First-trimester bleeding is frequent, occurring in 16–25% of pregnancies,1 and has been shown to be associated with increased risk of adverse pregnancy outcomes, including preterm delivery, prelabor premature rupture of membranes, small-for-gestational-age (SGA) fetal measurements, low birth weight, and fetal death, although evidence is equivocal.2–11 In a 2010 meta-analysis, first-trimester bleeding was associated with intrauterine growth restriction in only four of eight cohort studies with significant heterogeneity among studies.1 Furthermore, fetal growth restriction is typically defined as estimated fetal weight less than the 10th percentile, which does not discriminate between pathologic and constitutional growth restriction, because some fetuses are constitutionally small but normally grown, and low birth weight is more indicative of preterm delivery.7,12,13 Limitations in sample size, recall bias, and potential confounding resulting from pre-existing maternal conditions contribute to these mixed results. Additionally, little is known about the relation between first-trimester bleeding and individual fetal biometrics or the gestational timing of altered growth in association with early bleeding, which have the potential to inform the pathology of decreased estimated fetal weight or birth weight. The objective of the present study was to determine whether the presence and duration of first-trimester bleeding in nonobese, healthy women was associated with differences in fetal growth in a prospective cohort study.
MATERIALS AND METHODS
This study was a secondary analysis of the Eunice Kennedy Shriver National Institute of Child Health and Human Development Fetal Growth Studies–Singletons.14 The study recruited 2,334 nonobese women with healthy lifestyles (eg, nonsmokers, no illicit drug use) and low-risk medical and obstetric histories (eg, no chronic diseases) from 2009 to 2013 at 12 centers across the United States. Details of recruitment, inclusion and exclusion criteria, and data collection have been previously reported.14 Of relevance, baseline maternal characteristics and lifestyle were assessed during the enrollment interview between 10 0/7 weeks 0 days and 13 6/7 weeks of gestation. Women were included if they were between 18 and 40 years of age with prepregnancy body mass index (calculated as weight (kg)/[height (m)]2) 19.0–29.9 and were planning to deliver at one of the participating hospitals. Gestational age was determined based on last menstrual period and confirmed by first-trimester ultrasonography. Approval was received from the institutional review board for each of the participating sites, the data coordinating center, and the Eunice Kennedy Shriver National Institute of Child Health and Human Development; participants provided informed consent.
At enrollment, information on demographics, obstetric and medical histories, and lifestyle and health leading up to and during the first trimester of pregnancy was collected by in-person interview. Information included self-reported duration (cumulative number of days) and severity (mild, moderate, severe) of bleeding since the beginning of pregnancy. In-person interviews were conducted by research nurses at up to five follow-up study visits, and bleeding since the previous visit was assessed using the same metrics. After an enrollment ultrasonogram, women were randomly assigned to one of four ultrasound schedules with five planned, regularly spaced visits. Ultrasonograms were conducted by trained experts using standard operating procedures and identical equipment.15 Fetal ultrasound measurements obtained at each visit included biparietal diameter (calipers placed on the outer edge of the proximal skull edge and inner edge of the distal edge), head circumference, abdominal circumference, humerus length, and femur length. Estimated fetal weight was computed using a Hadlock16 formula and fetal head circumference, abdominal circumference, and femur length; head–abdominal circumference ratio was calculated to assess symmetry of fetal growth. Neonatal and maternal outcomes were assessed by abstraction of medical records. Hypertensive disorders were identified using postpartum diagnoses abstracted from the medical record.
The primary outcome assessed in this secondary analysis was the fetal growth pattern derived from calculated fetal weight and individual fetal biometrics across gestation. Secondary outcomes included mean birth weight and prevalence of SGA neonates defined as less than the 10th percentile for neonatal sex and gestational age at delivery using Duryea birth weight.17
Women were grouped into one of three categories by number of cumulative days of first-trimester (less than 14 weeks of gestation) bleeding: 0, 1, or more than 1 day. These categories were selected to distinguish the effects of prolonged bleeding from a single incidence (1 day) and because number of days provides a more objective measure than self-reported severity. Distribution of self-reported number of days of bleeding is shown in Figure 1. Maternal characteristics were compared among groups using χ2 or one-way analysis of variance for categorical or continuous data, respectively. Fetal biometric measurements and estimated fetal weight were log-transformed to stabilize variances across gestational age and to improve normal approximations for the error structures. For each biometric measurement (biparietal diameter, head circumference, abdominal circumference, humerus length, femur length, and head circumference–abdominal circumference ratio) and estimated fetal weight, fetal growth trajectories were created by fitting a linear mixed model with cubic splines for the fixed and a cubic polynomial for the random effects for each of the bleeding groups.18 Three-knot points (25th, 50th, 75th percentiles) were chosen at gestational ages that evenly divided the distributions. Percentiles were estimated based on assumed normal distribution of the random effects and error structure. Estimated trajectories (10th, 50th, and 90th percentiles) were determined across gestational age for each bleeding group. For estimated fetal weight and biometric measures, we tested for overall differences in the bleeding group trajectories using a likelihood ratio test. When the global test was significant (P<.05), we tested for week-specific differences among groups using Wald tests at each week of gestation. Significant differences were evaluated after adjustment for the following maternal characteristics, selected a priori based on causal diagrams and a review of confounders previously reported in the literature: maternal age, weight, height, parity, and self-identified racial–ethnic group.7–9,19 An additional model added neonatal sex as a covariate in the main analysis as a result of the difference in proportion of female to male neonates among groups. The Multivariate Imputation by Chained Equations software for R was used to impute (number of imputations =10) missing covariates when performing adjusted tests for week-specific differences in fetal growth trajectories.
Several sensitivity analyses were performed. First, the analysis was repeated by self-reported bleeding severity (mild, moderate to severe) compared with no bleeding. Moderate and severe bleeding were grouped as a result of the low frequency of severe bleeding, and association with number of days of bleeding was evaluated using a χ2 test. Next, fetal growth was evaluated among women with bleeding in the first trimester only, first and subsequent trimesters, or subsequent trimesters only compared with no bleeding. The final sensitivity analysis assessed the effects of first-trimester bleeding (stratified by number of days) on fetal growth among women who did not develop any complications and excluded pregnancies that resulted in stillbirth, preterm delivery at less than 37 weeks of gestation, hypertensive diseases, gestational diabetes, or neonatal conditions (eg, aneuploidy, anomalies, death). Poisson regression with a robust estimator of variance for standard error and adjustment for maternal age, height, weight, parity, and racial–ethnic group were used to evaluate the relative risk of delivering an SGA neonate in the 2,307 women included in the primary analysis and in the subset of 1,730 who delivered at term without complications. In these same groups, analysis of covariance was used to evaluate the association between categorical number of days of first-trimester bleeding and birth weight to confirm the differences in estimated fetal weight that were observed among the groups. The analysis of covariance was adjusted for the same covariates as the fetal growth analyses as well as gestational age at delivery to account for the effects of preterm and early term (37–38 weeks of gestation) delivery on birth weight.
All analyses were completed with the use of SAS 9.4 or R 3.3.1.
Of 2,334 women enrolled, 14 were found to be ineligible after enrollment. Three had no first-trimester bleeding data, two were missing information on number of days of bleeding, and eight had no ultrasound data; thus, a total of 2,307 women were included in the main analysis. In total, 410 women (17.8%) reported first-trimester bleeding. Of those, 176 (42.9%) bled for 1 day, and 234 (57.1%) bled for more than 1 day. Distribution of maternal characteristics is presented in Table 1. Bleeding was more likely for slightly older and heavier women and for those born outside the United States. Imputation was used for women missing data on height (n=15; 11 with no bleeding, 3 with 1 day, 1 with more than 1 day) and weight (n=4; all with no bleeding).
Figure 2 presents estimated fetal weight trajectories for the three groups, including 10th, 50th, and 90th percentiles. Estimated fetal weight for pregnancies with more than 1 day of bleeding differed significantly from no bleeding from 14 to 16 weeks of gestation and 35–39 weeks of gestation (global test P<.001). Magnitudes of difference in estimated fetal weight are presented in Table 2. The median estimated fetal weight for women with more than 1 day of bleeding was 2–3 g larger from 14 to 16 weeks of gestation than that of women with no bleeding, but became progressively smaller in the third trimester compared with women with no bleeding with estimated fetal weight 68–107 g smaller from 35 weeks through 39 weeks of gestation. There were no statistically significant differences in estimated fetal weight for 1 day of bleeding compared with no bleeding. The addition of neonatal sex as a covariate resulted in nonsignificant differences in estimated fetal weight from 14 to 16 weeks of gestation; the significant differences from 35 to 39 weeks of gestation remained unchanged, indicating that different proportions of females to males among the bleeding groups did not explain the observed decreases in fetal growth.
Differences in mean birth weight and prevalence of SGA or large-for-gestational-age neonates by number of days of bleeding are presented in Table 3. In addition to differences in estimated fetal weight during the third trimester, there were differences in the rate of SGA births among groups with 148 (8.5%), 9 (5.7%), and 33 (15.7%) SGA neonates delivered in the no bleeding, 1 day, and more than 1 day of bleeding groups, respectively. Relative risk of delivering an SGA neonate for women with more than 1 day of bleeding compared with no bleeding was 2.14 (95% CI 1.51–3.02) in the full cohort and 2.06 (95% CI 1.38–3.08) in the subset of uncomplicated pregnancies. The association between first-trimester bleeding and mean birth weight followed the same pattern as estimated fetal weight with overall mean birth weight 153 g smaller and mean birth weight at term 88 g smaller in the more than 1 day of bleeding group compared with no bleeding.
Figure 3 presents the growth trajectories for individual fetal biometric parameters in the main analysis. Biparietal diameter did not differ significantly among the groups (global P=.354). Head circumference differed significantly only at 14 weeks of gestation with slightly larger (1.4 mm) head circumference for women with more than 1 day of bleeding compared with no bleeding (global P=.019). Abdominal circumference for women with more than 1 day of bleeding was slightly larger than no bleeding at 14 and 15 weeks of gestation and became significantly smaller beginning at 34 weeks of gestation (global P<.001). Femur and humerus length trajectories differed significantly among groups early in the second trimester (14–16 weeks of gestation) only (global P<.001 for both) with slight increases in femur and humerus length for women with more than 1 day of bleeding compared with no bleeding. Head–abdominal circumference ratio differed significantly among the groups with smaller head–abdominal circumference ratio from 14 to 16 weeks of gestation and larger head–abdominal circumference ratio from 35 to 39 weeks of gestation for more than 1 day of bleeding compared with no bleeding (global P=.004).
Two previously excluded women with data on bleeding severity, but who were missing number of days, were included in the severity sensitivity analysis (n=2,309). Of 412 (17.8%) women who reported first-trimester bleeding, 367 (89.1%) described bleeding as mild and 45 (10.9%) described bleeding as moderate or severe. Estimated fetal weight trajectories differed significantly among groups from 34 to 36 weeks of gestation (global P=.003) and are shown in Figure 4A. Estimated fetal weight was lower from 34 to 36 weeks of gestation for women with mild bleeding compared with no bleeding. At 38 weeks of gestation, estimated fetal weight was 3,157 g for no bleeding, 3,092 g for mild bleeding, and 3,078 for moderate or severe bleeding. Bleeding severity was correlated with number of days of bleeding (P=.01, χ2 analysis) with 6.3% of 1 day and 14.1% of more than 1 day bleeding cases reported as moderate or severe.
For the analysis evaluating the effects of whether bleeding continued or resolved after the first trimester, 2,032 women were included after excluding 277 women who were without complete information on bleeding throughout pregnancy. Of women whose data were analyzed, 538 (26.5%) bled at any time during pregnancy. Of those, 276 (51.3%) bled in the first trimester only, 186 (34.6%) in subsequent trimesters only, and 76 (14.1%) in both the first and subsequent trimesters. Differences in estimated fetal weight or biometric parameters among groups did not reach statistical significance in this sensitivity analysis (Fig. 4B).
In the analysis limited to 1,730 women with pregnancies that did not develop complications (eg, gestational diabetes or hypertensive diseases) and delivered at term, the incidence of first-trimester bleeding was 16.5%. The pattern of decreased estimated fetal weight in association with first-trimester bleeding was greater in this group with significant differences beginning at 33 weeks of gestation for more than 1 day of bleeding (global P<.001) and estimated fetal weight 128 g smaller than the no bleeding group at 39 weeks of gestation compared with 106 g smaller in the full cohort (Fig. 4C). Differences in individual fetal biometrics in this cohort followed the same pattern as the full cohort.
In our prospective cohort study of healthy, low-risk, singleton pregnancies, first-trimester bleeding was common. More than 1 day of bleeding was associated with asymmetric, decreased fetal growth beginning in the third trimester and an increased risk of SGA at birth. Our finding that the association between first-trimester bleeding and decreased fetal growth was even stronger in the subset of women with uncomplicated pregnancies who delivered at term suggests that this mechanism of impaired fetal growth acts independently of potentially confounding obstetric complications. One day of bleeding was not associated with significant differences in fetal growth. Bleeding that continued into subsequent trimesters was also not associated with a greater magnitude of decreased growth, which may suggest that bleeding later in pregnancy is the result of a different mechanism.20
Our findings are consistent with previous studies that found an association between first-trimester bleeding and low birth weight (less than 2,500 g) or estimated fetal weight less than the 10th percentile for gestational age.10,21,22 Most previous studies investigated bleeding by severity and found an association between severe bleeding and low birth weight.4,5,7 Conversely, our study found that even mild bleeding was significantly associated with decreased fetal growth. These differences may be the result of lack of power or the subjective nature of reported severity. A novel contribution of our study is the assessment of the effects of bleeding based on duration, which is more objective and therefore potentially more clinically useful. Our finding that duration and severity of bleeding did not fully correlate suggests that the two measures are not interchangeable. The finding that 1 day of bleeding was associated with small, nonstatistically significant decreases in fetal growth, and accounts for nearly half of all incidents of first-trimester bleeding, provides an explanation for the numerous previous studies, which did not stratify by duration and found no significant association between bleeding and fetal growth.4,11,13,23
The increased head–abdominal circumference ratio observed in the more than 1 day of bleeding group supports previous hypotheses that first-trimester bleeding may be an early sign of placental dysfunction.7 Increased levels of hemoglobin and red blood cells were observed in the cord blood at birth for pregnancies with first-trimester bleeding but not bleeding associated with placenta previa.24,25 The authors hypothesized that increased red blood cell levels were an attempt to compensate for oxygen deficiency. Other theories postulating the mechanism explaining early bleeding as a sign of placental insufficiency include the finding that bleeding peaks at 5–8 weeks of gestation, coinciding with the luteal–placental shift and potentially indicating suboptimal placental development resulting in decreased levels of progesterone, which might be insufficient to prevent an episode of bleeding similar to menses.19
Limitations of this study include potential recall error as a result of the retrospective nature of data collection and the subjectivity of self-reported bleeding. However, there is no other established method of measuring vaginal bleeding in clinical practice, and bleeding was self-reported before the outcome was established, limiting its potential to bias the results of the study. An assessment of the accuracy of prospective diary or retrospective interview (end of the first trimester) to assess first-trimester bleeding demonstrated significant agreement between the two methods.26 Our study was limited by lack of information on the gestational week of bleeding and whether number of days was consecutive. Women in the study reported a wide range of number of days of bleeding, and there is potentially a biological difference between 2 or 3 days of bleeding compared with more, but this study lacked power to distinguish any difference based on number of days beyond 1 day of bleeding. Finally, because first-trimester bleeding was assessed at a gestational age ranging from 10 0/7 weeks to 13 7/7 weeks, there is potential to have missed some incidents of first-trimester bleeding that occurred after 10 0/7 weeks of gestation. However, 79.5% of enrollment interviews in this study occurred during 12 or 13 weeks of gestation, reducing some of this potential error.
In conclusion, more than 1 day of first-trimester vaginal bleeding was associated with asymmetric, decreased fetal growth independent of associated complications. Although the magnitude of this observed decrease in estimated fetal weight and birth weight was small, the difference in mean birth weight is comparable with differences in birth weight observed in pregnancies with maternal smoking, which range from approximately 50 to 200 g depending on maternal characteristics.27–29 It remains unknown whether bleeding is associated with short-term or long-term morbidity and if an additional intervention such as third-trimester ultrasonography would be of benefit. It is clinically relevant that bleeding that resolved by the second or third trimester is not necessarily exempt from decreased fetal growth. It is, however, reassuring that 1 day of first-trimester bleeding, which was the most common duration reported, was not associated with statistically significant patterns of decreased fetal growth.
1. Saraswat L, Bhattacharya S, Maheshwari A, Bhattacharya S. Maternal and perinatal outcome in women with threatened miscarriage in the first trimester: a systematic review. BJOG 2010;117:245–57.
2. Berkowitz GS, Harlap S, Beck GJ, Freeman DH, Baras M. Early gestational bleeding and pregnancy outcome: a multivariable analysis. Int J Epidemiol 1983;12:165–73.
3. Funderburk SJ, Guthrie D, Meldrum D. Outcome of pregnancies complicated by early vaginal bleeding. Br J Obstet Gynecol 1980;87:100–5.
4. Sipilä P, Sorri-Hartikainen AL, Oja H, Von Wendt L. Perinatal outcome of pregnancies complicated by vaginal bleeding. Br J Obstet Gynecol 1992;99:959–63.
5. Strobino B, Pantel-Silverman J. Gestational vaginal bleeding and pregnancy outcome. Am J Epidemiol 1989;129:806–15.
6. Sun L, Tao F, Hao J, Su P, Liu F, Xu R. First trimester vaginal bleeding and adverse pregnancy outcomes among Chinese women: from a large cohort study in China. J Matern Fetal Neonatal Med 2012;25:1297–301.
7. Weiss JL, Malone FD, Vidaver J, Ball RH, Nyberg DA, Comstock CH, et al. Threatened abortion: a risk factor for poor pregnancy outcome, a population-based screening study. Am J Obstet Gynecol 2004;190:745–50.
8. Williams MA, Mittendorf R, Lieberman E, Monson RR. Adverse infant outcomes associated with first-trimester vaginal bleeding. Obstet Gynecol 1991;78:14–8.
9. Yang JY, Savitz DA. The effect of vaginal bleeding during pregnancy on preterm and small-for-gestational-age births: US National Maternal and Infant Health Survey, 1988. Pediatr Perinatal Epidemiol 2001;15:34–9.
10. Mulik V, Bethel J, Bhal K. A retrospective population-based study of primigravid women on the potential effect of threatened miscarriage on obstetric outcome. J Obstet Gynaecol 2004;24:249–53.
11. Johns J, Jauniaux E. Threatened miscarriage as a predictor of obstetric outcome. Obstet Gynecol 2006;107:845–50.
12. Wilcox AJ. Intrauterine growth retardation: beyond birthweight criteria. Early Hum Dev 1983;8:189–93.
13. Davari-Tanha F, Shariat M, Kaveh M, Ebrahimi M, Jalalvand S. Threatened abortion: a risk factor for poor pregnancy outcome. Acta Med Iran 2008;46:314–20.
14. Buck Louis GM, Grewal J, Albert PS, Sciscione A, Wing DA, Grobman WA, et al. Racial/ethnic standards for fetal growth: the NICHD Fetal Growth Studies. Am J Obstet Gynecol 2015;213:449.e1–41.
15. Hediger ML, Fuchs KM, Grantz KL, Grewal J, Kim S, Gore-Langton RE, et al. Ultrasound quality assurance for singletons in the National Institute of Child Health and Human Development Fetal Growth Studies. J Ultrasound Med 2016;35:1725–33.
16. Hadlock FP, Harrist RB, Sharman RS, Deter RL, Park SK. Estimation of fetal weight with use of head, body, and femur measurements—a prospective study. Am J Obstet Gynecol 1985;151:333–7.
17. Duryea EL, Hawkins JS, McIntire DD, Casey BM, Leveno KJ. A revised birth weight reference for the United States. Obstet Gynecol 2014;124:16–22.
18. Pinheiro J, Bate D. Mixed effects models in S and S-Plus (statistics and computing). New York (NY): Springer; 2000.
19. Hasan R, Baird DD, Herring AH, Olshan AF, Jonsson Funk ML, Hartmann KE. Patterns and predictors of vaginal bleeding in the first trimester of pregnancy. Ann Epidemiol 2010;20:524–31.
20. Creasy RK, Resnik R, Green MF, Iams JD, Lockwood CJ, Moore TR. Creasy and Resnik's maternal-fetal medicine. 7th ed. Philadelphia (PA): Elsevier; 2014. p. 732.
21. Arafa M, Abdel-Fataah M, Abou Zeid HA, el-Khouly A. Outcomes of pregnancies complicated by early vaginal bleeding. East Mediterr Health J 2000;6:457–64.
22. Williams MA, Hickok DE, Zingheim RW, Mittendorf R, Kimelman J, Mahony BS. Low birth weight and preterm delivery in relation to early-gestation vaginal bleeding and elevated maternal serum alpha-fetoprotein. Obstet Gynecol 1992;80:745–9.
23. Tongsong T, Srisomboon J, Wanapirak C, Sirichotiyakul S, Pongsatha S, Polsrisuthikul T. Pregnancy outcome of threatened abortion with demonstrable fetal cardiac activity: a cohort study. J Obstet Gynecol (Tokyo 1995) 1995;21:331–5.
24. Walker J, Turnbull EP. Hemoglobin and red cells in the human fetus and their relation to the oxygen content of the blood in the vessels of the umbilical cord. Lancet 1953;265:312–8.
25. Turnbull EP, Walker J. The outcome of pregnancy complicated by threatened abortion. J Obstet Gynaecol Br Emp 1956;63:553–9.
26. Hasan R, Funk ML, Herring AH, Olshan AF, Hartmann KE, Baird DD. Accuracy of reporting bleeding during pregnancy. Paediatric Perinat Epidemiol 2010;24:31–4.
27. Butler NR, Goldstein H, Ross EM. Cigarette smoking in pregnancy: its Influence on birth weight and perinatal mortality. Br Med J 1972;2:127–30.
28. Conter V, Cortinovis I, Rogari P, Riva L. Weight growth in infants born to mothers who smoked during pregnancy. BMJ 1995;310:768–71.
© 2018 by American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.
29. Zheng W, Suzuki K, Tanaka T, Kohama M, Yamagata Z, et al. Association between maternal smoking during pregnancy and low birthweight: effects by maternal age. PLoS One 2016;11:e0146241.