| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Obstetric Imaging |
1 From the Departments of Radiology (R.S.B., P.W.C., V.A.F., R.A.F.) and Epidemiology and Biostatistics (R.S.B., P.B.), University of California, San Francisco, UCSF/Mt Zion Medical Center, 1600 Divisadero St, Box 1667, San Francisco, CA 94115; and Department of Obstetrics and Gynecology, Massachusetts General Hospital, Harvard Medical School, Boston, Mass (J.L.E.). Received May 2, 2001; revision requested June 4; revision received July 26; accepted August 24. R.S.B. supported in part by a Radiological Society of North America Nycomed-Amersham Research and Education Foundation grant. Address correspondence to R.S.B. (e-mail: rebecca.smith-bindman@radiology.ucsf.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Women (n = 321) who had singleton pregnancies and underwent two or more second- or third-trimester obstetric US examinations were included in a retrospective cohort analysis. Inadequate fetal growth was defined as growth at or below the 10th percentile. The relative risk of each poor outcome was calculated for fetuses with inadequate growth, compared with the risk for fetuses with normal growth.
RESULTS: Inadequate fetal growth was associated with 3.9 times the risk of a birth weight less than 2,500 g, 17.7 times the risk of a birth weight less than the 3rd percentile for gestational age, 2.3 times the risk of preterm birth, 2.6 times the risk of a long newborn hospital stay, and 3.6 times the risk of neonatal intensive care unit admission. After adjusting for confounding variables, including fetal weight, fetal growth remained a significant predictor of small birth size and poor outcomes. Inadequate growth predicted the risk of poor outcomes, even when gestational age was unknown. When inadequate growth was used to identify fetuses at risk, 21%–67% of neonates who were small at birth or had poor outcomes were identified at false-positive rates of only 5%–9%. For all outcomes, inadequate growth enabled identification of more fetuses with poor birth outcomes than low estimated fetal weight.
CONCLUSION: Morbidity is significantly increased among fetuses who demonstrate less than expected growth. Growth between two US examinations can be used to estimate the risk of neonatal morbidity, even when gestational age is unknown.
© RSNA, 2002
Index terms: Fetus, growth and development, 856.871 • Fetus, US, 856.1298 • Infants, newborn • Pregnancy, US, 856.1298
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Although many definitions of IUGR have been used, the most common definition is an estimated fetal weight at one point in time at or below the 10th percentile for gestational age (5). Hence, fetal size rather than fetal growth is commonly used to define IUGR. This definition is not ideal because most of the fetuses identified in this way are not at risk for neonatal morbidity but rather constitutionally small (7–9). Treatment strategies based on this definition will overdiagnose growth disturbances in small fetuses and underdiagnose growth disturbances in fetuses who are larger than an absolute size cutoff, but smaller than they constitutionally should be. Furthermore, for pregnant women who do not know the first day of their last menstrual period (LMP), a weight percentile cannot be calculated. For these women, IUGR is difficult to define and exclude. Thus, women with uncertain LMP dates may undergo close surveillance and antenatal testing because the fetal weight percentile cannot be accurately determined; thus, a growth disturbance may be difficult to exclude (10).
Compared with a single measurement of fetal size at one time during gestation (11–16), serial ultrasonographic (US) measurements of fetal size may be a more accurate method of identifying fetuses at risk for neonatal morbidity and mortality. The use of growth as a means of predicting neonatal outcomes might help identify fetuses at risk for poor outcomes independent of their absolute weight (8,9,17). Furthermore, in pregnant women who do not know their LMP date, normal fetal growth at two US examinations might indicate a low risk for neonatal morbidity and mortality and thus a reduced need for serial surveillance.
We performed this study to determine whether fetal growth at serial US examinations could predict neonatal morbidity and mortality independent of knowing a fetus’s gestational age or fetal weight. We hypothesized that inadequate growth at two US examinations would indicate a high risk of poor outcome independent of fetal weight.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Data Sources
The results of all US examinations performed at UCSF are available on a computerized US database (Aegis; Acuson, Mountain View, Calif) that includes US examination dates, LMP date, fetal biometric measurements (crown-rump length, head circumference, biparietal diameter, abdominal circumference, and femur length), the presence of fetal structural abnormalities, and clinical history. The gestational age of the fetus was calculated at the time of the initial US examination and used for all subsequent date determinations. In this database, gestational age was calculated by using the following criteria in descending order of importance: (a) known LMP date consistent within 7 days of first-trimester US or consistent within 14 days of second-trimester US, (b) first-trimester US, (c) second-trimester US (18), and (d) third-trimester US (18). All examinations were performed by using commercially available real-time US equipment (128 XP and Sequoia; Acuson).
The UCSF US database was linked with the UCSF obstetric database, which contains demographic information and details on the obstetric histories and neonatal outcomes for all births at UCSF. Variables abstracted from the obstetric database included race, height, prepregnancy weight, and substance abuse history of mother; sex, birth weight, and gestational age at birth of neonate; length of neonatal hospital stay; whether neonatal intensive care unit admission was required, whether neonatal assisted ventilation was required at birth, and whether fetal anomalies were present. In this database, birth outcomes are obtained by means of linkage with the UCSF neonatal database. Gestational age is determined by the senior on-call obstetrician, who reviews the patient’s records and makes a hierarchical decision by considering the following variables in descending order of importance: (a) in vitro fertilization date, (b) known regular LMP date consistent within 7 days of first-trimester US or within 14 days of second-trimester US, (c) first-trimester US, (d) second-trimester US, and (e) third-trimester US.
The estimated gestational age in the obstetric database was compared with that in the US database to ensure consistency. Inconsistent records were reviewed by one of the authors (R.S.B.), and the decision on how to date these pregnancies was made by two authors (R.S.B., J.L.E.). The most accurate method of pregnancy dating available in either database was used. Women for whom the gestational age of the fetus was known on the basis of in vitro fertilization, LMP dates, or first-trimester US were considered to have fetuses of known gestational age (n = 236 [73.5%]), and the remaining women were considered to have fetuses of unknown gestational age.
Calculation of Fetal Weight and Growth
Estimated fetal weight was calculated by using measurements of the head, abdomen, and femur (19). The details of estimating the percentiles of fetal weight are described elsewhere (20). In summary, the US data on 1,687 fetuses with known gestational age were used to estimate weight percentiles at each week of gestation. To construct the normal distribution of fetal growth, we used data on the 236 fetuses with serial US measurements of 13–38 weeks gestation that were obtained at examinations spaced at least 2 weeks and no more than 17 weeks apart and in whom the gestational age was known. Only one interval was used per fetus. If there was a choice of intervals, the US examination performed the earliest after 20 weeks gestation was chosen as the beginning of the interval and the subsequent examination was chosen as the end of the interval. This interval was chosen because we believed that it represented the first clinically relevant interval.
As expected, the rate of growth in grams per week is greater at later gestational ages. A scatterplot of growth versus midpoint of the US examination interval (not shown) showed a linear relationship but greater variance at later gestational ages. For example, the mean growth at 20 weeks gestation was 70 g/wk, whereas the mean growth at 30 weeks gestation was 175 g/wk (with correspondingly larger variance at 30 weeks). We therefore used the least-median-of-squares method to fit growth rate as a linear function of interval midpoint: growth = 10.8 g/d at 20 weeks gestation, with an increase of 1.35 g/d for each gestational week after week 20. For each fetus, the growth rate predicted by using the interval midpoint was subtracted from the observed growth rate during the interval as follows: normalized growth = observed growth - {10.8 + [(midpoint of interval - 20) x 1.35]}. With this technique, intervals that occurred at different times were placed on a comparable footing. We then determined weight percentiles by using these numbers. Inadequate growth was defined as growth at or below the 10th percentile. Normal growth was defined as growth in the 20th–80th percentiles.
Outcome Measures
The risks of the following neonatal outcomes were calculated by using data from the obstetric and neonatal archives: birth weight less than 2,500 g, birth weight less than 2,000 g, birth weight in the lower than 3rd percentile for gestational age, birth weight in the lower than 5th percentile for gestational age, preterm birth (ie, at less than 37 complete weeks of gestation), long neonatal hospital stay (>4 days), neonatal intensive care unit admission, and assisted ventilation required at birth. The first four outcomes are related to low birth weight, and the remaining outcomes more directly reflect neonatal morbidity and thus were grouped as poor neonatal outcomes. We included numerous neonatal outcomes because we wanted to encompass a broad range of potential adverse outcomes that might reflect neonatal morbidity (21).
Analyses
To assess whether the women who underwent two or more US examinations were similar to those who underwent only one US examination, we compared the demographic information and the prevalences of each outcome in women who underwent one US examination (n = 1,836) with these variables in the women who underwent multiple examinations (n = 321) by using a t test or 
2 test. The remaining analyses were performed in two groups of women, all of whom underwent two or more US examinations: those with fetuses of known gestational age (n = 236) and the entire sample of these women (n = 321). The latter group was analyzed by using a simulated unknown gestational age analysis.
Women with fetuses of known gestational age.—The relative risk of each outcome was calculated by dividing the risk of the outcome in the fetuses with inadequate fetal growth (growth at or below the 10th percentile) by the risk of that outcome in the fetuses with normal fetal growth (growth in the 20th–80th percentiles). To determine the accuracy of prenatal US in enabling the identification of fetuses with poor neonatal outcomes, we calculated the sensitivity, specificity, and positive and negative likelihood ratios for each neonatal outcome, defining an abnormal US result as fetal growth in the lowest 10th percentile. We also calculated the accuracy measures when a normal US result was defined as growth in the top half of the population of fetuses (ie, 50th percentile or greater). We calculated positive and negative likelihood ratios to determine the change in the risk of each outcome following a positive or negative US examination result.
To compare the accuracy of low estimated fetal weight and inadequate fetal growth in predicting poor neonatal outcomes, we generated receiver operating characteristic curves for each of these definitions of an abnormal US result. We calculated the partial areas under the receiver operating characteristic curves (22) within the clinically relevant range, defined as the part of the receiver operating characteristic curve that had a false-positive rate of 40% or less.
To determine whether the risks associated with inadequate fetal growth are equally important at different fetal weights, we compared the risks of poor neonatal outcomes associated with inadequate growth among different estimated fetal weight categories. For example, the risk of preterm birth was compared between fetuses with normal growth and those with inadequate growth in three estimated fetal weight categories: low estimated fetal weight (weight in the lowest quintile), average estimated fetal weight (weight between the 20th and 80th percentiles), and high estimated fetal weight (weight in the highest quintile). For this analysis only, to increase the sample size in each group, we defined inadequate growth as growth less than the 20th percentile.
We performed multivariate analysis to determine the association between inadequate fetal growth and neonatal outcome after adjusting for the following potential confounding variables: maternal age (<21 years, 21–35 years, or >35 years), height, weight, and race (white, African-American, Hispanic, other); body mass index; parity (nulliparous vs multiparous); maternal substance abuse history; fetal sex; and estimated fetal weight. For this multivariate analysis, fetal growth was dichotomized into inadequate (at or below 10th percentile) and normal (higher than 10th percentile) categories.
Simulated unknown gestational age analysis.—To evaluate the association between fetal growth and neonatal outcome in women who did not know their LMP date, we ignored the LMP dates for all women and performed simulated unknown gestational age analysis. This technique was used only in this analysis to increase the sample size of women with unknown LMP dates. For all women (n = 321) who underwent two or more US examinations, we calculated the relative risk of each outcome for fetuses with inadequate growth and for those with normal growth and calculated the accuracy measures as just described. We also used multivariate logistic regression analysis to determine the association between inadequate fetal growth and birth outcome for all women after adjusting for the potential confounding variables.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
|
|
We compared poor growth with low estimated fetal weight for enabling the identification of fetuses with poor neonatal outcomes. For each outcome that we evaluated, inadequate growth was significantly better than low estimated weight at enabling the detection of fetuses with poor outcomes (Fig 1) and had a correspondingly higher area under the receiver operating characteristic curves. For example, at every false-positive rate, poor growth enabled the identification of more fetuses with the outcome of a neonatal intensive care unit admission than did low estimated weight (area under curve, 0.53 vs 0.44) (Fig 1d).
|
|
|
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Although it is not surprising that fetal growth could predict poor neonatal outcomes, little previous research has been conducted to evaluate the association between fetal growth at second- and third-trimester prenatal US examinations and neonatal outcomes (23–25). It is interesting that among the large fetuses (those with estimated weights higher than the 80th percentile), inadequate growth was not associated with poor outcomes. This suggests that once a fetus reaches a certain size, continued growth may not be needed.
In clinical practice, many women do not know the first day of their last menstrual period, so practitioners cannot use this to establish the gestational age of the fetus. Accurate estimation of the gestational age enables the practitioner to evaluate the appropriateness of fetal size, antenatal surveillance, and delivery. If a woman who does not know her LMP presents for prenatal care within the first trimester of pregnancy, US is often performed and is a reliable method of determining the date of pregnancy onset (26). However, nearly 20% of women do not present for obstetric care until the second trimester of pregnancy or later (27), when US-based determination of pregnancy onset is less accurate. Therefore in these women, it is more difficult to estimate the due date or relative fetal size, and the management of a possible postterm gestation becomes complicated. Because an estimated fetal weight percentile based on the LMP-based gestational age cannot be generated for these women, growth restriction cannot be excluded, and as a result, these women commonly undergo intensive monitoring, repeat imaging, and close clinical follow-up. It would therefore be helpful to have a method of further stratifying the risk for poor neonatal outcomes for women in whom the gestational age of the fetus cannot be estimated.
We found that even when the gestational age is unknown, the fetal growth between two US examinations can be used to predict neonatal outcomes. Women whose fetuses demonstrate growth in the lowest decile are at a significantly increased risk for poor outcomes, and women whose fetuses demonstrate average or better growth are at a decreased risk for poor outcomes. The predicted risks of low birth weight and poor outcome may be significantly reduced (range, 10%–100%) if average or better fetal growth is identified. This reduction in predicted risk may allow practitioners to reduce the intensity of antenatal testing in this group of women.
There are several limitations to our analyses. The high prevalence of poor outcomes in our population reflects the nature of our practice at a tertiary referral hospital. For example, 21.8% of the women described in this report gave birth to babies who weighed less than 2,500 g, whereas the U.S. national average percentage of women who give birth to babies with birth weights this low ranges from 6.4% to 13.1%, depending on the racial group (27). In addition, 25.2% of the neonates described in this report were preterm, whereas the U.S. national average percentage of preterm infants ranges from 8.9% to 17.6%, depending on the racial group (27).
Poor outcomes were especially evident in the women who underwent more than one US examination. Although the relationship between inadequate growth and poor outcome is likely to exist in other populations, the threshold that separates women with increased risk from those with average risk might be different. The advantage of using this patient population is that it allowed us to see the association between fetal growth and outcomes, and this may have been more difficult to identify in a healthier population with fewer poor outcomes. Although the women included in this study were racially and ethnically diverse, the sample size was too small to determine whether normal fetal size differs according to racial group. However, race was included in all multivariate models, and including this variable did not nullify the importance of growth in predicting poor neonatal outcomes.
For simulated unknown gestational age analysis, we ignored the LMP dates for all the women. Because the neonatal outcomes of the women with unknown LMP dates were similar to those of the women with known LMP dates (20), we believe that increasing our sample size in this way did not bias our results. The US measurements described in this study were not scheduled for the purposes of this study, and the timing of these examinations may have been influenced by the existence of pregnancy difficulties. However, although such timing may have contributed to lower precision, the scheduling should not have created erroneous associations.
As a result of using existing clinical data, we could not determine the ideal time to perform fetal measurements or the ideal interval between US examinations to best predict fetal outcome. Although an interval of 8–10 weeks has been suggested as the statistically most reliable interval to measure fetal growth (28), it is not clear whether this interval is the most important clinically or the most effective for predicting neonatal outcomes.
Results of this report suggest that US measurements of fetal size are important predictors of birth outcomes and that fetal growth in particular can be used to help identify fetuses at risk for high neonatal morbidity. Inadequate fetal growth, as defined in this article, potentially could be used to focus resources on those pregnancies that are at highest risk. Fetuses with inadequate growth, for example, could be followed up with more frequent monitoring or antenatal testing. Conversely, fetuses with normal growth (despite a low estimated weight) might be followed up less intensively. Although few interventional procedures have been shown to improve the outcomes of fetuses with suspected growth disturbances (6,29), our study data suggest a mechanism by which growth disturbances might be better defined that might be used in future trials.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Author contributions: Guarantors of integrity of entire study, all authors; study concepts and design, all authors; literature research, R.S.B., J.L.E., V.A.F., R.A.F.; data acquisition, R.S.B.; data analysis/interpretation, all authors; statistical analysis, R.S.B., P.W.C., J.L.E., P.B.; manuscript preparation, definition of intellectual content, and editing, all authors; manuscript revision/review, P.W.C., R.S.B.; manuscript final version approval, R.S.B.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  V. L. Ward, M. Nishino, H. Hatabu, J. A. Estroff, C. E. Barnewolt, H. A. Feldman, and D. Levine Fetal Lung Volume Measurements: Determination with MR Imaging--Effect of Various Factors. Radiology, July 1, 2006; 240(1): 187 - 193. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B F Kalanda, S van Buuren, F H Verhoeff, and B J Brabin Anthropometry of fetal growth in rural Malawi in relation to maternal malaria and HIV status Arch. Dis. Child. Fetal Neonatal Ed., March 1, 2005; 90(2): F161 - F165. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Smith-Bindman, P. W. Chu, J. Ecker, V. A. Feldstein, R. A. Filly, and P. Bacchetti Adverse Birth Outcomes in Relation to Prenatal Sonographic Measurements of Fetal Size J. Ultrasound Med., April 1, 2003; 22(4): 347 - 356. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
