1 INTRODUCTION

Pregnancy is recognized as an acquired hypercoagulable state. Increased concentrations of coagulation factors, platelet activation, decreased fibrinolysis, and a reduction in free and total protein S activity are responsible for the hypercoagulability of pregnancy. The resultant increase in thrombin generation, however, predisposes to venous thromboembolism (VTE), which comprises deep vein thrombosis and its complication, pulmonary embolism. Over the past decade, risk factors that further increase this risk of VTE have been identified by a number of high-quality cohort and case-control studies from the developed world.1-6 In particular, a personal history of thrombosis and inherited thrombophilia are the most important risk factors for pregnancy-associated VTE.3 In addition, established risk factors include; age >35 years, medical comorbidities, significant pregnancy complications, and cesarean delivery.3, 4, 6 Accordingly, several evidence-based guidelines have been developed that recommend the use of pharmacological thromboprophylaxis during pregnancy and the postpartum period in women with an increased risk of thrombosis.7-11

In South Africa, a resource-limited setting, these guidelines have been increasingly adopted. Of particular concern, however, is the fact that human immunodeficiency virus (HIV) and its treatment, which are known risk factors for VTE, have received limited consideration in the current guidelines. In South Africa, particularly, this risk factor is of relevance, considering a HIV-adjusted maternal prevalence of 20.4%. Studies of non-pregnant HIV-infected individuals have described a two- to ten-fold increased risk of VTE.12 Epidemiologic studies have also evaluated the significance of HIV as a risk factor for VTE among pregnant and postpartum women.13 Unfortunately, the interpretation of these studies is limited by the small numbers (fewer than 10) of HIV-infected cases. Various risk factors predisposing to a pro-thrombotic and pro-inflammatory state have been described in association with HIV. These include opportunistic infections, malignancy, and immunodeficiency.14 In addition, several thrombophilias have also been observed in HIV; namely, the presence of a circulating lupus anticoagulant, activated protein C resistance, increased soluble plasminogen activator inhibitor-1, and decreased synthesis as well as an abnormal distribution between bound and free forms of protein S.15 Moreover, HIV-infected women are increasingly conceiving on antiretroviral therapy (ART). ART has been associated with an increased risk of stillbirths and preterm delivery, which are also predictors of thrombosis.16 Nevertheless, very little is currently known about the impact of HIV on the incidence of VTE in pregnant and postpartum women.

We therefore conducted a retrospective case-control study to evaluate the clinical spectrum as well as prepartum and postpartum risk factors for pregnancy-related VTE in African women to guide thromboprophylaxis in a resource-limited setting.

2 MATERIALS AND METHODS

A single-center study was conducted at the obstetric unit at Charlotte Maxeke Johannesburg Academic Hospital in Johannesburg, South Africa from July 1, 2017 to June 30, 2020. This is a specialist referral academic center for high-risk pregnant women including those with VTE. All cases of VTE during pregnancy or within 42 days postpartum were identified. Controls (without a history of arterial thrombosis or VTE) were selected from the population of patients presenting to the obstetric unit during the study period. The cases and selected controls were of African ethnicity, matched for gestation in a 1:5 ratio. During the study period, 98.9% of maternity admissions were of African ethnicity. Written informed consent was obtained from the cases, and the study protocol was approved by the Institutional Review Board (M 191038).

Data including demographics, detailed clinical information relating to diagnosis and associated prepartum and/or postpartum risk factors were collected from clinical records. The diagnoses of thrombosis were independently confirmed by three of the study investigators (ES, AG, and HR). VTE was diagnosed in patients with clinical, laboratory, and radiologic findings (compression ultrasound; ventilation-perfusion lung-scan; computed tomographic pulmonary angiography, adapted and standardized for pregnancy). All women with VTE in pregnancy were treated with dose-adjusted twice-daily low-molecular-weight heparin (enoxaparin) (anti-Xa level of 0.5–1.0 units/ml 3–4 h after the last dose) followed by warfarin in the postpartum period (International Normalized Ratio of 2–3). The median durations of treatment for deep vein thrombosis and pulmonary embolism were 6 months (interquartile range [IQR] 3 months) and 9 months (IQR 6 months), respectively.

In the subgroup of HIV-infected women, data on opportunistic infections and ART were collected. Associated prepartum risk factors were obtained. These included; age >35 years, personal history of VTE, self-reported family history of unprovoked or estrogen-related VTE in a first-degree relative, inherited thrombophilia, anti-phospholipid syndrome (APS), medical comorbidities (including at least one of the following: malignancy, heart disease, active systemic lupus erythematosus, inflammatory polyarthropathy, inflammatory bowel disease, chronic kidney disease, sickle cell disease, current intravenous drug user), chronic hypertension (systolic blood pressure >140 mm Hg and/or diastolic blood pressure >90 mm Hg), systemic infection (perioperative and opportunistic infection), HIV, parity three or more, multiple pregnancy, pre-eclampsia with fetal growth restriction (FGR), non-obstetric surgery, hospital admission or immobility (at least 3 days bed rest), gross varicose veins, hyperemesis gravidarum, in vitro fertilization, smoking and obesity (body mass index [BMI, calculated as weight in kilograms divided by the square of height in meters] ≥ 30). The following postpartum risk factors were also recorded; non-obstetric surgery, obesity (BMI ≥ 40), preterm delivery (<37 weeks), stillbirth (intrauterine fetal death >28 weeks), mode of delivery, prolonged labor (>24 h) and postpartum hemorrhage requiring transfusion.

During pregnancy, all women were screened for HIV when first presenting to prenatal care, with retesting at follow-up prenatal visits. A positive HIV test was further characterized with a CD4 count and HIV viral load. This was determined using the Cytomics FC 500 MPL Flow Cytometry System (Beckman Coulter, Brea, CA, USA) and the Cobas 6800/8800 system (Roche Diagnostics, Basel, Switzerland), respectively. The lower limit of detection for HIV RNA was 50 copies/ml.

In 82 cases (64.1%), inherited thrombophilia testing was performed for fibrinogen (STA-Liquid Fib; Diagnostica Stago), antithrombin (STA Stachrom kits; Diagnostica Stago), protein C (STA Stachrom kits; Diagnostica Stago), functional protein S (Staclot protein S; Diagnostica Stago), and Factor V Leiden and prothrombin G20210A mutations (LightCycler 480; Roche Diagnostics and DNA sequence analysis; Applied Biosystems). Only blood sampling 1-month after completion of anticoagulation treatment was analyzed, to exclude any pregnancy-related, acute thrombosis-related and anticoagulant-related alterations in coagulation and fibrinolysis.

Statistical analysis was performed using Statistica 13.2 software and SAS 9.1 software (SAS Inc., Cary, NC, USA). Normally distributed continuous data were presented as mean ± standard deviation and variables with non-Gaussian distribution as median (IQR). Categorical data were presented as frequencies and percentages. Comparisons of prepartum and postpartum risk factors in the cases and controls were performed using χ2 test or two-tailed Fisher's exact test. Subgroup analysis with multiple pairwise comparisons was subsequently performed applying the Bonferroni correction with a P value less than 0.0083 considered statistically significant. Univariate logistic regression analysis was conducted to predict thrombosis with the correspondent adjusted odds ratios (aOR) and 95% confidence intervals (CI). Multiple logistic regression analysis was then applied to build the models by including risk factors with a P value of less than 0.05, which were identified by univariate analysis. As a result of the limited number of cases, the explanatory variables were restricted to a maximum number of six per model. The resultant model was selected by exhibiting the highest −2 log likelihood value and area under the curve on receiver operating characteristic analysis. Two-tailed P value of less than 0.05 was considered statistically significant.

3 RESULTS

During the study period, 26 323 deliveries were registered. VTE was diagnosed in 128 women, of which 67 were deep vein thromboses, 58 were pulmonary embolisms, and three were both. The majority of thromboses were diagnosed during the postpartum period (n = 66, 51.6%) at a median of 6 days (IQR 12 days) and in the third trimester (n = 27, 21.1%) at 32 weeks (IQR 4 weeks) (Figure 1). Table 1 shows the characteristics of the study population. Inherited thrombophilia was identified in 17 (20.7%) cases, protein S deficiency was confirmed in 12 (14.6%), protein C deficiency in four (4.9%), and a rare prothrombin gene C20209T mutation was recorded in a single patient (1.2%) with a massive pulmonary embolism. All other screening results were negative. Forty-four (34.4%) women were HIV-infected; 16 (36.3%) with a median viral load of 3287 copies/ml (IQR 26 062 copies/ml) and 28 (63.6%) with virologic suppression. Complications of thrombosis included cor pulmonale (n = 2, 1.6%), post-thrombotic syndrome (n = 2, 1.6%), and maternal death as a result of right ventricular infarction and decompensated cardiac failure (n = 1, 0.8%). Patients who developed complications all tested positive for HIV. Additionally, four patients (3.1%) presented with a recurrent thrombosis after completion of anticoagulation during the study period. Of these, three (2.3%) were HIV infected; two with a deficiency of protein S despite virologic suppression and another not on ART with comorbid tuberculosis.

Flow diagram of study selection of patients with venous thromboembolism and controls

TABLE 1. Baseline characteristics of cases of venous thromboembolism and corresponding matched controls Characteristic Controls (n = 640) Cases with VTE (n = 128) P value Clinical Age, years 29 ± 7 30 ± 6 0.146 Parity 2 ± 1 2 ± 1 0.795 BMI 27.4 ± 5.1 27.9 ± 6.1 0.297 HIV-infected 114 (17.8) 44 (34.3) <0.001 Laboratory White cell count, ×109/L (ref: 3.9–12.6 × 109/L) 8.0 ± 2.6 10.2 ± 4.2 <0.001 Hemoglobin, g/L (ref: 116–164 g/L) 117.6 ± 16.3 106.7 ± 18.8 0.001 Platelet count, ×109/L (ref: 186–454 × 109/L) 249.4 ± 63.2 293.9 ± 107.0 0.003 Mean platelet volume, fL (ref: 7.3–11.3 fL) 10.4 ± 1.7 10.6 ± 1.2 0.547 Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by the square of height in meters); VTE, venous thromboembolism. a Values are given as mean ± SD or as number (percentage).

On univariate analysis, the following prepartum risk factors were independently associated with thrombosis; personal history of VTE, medical comorbidities, systemic infection, HIV, and hospital admission or immobility (Table 2). Postpartum risk factors associated with thrombosis included personal history of VTE, APS, medical comorbidities, systemic infection, HIV, pre-eclampsia with FGR, non-obstetric surgery, hospital admission or immobility, preterm delivery, prolonged labor, and postpartum hemorrhage. The most frequent medical comorbidity among the prepartum and postpartum cases was heart disease, which included peripartum cardiomyopathy (n = 10, 7.8%), dilated cardiomyopathy (n = 2, 1.6%), and pulmonary hypertension (n = 5, 3.9%).

TABLE 2. Prepartum and postpartum risk factors in the study patients Variable Controls Cases with VTE P value Prepartum risk factors (n = 310) (n = 62) Age > 35 years 63 (20.3) 16 (25.8) 0.336 Personal history of VTE 0 (0) 5 (8.1) <0.001 Positive family history for VTE 2 (0.6) 2 (3.2) 0.105 Inherited thrombophiliab 2 (14.3) 11 (25.6) 0.479 APSc 0 (0) 2 (6.7) 0.067 Medical comorbiditiesd 8 (2.6) 8 (12.9) <0.001 Chronic hypertension 20 (6.5) 7 (11.3) 0.186 Systemic infection 3 (1.0) 3 (4.8) 0.046 HIV 56 (18.1) 24 (38.7) <0.001 ART 52 (16.8) 21 (33.9) 0.443 Parity ≥3 74 (23.8) 15 (24.2) 0.957 Multiple pregnancy 6 (1.9) 2 (3.2) 0.527 Pre-eclampsia with FGRe 10 (5.3) 1 (3.1) 0.603 Hospital admission or immobility 4 (1.3) 4 (6.5) 0.021 Gross varicose veins 12 (3.9) 3 (4.8) 0.724 BMI ≥ 30 96 (31.0) 23 (37.1) 0.346 Postpartum risk factors (n = 330) (n = 66) Age > 35 years 69 (20.9) 14 (21.2) 0.956 Personal history of VTE 0 (0) 2 (3.0) 0.002 Positive family history for VTE 4 (1.2) 1 (1.5) 0.841 Inherited thrombophiliab 4 (13.3) 6 (15.4) 0.811 APSc 0 (0) 1 (3.2) 0.045 Medical comorbiditiesd 9 (2.7) 18 (27.3) <0.001 Chronic hypertension 42 (12.7) 14 (21.2) 0.074 Systemic infection 7 (2.1) 11 (16.7) <0.001 HIV 58 (17.6) 20 (30.3) 0.019 ART 48 (14.5) 17 (25.8) 0.817 Parity ≥3 90 (27.3) 23 (34.8) 0.215 Multiple pregnancy 16 (4.8) 5 (7.6) 0.370 Pre-eclampsia with FGR 34 (10.3) 19 (28.8) <0.001 Non-obstetric surgery 1 (0.3) 5 (7.6) 0.003 Hospital admission or immobility 11 (3.3) 22 (33.3) <0.001 Gross varicose veins 10 (3.0) 1 (1.5) 0.503 BMI ≥40 12 (3.6) 5 (7.6) 0.159 Preterm delivery 77 (23.3) 29 (43.9) <0.001 Stillbirth 23 (7.0) 8 (12.1) 0.160 Elective cesarean delivery 83 (25.2) 12 (18.2) 0.229 Emergency cesarean delivery 130 (39.4) 27 (40.9) 0.818 Prolonged labor 8 (2.4) 5 (7.6) 0.042 Postpartum hemorrhage 21 (6.4) 14 (21.2) <0.001 Abbreviations: APS, antiphospholipid syndrome; ART, antiretroviral therapy; BMI, body mass index (calculated as weight in kilograms divided by the square of height in meters); FGR, fetal growth restriction; VTE, venous thromboembolism. a Values are given as number (percentage). b 43 cases and 14 controls prepartum; 39 cases and 30 controls postpartum. c 39 cases and 59 controls prepartum; 31 cases and 116 controls postpartum. d Medical comorbidities refer to having at least one of the following: heart disease, active connective tissue disease, malignancy, diabetes mellitus with nephropathy or chronic kidney disease. e 32 cases and 188 controls prepartum.

Explanatory variables from the univariate analyses were selected to construct the multiple models. In the prepartum model, medical comorbidities, HIV, and hospital admission or immobility were identified as risk factors (Table 3). For the postpartum risk factors, two models were developed. Model one included the following variables: medical comorbidities, systemic infection, HIV, hospital admission or immobility, postpartum hemorrhage, and pre-eclampsia with FGR. Model two included medical comorbidities, systemic infection, HIV, hospital admission or immobility, postpartum hemorrhage, and preterm delivery. Collinearity was detected between significant postpartum variables, preterm delivery, and pre-eclampsia and FGR. Pre-eclampsia and FGR (OR 2.74, 95% CI 1.18–6.36, P > 0.019) showed a stronger association with postpartum VTE compared with preterm delivery (OR 1.62, 95% CI 0.81–3.25, P = 0.080) and was retained for further analysis.

TABLE 3. Univariate and multiple logistic regression analysis of venous thromboembolism Variable Univariate logistic regression Multiple logistic regression OR 95% CI P value Adjusted ORa 95% CI P value Prepartum (AUC 0.66, P = 0.372) Medical comorbiditiesb 5.59 2.01–15.54 <0.001 5.32 1.82–15.56 0.002 HIV infection 2.86 1.59–5.15 <0.001 2.84 1.50–5.41 0.001 Hospital admission or immobility 5.28 1.28–21.70 0.021 5.33 1.17–24.22 0.030 Postpartum (AUC 0.86, P = 0.172) Medical comorbiditiesb 13.38 5.68–31.47 <0.001 23.72 8.75–64.27 <0.001 HIV infection 2.04 1.12–3.70 0.019 3.20 1.49–6.87 0.003 Systemic infection 9.23 3.42–24.83 <0.001 4.48 1.28–15.68 0.019 Hospital admission or immobility 14.50 6.58–31.93 <0.001 13.18 5.04–34.49 <0.001 Postpartum hemorrhage 3.96 1.89–8.28 <0.001 4.38 1.75–10.97 0.003 Pre-eclampsia with FGR 3.52 1.86–6.68 <0.001 2.74 1.18–6.36 0.019 Abbreviations: AUC, area under the curve; CI, confidence interval; FGR, fetal growth restriction; OR, odds ratio. a Adjusted for age. b Medical comorbidities refer to having at least one of the following; heart disease, active connective tissue disease, malignancy, diabetes mellitus with nephropathy or chronic kidney disease.

Subsequently, prepartum and postpartum risk factors for VTE were analyzed according to HIV infection. Table 4 shows the characteristics of HIV-infected cases and controls compared with corresponding HIV-negative patients. Antepartum risk factors independently associated with HIV and thrombosis included medical comorbidities and chronic hypertension. The following postpartum risk factors were also associated with HIV and thrombosis; namely, a personal history of VTE, medical comorbidities, systemic infection, hospital admission or immobility, and postpartum hemorrhage. However, there was no difference in the rates of stillbirth between the groups (P = 0.154). Furthermore, the risk of thrombosis was not significantly associated with immune suppression, as evidenced by viral load and CD4 count, opportunistic infections, and ART (Table 5).

TABLE 4. Prepartum and postpartum risk factors in the study participants according to HIV infection Variable HIV-infected controls HIV-negative controls HIV-infected with VTE HIV-negative with VTE P value Prepartum risk factors (n = 56) (n = 254) (n = 24) (n = 38) Age >35 years 28 (50.0)* 35 (13.8) 8 (33.3) 8 (21.1) <0.001 Personal history of VTE 0 (0)** 0 (0) 1 (4.2) 4 (10.5) <0.001 Positive family history for VTE 1 (1.8) 1 (0.4) 1 (4.2) 1 (2.6) 0.222 Inherited thrombophiliab 2 (100.0)* 0 (0) 6 (33.3) 5 (20.0) 0.009 Protein S 2 (100.0)* 0 (0) 5 (27.8) 4 (16.0) 0.007 Protein C 0 (0) 0 (0) 1 (5.6) 0 (0) 0.531 Prothrombin gene mutation 0 (0) 0 (0) 0 (0) 1 (4.0) 0.729 APSc 0 (0) 0 (0) 0 (0) 2 (7.1) 0.084 Medical comorbidities 1 (1.8) 7 (2.8)*** 4 (16.7) 4 (10.5) 0.002 Heart disease 0 (0)**** 2 (0.8)*** 3 (12.5) 2 (5.3) <0.001 Active connective tissue disease 0 (0) 3 (1.2) 1 (4.2) 2 (5.3) 0.156 Diabetes mellitus with nephropathy 1 (1.8) 2 (0.8) 0 (0) 0 (0) 0.764 Chronic hypertension 5 (8.9) 15 (5.9)*** 6 (25.0) 1 (2.6) 0.004 Systemic infection 2 (3.6) 1 (0.4) 1 (4.2) 2 (5.3) 0.048 Parity ≥3 21 (37.5) 53 (20.9) 9 (37.5) 6 (15.8) 0.013 Multiple pregnancy 1 (1.8) 5 (2.0) 1 (4.2) 1 (2.6)