Ferres, Millie A. MD; Olivarez, Sofia A. MD; Trinh, Victoria; Davidson, Christina MD; Sangi-Haghpeykar, Haleh PhD; Aagaard-Tillery, Kjersti M. MD, PhD
Venous thromboembolic disease, including pulmonary embolus and deep vein thrombosis (DVT), remains the leading cause of maternal morbidity and mortality in developed countries.1–4 The maternal mortality ratio due to thromboembolic disease has changed little over the past 40 years, whereas all other causes of maternal death have declined during this time.3,4 There is a near general agreement that the incidence of venous thromboembolism, especially pulmonary embolus and death, is higher during the postpartum than the antepartum period,2 and cesarean delivery alongside length of labor are important risk factors.5 Other investigators have observed that although pregnancy itself is associated with a 2.5- to 10-fold increase in the risk of venous thromboembolism when compared with age- and parity-matched controls, this risk increases to 6.7- to 22-fold higher after cesarean delivery.5,6
Thromboprophylaxis through the postpartum interval is a known effective prevention for development of embolic event in women with a history of venous thromboembolism.7–12 Cochrane Database and expert series reviews have consistently demonstrated a decrease in the risk of thromboembolic disease with subcutaneous heparin or low molecular weight heparin after major surgery.13–15 However, to date, no adequately powered clinical trials of thromboprophylaxis after cesarean delivery are available to thoroughly evaluate risk-benefit ratios.7–10,16,17 As a result, the majority of current recommendations are extrapolated from studies in nonpregnant women; are based on epidemiologic data, expert opinion, and consensus; or both.7
Based on the review of the literature and the American College of Chest Physicians Evidence, in 2007 the Perinatal Guidelines Committee at our institution established a consensus protocol for postpartum thromboprophylaxis. Given our care system and patient population, we were in a unique position to capture postpartum follow-up data and study the cohort of individuals who received postpartum thromboprophylaxis compared with those who did not. We hypothesized that compliance with administration of enoxaparin among at-risk gravid women would not be associated with a significant increased risk of secondary wound complications or postoperative bleeding but would decrease the rate of venous thromboembolism. Our objective was to estimate the rate of wound complications associated with protocol-driven postcesarean enoxaparin thromboprophylaxis.
MATERIALS AND METHODS
We employed a retrospective cohort design with patients from the Harris County Hospital District through Ben Taub General Hospital after institutional review board approval from Baylor College of Medicine with subsequent approval by the Harris County Hospital District. Our convenience cohort (23-month interval) was comprised of at-risk gravid women who received guideline-recommended thromboprophylaxis (at-risk, protocol-compliant “cases”) and their similarly at-risk cohort who did not (at-risk, protocol-noncompliant “controls”). Our primary measures were the rate of wound separation, wound infection, and rehospitalization. Our secondary measures were the rate of venous thromboembolism. Subsequent stratification by virtue of additional maternal comorbidities (body mass index [BMI, calculated as weight (kg)/[height (m)]2], maternal diabetes, and hypertensive disorders) was performed to address for the potential of further confounding.
All at-risk individuals were identified after extensive review of every cesarean delivery that occurred at Ben Taub General Hospital from the time of protocol implementation into standard clinical care through the most current completed postpartum interval (December 2007 through October 2009). No alterations in clinical care were undertaken as part of this study, and all practicing physicians were aware of guideline implementation. Cohorts were derived from among those women meeting at-risk criteria for consideration of thromboprophylaxis: “protocol-noncompliant controls” were gravid women defined as at-risk but who did not receive enoxaparin thromboprophylaxis, and “protocol-compliant cases” were gravid women who were similarly at-risk and did receive postpartum thromboprophylaxis. Follow-up incidence of secondary complications was queried through the Harris County Hospital District electronic medical record system, which incorporates three hospitals and more than 38 community and mobile health clinics.
According to the perinatal guidelines at Ben Taub General Hospital (established by the Perinatal Committee of the Department of Obstetrics and Gynecology in Baylor College of Medicine), women undergoing cesarean delivery who were older than 35 years of age or had a BMI greater than 30 at that time, or both, were considered at moderate to high risk for venous thromboembolism. These women therefore met criteria for thromboprophylaxis (Appendix 1, available online at http://links.lww.com/AOG/A210).7 “Protocol-compliant at-risk subjects” constituted the cohort who were administered a subcutaneous prophylactic dose of low molecular weight heparin (eg, 40 mg of enoxaparin [Lovenox]) 6 hours after cesarean delivery. A repeated prophylactic dose of low molecular weight heparin was given daily until discharge from the hospital. If contraindications to the use of heparin were present (examples include postpartum hemorrhage or complications of regional anesthesia), graduated compression stockings were used as an alternative.
After institutional review board approval from at Baylor College of Medicine and the Harris County District, a complete list of all the cesarean deliveries that occurred after protocol establishment over a 23-month interval (December 1, 2007, through October 31, 2009) was obtained from the Data and Logistics from the Women, Pediatrics and Infants Service at Ben Taub General Hospital. Data from their antenatal, intrapartum, and postpartum course were abstracted from their electronic medical records (Epic), and in 1 of every 10 participants confirmed with “hard copy” (ie, paper copy) chart reviews. Variables extracted included maternal age, maternal weight and height on admission, labor and delivery course (ie, prolonged labor, estimated blood loss), maternal comorbidities (ie, chronic hypertension, diabetes mellitus, maternal tobacco use; as well as other comorbidities, such as antiphospholipid syndrome, systemic lupus erythematosus, and so forth), and enoxaparin use (with confirmation of medicine administration against the nursing records in the medicine administration record). Over the 3-month interval after their delivery, data pertaining to their postpartum course and potential for complications—that is, postpartum hormonal contraception, emergency room visits, clinic follow-ups, and readmission to the hospital for wound complications or thromboembolic disease—were also collected.
Diagnosis and identification of wound complications (wound hematoma, separation or dehiscence with or without infection) in both cohorts was defined by strictly documented physical examinations or wound description in hospital discharge notes, wound check visits, readmission summary, and operative notes. We included all wound complications, from minor disruptions to wound dehiscence (wound separation of 1–2 cm) to major wound complications requiring surgical exploration, incision and drainage, or placement of wound vac and debridement or a combination thereof. The diagnosis and identification of DVT or pulmonary embolism was based on hospital discharge or readmission diagnosis and summary, with detailed confirmation or review of radiographic evidence and treatment by at least two study investigators. All data were obtained from emergency medical records and review of Harris County admission records. For confirmation against data extracted from the emergency medical records, the charts were pulled for 1 of every 10 individuals such that validation with hard-copy chart review and complete data abstraction were performed in an a priori defined 10% of patients.
Data were analyzed in univariable and multivariable models. The primary outcome was the singular and composite rate of wound complications (wound hematoma, wound separation or dehiscence, with or without infection, rehospitalization for any of these wound complications, or a combination thereof) in at-risk patients receiving postcesarean prophylaxis (“cases”) as compared with those who did not (protocol-noncompliant “controls”). The secondary outcome was the occurrence of DVT or pulmonary embolism in these patients. Summary measures included the rate (incidence) and prevalence of both primary and secondary outcomes, where our denominator was the number of women who received enoxaparin and our numerator was the number of women who returned with complications (singular and composite). Statistical significance was derived using χ2 or Fisher exact test for discrete variables (wound separation, wound hematoma, rehospitalization, and a composite outcome of all three).
To examine the modification of the effect of maternal BMI on each outcome, we stratified the analysis based on BMI (less than 30, 30–35, higher than 35). Logistic regression analysis was employed to estimate odds ratios for various outcomes of interest by enoxaparin use adjusted for potential confounders, namely maternal age, maternal BMI, diabetes, and hypertension. SAS version 9.2 was used for all analyses. A nominal P<.05 was considered significant.
From December 1, 2007, to October 20, 2009 (23 months), 2,509 patients underwent cesarean delivery in our institution. After identification of inclusion criteria by risk assessment, a total of 1,677 patients (67%) met protocol at-risk criteria by age or BMI alone or both, whereas 832 did not meet protocol at-risk criteria. Of the 1,677 women who met criteria for thromboprophylaxis, 653 (39%) received enoxaparin (protocol-compliant “cases”) and 1,024 (61%) did not (protocol-noncompliant “controls”). Table 1 shows the patient characteristics of the protocol-compliant cases and protocol-noncompliant controls cohorts. The mean maternal age (30.7 compared with 29.2 years, P<.001) and BMI (35.7 compared with 33.7, P<.001) were significantly higher in the protocol-compliant cases cohort, as was the presence of diabetes (P=.002) and hypertension (P=.03).
Table 2 summarizes the primary outcome in a univariate analysis among the cohorts, as singular and composite morbidity. A total of 55 (8.5%) wound complications (separation, hematoma, or both) were identified among protocol-compliant cases, with 48 (4.7%) wound complications having occurred among protocol-noncompliant controls. Wound separations were significantly more common in the protocol-compliant cases (6.8% compared with 3.6%, P=.003), whereas no significant difference was noted in the rate of wound hematomas (P>.06). Because the rate of rehospitalization was higher in the protocol-compliant cases cohort (2.1% compared with 0.8%, P=.017), so was the overall composite morbidity (8.9% compared with 4.8%, P=.002).
A total of seven venous thromboembolic events were identified in the entire at-risk cohort (Table 3). Two pulmonary embolisms (0.3%) were identified in the enoxaparin cohort, whereas five (0.5%) events occurred in the at-risk protocol-noncompliant controls cohort, including four pulmonary embolisms (0.4%; P=.99) and one DVT (0.1%; P=.99). No superficial vein thrombosis was identified in either cohort.
Based on our univariable analysis with significant findings as documented in Tables 1 and 2, we next sought to control for significant potential confounders (such as BMI) in the rate of our primary outcome. We therefore stratified our analysis by virtue of maternal BMI. As reported in Table 4, we observed significant association for the occurrence of wound separation in morbidly obese women whose BMI strata exceeded 35 (P = .03), with the composite morbidity reaching significance only in the Class III obese strata (P = .049). Using logistic regression models adjusting for BMI, maternal age, diabetes, and chronic hypertension, enoxaparin use was found to be associated with wound separation at an odds ratio of 1.66 (P=.04) and overall composite morbidity at an odds ratio of 1.69 (P=.01) as observed in Table 5.
Venous thromboembolism persists as a current leading cause of maternal morbidity and mortality worldwide, and the leading cause of maternal death in the developed world. The incidence of pregnancy-related thromboembolism in two large population-based cohorts continues at 0.7 to 1.3 per 1,000 deliveries.1–4 For these reasons it has been advocated that an at-risk protocol be established for the administration of thromboprophylaxis in women at moderate to high risk for the development of venous thromboembolism after cesarean delivery.7 Such protocols are consistent with current guidelines in other surgical fields.13–15 Despite the near-uniform recommendations and establishment of protocols designed for the prevention of venous thromboembolism, obstetricians have been reluctant to initiate widespread use for fear of secondary mortality.8–11 We therefore sought to investigate the risk of primary wound complications among at-risk women receiving thromboprophylaxis compared with those who did not.
There are a number of noted strengths to our study. First, we took advantage of our large at-risk population. In the first 2 years (23 months) of the protocol, 2,509 cesarean deliveries were performed, of which 1,677 patients met moderate to high-risk criteria by BMI or maternal age alone, or both. This is in accordance with other similarly at-risk populations. Second, we performed extensive data abstraction, with the use of our County-wide (Harris County Hospital District) electronic medical record system, Epic. Third, because of leveraged access to care, our study population arises from a wide catchment area and broad community-based clinics. As of 2009, Harris County Hospital District provided care to nearly a quarter of the entire population of Harris County (the third most populous county in the United States). We were therefore able to ascertain compliance with follow-up with the incorporation of all our Harris County Hospital district care centers utilizing our emergency medical record system. This includes three hospitals and more than 38 community and mobile health clinics. We were therefore in a unique position to ascertain postoperative complications with measured follow-up exceeding 64% of participants with documented postpartum interval visits (data not shown).
There are several potential weaknesses to our study. First, we are reporting on our observations after enactment of perinatal guidelines and not strictly adhered study protocol, thus lending our study population to noncompliance and selection bias. Indeed, we observed that the percent of protocol noncompliance reached 53% in the first year and 72% in the second year, with the overall compliance rate of 38%. Because it was outside the scope of our analysis, we did not specifically characterize the rationale documented for protocol noncompliance (eg, intraoperative bleeding, postpartum hemorrhage, arteriovenous malformation, attending preference, and so forth) in this study. Second, despite our wide catchment arena for follow-up, we could document only 64% postpartum interval visits among study participants.
Nevertheless, several key observations from our study remain. First, there was an increase in the rate of wound separation in the protocol-compliant cases cohort (6.8% compared with 3.6%, P=.003). The rate of wound separation failed to reach significance when stratified by maternal obesity status except among the morbidly obese (BMI more than 35: 7.4% compared with 2.6%, P=.03; Table 3). Conversely, despite our protocol-noncompliant controls being at “lesser risk” (but still at-risk), the unadjusted rate of thrombosis was actually higher than that observed among the higher-risk protocol-compliant cases cohort (0.3% compared with 0.5%), with three additional thrombosis events in the individuals who should have gotten enoxaparin but did not. We find these latter observations to be potentially noteworthy for they suggest that in our cohort enoxaparin use was nonsignificantly associated with three fewer venous thromboembolic events. However, the only significant observation in our study remains an observed increased risk of wound separation in protocol-compliant cases.
We can further consider our study in the context of the feasibility of a nationwide study designed to test the weighted efficacy of risk-based thromboprophylaxis. Using the two sided Mantel-Haenszel test and building on our study observations with a rate of P1=.003 in the protocol-compliant cases cohort and P2=.005 in the protocol-noncompliant controls cohort, the difference in rate of outcome will reach significance, with 80% power, if 31,000 at-risk individuals were enrolled, with 15,500 individuals in each arm. Assuming a U.S. cesarean delivery rate of 29.4%, where 40% meet at-risk criteria (by BMI or age), and approximately 4.96 million live births per year (245,000 occurring in four of the most populous counties), such a trial could potentially be completed in less than a year for such populous counties alone. It bears mention that such a power analysis must be considered with caution, as our difference in the rate of venous thromboembolic events did not reach significance.
In summary, significant progress has been made in recent years in the prevention and management of thromboembolic events in the nonobstetric surgical population, and strong clinical guidelines have been established recently. However, our study notes that there is an increased risk of wound separation associated with enoxaparin use. This observed risk is not accompanied by a significant demonstration of prevention of thrombosis, which obligates a larger sample size. For this reason, we emphasize the need for a randomized controlled trial designed to specifically measure such risk alongside cost-assessment profiling. Such a trial would allow the establishment of evidence-based guidelines regarding thromboprophylaxis in pregnancy and the postpartum interval.
1. Palareti G. Pregnancy and venous thrombosis. Haematol Rep 2005;1:13–7.
2. Blandon M, Perrier A, Nendaz M, Righini M, Boehlen F, Boulvain M, et al. Thromboprophylaxis with low-molecular-weight heparin after cesarean delivery. Thromb Haemost 2010;103:129–37.
3. Chang J, Elam-Evans LD, Berg CJ, Herndon J, Flowers L, Seed KA, et al. Pregnancy-related mortality surveillance–United States, 1991–1999. MMWR Surveill Summ 2003;52:1–8.
4. Alexander S, Wildman K, Zhang W, Langer M, Vutuc C, Lindmark G. Maternal health outcomes in Europe. Eur J Obstet Gynecol Reprod Biol 2003;111(suppl 1):S78–87.
5. Gherman RB, Goodwin TM, Leung B, Byrne JD, Hethumumi R, Montoro M. Incidence, clinical characteristics, and timing of objectively diagnosed venous thromboembolism during pregnancy. Obstet Gynecol 1999;94(5 pt 1):730–4.
6. Heit JA, Kobbervig CE, James AH, Petterson TM, Bailey KR, Melton LJ 3rd. Trends in the incidence of venous thromboembolism during pregnancy or postpartum: a 30-year population-based study. Ann Intern Med 2005;143:697–706.
7. Bates SM, Greer IA, Pabinger I, Sofaer S, Hirsh J; American College of Chest Physicians. Venous thromboembolism, thrombophilia, antithrombotic therapy, and pregnancy: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines (8th Edition). Chest 2008;133(6 suppl):844S–86S.
8. Pomp ER, Lenselink AM, Rosendaal FR, Doggen CJ. Pregnancy, the postpartum period and prothrombotic defects: risk of venous thrombosis in the MEGA study. J Thromb Haemost 2008:6:632–7.
9. Nelson-Piercy C. Low molecular weight heparin for obstetric thromboprophylaxis. Br J Obstet Gynaecol 1994;101:6–8.
10. Gates S, Brocklehurst P, Ayers S, Bowler U; Thromboprophylaxis in Pregnancy Advisory Group. Thromboprophylaxis and pregnancy: two randomized controlled pilot trials that used low-molecular-weight heparin. Am J Obstet Gynecol 2004;191:1296–303.
11. James AH, Jamison MG, Brancazio LR, Myers ER. Venous thromboembolism during pregnancy and the postpartum period: incidence, risk factors, and mortality. Am J Obstet Gynecol 2006;194:1311–5.
12. De Stefano V, Martinelli I, Rossi E, Battaglioli T, Za T, Mannuccio Mannucci P, et al. The risk of recurrent venous thromboembolism in pregnancy and puerperium without antithrombotic prophylaxis. Br J Haematol 2006;135:386–91.
13. Colditz GA, Tuden RL, Oster G. Rates of venous thrombosis after general surgery: combined results of randomised clinical trials. Lancet 1986;2:143–6.
14. Clagett GP, Reisch JS. Prevention of venous thromboembolism in general surgical patients: results of a meta-analysis. Ann Surg 1988;208:227–40.
15. Collins R, Scrimgeour A, Yusuf S, Peto R. Reduction in fatal pulmonary embolism and venous thrombosis by perioperative administration of subcutaneous heparin: overview of results of randomized trials in general, orthopedic, and urologic surgery. N Engl J Med 1988;318:1162–73.
16. Burrows RF, Gan ET, Gallus AS, Wallace EM, Burrows EA. A randomised double-blind placebo controlled trial of low molecular weight heparin as prophylaxis in preventing venous thrombotic events after caesarean section: a pilot study. BJOG 2001;108:835–9.
17. Sharma S, Monga D. Venous thromboembolism during pregnancy and the post-partum period: incidence and risk factors in a large Victorian health service. Aust N Z J Obstet Gynaecol 2008;48:44–9.