Tuuli, Methodius G. MD, MPH; Cahill, Alison MD, MSCI; Stamilio, David MD, MSCE; Macones, George MD, MSCE; Odibo, Anthony O. MD, MSCE
Adverse perinatal outcomes including small for gestational age (SGA), preterm delivery, and preeclampsia are associated with increased neonatal mortality and morbidity as well as complications later in adult life.1–3 Increasing evidence suggests that these adverse outcomes have their origins, at least in part, in early pregnancy.4 One of the earliest large-scale studies reported that first-trimester crown-rump length 2 to 6 days smaller than expected was associated with a significantly increased risk of low birth weight (LBW) less than 2,500g, SGA at delivery, and preterm delivery before 24 and 32 weeks of gestation.5 Other reports have implicated second-trimester fetal growth restriction and reduced first-to-second-trimester fetal growth velocity as risk factors for adverse perinatal outcomes.6–8 A recently published prospective study found early fetal growth restriction to be associated with growth abnormalities at birth as well as differential growth in early childhood.9
Taken together, these studies suggest that late-pregnancy complications in many women are established early in pregnancy and that such pregnancies can be identified by suboptimal early fetal growth. However, these studies have not comprehensively evaluated potential measures of early fetal growth restriction to determine the threshold values that would result in optimal prediction of pregnancies at risk for adverse outcomes. In particular, the comparative screening efficiency of first- and second-trimester fetal growth restriction and first-to-second-trimester growth lag has not been explored.
The objective of this study was to estimate the optimal definitions and comparative efficiency of first- and second-trimester fetal growth restriction and first-to-second-trimester fetal growth lag for predicting adverse pregnancy outcomes including SGA at delivery, LBW, preterm delivery, stillbirth, and preeclampsia.
MATERIALS AND METHODS
This is a retrospective cohort study from our prospectively compiled perinatal database (January 1990 to December 2009). Approval for the study was obtained from our institutional review board. Maternal demographics and medical and obstetric history were obtained using a self-report questionnaire at the initiation of prenatal care and at the time of initial ultrasound examinations. Ultrasound findings and obstetric outcomes were entered into the database by trained research nurses. For women who delivered outside of our facility, outcomes were obtained by contacting the patient and referring physician. Ultrasound examinations were performed by certified dedicated obstetric sonographers. Final diagnoses and interpretations were made by fellowship trained Maternal Fetal Medicine attending physicians.
Women with pregnancies dated by known last menstrual periods that were consistent with first-trimester ultrasound measurements were identified for this study. We included only pregnancies that were evaluated in both the first and second trimesters in our ultrasound unit. Multiple gestations and pregnancies with major structural and chromosomal anomalies were excluded. We included all women meeting the inclusion criteria. No a priori sample size calculation was performed.
Fetal crown-rump lengths were measured at 10–14 weeks of gestation. Second-trimester ultrasound examinations were performed at 18–22 weeks. This involved detailed anatomic survey for structural anomalies and measurement of fetal biometry including head circumference, biparietal diameter, abdominal circumference, and femur length. Head dimensions were measured from leading edge to leading edge for biparietal diameter and from the outer margin to outer margin for the occipito-frontal diameter at the level of the cavum septum pellucidum. Head circumference was calculated from the biparietal diameter and occipito-frontal diameter. Abdominal circumference was measured at the level of the umbilical vein using the anterio-posterior and transverse diameters. Femur length was measured from the proximal to the distal metaphysis according to the technique described by O'Brien and Queenan.10 Estimated fetal weights were calculated from fetal biometry using the Hadlock formula.11 Crown-rump length at 10–14 weeks and the estimated fetal weight at 18–22 weeks for each fetus were converted to gestational age–adjusted Z-scores (crown-rump length Z-score and estimated fetal weight Z-score, respectively).
First-to-second-trimester fetal growth was calculated as the difference between gestational age in the first trimester estimated from the crown-rump length using the formula by Robinson and Fleming (GA1)12 and second-trimester gestational age estimated from the head circumference (GA2) using the formula by Altman and Chitty.13 This was then converted to gestational age–adjusted fetal growth by dividing the difference in days between the first- and second-trimester gestational ages by the number of calendar days between the two ultrasound examinations (interval). That is, gestational age–adjusted fetal growth=(GA2−GA1)/interval. By so doing, gestational age–adjusted first-to-second-trimester growth of 1 corresponds to no growth deviation, more than 1 represents faster growth than expected, and less than 1 represents slower growth than expected. The first-to-second-trimester growth was also converted to Z-scores (growth Z-scores).
The first- and second-trimester fetal growth restriction and fetal growth lag were defined using different Z-score cutoffs (less than −1.0, −1.5, −2.0 and −2.5). Because the distribution of the Z-scores followed the standard normal distribution, the anticipated screen positive rates for these cutoffs were 15.9%, 6.7%, 2.3%, and 0.6%, respectively.14 Receiver-operating characteristics curves were constructed for the primary outcome (SGA) and crown-rump length Z-score, estimated fetal weight Z-score, and growth Z-score. These were then used to identify the cutoffs for these measures of fetal growth restriction that optimized sensitivity and specificity.
The primary outcome was SGA at delivery, defined as birth weight less than the 10th percentile for gestational age on the Alexander growth curve.15 Secondary outcomes were low birth weight (LBW; birth weight less than 2,500g), stillbirth (fetal death at 20 weeks or later), preterm delivery (less than 32 weeks), neonatal intensive care unit (NICU) admission, and preeclampsia (defined according to guidelines of the International Society for the Study of Hypertension in Pregnancy).16 Crude and adjusted odds ratios (OR) with 95% confidence intervals (CIs) were calculated for each outcome. The adjusted ORs were obtained using multivariable logistic regression. We selected candidate variables for the regression models based on biological plausibility, factors identified in the literature, and results of our univariable analysis. The number of variables in each model was then reduced using backward elimination. The Wald test was used during the model building process to assess effects of removing different predictor variables. Using these, the modeling for each of the three measures of fetal growth restriction and six outcomes took several steps. Goodness-of-fit for each final model was assessed using the Hosmer-Lemeshow lack-of-fit test.17 For the three measures of early fetal growth restriction and the six outcomes examined, a total of 18 Hosmer-Lemeshow tests were performed.
The screening efficiency of each measure of early fetal growth restriction was estimated by calculating sensitivities, specificities, positive predictive values, negative predictive values, positive likelihood ratios, and negative likelihood ratios. The efficiencies of the different measures were tested by comparing the areas under the receiver-operating characteristics curves using the nonparametric method described by DeLong et al.18 To estimate whether first- and second-trimester fetal growth restriction have differential effects on perinatal outcomes, we calculated crude and adjusted odds ratios for the primary outcome among categories of fetuses based on the presence or absence of first- and second-trimester fetal growth restriction compared with controls. We also compared the rates of NICU admission in these categories using χ2 and multivariable logistic regression analyses.
All statistical analysis was performed using STATA 10.0. Tests with P<.05 were considered statistically significant.
A flow chart of study participants is shown in Figure 1. A cohort of 8,978 singleton pregnancies without major structural or chromosomal anomalies and with ultrasound examinations at both 10–14 and 18–22 weeks were identified. Seventy-three pregnancies (0.8%) resulted in spontaneous abortions, 23 (0.3%) were elective terminations, 15 (0.2%) were stillbirths, and 434 (4.8%) had no outcome information. Of the remaining 8,433 live births, 551 (6.5%) neonates were SGA and 7,882 were not. The mean maternal age was 34.5±5.5 years, and more than half were white (68.2%). The mean gestational age of the entire cohort was 11.6±0.9 weeks of gestation at the first-trimester ultrasound examination and 19.5±1.0 weeks at the second-trimester ultrasound examination. Women with SGA neonates were more likely to be African American, smoke cigarettes, and have chronic hypertension. SGA neonates were also more likely to be female (Table 1). Plots of the Z-scores for crown-rump length, estimated fetal weight, and interval growth followed the standard normal distribution with means of approximately 0 and standard deviations of 1.0.
The area under the curve for predicting SGA was significantly higher for estimated fetal weight Z-scores than crown-rump length Z-scores and interval growth lag Z-scores (0.70 compared with 0.59 and 0.66, nonparametric test P<.001) (Fig. 2). Performance of a range of cutoffs of crown-rump length, estimated fetal weight, and interval growth Z-scores for predicting SGA are shown in Table 2. The risk of SGA increased with both increasing severity of second-trimester fetal growth restriction and interval growth lag (except for Z score less than −2.5), whereas only crown-rump length Z-score less than −1 was significantly associated with SGA. For crown-rump length Z-score less than −1, the sensitivity, specificity and area under the receiver-operating characteristics curve were 21.4%, 83.4%, and 0.53, respectively. Estimated fetal weight Z-score less than −1 also emerged the optimal cutoff of second-trimester fetal growth restriction for predicting SGA with sensitivity, specificity, and area under the receiver-operating characteristics curve of 37.2%, 85.5%, and 0.61, respectively. Interval fetal growth Z-score less than −1 was also the optimal cutoff of first-to-second-trimester fetal growth lag for predicting SGA. The associated sensitivity, specificity, and area under the receiver-operating characteristics curve were 25.6%, 89.5%, and 0.57, respectively (Table 3).
Risk estimates for SGA and other adverse perinatal outcomes associated with early fetal growth restriction are shown in Table 4. First-trimester fetal growth restriction was associated with a significantly increased risk of SGA (8.2% compared with 5.9%, adjusted OR 1.41 [95% CI 1.13–1.74]). Second-trimester fetal growth restriction was associated with an even higher risk of SGA (14.7% compared with 4.7%, adjusted OR 3.44 [2.85–4.15]). The risk of SGA associated with first-to-second-trimester growth lag was intermediate between first- and second-trimester fetal growth restriction (adjusted OR 2.61 [95% CI 2.0–3.25]).
All measures of early fetal growth restriction were also associated with LBW. On the other hand, whereas second-trimester fetal growth restriction and first-to-second-trimester fetal growth lag were associated with an increased risk of stillbirth and preterm delivery, first-trimester fetal growth restriction was not associated with a significantly increased risk of these outcomes. In addition, none of the measures of early fetal growth restriction was associated with a significant increase in the risk of preeclampsia.
Four categories of fetuses based on the presence or absence of first- and second-trimester fetal growth restriction are shown in Table 5. Compared with controls, normal first-trimester growth with second-trimester fetal growth restriction was associated with the highest risk for SGA (16.5% compared with 4.9%, adjusted OR 3.60 [95% CI 2.87–4.51]). Fetuses with both first- and second-trimester fetal growth restriction were also at a high risk for SGA (13.9% compared with 4.9%, adjusted OR 3.14 [95% CI 2.38–4.16]). The risk of SGA among fetuses with first-trimester fetal growth restriction but normal second-trimester growth (catch-up growth) was not significantly different from controls (5.0% compared with 4.9%, adjusted OR 0.99 [95% CI 0.71–1.39]). The pattern of NICU admission among the four categories was similar to the associations with SGA (Fig. 3). Notably, the risk of NICU admission among fetuses with both first- and second- trimester fetal growth restriction was similar to the risk among fetuses with normal first-trimester growth but second-trimester fetal growth restriction (6.1% compared with 7.0%, adjusted OR 0.99 [0.59–1.67]).
We used a large, single-center, prospectively compiled, comprehensive perinatal database to identify the optimal definitions of early fetal growth restriction for predicting SGA and other adverse perinatal outcomes. We also evaluated the comparative screening efficiency of the optimal definitions of first- and second-trimester fetal growth restriction and first-to-second-trimester growth lag for predicting SGA at delivery. Each of these measures of suboptimal fetal growth was found to be associated with a significantly increased risk of SGA and low birth weight. Fetal growth restriction detected for the first time in the second trimester was associated with the highest risk of SGA at delivery. The screening efficiency of second-trimester fetal growth restriction for SGA was superior to first-trimester fetal growth restriction and first-to-second-trimester growth lag. Second-trimester fetal growth restriction and first-to-second-trimester growth lag were also associated with a significantly increased risk of stillbirth and preterm delivery at less than 32 weeks. Fetuses with evidence of fetal growth restriction in both the first and second trimesters were at increased risk, not only for SGA at delivery, but also for NICU admission, suggesting suboptimal growth rather than constitutional smallness.
The superiority of second-trimester fetal growth restriction for predicting SGA at birth is consistent with the current understanding of the control of fetal growth physiology.19 Major influences on first-trimester growth are thought to be from innate genetic factors, with a smaller percentage contribution from intrauterine environment and uterine-placental interchange factors. Transition to the second trimester and beyond is associated with an increasing role of environmental factors and uterine-placental exchange. Second-trimester growth is therefore more reflective of both genetic, environmental and uterine-placental factors that will determine the ultimate birth weight.
Whereas many studies have reported an association of each of these measures of fetal growth restriction and adverse perinatal outcomes, our analysis directly compared the screening efficiencies of these measures to each other. The large sample size allowed us to perform subgroup analyses for even the relatively rare perinatal outcomes. Our findings suggest that suboptimal early fetal growth may be in the pathway that links risk factors early in pregnancy to eventual pregnancy outcomes. The positive predictive values for predicting SGA and other adverse outcomes are modest. For example, the 14.7% positive predictive value of second-trimester fetal growth restriction for predicting SGA at delivery implies that only about 15 of every 100 women predicted to be SGA fetuses will actually have SGA neonates at delivery. However, the potential use of early fetal growth restriction as a screening tool for SGA has a number of attractive features. Crown-rump length and estimated fetal weight are measured routinely at first- and second-trimester ultrasound examinations, respectively. They are therefore readily available at no extra cost or patient inconvenience. The high negative predictive values mean that women can be reassured to a significant degree if fetal measurements are normal at both first- and second-trimester ultrasound examinations. For example, the 95.3% negative predictive value of second-trimester fetal growth restriction for SGA suggests that only 5 fetuses out of 100 predicted not to be SGA will be SGA at delivery. Finally, the combination of these measurements with other independent predictors has the potential to improve screening. This would aid the testing and use of potential interventions to improve outcomes for pregnancies identified to be at high risk for complications.
Previous studies have used inconsistent definitions of fetal growth restriction and outcomes including SGA and preterm delivery. The different methodologies notwithstanding, most studies to date have produced results similar to ours. Smith et al used first-trimester growth lag in days to define fetal growth restriction,5 and Mook-Kanamori et al defined first-trimester fetal growth restriction as gestational age–adjusted crown-rump length in the lowest 20% of the population.9 Leung et al used crown-rump length Z-scores to define first-trimester fetal growth restriction and noted that crown-rump length Z-scores and log-PAPP-A multiples of the median were the only independent predictors of SGA in multivariable analysis.20 The area under the receiver-operating characteristics curve of crown-rump length Z-score for predicting SGA was 0.59, exactly the same as in this present study.20 In contrast, Pedersen et al found no association between crown-rump length less than the 10th percentile and SGA at birth.6 The authors suggested this may be the result of overestimation of gestational age at the time of the first-trimester ultrasound examination rather than slow growth. The relatively small sample size and the use of customized birth weights to define SGA in that study may have blunted the association. It is reassuring that results of our second-trimester fetal growth restriction analysis are consistent with those of previous studies.7,8,21
There are limitations that should be considered when interpreting our data. Pregnancies in this cohort were dated by last menstrual period and confirmed by first-trimester ultrasound examination. There is a potential for inaccurate assessment of gestation age using this method, considering that menstrual periods are of varied lengths and the timing of conception for spontaneous pregnancies cannot be absolutely known. In addition, there is the potential for recall bias with self-reported last menstrual period.22 However, because these dating errors occurred before the outcomes, bias is most likely to be nondifferential, which would blunt the magnitude of association rather than produce an invalid one. Results of secondary analysis of almost 1,000 women in the FASTER trial who conceived by assisted reproductive technology and for whom the day of conception was known, confirmed the relationship between first-trimester fetal growth restriction and SGA.23 This and the results of other studies suggest that the associations between early fetal growth restriction and adverse outcomes persist even when the exact timing of conception is unknown. The retrospective nature of our study makes it vulnerable to inaccuracies in coding and imputation of pregnancy, neonatal, and sonographic data. Fortunately, our perinatal database has been validated in previous studies and misclassification found to be minimal.24–26 Finally, as with any study, it is important for readers to keep generalizability of our findings in mind. This study cohort is from our center where we see referrals in addition to our regular patients. Therefore, it is likely that they may be at a higher risk of adverse outcomes. Thus, negative and positive predictive values of measures of early fetal growth restriction for predicting adverse perinatal outcomes could be affected by the higher prevalence of these outcomes while sensitivities and specificities would be unaffected.
An important challenge in assessing fetal growth is the differentiation of fetuses failing to reach their genetic growth potential from those that are constitutionally small. Mook-Kanamori et al reported that first-trimester fetal growth restriction was associated with a more rapid growth in the first 2 years of life.9 This implicates the intrauterine environment and not fetal constitution as responsible for the smaller-than-expected fetal size. Using NICU admission as a proxy for pathologic SGA, we noted that fetuses that were smaller than expected at both first- and second-trimester ultrasound examinations and would have been considered to be constitutionally small, were at the same high risk for NICU admission as those that were normally grown in the first trimester but fell off their growth curve by the second trimester. Another important consideration is the definition of SGA used in this study. A cutoff of birth weight less than the 5th percentile would identify a group with more severe growth restriction and, thus, define a population with a higher proportion of pathological growth. However, this trade-off would miss identifying neonates with pathological growth in the 5th to 10th percentile range. We chose to use a less restrictive definition of SGA to identify a greater proportion of neonates with pathologic growth. However, when we repeated the analysis with SGA defined as less than the 5th percentiles our conclusions were unchanged.
In conclusion, our data confirm the association between early fetal growth restriction and adverse pregnancy outcomes. We have demonstrated that although the predictive ability of the three measures of early fetal growth restriction for adverse perinatal outcomes is modest, second-trimester fetal growth restriction is superior to first-trimester fetal growth restriction and first-to-second-trimester growth lag. Central to our ability to identify early fetal growth restriction is reliable pregnancy dating. This should provide another incentive for ultrasound examinations early in pregnancy, especially for pregnancies at risk for fetal growth restriction. Combining these ultrasound measures of early fetal growth with other independent predictors may improve the modest screening efficiencies observed. Our data add to existing literature by identifying second-trimester fetal growth restriction as the superior measure to identify fetuses at risk for SGA. For fetuses at the highest risk for SGA such as those with normal first-trimester growth but second-trimester fetal growth restriction and those with both first- and second-trimester fetal growth restriction, a reasonable approach would be to perform a third-trimester ultrasound examination and initiate appropriate monitoring and interventions for those found to have evidence of growth restriction.
1.McIntire DD, Bloom SL, Casey BM, Leveno KJ, Birth weight in relation to morbidity and mortality among newborn infants. N Engl J Med 1999;340:1234–8.
2.Barker DJ. Adult consequences of fetal growth restriction. Clin Obstet Gynecol 2006;49:270–83.
3.Gluckman PD, Hanson MA, Cooper C, Thornburg KL. Effect of in utero and early-life conditions on adult health and disease. N Engl J Med 2008;359:61–73.
4.Smith GC. First-trimester determination of complications of late pregnancy. JAMA 2010;303:561–2.
5.Smith GC, Smith MF, McNay MB, Fleming JE. First-trimester growth and the risk of low birth weight. N Engl J Med 1998;339:1817–22.
6.Pedersen NG, Figueras F, Wojdemann WR, Tabor A, Gardosi J. Early fetal size and growth as predictors of adverse outcome. Obstet Gynecol 2008;112:765–71.
7.Pedersen NG, Wojdemann KR, Scheike T, Tabor A. Fetal growth between the first and second trimesters and the risk of adverse pregnancy outcome. Ultrasound Obstet Gynecol 2008;32:147–54.
8.Rasmussen S, Kiserud T, Albrechtsen S. Foetal size and body proportion at 17–19 weeks of gestation and neonatal size, proportion, and outcome. Early Hum Dev 2006;82:683–90.
9.Mook-Kanamori DO, Steegers EA, Eilers PH, Raat H, Hofman A, Jaddoe VW. Risk factors and outcomes associated with first-trimester fetal growth restriction. JAMA 2010;303:527–34.
10.O'Brien GD, Queenan JT. Growth of the ultrasound fetal femur length during normal pregnancy. Part I. Am J Obstet Gynecol 1981;141:833–7.
11.Hadlock FP, Harrist RB, Sharman RS, Deter RL, Park SK. Estimation of fetal weight with the use of head, body, and femur measurements–a prospective study. Am J Obstet Gynecol 1985;151:333–7.
12.Robinson HP, Fleming JE. A critical evaluation of sonar “crown-rump length” measurements. Br J Obstet Gynaecol 1975;82:702–10.
13.Altman DG, Chitty LS. New charts for ultrasound dating of pregnancy. Ultrasound Obstet Gynecol 1997;10:174–91.
14.Kirkwood BR, Sterne JAC. Essential medical statistics. 2nd ed. Oxford: Blackwell; 2003. p. 470–471.
15.Alexander GR, Himes JH, Kaufman RB, Mor J, Kogan M. A United States national reference for fetal growth. Obstet Gynecol 1996;87:163–8.
16.Brown MA, Lindheimer MD, de Swiet M, Van Assche A, Moutquin JM. The classification and diagnosis of the hypertensive disorders of pregnancy: statement from the International Society for the Study of Hypertension in Pregnancy (ISSHP). Hypertens Pregnancy 2001;20:IX–XIV.
17.Hosmer DW, Lemeshow S. Applied logistic regression. 2nd ed. New York (NY): Wiley; 2000.
18.DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1998;44:837–45.
19.Mongelli M, Gardosi J. Fetal growth. Curr Opin Obstet Gynecol 2000;12:111–5.
20.Leung TY, Chan LW, Leung TN, Fung TY, Sahota DS, Spencer K, et al. First-trimester combined screening for trisomy 21 in a predominantly Chinese population. Ultrasound Obstet Gynecol 2007;29:14–7.
21.Nakling J, Backe B. Adverse obstetric outcome in fetuses that are smaller than expected at second trimester routine ultrasound examination. Acta Obstet Gynecol Scand 2002;81:846–51.
22.Savitz DA, Terry JW Jr, Dole N, Thorp JM Jr, Siega-Riz AM, Herring AH. Comparison of pregnancy dating by last menstrual period, ultrasound scanning, and their combination. Am J Obstet Gynecol 2002;187:1660–6.
23.Bukowski R, Smith GC, Malone FD, Ball RH, Nyberg DA, Comstock CH, et al. Fetal growth in early pregnancy and risk of delivering low birth weight infant: prospective cohort study. BMJ 2007;334:836.
24.Tuuli MG, Dicke JM, Stamilio DM, Gray DL, Macones GA, Rampersad R, et al. Prevalence and likelihood ratios for aneuploidy in fetuses diagnosed prenatally with isolated congenital cardiac defects. Am J Obstet Gynecol 2009;201:390.e1–5.
25.Dicke JM, Blanco VM, Yan Y, Coplen DE. The type and frequency of fetal renal disorders and management of renal pelvis dilatation. J Ultrasound Med 2006;25:973–7.
26.Goetzinger KR, Stamilio DM, Dicke JM, Macones GA, Odibo AO. Evaluating the incidence and likelihood ratios for chromosomal abnormalities in fetuses with common central nervous system malformations. Am J Obstet Gynecol 2008;199:285.e1–6.
Figure. No caption available.