Association Between Newborn Birth Weight and the Risk of Postpartum Maternal Venous ThromboembolismCLINICAL PERSPECTIVE
A Population-Based Case–Control Study
Background—Postpartum venous thromboembolism (VTE) is a potentially fatal and preventable event leading to substantial short- and long-term morbidity. We sought to evaluate whether the delivery of term newborns of low or high birth weight was associated with greater risks of VTE.
Methods and Results—In a population-based case–control study conducted in Washington State from 1987 through 2011, cases of hospitalized VTE within 3 months of delivery were identified by using selected International Classification of Diseases, Ninth Revision, Clinical Modification codes. Controls were randomly selected postpartum women without VTE, matched on birth year. Birth weight and other maternal and pregnancy characteristics were extracted from birth certificate data. Among term live singleton deliveries, we compared the risk of VTE for mothers of newborns of low and high birth weights (<2500 g and >4000 g, respectively) versus mothers of newborns of normal birth weight (2500–4000 g). Logistic regression models were adjusted for maternal age, race, education, body mass index, parity, delivery methods, gestational length, smoking, gestational diabetes mellitus, and preeclampsia. Patients with VTE (n=547) were older, had a higher body mass index, and experienced more pregnancy-related complications than controls (n=9482). In comparison with mothers of newborns with normal birth weight, mothers of newborns with low birth weight had a 3-fold increased risk of VTE, which persisted after multivariable adjustment (odds ratio, 2.98; 95% confidence interval, 1.80–4.93). Mothers of newborns with high birth weight had only a slightly increased risk of VTE, which was attenuated after multivariable adjustment (odds ratio, 1.26; 95% confidence interval, 0.99–1.61).
Conclusions—The delivery of a newborn with low birth weight is associated with a 3-fold increased risk of maternal postpartum VTE. This should be considered when assessing VTE risk at delivery.
The hypercoagulable state and altered venous flow during pregnancy substantially increase the incidence of venous thromboembolism (VTE).1,2 The greatest risk lies in the postpartum period, with estimated absolute risks of 0.5 to 2/1000 pregnancies,3–5 and thromboprophylaxis at the time of delivery is expected to reduce the occurrence of pulmonary embolism (PE) and deep vein thrombosis (DVT). Several studies have already identified risk factors for pregnancy-related VTE, including maternal characteristics (advanced age, greater body mass index [BMI], black race, thrombophilia)5–7 and obstetric characteristics (cesarean delivery, preterm delivery, preeclampsia, postpartum hemorrhage or infection).5,8,9 They help care providers to target the use of thromboprophylaxis to women at risk to maximize its benefit.
Clinical Perspective on p 1476
Few studies have evaluated newborn birth weight (BW) as a risk factor.8,10,11 Although an association of intrauterine growth restriction (IUGR) with postpartum VTE has been suggested, whether low BW is a risk factor independently of preterm deliveries remains largely unknown. Furthermore, the association of deliveries of large (macrosomic) newborns at term with the risk of postpartum VTE has not been reported. An enhanced understanding of these associations may improve the risk stratification of VTE in the postpartum period.
For this population-based, case–control study of women with term, singleton deliveries, we hypothesized greater risks of VTE in mothers of newborns of low BW, perhaps through a thrombotic predisposition also responsible for some placental vascular insufficiency, and in mothers of newborns of high BW, through possible greater venous compression and trauma at delivery, in comparison with mothers of newborns of normal BW.
This retrospective case–control study included all births in Washington State from January 1, 1987 to December 31, 2011. All births were identified from birth certificates, resulting in a well-defined population-based study. Birth certificate data were linked to the Comprehensive Hospital Abstract Reporting System registry that reports International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) discharge codes from all nonfederal hospitals in Washington State. This study was approved by the Washington State Institutional Review Board (D-022713-H).
Case and Control Selection
Cases were defined as women who had experienced an incident hospitalized DVT or PE at delivery or within the 3 months following delivery. We identified 1052 cases with at least one ICD-9-CM code for DVT or PE at a delivery hospitalization or during the postpartum period (415.11/19, 673.22/24/82 for PE or 451.1x/2x/81/9, 453.1/2/4x/8x/9, 671.42/44/92/94 for DVT) and without a previous ICD-9-CM code for VTE (since 1987). We subsequently excluded 82 cases with codes 673.82, 671.92/4, and 453.8x (after 2009) to restrict our cases to the ICD-9-CM codes with the greatest documented validity (or positive predictive value [PPV]).7,12,13 To further maximize the PPV of DVT ICD-9-CM codes, we excluded 255 cases who had a DVT ICD-9-CM code in a secondary position (not the first position for rehospitalization after delivery or not the first 3 positions for the delivery hospitalization, when first codes relate to the delivery itself) or whose hospitalization did not last at least 2 days.12,13 With this algorithm, we conducted a small validation study of 28 subjects within Group Health Cooperative, a large integrated healthcare organization in western Washington State, which yielded PPVs of 90% for PE codes 673.22/24 (95% confidence interval [CI], 67%–99%) and of 78% for DVT codes 671.42/44 in primary positions (95% CI, 40%–97%).7
Controls were defined as women without VTE within the 3 months following delivery. We randomly sampled 10 520 women who had delivered in Washington State without ICD-9-CM codes for DVT or PE since 1987 until 3 months after a delivery, frequency-matched to cases on birth year with a 10:1 ratio.
Further exclusions were applied to women with a previous history of VTE documented by a ICD-9-CM code V12.51/52/55 (14 cases and 9 controls) and women with missing data for BW or gestational age (16 cases and 147 controls). Finally, to restrict to live singleton term newborns, we excluded 138 cases and 882 controls with a preterm delivery (<37 weeks of gestational age), fetal deaths, or multiple pregnancies (twins or triplets), leaving 547 cases and 9482 controls for the analysis.
Exposure and Other Variables
We extracted data for the primary exposure (BW) and other maternal, obstetric, and newborn characteristics from birth certificates. BW was categorized into low (<2500 g), normal (2500–4000 g), and large BW (>4000 g), following the Centers for Disease Control and Prevention categories.14 Birth certificate data on maternal postpartum infections and hemorrhages were augmented with appropriate ICD-9-CM codes. Gestational age was estimated from the age reported by the certifier on the birth certificate or, if unavailable, calculated from the reported date of the last menstrual period.
Missingness was substantial for maternal BMI (58.8%), because maternal weight and height were not recorded before 1992 and 2003, respectively. Apart from maternal education (18.8% missing), all other covariates had <5% missing data. All missing data were multiply imputed by using multivariable imputation by chained equations (20 imputations). Therefore, presented multiadjusted model estimates, confidence intervals, and P values were based on multiple imputations.
Relative risks were estimated by odds ratios from multivariable logistic regression models. In primary analyses, we compared the risk of VTE between categories of newborn BW with 3 models. The first model was adjusted for the design variable (birth year), maternal age (categorical variable: <18, 18–29, 30–34, 35–39, and >40 years), and race/ethnicity (white, black, Asian, Hispanic, other). The second model was further adjusted for the following plausible confounding variables, selected before analyses: parity (first delivery versus second or greater delivery), smoking status at the start of pregnancy, gestational hypertension and preeclampsia, preexisting and gestational diabetes mellitus, delivery methods (vaginal delivery, assisted vaginal delivery, and cesarean delivery), gestational age (continuous), and maternal education status (high school diploma, less than high school, greater than high school). The third and most complete model further adjusted for maternal BMI at the start of pregnancy (<18.5 kg/m2, 18.5–25 kg/m2, 25.1–30 kg/m2, >30 kg/m2),
Secondary analyses evaluated risk separately for PE (including PE and concomitant DVT) and DVT-alone outcomes. We conducted analyses stratified by delivery method (cesarean delivery versus unassisted or assisted vaginal delivery), with interaction P values from global Wald tests of multiplicative interaction terms of BW and delivery method. We also restricted analyses to white women to reduce heterogeneity by race/ethnicity.
In sensitivity analyses, we restricted the sample to women from the 10 most recent years of the study period (2002–2011) to assess the influence of possible secular trends (n=275 cases, 4740 controls). To assess the potential influence of BMI, for which more than half of the women had missing data, on the observed associations, we also conducted a multivariable regression analysis with and without adjustment for BMI on a restricted sample of women with complete information for BMI (n=237 cases, 3894 controls).
In an exploratory analysis, we graphically assessed the multiadjusted association of BW as a continuous variable with the risk of postpartum VTE by using a nonimputed model of restricted cubic splines with 5 knots between 1500 and 5000 g.
Statistical significance was determined by using a 2-sided α-level of 0.05. Stata 11.0 (Stata Corporation, College Station, TX) was used for all analyses.
We identified 547 women who experienced a VTE in the 3 months following delivery (326 DVT and 221 PE with or without DVT) and 9482 women who did not experience a VTE, all of whom had given birth to live singletons at term between 1987 and 2011. Most patients were rehospitalized for VTE after discharge from the delivery hospitalization (63%), a median of 13 days after delivery.
The prevalence of known risk factors for postpartum VTE was higher in cases than controls (Table 1). In particular, cases were more likely to be of black race, to be obese, to have gestational diabetes mellitus, pre-eclampsia, postpartum hemorrhage or maternal transfusion, postpartum infection, and delivery via cesarean delivery.
Among controls, mothers of newborns with low BW (1.5%) had a higher prevalence of low BMI (<18.5 kg/m2, 14% versus 3.9%), smoking (31.3% versus 14.1%), preeclampsia (10.9% versus 4.6%), and placental abruption (2.8% versus 0.5%), and a greater proportion were of Hispanic ethnicity (21.1% versus 15.0%) than mothers with normal BW newborns. Mothers of newborns with high BW (12.9%) had a higher prevalence of obesity (36.1% versus 21.3%, >30 kg/m2), diabetes mellitus (6.4% versus 3.5%), cesarean delivery (29.3% versus 20.4%), and non-Hispanic white race (79% versus 70.0%) than mothers of normal BW newborns. Maternal age was similar across BW categories. Reported lacerations of third or fourth degree were uncommon, from 0% in controls with newborns with low BW to 1.6% in controls with newborns of large BW.
In comparison with mothers of normal BW newborns, mothers of low BW newborns (<2500 g) were at an ≈3-fold greater risk of postpartum VTE (odds ratio [OR], 2.96; 95% CI, 1.86–4.78), following adjustment for age and race (Table 2). This estimate remained similar to additional adjustment for a priori planned covariates, including with or without BMI (adjusted OR, 2.98; 95% CI, 1.80–4.93 with BMI). We estimated a 60% greater risk of postpartum VTE for mothers of high BW newborns (>4000 g) in the age- and race-adjusted model (adjusted OR, 1.56; 95%, CI 1.24–1.96). This risk was attenuated to a 30% greater risk, with borderline statistical significance, after adjusting for confounders (adjusted OR, 1.26; 95% CI, 0.99–1.61). Cesarean delivery appeared to be the strongest confounder of the latter association.
In secondary analyses, the delivery of a low BW newborn was associated with similar estimates for the risk of postpartum PE and DVT alone (adjusted OR, 3.08; 95% CI, 1.55–6.15 and adjusted OR, 2.84; 95% CI, 1.51–5.37, respectively, Table 3). Delivery of a large BW newborn, in comparison with the delivery of a normal weight newborn, was associated with a greater risk of postpartum DVT (OR, 1.40; 95% CI, 1.04–1.87) but not with PE (OR, 1.07; 95% CI, 0.73–1.57).
We found no statistical evidence of a difference in the BW-VTE association by delivery method (cesarean delivery versus vaginal delivery, Wald P value for interaction=0.12). Delivery of a low BW newborn was associated with a 4-fold risk of postpartum VTE among women who delivered vaginally (OR, 3.92; 95% CI, 2.12–7.27) and a 2-fold risk among women who delivered via cesarean delivery (OR, 2.07; 95% CI, 0.96–4.48). Delivery of a high BW newborn was associated with a 1.5-fold risk of postpartum VTE among women who delivered vaginally (OR, 1.56; 95% CI, 1.12–2.16) and we found no evidence of an association between delivery of a high BW newborn and postpartum VTE among women who delivered via cesarean delivery (OR, 1.05; 95% CI, 0.75–1.47). When restricted to white women, multivariable adjusted ORs associated with the risk of postpartum VTE were 3.85 (95% CI, 2.10–7.07) for mothers of low BW newborns and 1.40 (95% CI, 1.07–1.84) for mothers of large BW newborns.
In sensitivity analyses, when restricting to the study period from 2002 to 2011, we observed multivariable adjusted ORs of 2.69 (95% CI, 1.36–5.30) and 1.25 (95% CI, 0.88–1.76) for mothers of low and large BW newborns, respectively. Multivariable regression models of women with complete BMI data yielded similar ORs with and without adjustment for BMI.
The restricted cubic spline analysis of BW as a continuous variable is shown in the Figure.
In this population-based, case–control study of VTE in Washington State, we found that the risk of maternal VTE following delivery of a low BW (<2500 g) newborn was 3-fold greater than following the delivery of a normal BW newborn. Importantly, this increased risk appeared to be independent of other known risk factors of postpartum VTE. Contrary to our hypothesis, we found only weak evidence of a greater risk of VTE associated with delivery of a large BW newborn (>4000 g).
Previous studies have evaluated the risk of postpartum VTE in women of newborns with IUGR, as defined by low BW for gestational age. Danilenko-Dixon et al15 found no association between BW as a linear variable and the risk of postpartum VTE. However, the power of this case–control study was limited (56 cases), a nonlinear association was not tested, and mothers of preterm newborns were not excluded. In a Norwegian hospital-based, case–control study by Jacobsen et al,10 mothers of newborns with IUGR were at 3.8-fold risk of postpartum VTE. Weaker associations were reported in a Canadian retrospective cohort by using administrative data (International Classification of Diseases, Ninth and Tenth Revision codes) for both the outcome and the identification of IUGR (relative risk of 1.4),11 and in an Australian retrospective cohort of a similar design to our study (relative risk of 1.2).8 Nondifferential misclassification of variables, in particular, of ICD-9-CM–coded IUGR, and the absence of restriction to singleton newborns may account for these weaker signals. Our study, with less potential for misclassification, suggests that IUGR, simply measured by a weight of <2500 g at term, is a strong and independent risk factor for postpartum VTE.
Biological explanations for this observed association exist, of which placental insufficiency is most plausible. Previous knowledge indicates that ischemic placental disease may be linked to both low BW and an increased venous thrombotic risk.3,16 Other conditions of ischemic placental disease, preeclampsia and placental abruption, are also consistently documented risk factors for maternal postpartum VTE.3,5 Furthermore, histological placental examinations of women with growth restriction provide evidence of both maternal and fetal vascular obstruction.17 Therefore, we could postulate the presence of thrombophilic traits leading to both ischemic placental disease and an increased venous thrombotic risk. Since the most common genetic thrombophilia variants, Factor V Leiden and Factor II G20210A polymorphisms, do not lead to growth restriction,18 unidentified genetic variants or acquired factors, such as antiphospholipid antibodies, may play a role. Antiphospholipid syndrome is linked to both VTE risk and placental insufficiency, leading to a preterm delivery according to the current diagnostic criteria.19 We were unable to assess the role of antiphospholipid antibodies as a mediating factor in our sample of term births, because a specific ICD-9-CM code does not exist. Furthermore, we observed a very low prevalence of ICD-9-CM codes for systemic lupus erythematosus (0.1% among controls), an important cause of secondary antiphospholipid syndrome, suggesting a minor mediating role or a partial capture of this diagnosis.
Mothers of macrosomic newborns were not at a substantially increased risk of VTE in our analysis. Because previous studies had not restricted their sample to term deliveries, the interpretation of the risk estimates for mothers of large-for-gestational-age newborns is difficult.8,11 Our secondary analyses may suggest that the slightly increased risk of postpartum VTE is restricted to DVT. This finding, if true, could result from a greater compression of iliac vessels with greater newborn BW, leading to DVT preferentially and perhaps a lower potential for embolism. Until our findings are further evaluated in independent samples, macrosomia should only be viewed as a possibly minor risk factor for postpartum VTE.
Clinical implications of our findings lie in a more precise identification of women at risk for postpartum VTE, thus with a greater potential benefit from thromboprophylaxis. Experts’ opinion from the American College of Chest Physicians guidelines advise to use prophylaxis in women with an absolute risk of postpartum VTE of >3 per 100 pregnancies,20 but a previous decision analysis model suggested benefit at much lower thresholds (0.2%–0.3%).21 Assuming a baseline VTE risk of 0.1%, delivering a macrosomic newborn may be considered a minor risk factor (relative risk of 1.3), not large enough by itself to indicate thromboprophylaxis. In mothers delivering a newborn of low BW, however, a 3- to 4-fold increase of the baseline VTE risk may be substantial enough to warrant the use of thromboprophylaxis. Such conclusions remain speculative, because many areas of uncertainty exist, such as the benefit and bleeding risk associated with low-molecular-weight heparin and its cost-effectiveness at delivery.
The validity of our results greatly depends on the validity of measurement of both BW and VTE, which were not evaluated in our study but have been assessed in previous validation efforts, using medical records review as the gold standard. Among 11 US health plans between 2001 and 2007, the identification of newborns of low BW by using birth certificates had excellent PPV (n=209; PPV, 99%; 95% CI, 96–100), similar to numbers reported by the Centers for Disease Control and Prevention in 2 US states in 2009 to 2011.22 Previously, another validation study performed in 1989 in North Carolina had also reported an excellent agreement (100%) for BW.23 For VTE, published PPVs for pregnancy-related (673.22/24) and nonpregnancy-related (415.11/19) ICD-9-CM codes of PE are strong, from 79% through 100%.12,13 Because the validity of DVT ICD-9-CM codes is lower,12,13 only specific codes in primary positions with an at least 2-day hospitalization were considered eligible as cases in our study. Even though most DVT codes were “deep phlebothrombosis, postpartum” 671.42/44, with possible misdiagnoses from septic pelvic thrombosis, we found a PPV of 78% in our limited validation substudy. Nevertheless, possible misclassification of DVT highlights the importance of our secondary analyses stratifying outcomes into PE and DVT, where results were substantially similar for mothers of newborns with low BW.
The strengths of our study include its population-based design, with consequent good generalizability, the high validity of BW measurements, and the large sample size. We also acknowledge several limitations: the nonadjudication of VTEs that were limited to hospitalized events, the possibility of residual confounding, the lack of data on thrombophilia and the use of thromboprophylaxis, and the incomplete data on BMI, leading to its imputation and not allowing the stratification of results by maternal BMI.
In conclusion, our findings suggest that mothers delivering term newborns of low BW have a 3-fold increased risk of postpartum VTE, and this risk should be considered when prescribing thromboprophylaxis in women after delivery. For women delivering term newborns with high BW, the relative risks were low and less suggestive of a clinically relevant association.
We thank Bill O’Brien for his technical assistance in preparing the data for this analysis.
Sources of Funding
Dr Blondon was supported by a fellowship for advanced researchers from the Swiss National Science Foundation through the Swiss Foundation for Grants in Biology and Medicine. This sponsor had no role in the development of the research or of the manuscript.
- Received August 12, 2014.
- Accepted February 13, 2015.
- © 2015 American Heart Association, Inc.
- Sultan AA,
- Tata LJ,
- West J,
- Fiaschi L,
- Fleming KM,
- Nelson-Piercy C,
- Grainge MJ.
- White RH,
- Garcia M,
- Sadeghi B,
- Tancredi DJ,
- Zrelak P,
- Cuny J,
- Sama P,
- Gammon H,
- Schmaltz S,
- Romano PS.
- Dalenius K,
- Borland E,
- Smith B,
- Polhamus B,
- Brummer-Strawn L.
- Danilenko-Dixon DR,
- Heit JA,
- Silverstein MD,
- Yawn BP,
- Petterson TM,
- Lohse CM,
- Melton LJ 3rd.
- Rodger MA,
- Walker MC,
- Smith GN,
- Wells PS,
- Ramsay T,
- Langlois NJ,
- Carson N,
- Carrier M,
- Rennicks White R,
- Shachkina S,
- Wen SW.
- Bates SM,
- Greer IA,
- Middeldorp S,
- Veenstra DL,
- Prabulos AM,
- Vandvik PO
- Andrade SE,
- Scott PE,
- Davis RL,
- Li DK,
- Getahun D,
- Cheetham TC,
- Raebel MA,
- Toh S,
- Dublin S,
- Pawloski PA,
- Hammad TA,
- Beaton SJ,
- Smith DH,
- Dashevsky I,
- Haffenreffer K,
- Cooper WO.
This population-based observational study aims to determine the associations between the birth weights of term newborns and the risk of postpartum venous thromboembolism in their mothers, after accounting for other thrombotic risk factors. In comparison with mothers of newborns with normal birth weights, mothers of newborns with low birth weights (<2500 g) were at greater thrombotic risk (3-fold). Only weak evidence supported a slightly increased risk (1.3-fold) for mothers of newborns with large birth weights (>4000 g). These results suggest that birth weight should be considered when determining the benefit of thromboprophylaxis after term delivery, on top of already recognized risk factors such as obesity, preeclampsia, or the use of cesarean delivery. Clinicians should be more inclined to prescribe heparin or mechanical thromboprophylaxis to mothers of newborns of low birth weights. On the other hand, having delivered a macrosomic newborn should not meaningfully influence the decision to prescribe thromboprophylaxis.