Abstract
Objective
To explore the perioperative risk factors for predicting postpartum pulmonary embolism (PE) in Taiwanese women with Cesarean section (CS) delivery.
Methods
Data from Taiwan Longitudinal Health Insurance Database were analyzed. All CS women (2002–2007) in Taiwan, according to Diagnosis-Related Group codes, were included. Women having postpartum PE were identified by the diagnosis codes of PE from the medical records within 40 days after CS. Risk factors were analyzed using multivariate logistic regression.
Results
A total of 285,043 women who received CS between 2002 and 2007 were analyzed. Among them, 44 women were diagnosed as having postpartum PE. The overall incidence of postpartum PE was 0.154 per 1000 CS women. Analysis revealed that the perioperative risk factors for predicting postpartum PE in CS women included chronic heart disease (adjusted odds ratio [OR] = 89.92, 95% confidence intervals [CI] = 41.34–195.60, P < 0.001), systemic lupus erythematosus (adjusted OR = 45.05, 95% CI = 7.56–268.40, P < 0.001), postpartum hemorrhage (adjusted OR = 3.20, 95% CI = 1.10–9.31, P = 0.033), postpartum blood transfusion (adjusted OR = 8.92, 95% CI = 4.17–19.09, P < 0.001) and postpartum infection (adjusted OR = 7.13, 95% CI = 2.93–17.38, P < 0.001). Of note, anesthetic mode was not a risk factor for predicting postpartum PE in CS women, as women receiving general anesthesia for CS delivery were not associated with an increased risk of developing postpartum PE comparing to those who received neuraxial anesthesia (adjusted OR = 1.28, 95% CI = 0.52–3.14, P = 0.591).
Conclusions
Chronic heart disease, systemic lupus erythematosus, postpartum hemorrhage, postpartum blood transfusion and postpartum infection, but not anesthetic mode, were strong perioperative risk factors for predicting postpartum PE in Taiwanese CS women.
Keywords
Cesarean section; Perioperative risk factors; Postpartum; Pulmonary embolism;
1. Introduction
Pulmonary embolism (PE), a severe form of venous thromboembolism and a rare but potentially fatal disorder, remains a major cause of maternal mortality despite the tremendous progresses in modern obstetrical practices.1,2 The risk of PE in pregnant women is four to five times higher than those who are not pregnant.3,4 The risk of PE further increased during the postpartum period.3,4 What's more, the incidence of postpartum PE is much higher with Cesarean section (CS) delivery than with vaginal delivery.5,6 Many risk factors for PE have been identified for women during pregnancy and the puerperium period.7 However, PE was thought to be relatively uncommon for Asian population and very few data are available regarding the perioperative risk factors of postpartum PE in Asian women undergoing CS delivery.
The etiology of PE during pregnancy is multifactorial. Crucial mechanisms include gradual activation of the hemostatic system, progressive development of hormonally-induced regional stasis of blood flow and pelvic venous compression from the enlarging uterus.8,9 The impacts of these mechanisms reach a peak in late pregnancy.10 For women undergoing CS delivery, surgery can further enhance hypercoagulability, as tissue damage, catecholamine release and systemic inflammatory responses resultant from surgery synergistically activate coagulation factors and platelets.11–13 Surgery can also decrease deep vein blood flow in legs and aggravate regional stasis.14 These mechanisms provide rational explanations for the clinical observation of higher incidences of postpartum PE in CS women.
Identification of high risk patients and provision of timely therapeutic interventions is the key element for successful management of PE.10 Therefore, we conducted this large, comprehensive population-based study aimed at identifying the perioperative risk factors for predicting postpartum PE in Taiwanese CS women.
2. Materials and methods
2.1. Ethics
Ethical approval for this study (Protocol Number: 02-X11-025) was provided by the Institutional Review Board of Taipei Tzu Chi Hospital, Taipei, Taiwan. This study complies with “Personal Information Protection Act” in Taiwan and was exempt from full review by the Institutional Review Board of Taipei Tzu Chi Hospital.
2.2. Data source
This study used anonymized data from the Taiwan National Health Insurance Research Database (NHIRD) that was published by the National Health Research Institutes (NHRI) and released for public access for research purposes. The National Health Insurance (NHI) program in Taiwan is a single-payer payment system with the government as the sole insurer. The NHI program provides a comprehensive benefit package covering preventive, dental and medical services to all the 23 million residents of Taiwan. The NHI program provides universal coverage and access to any medical institution of the individual patient's choice. According to the NHRI, all data that could be used to identify the patients or the care providers or medical institutions were encrypted before entry into the NHIRD, and were further encrypted by the NHRI before being released to researchers.
2.3. Study sample
In this analysis, the CS cohort was defined from medical records, between January 2002 and December 2007, as all women ascribed to Diagnosis-Related Group (DRG) codes of CS (0371A) and maternally requested CS (0373B). Those who aged under 16 or over 50 years were excluded from the data set. Previous data indicated that the risk for postpartum venous thromboembolism remains high for up to 6 weeks after delivery.4 Thus, the follow-up for CS cohort was conducted up to 40 days after CS or until their withdrawal from the NHI program.
2.4. Definition of variables
The diagnosis of PE was identified from the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnostic codes (Table 1). CS women with the PE diagnostic code on their discharge with CS history within 40 days were identified as the PE CS cohort and those without the PE diagnostic code were identified as the non-PE CS cohort. Based upon previous studies reporting the risk factors of PE or venous thromboembolism,5,6 we also extracted and factored into the analysis information regarding CS women's demographic characteristics, medical comorbidities, pregnancy characteristics, delivery complications and anesthetic mode from NHIRD by using the ICD-9-CM diagnosis and order codes (Table 1). For factors regarding pregnancy characteristics and delivery complications, we only extracted those at the time of the index pregnancy or the index CS. Postpartum blood transfusion and postpartum infection (including urinary tract infection) were recognized as the related diagnosis codes on the discharge records from the index CS or within 40 days after the index CS.
2.5. Statistical analyses
We performed Chi-square test to compare the differences in sample characteristic distributions, including baseline demographic characteristics, medical comorbidities, pregnancy characteristics, anesthetic mode and delivery complications, between the PE CS cohort and the non-PE CS cohort. All variables with a P value of less than 0.2 in univariate logistic regression were subsequently included in the multivariate logistic regression models to identify the predicting factors for postpartum PE in CS cohort. The Statistical Package for the Social Sciences, version 16.0, for windows (SPSS, Inc., Chicago, IL, USA) was used for statistical analyses.
3. Results
A total of 44 patients were identified as the PE CS cohort and 284,999 were identified as the non-PE CS cohort (Fig. 1). The overall incidence rate of postpartum PE was 0.154 per 1000 CS women.
Download full-size image
The differences in baseline demographic characteristics, medical comorbidities, pregnancy characteristics, anesthetic mode and delivery complications between the PE and non-PE CS cohorts were summarized in Table 2. Compared to the non-PE CS cohort, the PE CS cohort had significantly higher incidences of chronic heart disease (including rheumatic heart disease, chronic rheumatic pericarditis, endocarditis and diseases of aortic valve, mitral valve, tricuspid valve or other endocardial structure) (P < 0.001), systemic lupus erythematosus (P < 0.001), preeclampsia (P = 0.004), early or threatened labor (P = 0.003), postpartum hemorrhage (P < 0.001), postpartum blood transfusion (P < 0.001), postpartum infection (including urinary tract infection) (P < 0.001) and receiving general anesthesia (P < 0.001).
All variables that presented in Table 2 with a P value of less than 0.2 in univariate logistic regression were further analyzed to elucidate the risk factors for predicting postpartum PE in CS cohort (Table 3). Analysis revealed that the crucial risk factors for predicting postpartum PE in CS women included chronic heart disease chronic heart disease (adjusted odds ratio [OR] = 89.92, 95% confidence intervals [CI] = 41.34–195.60, P < 0.001), systemic lupus erythematosus (adjusted OR = 45.05, 95% CI = 7.56–268.40, P < 0.001), postpartum hemorrhage (adjusted OR = 3.20, 95% CI = 1.10–9.31, P = 0.033), postpartum blood transfusion (adjusted OR = 8.92, 95% CI = 4.17–19.09, P < 0.001) and postpartum infection (adjusted OR = 7.13, 95% CI = 2.93–17.38, P < 0.001). Of note, preeclampsia, early or threatened labor and general anesthesia were not significant risk factors for predicting postpartum PE in CS women (all P > 0.05, Table 3).
4. Discussion
Impacts of a wide range of potential risk factors, including demographic characteristics, medical comorbidities, pregnancy characteristics, delivery complications, and anesthetic mode, were comprehensively investigated in this population-based study. Data from this large, national cohort study of nearly 300,000 CS deliveries over a 6-year period revealed that the overall incidence rate of postpartum PE was 0.154 per 1000 CS deliveries in Taiwan. This study further demonstrated that chronic heart disease, systemic lupus erythematosus, postpartum hemorrhage, postpartum blood transfusion and postpartum infection were strong perioperative risk factors for predicting postpartum PE in CS women. These data therefore can provide the most relevant information regarding the risk factors in predicting postpartum PE in Taiwanese CS women and the clinical significance of these data should be profound.
As aforementioned, CS women with chronic heart disease or systemic lupus erythematosus were noted to have significantly higher risk for postpartum PE in this study. Evidence of sustained active inflammation was observed in patients with chronic heart disease, especially rheumatic heart disease.15 Similar to chronic heart disease, sustained active inflammation is also observed in patients with systemic lupus erythematosus.16 Previous data revealed that patients with systemic lupus erythematosus had an increased risk of developing thrombosis.16 The underlying mechanisms are thought to involve platelet activation, complement consumption and augmenting platelet complement deposition.17 As mentioned previously, pregnancy induced a prothrombotic environment.8–10 It is thus reasonable to observe clinically that pregnancy can significantly increase the risk of thrombosis in chronic heart disease and systemic lupus erythematosus patients.6 Of particular relevance to it is the demonstration that coincident inflammation and hypercoagulability increased the risk of thrombosis.18 Since surgery aggravated inflammation,11,12 it is reasonable to observe our data that CS women with chronic heart disease or systemic lupus erythematosus had an increased risk of developing postpartum PE.
These above-mentioned data, consistent with those previous ones,5,6 confirm the concept that medical comorbidities, including chronic heart disease and systemic lupus erythematosus, do contribute to the development of postpartum PE in Taiwanese CS women. However, in contrast to those previous data,5,6 this study revealed that the medical comorbidities of diabetes mellitus, chronic hypertension, varicose vein of lower extremities, sickle cell disease, inflammatory bowel disease, obesity and chronic obstructive pulmonary diseases (i.e. the proxy of smoking) did not contribute to the development of postpartum PE in Taiwanese CS women. The mechanisms underlying the discrepancy between our data and those previous ones5,6remain un-studied. More studies are needed before further conclusion can be drawn. Nevertheless, one possible explanation for our observation may be that the sample size of the PE CS cohort was quite small and the case numbers of the PE CS cohort with these medical comorbidities was even smaller. For instance, the case number of diabetes mellitus, chronic hypertension, varicose veins of lower extremities or inflammatory bowel disease of the PE CS cohort was zero. The other possible explanation may be that codes utilized by the NHIRD database did not allow us to track body mass index and smoking status. Alternatively, we employed the codes of obesity and chronic obstructive pulmonary diseases (as the proxies of body mass index and smoking, respectively) to help elucidate the possible impacts of these two factors. Therefore, it is possible that our analysis underestimates the impacts of these medical comorbidities.
Our analysis also revealed that postpartum hemorrhage, postpartum blood transfusion or postpartum infection (including urinary tract infection) had significantly higher risks of developing postpartum PE than CS women without those above-mentioned conditions. Acute hemorrhage can induce hypercoagulability due to decreased anticoagulant factors along with exposure of tissue factor at the vascular injury site, especially in the setting of prolonged or major hemorrhage.19 Previous data also indicated that postpartum hemorrhage is a strong risk factor for thrombosis in postpartum women.7 In line with this notion, it is thus reasonable to observe our data that postpartum hemorrhage is a strong risk factor for predicting postpartum PE in CS women. Moreover, postpartum blood transfusion was usually employed in instances where CS women experienced severe postpartum hemorrhage. On top of that, blood transfusion with a pro-inflammatory nature may further disrupt the balance of coagulation system and facilitate hypercoagulable status.20 Take above mentioned into account, it is thus reasonable to observe that postpartum blood transfusion is also a strong risk factor for predicting postpartum PE in CS women. As above-mentioned, postpartum infection was also a strong risk factor in predicting postpartum PE in CS women. Infection can trigger inflammatory host response and expression of inflammatory mediators and subsequently induce hypercoagulability in patients with infection.21 In line with this notion, it is thus reasonable to observe our data that CS women with postpartum infection were at a higher risk of having postpartum PE than those who without postpartum infection.
Hypercoagulability can also be observed in patients with preeclampsia and early or threatened labor.22–24 However, to our surprise, our analysis revealed that neither preeclampsia nor early or threatened labor was a risk factor for predicting postpartum PE in CS women. More studies are needed before further conclusion can be drawn. Nevertheless, one possible explanation may be that both preeclampsia and early or threatened labor present as serious but acute pathologies which are abruptly terminated by delivery. As such the short-term pathologies may not contribute significantly to the development of postpartum PE in CS women.
Previous data revealed that patients receiving neuraxial anesthesia had lower surgical stress and catecholamine levels than patients receiving general anesthesia.25,26Moreover, blood flow in lower limbs was increased during neuraxial anesthesia.25,26 It is thus reasonable to observe previous data that the risk of PE in patients receiving neuraxial anesthesia were significantly lower than those who receiving general anesthesia.27 However, to our surprise, our data revealed that anesthetic modes did not exert significant impacts on the risk of developing postpartum PE in CS women. The mechanisms underlying the discrepancy between our data and those previous ones27 remain to be elucidated. Nevertheless, those previous data were obtained from surgeries other than CS delivery and most of the patients were Caucasian with both genders (i.e., male and female).27 In contrast, this study included only Asian (Taiwanese) female patients undergoing CS delivery. It is likely that the differences in surgery, population and/or gender might contribute to the observed discrepancy between our data and those previous ones.27 More studies are needed before further conclusions can be drawn.
We acknowledged that the NHIRD database imposes some limitations on the analysis. Firstly, the codes utilized by the Taiwanese NHIRD database did not reveal the extent of blood loss at postpartum hemorrhage. Secondly, there is difficulty establishing exacting temporality between PE and the other risk factors identified in the same hospital admission period (e.g., postpartum PE and postpartum urinary tract infection) using codes of the NHIRD database. However, influence of this effect is probably minimal since our sample size is relative large while the maximum temporal disparity can only be 40 days. Thirdly, PE is a potentially fatal postpartum complication. Knowing the mortality rate of postpartum PE is important. However, NHIRD database provides only claim data and the data of mortality are not available in this database. Therefore, this study cannot perform further investigation in this regard. More studies are needed before we can conclude on the mortality rate as well as the causes of mortality and the risk factors of mortality caused by PE.
In conclusion, chronic heart disease, systemic lupus erythematosus, preeclampsia, postpartum hemorrhage, postpartum blood transfusion and postpartum infection, but not anesthetic mode, were strong perioperative risk factors for predicting postpartum PE in Taiwanese CS women.
Conflicts of interest
The authors state that there is no financial and non-financial conflict of interest upon the publication of this study.
Funding/support statement
This study was supported by grants from Taipei Tzu Chi Hospital [TCRD-TPE-104-30and TCRD-TPE-104-RT-1 (2)] awarded to C.J. Huang.