pregnancy.1 Management of anticoagulation around the time of delivery is a particular challenge. Our goal was to have a vaginal delivery and avoid caesarean section, unless there were definite obstetric indications. The high rate of secondary PPH and other bleeding following both vaginal delivery and caesarean section was frequently associated with overlap of the therapeutic effects of LMWH and warfarin. Subcutaneous LMWH has a more prolonged anticoagulant action19 than intravenous UFH and its anticoagulant effect may continue once warfarin is therapeutic. In addition, if bleeding complications develop, the anticoagulant effect of LMWH, unlike that of UFH, is not fully reversed with protamine.19 For these reasons, we consider it preferable to use intravenous UFH postpartum until warfarin levels are therapeutic instead of restarting LMWH. In addition, it may be helpful to stop low-dose aspirin 5 to 7 days before planned delivery to reduce the risk of postpartum bleeding.
Overall, we had a high rate of adverse outcomes in the 35 pregnancies continuing after 20 weeks’ gestation, with a quarter of the babies being born preterm and a quarter were SGA. SGA pregnancies were not increased in pregnancies where women predominantly used enoxaparin but occurred in 58% of warfarin-exposed pregnancies. This association between warfarin and SGA has been previously reported,20 but this is the first study that has used customised birthweight centiles to define SGA. Our study was not sufficiently powered to determine whether other risk factors for SGA, such as maternal smoking were confounding factors. Major adverse fetal outcome attributable to warfarin occurred in four of the 13 (31%) pregnancies where women took warfarin, consistent with data from previous studies.21
Given the high risk for both mother and baby, women with mechanical heart valve replacement must be fully informed of the maternal and fetal risks with different anticoagulant regimens during pregnancy and also given the option of not continuing the pregnancy. While the focus for clinicians looking after pregnant women with mechanical heart valves may be to prevent maternal thromboembolic complications, the overriding concern for many women is to avoid any harm to their unborn child, even when this places her health at risk. The importance of considering women’s preference in decisions about anticoagulation is emphasised in the recent American College of Chest Physicians guidelines.3 Compliance with twice daily injections and frequent blood tests for anti-Xa levels throughout pregnancy is critical to provide safe management of these women. In women who are compliant, a low rate of valve thrombosis can be achieved that is acceptable to clinicians, together with a high rate of baby survival.
Our data suggest that therapeutic dose enoxaparin in combination with low-dose aspirin can be used for prevention of valve thrombosis in pregnant women with mechanical heart valves. If enoxaparin is the chosen anticoagulant, close clinical follow up with a dedicated, well-resourced multidisciplinary team and access to appropriate laboratory testing are essential. While maternal safety may be improved with warfarin, in our series, women who took predominantly oral anticoagulant therapy during pregnancy had poor fetal outcomes.
Conflict of interest
The authors have no conflict of interest to disclose.
Contribution to authorship
C.M designed the study database, carried out data collection, analysed the data, wrote the first draft of the paper and all subsequent revisions. L.M. provided clinical input and contributed to data analysis and to the writing of the paper. R.N assisted with the design of the database and data analysis and contributed to the writing of the paper.
Details of ethics approval
The study received expedited ethics approval from the Northern Regional Ethics Committee, Auckland, New Zealand. Study approval number NTX/07/104/EXP.
The study received no funding.
The authors would like to thank Dr Carl Eagleton for providing details relating women under his care during pregnancy and also Eliza Chan for providing statistical analysis.