Several important physiologic changes during pregnancy increase the risk for development of venous thromboembolism, including changes in the hemostatic system causing hypercoagulability, hormonally induced changes in venous outflow, mechanical obstruction by the uterus, and vascular injury.1,2 Pregnancy-related hypercoagulability is maximal immediately postpartum, conferring an increased risk for venous thromboembolism during that time.2 Previous studies have documented an increased risk for venous thromboembolism of 2.5–84 times during the first 6 weeks postpartum compared with nonpregnant, nonpostpartum, reproductive-aged women.2,3 However, very few studies have reported the incidence by week postpartum, instead reporting an aggregate risk. The normalization of different parts of the coagulation system occurs at varying times during the postpartum period with some factors returning to normal immediately after delivery and others not until 6–12 weeks postpartum.4,5 Little is known about how this varying return to normal function may affect rates of venous thromboembolism over time. In addition, previous reports have not examined how risk factors for venous thromboembolism might affect rates of venous thromboembolism at different times postpartum. More precise information about weekly incidence and risk factors might allow for more informed decisions regarding postpartum prophylaxis or initiation of certain medications that may pose additional risk for venous thromboembolism such as estrogen-containing contraceptives.
The objectives of this study were to calculate incidence of venous thromboembolism among postpartum women by week postpartum through week 12. In addition, we sought to characterize associated factors for the development of postpartum venous thromboembolism and how these factors affect risk of venous thromboembolism at different times postpartum.
MATERIALS AND METHODS
Data for this study were obtained from the Truven Health MarketScan Commercial and Multi-State Medicaid databases, which contain individual-level, deidentified health care claims information from employers, health plans, hospitals, and Medicaid programs across the United States.6 Both databases provided information on health care services used in inpatient and outpatient settings as well as linked information on filled outpatient prescription drug claims. The quality of MarketScan data is enhanced through extensive validity and reasonableness checks.6,7 Because the data are deidentified, the Centers for Disease Control and Prevention deemed this study to be exempt from review by an institutional review board.
We restricted our analysis to women aged 15–44 years with information on pharmaceutical claims who had a delivery hospitalization, defined as any inpatient hospital admission record with an International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) code for a delivery-related diagnosis or procedure or a Current Procedural Terminology code or a diagnosis-related group code for vaginal or cesarean delivery (Table 1). Women were excluded if the diagnosis included codes or procedures for hydatidiform mole, ectopic pregnancy, other abnormal products of conception, or abortion (Table 1). Delivery hospitalizations were identified among women aged 15–44 years during the years 2005–2011; for women with more than one delivery, only the first delivery was included. The date of first report of delivery diagnoses or procedures was considered to be day 0 and women were followed through 12 weeks after this date, because most hemostatic factors normalize by 12 weeks.5 Women with deliveries occurring after September 30, 2011, were excluded to allow for 12 weeks of follow-up for all women.
Instances of deep venous thrombosis (DVT) and pulmonary embolism (PE) were identified using ICD-9-CM codes (Table 1). A unique identifier was assigned to each individual and therefore, even if venous thromboembolism was coded multiple times, each venous thromboembolism was counted only once. All inpatient DVT or PE diagnoses were considered valid because the diagnostic codes are recorded at hospital discharge and would therefore only be reported for confirmed diagnoses and not for suspected diagnoses before evaluation. Venous thromboembolism codes found solely in outpatient claims data may include codes for both suspected and confirmed diagnoses; therefore, outpatient venous thromboembolism codes were considered valid only if a filled anticoagulant prescription claim was noted within 90 days of the venous thromboembolism diagnosis, because it is unlikely that individuals would be prescribed anticoagulation without a confirmed diagnosis.8 We excluded women who were prescribed anticoagulants during the 6 months preceding the delivery admission, because these women would have been at very low risk for development of new venous thromboembolism, to calculate a true incidence proportion among the population at risk for venous thromboembolism. To assess the timing of the venous thromboembolism diagnosis, the 12-week postpartum period was categorized into 1-week intervals starting with the date of delivery. Venous thromboembolism events were assigned to a postpartum week according to the date of the first reported postpartum venous thromboembolism diagnosis.
Certain chronic medical conditions previously reported to be associated with increased postpartum venous thromboembolism risk9 were identified using ICD-9-CM codes noted during the 12 months before the delivery and included chronic diabetes, chronic hypertension, obesity, sickle cell disease, and lupus erythematosus (Table 1). Certain pregnancy-related conditions associated with increased postpartum venous thromboembolism risk9–11 were identified using ICD-9-CM or Current Procedural Terminology codes reported any time within the 40 weeks before the delivery admission or during the 12-week postpartum interval (Table 1). These conditions included gestational diabetes, preeclampsia, eclampsia, multiple gestation, blood transfusion, anemia, hysterectomy, antepartum hemorrhage, postpartum hemorrhage, postpartum infection, and other severe complications (ie, heart failure, pulmonary edema, adult respiratory distress syndrome, disseminated intravascular coagulation, shock, sepsis, cerebrovascular disorders, ventilation, peripartum cardiomyopathy, cardiovascular conditions [including myocardial infarction]). Smoking in the 40 weeks before the delivery admission or postpartum period was also assessed.9,10 To improve the accuracy of all outpatient diagnoses for chronic medical conditions, we considered an outpatient diagnosis valid only if the diagnosis appeared two or more times and was reported at least 30 days apart. This approach can increase the specificity of outpatient diagnoses and potentially avoid overdiagnosis of women evaluated and ruled out for the disease.12 Outpatient procedures (hysterectomy, transfusion, or ventilation) were not subject to this restriction because a woman would not likely have more than one occurrence of these procedures. We considered all inpatient diagnoses to be valid, even if they only appeared once, because the diagnostic codes are assigned at the time of hospital discharge and therefore likely to be confirmed diagnoses.
The incidence proportion of venous thromboembolism among postpartum women was calculated by dividing the number of postpartum diagnoses of venous thromboembolism by the total number of deliveries that contributed to the time period. For each characteristic and risk factor, logistic regression was used to calculate odds ratios and 95% confidence intervals for venous thromboembolism during 1–12 weeks postpartum compared with no venous thromboembolism. Models were constructed separately for women with private insurance and Medicaid because the commercial databases include women across the United States, whereas the Medicaid database includes women in 10 unidentified states. Furthermore, race and ethnicity is only available in the Medicaid database but not available in the commercial data set. To examine whether different risk factors have an effect on venous thromboembolism at different time periods postpartum, we divided the postpartum period into early (1–3 weeks postpartum) and late (4–12 weeks postpartum) periods. We chose these cutoffs because weeks 1–3 demonstrated the highest venous thromboembolism incidence and weeks 4–12 demonstrated lower and stable incidence rates. Because week 1 may have included some antepartum venous thromboembolisms not diagnosed until after delivery,11 we also performed the analyses defining early venous thromboembolisms as week 1 postpartum and late venous thromboembolisms as weeks 2–12 postpartum, and results did not substantially change. We used multinomial logistic regression to evaluate the factors associated with venous thromboembolism during the two time periods, compared with those without venous thromboembolism, using the following response categories: venous thromboembolism in 1–3 weeks postpartum, venous thromboembolism in 4–12 weeks postpartum, and no venous thromboembolism in the postpartum period. We did not include lupus, sickle cell disease, blood transfusion, and hysterectomy in the models as a result of small numbers of women with these diagnoses. SAS 9.3 was used for all analyses. All analyses were independently conducted by two statisticians.
During 2005–2011, there were a total of 2,456 venous thromboembolisms identified during weeks 1–12 postpartum among women with private insurance (1,550 DVT and 906 PE±DVT) and 1,713 venous thromboembolisms among women with Medicaid (1,039 DVT and 674 PE±DVT). Overall, the incidence proportions were 16 per 10,000 deliveries among women with private insurance and 17 per 10,000 deliveries among women with Medicaid. The incidence proportion was highest during the first week postpartum (9.2 per 10,000 deliveries among women with private insurance; 9.4 per 10,000 deliveries among women with Medicaid) (Table 2; Fig. 1). By week 4, the incidence proportions decreased to 0.9 per 10,000 deliveries among women with private insurance and one per 10,000 deliveries among women with Medicaid. Proportions continued to decline through week 7 and remained low and steady through the 12th week at approximately 0.1–0.2 per 10,000 deliveries. When the venous thromboembolism cases were separated into cases of DVT and PE±DVT, the incidence proportions of both DVT and PE were similarly highest during the first week postpartum and decreased steadily through the 12th week (data not shown).
We examined the association between selected characteristics, conditions, and delivery complications and the occurrence of postpartum venous thromboembolism (Table 3). Older age, obesity, and smoking were associated with an increased risk of venous thromboembolism after adjusting for all other covariates in the table. Black race was associated with an increased risk of venous thromboembolism in the Medicaid database; race or ethnicity information was not available in the private insurance database. Chronic diabetes, chronic hypertension, gestational diabetes, or multiple births were not associated with risk of venous thromboembolism. Obstetric procedures or complications such as cesarean delivery, preeclampsia, anemia, antepartum hemorrhage, postpartum hemorrhage, postpartum infection, or other severe complications were associated with an increased risk of venous thromboembolism.
Risk factors for early venous thromboembolism (1–3 weeks postpartum) and late venous thromboembolism (4–12 weeks postpartum) compared with no venous thromboembolism are shown in Table 4. After adjusting for all other covariates, increased risk with increasing age was seen only among women with early venous thromboembolism and not among women with late venous thromboembolism. Obesity was a risk factor for both early and late venous thromboembolisms. In contrast, smoking was significantly associated with an increased risk for late but not early venous thromboembolism. Cesarean delivery, preeclampsia, anemia, antepartum hemorrhage, postpartum hemorrhage, postpartum infection, and other severe complications were found to confer increased risk for both early and late postpartum venous thromboembolisms. For late postpartum venous thromboembolisms, these obstetric factors increased risk by 1.4- to 16.3-fold among those with private insurance and 1.3- to 18.9-fold among those with Medicaid. Results were largely consistent between women with private insurance and women with Medicaid.
This analysis examined weekly incidence of postpartum venous thromboembolism in a nationwide database. Very few studies have examined weekly incidence of venous thromboembolism postpartum, reporting a range of one to 357 per 10,000 deliveries during postpartum week 1.11,13 Our incidence proportions through 12 weeks postpartum are lower than those reported by Heit and Sultan13,14 but higher than those reported by Jacobsen,3,11 possibly as a result of differences in timeframes, countries, and methods of ascertainment. Our results are similar to other studies that found the highest rates during the first few weeks postpartum.11,13,14 Our overall incidence proportions of 16 (among private insurance) and 17 (among Medicaid) per 10,000 women are consistent with other published reports.3,14
Our results also agree with several other studies that have identified women with pregnancy complications to be at higher risk for postpartum venous thromboembolism.9,11,15 The contribution of pregnancy-related risk factors to venous thromboembolism formation is likely mediated through their effects on coagulability and inflammation.16 In addition, acute perturbations such as hemorrhage or infection may compound the changes associated with the postpartum hemostatic system to further increase venous thromboembolism formation.
Our analysis showed that obstetric factors and complications conferred an increased risk for both early and late postpartum venous thromboembolisms and that the elevated risk for venous thromboembolism among women with obstetric complications may extend through at least 12 weeks postpartum. This may have implications for length of postpartum venous thromboembolism prophylaxis and other preventive measures17 as well as other postpartum decisions such as choice of contraceptive method. By 12 weeks, much of this remaining increase in risk may be attributable to the obstetric complication or other risk factor rather than postpartum status.
This analysis has several strengths. The large size of the databases allows for identification of relatively rare events such as venous thromboembolisms. Linking several databases including inpatient, outpatient, and pharmacy information ensures a more complete capture of venous thromboembolism diagnoses and more complete follow-up of individuals. Examination of rates and characteristics among women with both private insurance and Medicaid and the similarity of findings between the two populations increase the generalizability of our results to a broader population.
Some limitations should be considered before generalizing these results. First, because venous thromboembolism diagnoses were not validated using medical records, it is likely that our findings were affected by some degree of misclassification. For venous thromboembolism claims reported only in the outpatient setting, we included only those with a filled anticoagulant coagulant prescription within 90 days of the venous thromboembolism diagnosis to decrease overestimation of venous thromboembolism. However, it is possible that we included some instances of women who were started on anticoagulation and later discontinued if venous thromboembolism diagnosis was ruled out. It is also possible that we missed some instances of venous thromboembolism if they were diagnosed in other settings and not associated with an insurance claim. Second, the overall validity of diagnostic codes in administrative databases can be variable; thus, our estimates of the association between venous thromboembolism and the risk factors examined in the present study may understate or overstate the true association. In one study of hospital discharge data, the validity of diagnostic codes for certain chronic conditions such as obesity, diabetes, and hypertension was variable; however, the validity of pregnancy-related comorbidities such as cesarean delivery, multiple gestation, hemorrhage, infection, and preeclampsia was high.18 Third, we were not able to look at certain factors associated with increased or decreased risk for venous thromboembolism. We were unable to examine prior venous thromboembolism or thrombophilia, because these women would likely have been anticoagulated during pregnancy and therefore excluded from our cohort. In addition, we were unable to examine the effect of peripartum pharmacologic thromboprophylaxis, because we did not have information on inpatient medications, and nonpharmacologic thromboprophylaxis, because this information is not captured by diagnostic codes. We were also unable to examine the effect of race or ethnicity among women with private insurance because this information was not available. Fourth, by examining only the first delivery hospitalization, we may have underestimated venous thromboembolism prevalence.
These results add to our understanding of the timing of and risk factors for postpartum venous thromboembolism. The risk of postpartum venous thromboembolism is highest in the immediate postpartum period, and older women and women with obstetric complications are at higher risk. In addition, obstetric factors and complications confer an elevated risk during both the early and later postpartum period. These results underscore the need for clinicians to have heightened awareness of the elevated and prolonged risk of postpartum venous thromboembolism, particularly among women with obstetric complications.
1. James AH. Pregnancy-associated thrombosis. Hematology Am Soc Hematol Educ Program 2009:277–85.
2. Thornton P, Douglas J. Coagulation in pregnancy. Best Pract Res Clin Obstet Gynaecol 2010;24:339–52.
3. Jackson E, Curtis K, Gaffield M. Risk of venous thromboembolism during the postpartum period: a systematic review. Obstet Gynecol 2011;117:691–703.
4. Epiney M, Boehlen F, Boulvain M, Reber G, Antonelli E, Morales M, et al.. D-dimer levels during delivery and the postpartum. J Thromb Haemost 2005;3:268–71.
5. Hellgren M. Hemostasis during normal pregnancy and puerperium. Semin Thromb Hemost 2003;29:125–30.
6. Hansen LG, Chan S. Health Research Data for the Real World: The MarketScan Databases. Bethesda (MD): Truven Health Analytics; 2012.
7. Truven Health Analytics. Multi-state Medicaid database user guide, data year 2011 edition. Available at: http://marketscan.truvenhealth.com/marketscanportal/
. Retrieved March 13, 2014.
8. Boulet SL, Grosse SD, Hooper WC, Beckman MG, Atrash HK. Prevalence of venous thromboembolism among privately insured US adults. Arch Intern Med 2010;170:1774–5.
9. 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.
10. Danilenko-Dixon DR, Heit JA, Silverstein MD, Yawn BP, Petterson TM, Lohse CM, et al.. Risk factors for deep vein thrombosis and pulmonary embolism during pregnancy or post partum: a population-based, case-control study. Am J Obstet Gynecol 2001;184:104–10.
11. Jacobsen AF, Skjeldestad FE, Sandset PM. Incidence and risk patterns of venous thromboembolism in pregnancy and puerperium—a register-based case-control study. Am J Obstet Gynecol 2008;198:233.e1–7.
12. Hebert PL, Geiss LS, Tierney EF, Engelgau MM, Yawn BP, McBean AM. Identifying persons with diabetes using Medicare claims data. Am J Med Qual 1999;14:270–7.
13. 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.
14. Sultan AA, West J, Tata LJ, Fleming KM, Nelson-Piercy C, Grainge MJ. Risk of first venous thromboembolism in and around pregnancy: a population-based cohort study. Br J Haematol 2012;156:366–73.
15. Sultan AA, Tata LJ, West J, Fiaschi L, Fleming KM, Nelson-Piercy C, et al.. Risk factors for first venous thromboembolism around pregnancy: a population-based cohort study from the United Kingdom. Blood 2013;121:3953–61.
16. Ageno W, Becattini C, Brighton T, Selby R, Kamphuisen PW. Cardiovascular risk factors and venous thromboembolism: a meta-analysis. Circulation 2008;117:93–102.
17. 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(suppl):844S–886S.
18. Yasmeen S, Romano PS, Schembri ME, Keyzer JM, Gilbert WM. Accuracy of obstetric diagnoses and procedures in hospital discharge data. Am J Obstet Gynecol 2006;194:992–1001.