Risk of Postoperative Venous Thromboembolism Among Pregnant Women

Venous thromboembolism (VTE) is a critical complication after surgery. Although pregnancy is a known risk factor of VTE, available data on the risk of postoperative VTE are scarce. Using the American College of Surgeons National Surgical Quality Improvement Program database between 2006 and 2012, we matched 2,582 pregnant women to 103,640 nonpregnant women based on age, race, body mass index, and modi ﬁ ed Rogers score. Pregnant women, compared with matched nonpregnant women, experienced higher incidence of VTE (0.5% vs 0.3%; odds ratio 1.93, 95% con ﬁ dence interval 1.1 to 3.37, p [ 0.02). Pregnant women also showed higher risk of pneumonia, ventilator dependence ‡ 48 hours, bleeding, and sepsis than did the counterparts. In conclusion, pregnancy was associated with higher risk of VTE after surgery as well as other postoperative complications. The absolute risk difference was small, and careful evaluation against the potential risk and bene ﬁ t should be given when surgical treatment is considered among pregnant women. (cid:1) 2017 Elsevier Inc. All rights reserved. (Am J Cardiol 2017;120:479 e 483) Venous thromboembolism (VTE) is a critical complication after surgery. Although pregnancy increases the risk of VTE 4- to 5-fold, 1 there are scarce data on the risk of VTE among pregnant women undergoing surgery. Therefore, we examined the association of pregnancy with postoperative VTE by using the American College of Surgeons National Surgical Quality Improvement Program (NSQIP) data. We also examined the association with other postoperative complications because the risk of

Venous thromboembolism (VTE) is a critical complication after surgery. Although pregnancy is a known risk factor of VTE, available data on the risk of postoperative VTE are scarce. Using the American College of Surgeons National Surgical Quality Improvement Program database between 2006 and 2012, we matched 2,582 pregnant women to 103,640 nonpregnant women based on age, race, body mass index, and modified Rogers score. Pregnant women, compared with matched nonpregnant women, experienced higher incidence of VTE (0.5% vs 0.3%; odds ratio 1.93, 95% confidence interval 1.1 to 3.37, p [ 0.02). Pregnant women also showed higher risk of pneumonia, ventilator dependence ‡48 hours, bleeding, and sepsis than did the counterparts. In conclusion, pregnancy was associated with higher risk of VTE after surgery as well as other postoperative complications. The absolute risk difference was small, and careful evaluation against the potential risk and benefit should be given when surgical treatment is considered among pregnant women. Ó 2017 Elsevier Inc. All rights reserved. (Am J Cardiol 2017;120:479e483) Venous thromboembolism (VTE) is a critical complication after surgery. Although pregnancy increases the risk of VTE 4-to 5-fold, 1 there are scarce data on the risk of VTE among pregnant women undergoing surgery. Therefore, we examined the association of pregnancy with postoperative VTE by using the American College of Surgeons National Surgical Quality Improvement Program (NSQIP) data. We also examined the association with other postoperative complications because the risk of postoperative VTE may be induced by other complications (i.e., infections and bleeding) and vice versa.

Methods
An exemption from institutional review board review for the use of the NSQIP patient-level data was obtained at the University of California, Irvine Medical Center. We identified 269,104 women (including 3,719 pregnant women) who were aged <50 years and who underwent surgery using the NSQIP database between 2006 and 2012, when data on pregnancy are available ( Figure 1). Clinical data on preoperative risk factors and laboratory values, type of surgery, and 30-day postoperative mortality and morbidity were collected by trained nurses at >500 hospitals across North America. After excluding patients with missing information for relevant clinical factors, we identified 192,554 patients including 2,596 pregnant women. We calculated modified Rogers score where we excluded hematocrit, serum bilirubin, and serum albumin from original Rogers score 2 because these factors are influenced by pregnant status, and hence considered intermediates rather than confounders. Primary outcome was postoperative VTE defined as either venous thromboses or pulmonary embolism during 30 days after surgery. Secondary outcomes included other postoperative complications such as surgical site infection, pneumonia, ventilator dependence >48 hours, bleeding, and sepsis (see Supplementary Text for detail). We estimated odds ratios (ORs) of postoperative complications using logistic regression. Age, race, and body mass index were adjusted for in the complete case analysis. Additionally, pregnant women were matched to nonpregnant women based on age, race, body mass index, and modified Rogers score by coarsened exact matching with Sturge's rule. 3e5 First, we coarsened age and modified Rogers score into categories using cut-points of 20 and 35 years and 3, 14, 16, and 19 points, respectively. We then sorted all patients by each stratum of the age category, race, and the modified Rogers score category. Within each stratum that included at least 1 patient in each group, pregnant patients were given a weight of 1, and nonpregnant women were given a weight that equalized the ratio of sum of weights in each group of the stratum to the ratio of total matched patients on each group in the complete case cohort. The comparison of clinical characteristics between matched cohort and unmatched cohort was listed in Supplementary Table 1. The association between pregnancy and each postoperative complication was then estimated in the matched cohort without adjustment. As a sensitivity analysis, we further added emergent surgery into matching variables.
Between-group differences were evaluated by standardized difference because of the large sample size of this study; 0. 8 large, medium, and small differences. 6,7 Statistical significance was defined as a p value of <0.05. All analysis was performed using Stata 13.1 (StataCorp LP, College Station, Texas).

Results
Among 2,596 pregnant and 189,958 nonpregnant women in the complete case cohort, 2,582 (99.5%) pregnant women were matched to 103,640 nonpregnant women. There were large differences in age, body mass index, and modified Rogers sore between pregnant versus nonpregnant women in the complete case cohort, and these differences were diminished in the weighted matched cohort (Table 1). Among the components of modified Rogers score, pregnant women had lower preoperative serum sodium levels and higher prevalence of emergent surgery in both the entire and weighted matched cohort. The incidence of VTE among pregnant and nonpregnant patients was 0.5% versus 0.4% in the complete case cohort and 0.5% versus 0.3% in the matched cohort, respectively ( Table 2). After matching, the incidence of other postoperative complications also decreased among nonpregnant women.
Pregnant women, compared with nonpregnant women, showed higher risk of VTE and other complications except for surgical site infection after adjustment for age, race, and body mass index in the matched cohort ( Figure 2

Discussion
This study demonstrated the association of pregnancy with increased risk of postoperative VTE. The risk of other postoperative complications (i.e., pneumonia, ventilator dependency >48 hours, bleeding, and sepsis) was also significantly higher among pregnant patients than nonpregnant female patients.
The absolute rate of postoperative VTE in this study was consistent with previous studies. 8e10 A recent US population-based study by Abbasi et al 9 also showed that among pregnant women who developed acute appendicitis during the hospital stay for delivery, the risk of postoperative VTE was 60% higher than nonpregnant women after adjustment for age, race, obesity, income, insurance, and hospital type. Our study included all types of surgical procedures that were ever performed for pregnant women during the antepartum period, and confirmed the association between pregnancy and postoperative VTE even after matching on factors related to VTE.
Previous studies also showed that pregnant women had higher risk of postoperative adverse events including

480
The American Journal of Cardiology (www.ajconline.org) infection, sepsis, pneumonia, and mortality. 8,9 Patients with severe complications, especially those who are ventilator dependent, usually require complete bed rest, and immobility is a major cause of postoperative VTE. 11 Additionally, infection and systemic inflammation may increase the risk of VTE through Virchow triad (i.e., stasis of   blood flow, endothelial injury, and hypercoagulability). 12,13 Bleeding activates coagulation system, and transfusion may also increase the risk of VTE. 14e16 Meanwhile, pulmonary embolism may cause other complications including respiratory dysfunction and sepsis in turn, and anticoagulant therapy increases bleeding risk, creating a vicious cycle among these complications. Hypernatremia is an index of dehydration which may lead to postoperative VTE, 17 as included in Rogers score. 2 Despite the physiological increase in body fluid during the antepartum period as partly reflected by lower serum sodium levels, pregnancy was associated with postoperative VTE. These results suggested the involvement of other factors such as increased procoagulant factors (i.e., fibrinogen, factor VIII, and von Willebrand factor) 18 and reduced anticoagulant factors (i.e., protein C and protein S), 19 as well as venous compression by enlarged uterus. 2 The higher prevalence of emergent surgery among pregnant women may reflect an effort of physicians to avoid surgical complications. However, delayed treatment may result in adverse consequences. Abbasi et al 9 reported that conservative treatment was more common than expected and associated with septic shock, peritonitis, and VTE among pregnant women. We also found that the risk of pneumonia, ventilator dependence, and sepsis was attenuated after adding emergent surgery into matching variables, suggesting the contribution of preoperative conditions to these complications. The postoperative VTE risk remained significant, but the absolute rate difference was 0.2%. Therefore, surgical procedures for pregnant women should be considered with the risk-benefit balance between surgical versus conservative treatment on an individual basis. Further studies are necessary to identify high-risk patients who need careful monitoring and aggressive prophylaxis.
Several limitations should be noted. First, there may be residual confounding and/or effect modification by type of surgery. 20 However, type of surgery was contained in Rogers score, 2 and we selected those ever performed in pregnant women. Second, there may also be unmeasured confounding by hereditary thrombophilia (i.e., antithrombin, protein C, or protein S deficiency), 21 smoking, a history of VTE, and the use of prophylactic and therapeutic drugs and devices. Third, the incidence of VTE may be underestimated in this study because the NSQIP database contains only clinically observed outcomes within 30 days after surgery. Indeed, a previous study showed that the risk of VTE among middle aged women was substantially increased in the first 12 postoperative weeks. 10 Finally, gestational period was not recorded in the NSQIP database. The incidence rate of venous thromboses may be consistent across the trimesters, 22 but the physiological changes in the course of pregnancy may alter the risk of VTE among pregnant women, which need further investigation.