- Email: [email protected]
Fetal growth and the etiology of preterm delivery
Fetal Growth Deliverv
and the Etiology
of Preterm
J
MARY L. HEDIGER, PhD, THERESA 0. SCHOLL, PhD, JOAN I. SCHALL, PhD, LAURIE W. MILLER, BS, RDMS, AND RICHARD L. FISCHER, MD Objective:
To
confirm
with fetal growth various etiologies
that
restriction of preterm
preterm
same degree and type of FGR. Methods: Two hundred ninety gravidas who had routine initial also had subsequent ultrasound gestation. Fetal tween preterm deliveries, obstetric (PROM),
delivery
(FGR), delivery
is
associated
and to determine if the are associated with the young, primarily minority ultrasound examinations examinations at 32 weeks’
growth characteristics (less than 37 weeks’
were compared gestation) and
and among preterm deliveries indications, premature rupture and spontaneous preterm labor.
with of
beterm
medical membranes
or
Results: Forty-six infants (15.9%) were born preterm. At 32 weeks’ gestation, all fetuses later delivered preterm were already smaller than fetuses later delivered at term (P < .05) for all dimensions: circumference (AC), length (FL). However,
head circumference biparietal diameter after stratifying
(HC), (BPD), by cause
abdominal and femur of preterm
delivery for those fetuses later delivered for medical or obstetric indications, we found that only AC was decreased (P < .Ol) and that the HC-AC ratio was elevated (asymmetric FGR). Neonates delivered after unsuccessfully PROM or preterm labor were symmetrically smaller characteristics (HC, AC, BPD, and FL).
treated in all
Conclusion: By 32 weeks’ gestation, fetuses later delivered preterm are already significantly smaller than fetuses later delivered at term. However, when stratified by the etiology of preterm obstetric
delivery, indications
infants delivered had asymmetric
preterm growth
for medical or patterns, which
suggests a growth failure late in pregnancy. Infants delivered preterm after PROM or after failed or no tocolysis for spontaneous preterm labor were proportionately smaller, implying an overall slowing of growth that early in pregnancy and possibly demonstrate stress. (Obstet Gynecol 1995;85:175-82)
From
may originate a more chronic
the Departments of Obstetrics and Gynecology, The University and Dentistry of New Jersey-School of Osteopathic Medicine and Robert Wood Johnson Medical School, Camden, New Jersey. Supported by grant no. HD18269 fr om the National lnstitute of Child Health and Human Development.
of Medicine
VOL.
85,NO.Z,FEBRUARY
1995
Preterm delivery is a major contributor to neonatal and infant mortality and morbidity, but the etiology and risk factors for preterm delivery are not well established. Some of the risk factors associated with preterm delivery are urogenital infection, poor placentation or vasculopathy, multiple gestation, cigarette smoking, substance abuse, black ethnicity, and, possibly, poor diet and nutrition.’ However, identification of risk factors has been complicated by the different etiologic pathways culminating in preterm delivery: spontaneous preterm labor, preterm premature rupture of membranes (PROM), and numerous medical or obstetric indications and complications (eg, severe preeclampsia, placental abruption, or chorioamnionitis).‘,’ The risk factors that may be associated separately with each of the various pathways, especially idiopathic preterm labor and PROM, are also not well known. It has been postulated that, whatever the cause, preterm delivery is characterized by suboptimal fetal growth.” It follows, then, that birth weight for gestational age standards generated from birth weights at preterm gestations may underestimate the size of fetuses that will subsequently be delivered at term, thereby underestimating the number of preterm infants who are born with fetal growth restriction (FGR) and who are at risk for major morbidity. Even given the widespread use of obstetric ultrasonography during pregnancy, only a few studies3-s have examined the relation between preterm delivery and fetal growth. Although previous studies3-s have evaluated the association between preterm delivery and fetal growth, they pose a number of methodologic concerns. In several, growth measurements were determined by ultrasound scan either just preceding delivery., after PROM, or at admission for preterm labor. The problem with cranial measurements obtained after rupture of membranes is that fetuses may have artifactually decreased biparietal diameters (BPD) (dolichocephaly) and may not truly be growth restricted.5’9 Furthermore,
0029-7844/95/$9.50 0029-7844(94)00365-K
175
none of the studies included controls who had no risk factors for preterm delivery and who later delivered at term. In only one of the studies of preterm delivery and fetal growth’ was there any control for other factors known to affect fetal growth,“‘,” such as maternal age, parity, ethnicity, maternal pregravid body mass index (BMI), maternal height, cigarette smoking, or fetal sex. Although there are different proximate causes of preterm delivery, it is not known if each is associated with the same degree and type of FGR.” The objective of this study was to confirm, in a controlled prospective cohort study of primarily young, minority gravidas, that suboptimal fetal growth is characteristic of preterm delivery, and to determine if the various etiologies of preterm delivery are associated with the same degree and type of FGR.
Materials and Methods Thirty percent of gravidas enrolled from October 1990 through November 1993, in a larger, ongoing study of nutrition and growth in adolescent (younger than 19 years) and older gravidas (between 19-29 years), were also recruited to have a subsequent ultrasound examination at 32 weeks’ gestation. All subjects were initially enrolled at entry to prenatal care under the same protocol from an urban clinic in Camden, New Jersey. Gravidas with verified chronic or metabolic disease that could affect nutrition or fetal outcome-such as congenital heart disease, tuberculosis, chronic hypertension, insulin-dependent diabetes mellitus, hemoglobinopathies, cancer, lupus, seizures, drug or alcohol abuse, and psychiatric disorders-were not enrolled in the larger study. Every third subject (290 of 977) enrolled in the larger study during this time period was recruited within a month of entry to prenatal care to have a real-time and Doppler ultrasound scan for research purposes at 32 weeks’ gestation. Funding was available only to perform ultrasound scans at 32 weeks on this proportion of the larger cohort. At about 7 months’ gestation (28 weeks), the subjects identified previously were scheduled for obstetric ultrasound examinations at 32 weeks. Only the results of the real-time ultrasound scans for fetal growth are presented here. The protocol met the guidelines of the Institutional Review Board of the University of Medicine and Dentistry of New Jersey. There were no differences in ethnic distribution between those recruited for an ultrasound scan at 32 weeks’ gestation and the cohort as a whole. However, the 290 subjects enrolled for later ultrasound were slightly, but significantly, younger (17.4 I 2.4 versus 17.9 2 2.6 years, P < .Ol), and consequently there was a higher percentage of primiparas among those enrolled
176
Hediger
et al
Fetal
Growth
and Delioery
for later ultrasound scans (61.4 versus 49.9% primiparas, P < .OOl). The gravidas were interviewed, and anthropometric data, substance use details, and sociodemographic information were obtained at entry to prenatal care and up dated at 28-week and 36-week prenatal examinations. Maternal weight (kg) was measured at each prenatal visit on a beam balance scale, and height (cm) was measured at entry to care using standard anthropometric techniques.” Pregravid weight was obtained by recall at entry to prenatal care. Several studies’-%I5 have addressed the issue of the reliability of a self-reported weight in adolescents and adults and found that the weight reported is sufficiently reliable (Y > 0.90) for research purposes, except for a tendency for very overweight individuals to underestimate their true weight. From the recalled pregravid weight and measured height, we computed pregravid BMI as weight (in kg) divided by height squared (m*). Categories of low pregravid BMI or underweight (BMI less than 19.8) and overweight (BMI greater than 26.0) were based on the BMI criteria proposed by the Institute of Medicine.‘” Adequacy of total gestational weight gain was determined using a published schedule of weight gain for gestational age.” For example, using this schedule, total gestational weight gains less than 7.4 kg for a delivery at 32 weeks’ gestation, 9.0 kg at 36 weeks, 9.7 kg at 38 weeks, and 9.9 kg at 40 weeks were all classified as inadequate. Information on current and past pregnancy outcomes and complications of pregnancy were abstracted from the prenatal record, delivery record, delivery room logbooks, and the infants’ charts. Length of gestation was estimated from the mother’s last menstrual period (LMP) and confirmed or modified by a routine first- or second-trimester ultrasound examination, which was usually performed within a week of entry to prenatal care. For initial ultrasound scans to confirm or establish dates at or before 12 weeks’ gestation, crown-rump length was measured; BPD measurements were the basis for dating for second-trimester ultrasound examinations. If the crown-rump length measurement1a*i9 was within 7 days of menstrual age or the BPD measurement*” was within 10 days of menstrual age, then the estimated date of delivery was based on the LMP. If these limits were exceeded, then the estimated date of delivery was based on the crown-rump length or BPD. Eighty gravidas (27.6%) had their initial ultrasound scans at or before 12 weeks’ gestation; of these, 27.5% (22 of 80) had their estimated dates of delivery modified based on the initial ultrasound. Of the remainder with second-trimester ultrasound examinations, 32.9% (52 of 158) had their estimated dates of delivery modified based on ultrasound scans between 13-20 weeks’ ges-
Obstetrics
& Gynecology
tation, and 32.7% (17 of 52) had modifications when measured after 20 weeks’ gestation. There was no statistically significant difference (,$ = 0.77, P = .68) in the percentage of dates modified based on the timing of the initial ultrasound. The pregnancy outcomes we examined included two estimates of preterm delivery (before 37 completed weeks’ gestation): preterm based on the ultrasoundmodified LMP and preterm from the LMP unmodified by ultrasound. Low birth weight (LBW) was defined as less than 2500 g, and small for gestational age (SCA) status was defined as below the tenth percentile for standards. Small for gestational age infants were categorized using birth weight for gestational age standards, adjusting birth weight for black ethnicity, maternal parity, and fetal sex.*l Preterm deliveries, based on the ultrasound-modified LMP, were also categorized by etiologic pathway.* Medical indications included those delivered for complications that could potentially compromise maternalfetal outcome (eg, placental abruption, severe preeclampsia, nonreassuring fetal heart rate patterns, and a diagnosis of chorioamnionitis, oligohydramnios, or FGR). Premature rupture of membranes was defined as rupture of the chorioamniotic membranes before onset of labor, regardless of gestational age. Preterm labor was diagnosed based on the persistence of uterine contractions (four in 20 minutes or eight in 60 minutes), with either documented cervical changes or effacement of 80% or dilatation of 2 cm or more.22 Tocolysis to arrest preterm labor was attempted, except when cervical dilatation exceeded 4 cm or there were other relative contraindications for tocolysis, such as rupture of membranes. Usual treatment for preterm labor included hospitalization, intravenous hydration, and administration of magnesium sulfate to arrest labor, then maintenance on oral terbutaline. Tocolytic therapy was discontinued at 36-37 weeks’ gestation, and labor was allowed to proceed to delivery any time after termination of therapy. The obstetric ultrasound scan to measure fetal growth was scheduled for 32 weeks’ gestation based on the ultrasound-modified LMP. Fetal growth indices were measured by ultrasound with the Acuson 128XPlO (Acuson Corporation, Mountain View, CA) or General Electric Model RT3200S (General Electric Medical Systems, Milwaukee, WI). Following the guidelines of the American Institute of Ultrasound in Medicine for thirdtrimester sonography,23 the measurements obtained consisted of head circumference (HC), abdominal circumference (AC), BPD, and femur length (FL). Femur length was measured as the length of the femoral diaphysis. From the measurements of HC and AC, when available, we derived the HC-AC ratio, which is
VOL.
85, NO.
2, FEBRUARY
1995
used primarily to indicate asymmetric FGR.24 Estimated fetal weight (EFW) was calculated using the equation of Shepard et alz5: log (EFW) = -1.7492 + 0.166 (BPD) + 0.046 (AC) - 2.646 (AC X BPD)/lOOO. We performed several different kinds of statistical analyses on the collected data. Simple univariate comparisons between groups based on timing of delivery (preterm compared with term) were calculated by t test for continuous variables (age, pregravid BMI, maternal height, weeks of gestation at ultrasound) and by 2 or Fisher exact test for percentages (ethnicity, Medicaid status, smoking, inadequate weight gain, and pregnancy outcomes). The individual fetal measurements (HC, AC, BPD, and FL), HC-AC ratio, and EFW were analyzed in two ways. The biometric indices and the rates of growth (mm/week) from the initial or subsequent ultrasound scan (16-28 weeks’ gestation) to the 32-week scan were compared between the preterm and term groups using analysis of covariance, adjusting the measurements or rates for factors known to affect fetal growth (gestational age at ultrasound, maternal age, black ethnicity, low pregravid BMI, maternal height, tocolysis for preterm labor, parity, the number of cigarettes smoked per day, and fetal sex).2h The results of the analyses of covariance are presented as adjusted least square means + standard error of the mean. Multiple regression analyses predicting the individual fetal growth characteristics were constructed by using both continuous variables (gestational age at ultrasound, maternal age, maternal height, and number of cigarettes smoked per day) and indicator variables (preterm delivery, cause of preterm delivery, black ethnicity, low pregravid BMI, tocolysis for preterm labor, parity, and fetal sex). We did this to allow direct estimation of the magnitude of the indicator effects (such as preterm delivery and cause of preterm delivery) on the growth characteristics (in millimeters) relative to the controls (infants born at term) and model interactions.2h We performed a separate analysis using each growth characteristic (HC, AC, HC-AC ratio, BPD, FL, and EFW) as the dependent variable, and the entire sample was used in all analyses. The results from the multiple regression analyses are presented as the regression coefficient -t standard error associated with preterm delivery (preterm in relation to term controls) or the cause of preterm delivery (separate indicator variables for each cause in a combined model in relation to term controls). Regression coefficients for preterm delivery were considered significant if they differed from zero at an alpha level of .05, using a two-tailed test. Because the expectation based on the overall regressions was for diminished growth with preterm
Hediger
et al
Fetal GYCJZJ.#-I and Delivery
177
Table
1. Sample
Characteristics
by Timing
of Delivery
Preterm No. of patients Age (y) Body mass index (kg/& Height (cm) Initial ultrasound (wk) 32.week ultrasound (wk) Ethnicity (%) her& Rican Black White Medicaid recipient (% ) Primipara (%I Smokers (%) Inadequate weight gain (%I
Table
46 (15.9%) 17.1 5 2.3 22.9 2 5.4 160.7 2 6.6 16.7 -t 5.3 32.0 + 1.1
244 (84.1%) 17.5 2 2.4 23.9 k 5.5 161.1 k 6.1 16.0 t 5.0 32.0 -t 0.8
43.5 50.0 6.5 97.8 63.0 21.7 26.1
38.5 53.5 7.0 95.9 60.7 20.5 20.9
Data are presented as N, mean 2 standard deviation, or % There are no significant differences for any of the variables between the two groups.
delivery, subgroup
we used a one-tailed test of significance analyses (cause of preterm delivery).
for the
Results Forty-six of the 290 gravidas (15.9%) delivered preterm (Table 1). The average age of the cohort was young (17.4 t 2.4 years); 22.8% were between 13-15 years old at entry to care, 50.0% were between 16-18, and 27.2% were between 19-29. There were no differences in average gestational age at entry, pregravid BMI (kg/m2), and maternal height between gravidas delivering preterm and those delivering at term. There were no differences in the gestational age when the initial ultrasound examination was performed, nor in the timing of the 32-week ultrasound. Reflecting the clinic catchment area, the sample primarily represented minority women (Table 1). Overall, 39.3% were Hispanic (Puerto Rican), 53.8% black, and 6.9% white. There were no significant differences in the percentages of Medicaid recipients, primiparas, smokers, or inadequate weight gain between those delivering preterm and those delivering at term. The percentages of delivery indications and adverse outcomes differed between the preterm and term groups (Table 2). Well over half (31 of 46, 67.4%) of the preterm deliveries were preceded at some time by preterm labor; only 11.5% of the term deliveries were complicated by preterm labor. Low birth weight (less than 2500 g) was more common among infants who were delivered preterm, and SGA was about three times more likely among preterm infants. The growth measurements determined by ultrasound at 32 weeks’ gestation by the timing of delivery showed that all were significantly reduced with preterm deliv-
178
Hediger
et al
Fetal
Growth
and Delivery
Outcomes
2. Pregnancy
Term
by Timing
of Delivery Term
Preterm No. of patients Time to delivery from 32.mreek ultrasound (wk) Gestation (ruk) Birth weight (g) Male (%I Spontaneous preterm labor (%) Premature rupture of membranes (‘~1 LOXV birth tveight (%) Small for gestational age (W)
P
46 (15.9%) 3.1 t 1.4
244 (84.1%) 7.0 2 1.4
34.9 2 1.3 2401.4 ? 503.6 60.9 67.4
39.0 2 1.3 3223.0 i 461.2 45.1 11.5
.049 1.0001
39.1
3.3
< .OOOl
52.2 13.0
5.7 4.5
i .OOOl ,036
Data are presented as mean 2 standard deviation or %. Comparisons between percentages are by ,$ analysis or Fisher exact test.
ery compared with those who delivered at term (Table 3). However, the HC-AC ratio was not elevated, and was at the mean for the norm (1.06) at 32 weeks‘ gestation for both groups.21 The discrepancy in fetal growth, attributable to preterm delivery, was determined by regression analyses at 32 weeks’ gestation using both the ultrasoundmodified estimate of gestation (LMP confirmed or modified by initial ultrasound) and gestation estimated from the LMP unmodified by ultrasound (Table 4). There was good concordance using the two methods; 93.5% of the deliveries were classified identically as preterm or term. In nine cases (3.1%), a delivery was preterm from the ultrasound-modified estimate and term from the LMP, and in ten cases (3.4%), the reverse was true. Compared with the term infants, fetuses later delivered preterm, based on the ultrasound-modified LMP, were already significantly smaller in all dimensions at 32 weeks’ gestation. A slightly higher percentage of infants
Table
3. Growth Timing
Characteristics of Delivery
289.6 272.8 1.07 80.0 60.6 1833.7
? +? 2 t ?
Gestation
by
Term
Preterm HC (mm) AC (mm) HC-AC ratio BPD (mm) FL (mm) EFW (g)
at 32 Weeks’
1.8 2.5 0.008 0.5 0.4 41.8
294.8 279.3 1.06 81.6 61.7 1954.7
k 2 2 2 k 2
P 0.8 1.0 0.004 0.2 0.2 17.8
,008 ,016 NS ,006 ,013 ,009
HC = head circumference; AC = abdominal circumference; NS = not significant; BPD = biparietal diameter; FL = femur length; EFW = estimated fetal weight. Data are presented as least square means ? stand error of the mean from models adjusting for gestational age at ultrasound, maternal age, black ethnicity, low’ pregravid body mass index, maternal height, tocolysis for preterm labor, parity, cigarettes smoked per day, and fetal sex.
Obstetrics
G Gynecology
were preterm from the LMP unmodified by ultrasound examination. Although the discrepancy in fetal growth with preterm delivery from the LMP unmodified by ultrasound was smaller because of misclassification (clearly term infants were reassigned to the preterm group), the same pattern of results was found. Head circumference and BPD were significantly smaller (P < .05). Furthermore, the exclusion of the measurements from the 17 fetuses whose estimated dates of delivery may have been biased because they were modified from the LMP based on initial ultrasound scans obtained after 20 weeks’ gestation did not change the finding of reduced fetal size with later preterm delivery. Head circumference (-5.28 ? 2.15 mm, P = .014), AC (-6.17 -C 2.78 mm, P = .027), BPD (-1.63 -C 0.63 mm, P = .Oll), and FL (-1.15 -t 0.45 mm, P = .Oll) were all still significantly smaller among those fetuses later delivering preterm. A total of 54.5% (158 of 290) of the sample had ultrasonography performed between 16-28 weeks’ gestation, either to establish dates (136 of 158,86.1%) or for diagnostic purposes after an earlier ultrasound (22 of 158,13.9%). The sample with ultrasound scans between 16-28 weeks was representative of the total sample. There were no differences in background characteristics (maternal age, pregravid BMI, ethnicity, parity, smoking, and Medicaid status) between those who had an ultrasound examination at 16-28 weeks and those who only had an ultrasound to confirm dates before 16 weeks. Between the ultrasound examinations at 16-28 weeks and later at 32 weeks, there were no differences in the length of the interval to indicate whether the infant was subsequently born preterm or at term (Table 5). There were no differences in AC, BPD, or FL between the two groups, although there was a tendency for FL to be
Table
4.
Regression Coefficients Preterm Delivery
at 32 Weeks’ Gestation Estimate
Ultrasound-modified LMP No. of preterm HC (mm)
deliveries
46 (15.9%)
of gestational LMP
for
age unmodified
by ultrasound
AC (mm) BI’D (mm)
~5.27 -6.51 ~1.63
k 1.99* -c 2.69’ k 0.59’
47 (16.2%) -4.28 2 1.96” ~0.68 2 2.67 -1.15 i 0.58’
FL (mm)
-1.04
2 0.42’
~0.61
? 0.41
LMP = last menstrual period; all other abbreviations as in Table 3. Data are presented as p 2 standard error, adjusting for gestational age at ultrasound, maternal age, black ethnicity, low pregravid BMI, maternal height, tocolysis for preterm labor, parity, cigarettes smoked per day, and fetal sex. *P < .Ol. ’ Significantly different from zero at P < .05.
VOL.
85,
NO.
2, FEBRUARY
1995
Table
5.
Rates of Fetal Growth
bv Timing
of Delivery P
Preterm
Term
No. of patients
30 (19.0%)
128 (81.0%)
Initial
ultrasound
21.0 i 4.0
19.9 2 3.2
NS
(wk)* Interval (wk)* Initial ultrasound
11.1 2 4.4
12.1 i 3.3
NS
156.4 2 1.9
156.6 -t 0.9
NS
48.6 Z 0.5 32.9 2 0.5
48.4 t 0.3 33.7 t 0.2
NS NS
9.1 i 0.4 2.5 2 0.07
10.2 2 0.2 2.8 i 0.03
,008 .0004
2.2 i 0.06
2.3 i 0.03
characteristic AC BPD FL Rates of fetal (mm/wk)* AC
(mm)+
growth
BPD FL
.05
NS = not significant; all other abbreviations as in Table 3. * Data are presented as mean i standard deviation. Comparisons are by t test. ’ Data are presented as least square mean ? standard error of the mean, adjusting for gestational age at initial ultrasound, maternal age, black ethnicity, low pregravid body mass index, maternal height, parity, cigarettes smoked per day, and fetal sex. Instead of gestational age at initial ultrasound, the rates of fetal growth are adjusted for the interval between the initial and 32.week ultrasound examinations.
shorter among those later delivered preterm. However, for AC and BPD, the rates of fetal growth from 20-32 weeks (mm/week) were significantly lower (P < .Ol) for the preterm group, and the difference in the rates for FL was also significant (P = .05), indicating a generally slower rate of overall growth during this interval for infants later delivered preterm. Multiple regression analyses were used to estimate the effects of different causes of preterm delivery on individual characteristics. Of the 46 preterm deliveries, eight (17.4%) were delivered for medical or obstetric indications (two with placental abruption, one with severe preeclampsia, three with FGR and oligohydramnios, and two with chorioamnionitis), 16 (34.8%) delivered after PROM with no evidence of infection, and 22 (47.8%) after spontaneous preterm labor. This latter group was further divided into the 14 who delivered after failed or no tocolysis and the eight who delivered at 36 weeks’ gestation after earlier preterm labor (31-32 weeks’ gestation) had been treated successfully. At 32 weeks’ gestation, both the HC and FL of those who later delivered for medical or obstetric indications did not differ significantly from the term controls, but AC was significantly smaller (P < .Ol), and the HC-AC ratio was elevated (Table 6). This is the pattern seen typically with asymmetric or disproportional types of FGR (head-sparing). However, fetuses that were later delivered after PROM or spontaneous preterm labor and unsuccessful tocolysis were smaller in all dimensions, particularly HC and FL, and the HC-AC ratio was not elevated. This pattern implies a more proportional
Hediger
et al
Fetal
Growth
and Delivery
179
Table
6. Regression Coefficients for Preterm Delivery for Fetal Growth Parameters at 32 Weeks’ Gestation by Cause of Preterm Delivery
Medical/ obstetric indications (N = 8) HC (mm) AC (mm) HC-AC ratio BPD (mm) FL (mm) EFW (g)
Preterm labor failed tocolysis (N = 14)
Premature rupture (N = 16)
-2.06 2 4.31 -11.28 -15.21 ? 5.89* -7.42 0.06 t 0.02* -0.01
2 3.16* 2 4.31+ k 0.01
-2.80 + 1.28’ ~2.71 PO.30 2 0.91 -2.01 -259.5 5 99.5) ~168.9
% 0.93* ~1.64 + 0.67* -1.30 5 72.8+ ~137.3
Preterm labor after tocolysis (N = 8)
~6.30 i 3.30' 5.51 i 4.32 ~6.59 5 4.51 4.58 2 5.90 0.003 2 0.02 0.001 f 0.02 -+ 0.98’ 1.79 5 7.28 2 0.70+ 0.62 2 0.91 Z 76.2+ 152.5 t- 99.8
All abbreviations as in Table 3. Data are presented as 0 t standard error, adjusting for age at ultrasound, black ethnicity, low pregravid body maternal height, tocolysis for preterm labor, maternal cigarettes smoked per day, and fetal sex. *P < .Ol, one-tailed test. + Significantly different from zero at P < .05, one-tailed
gestational mass index, age, parity,
test.
or symmetric slowing of fetal growth (stunting). The neonates born preterm after successful tocolysis for earlier preterm labor did not differ from the controls in regard to their biometric characteristics.
Discussion Consistent with our previous study27 with similar young, primarily minority cohorts, the percentage of preterm deliveries was quite high (nearly 16%). We found that infants who were delivered preterm were already significantly smaller in all fetal growth dimensions by 32 weeks’ gestation. The diminished fetal growth was attributable to slower rates of growth from about 16-32 weeks’ gestation. Fetal growth was diminished with preterm delivery, either estimating length of gestation from the LMP confirmed or modified by an initial ultrasound scan or from the LMP unmodified by ultrasound when the dates were discordant. The findings are also consistent with and expand on the results from previous reports of fetal growth and preterm delivery in older maternal cohorts (approximately 25 years old). In one of the earliest studies using ultrasound, Tamura et al4 evaluated prospectively 148 fetuses at risk for preterm delivery, the majority of which (63%) were measured within 7 days of delivery. They found that an increased percentage had AC and BPD below the tenth percentile. This percentage was more than three times the expected amount. Two studies examined altered growth with preterm labor. Westgren et al6 examined the FL-AC ratio, which is presumed to be independent of gestational age, in 82 fetuses of gravidas hospitalized for preterm labor. Thirty-
180
Hediger
et al
Fetal
Growth
and Delivery
nine infants delivered preterm after failed tocolysis, whereas 43 went on to deliver at term. They found that the FL-AC ratio was significantly increased among those with failed tocolysis, implying that the growth of the fetuses delivering preterm was already suboptimal at the time when preterm labor was diagnosed. In a similar study, MacGregor et al’ performed ultrasound scans on admission for preterm labor in 78 pregnancies. They found that at 31 weeks’ gestation, significant percentages of nearly all growth measurements (BPD, HC, AC, and FL) were below the 25th and tenth percentiles (compared with standards) among the 48 who delivered before 36 weeks’ gestation compared with the 30 who delivered after 36 completed weeks, their cutoff for preterm delivery. Weiner et al5 used extant charts to estimate fetal weights from ultrasound scans and compared EFW at various gestational ages to published birth weight for gestational age norms.” They found that the birth weights of preterm infants were significantly smaller than the EFW of same-age infants who delivered at term. Secher et al7 and Ott3 drew the same conclusion based on similarly designed studies, comparing EFW derived from ultrasound scans to birth weights or sonographically derived standards. We know of no previous studies that have considered the type of FGR in relation to the cause of preterm delivery. In our sample, 17.4% of the infants were delivered principally for medical or obstetric indications, 34.8% for PROM, and 47.8% for preterm labor. These percentages are similar to those reported in studies including minority women.2,28,29 However, we recognize that the causes of preterm delivery may not be mutually exclusive in many individual cases and, likewise, the type of FGR may be relative. Nevertheless, when stratified by principal cause of preterm delivery, infants delivered preterm for medical or obstetric indications tended to have asymmetric FGR at 32 weeks’ gestation (ie, only AC was smaller). This implies a growth failure at a time late in the second and into the third trimester, when the growth velocities for most bone dimensions (eg, HC, BPD, and FL) are decreasing and fetal weight is still increasing.30J31 Severe preeclampsia, chorioamnionitis, and other medical or obstetric complications, which may result in reduced uterine blood flow and placental perfusion with onset late in pregnancy, seem to be associated with a relatively asymmetric FGR as early as 32 weeks‘ gestation. On the other hand, infants delivered preterm after PROM or after failed or no tocolysis for spontaneous preterm labor tended to be proportionately or symmetrically smaller. Both bone (HC, BPD, and FL) and soft-tissue dimensions (AC) were diminished relative to the controls, which suggests a different pattern of
Obstetrics
t3 Gynecology
altered fetal growth (ie, an overall slowing of growth that may originate earlier in pregnancy or possibly reflect a more chronic stress). Based on EFW at 32 weeks’ gestation, the infants later born preterm with PROM or spontaneous preterm labor were not as severely growth restricted as those delivered with medical or obstetric indications, but their growth was suboptimal relative to the controls. However, because our study design precluded very early preterm deliveries, it is possible that the most severely growth restricted fetuses are born preterm with PROM or spontaneous preterm labor at very early gestational ages (less than 33 completed weeks). Factors associated with slowed fetal growth might also be associated with preterm labor or PROM. One such factor may be poor maternal nutrition (ie, diets low in energy and/or essential vitamins and minerals). Early poor nutrition may affect placental development and vascularization, thereby restricting fetal growth by affecting the integrity of the chorioamniotic membranes (leading to PROM)“’ or by increasing susceptibility to infection. We have found in previous observational studies”““4 of both young and older gravidas that low intakes of iron and zinc are related to an increased risk of preterm delivery. For zinc, the risk of preterm delivery with low dietary intake was particularly strong (threefold increased risk) for those whose rupture of membranes preceded labor.“4 Slowed fetal growth in association with preterm delivery is also consistent with findings among both adolescent and adult gravidas that the risk of preterm delivery is increased with low rates of gestational weight gain in the last half of pregnan3537 The low rates of gestational gain may reflect a CY. slowed fetal growth and thus be a marker for impending preterm birth. These results suggest that current birth weight for gestational age standards and fetal growth curves based on the birth weights of delivered infants probably underestimate the growth of infants who go to term and underestimate the proportion of preterm infants who are actually growth restricted. The factors culminating in spontaneous preterm labor or PROM appear to be associated with a proportional slowing of fetal growth; although not necessarily as clinically severe as late asymmetric restriction, this is nevertheless characterized by slower rates of growth. This discrepancy in growth, associated with spontaneous preterm labor or PROM, is already significant by the third trimester of pregnancy. Consequently, research into the etiology of preterm delivery and interventions designed to prevent preterm delivery, especially that resulting from spontaneous preterm labor or PROM, may have to focus on factors (such as periconceptional nutrition) that can be modified early in pregnancy. Interventions3’ aimed
VOL.
85, NO.
2, FEBRUARY
1995
later in pregnancy may affect only those pregnancies in which there has not already been an early insult characterized by a long-term slowing of fetal growth, accounting for ineffectiveness in decreasing the incidence of preterm birth.
References 1
Berkowitz GS, Papiernik miol Rev 1993;15:414&43.
2
Savitz DA, Blackmore CA, Thorp JM. Epidemiologic tics of preterm delivery: Etiologic heterogeneity.
characterisAm J Obstet
Gynecol 1991;164:467-71. Ott WJ. Intrauterine growth retardation J Obstet Gynecol 1993;168:1710-7.
delivery.
3 4
5
E. Epidemiology
Tamura RK, Sabbagha RE, Depp ML. Diminished growth in fetuses labor or rupture of membranes.
of preterm
birth.
and preterm
Depp R. A hypothetical growth in infants deliv-
ered preterm. Obstet Gynecol 1985;65:323-6. 6. Westgren M, Beall M, Divan M, Platt L. Fetal femur abdominal circumference ratio in preterm labor patients
7
successful NJ,
retardation 20. 8. MacGregor
9
Kern
tocolytic
therapy.
Hansen
in preterm
J Ultrasound
P, Thomsen
infants.
SK, Sabbagha
Am
R, Vaisrub N, Dooley SL, Socol born preterm after spontaneous Am J Obstet Gynecol 1984;148:
1105-10. Weiner CP, Sabbagha RE, Vaisrub N, model suggesting suboptimal intrauterine
without 243-5. Secher
Epide-
BL,
Br J Obstet
RE, Tamura
Med
Keiding Gynaecol
RK, Pielet
length/ with and
N.
1986;5: Growth
1987;Y4:115-
SN, Feigenbaum
SL. Differing fetal growth patterns in pregnancies preterm labor. Obstet Gynecol 1988;72:834-7. Wolfson RN, Zador IE, Halvorsen P, Andrews
complicated B, Sokol
by RJ.
Biparietal diameter in premature rupture of membranes: Errors in estimating gestational age. JCU 1983;11:371-4. CR, Goldenberg RL, Hoffman HJ, Cliver SP, 10. Davis RO, Cutter Brumfield CG. Fetal biparietal, head circumference, abdominal circumference and femur length: A comparison by race and sex. J Reprod Med 1993;38:201-6. 11. Kramer MS, Olivier M, McLean FH, Dougherty Usher RH. Determinants of fetal growth and body Pediatrics 12. Cameron F, Tanner Plenum 13. Stewart
GE, Wilhs DM, proportionality.
1990;86:18-26.
N. The methods JM,
eds.
Human
of auxological anthropology. growth. 3. Methodology.
In: Falkner New, York:
and validity
weight
Press, 1986:3-46. AL. The reliability
of self-reported
and
height. J Chron Dis 1982;35:295-309. 14. Palta M, Prineas RJ, Berman R, Hannan I’. Comparison of selfreported and measured height and weight. Am J Epidemiol 1982;115:223-30. 15. Stevens-Simon C, McAnarney pregnant adolescents estimate Adolesc 16. Institute Weight
ER, Coulter MI’. How accurately their weight prior to pregnancy?
Health Care 1986;7:250&4. of Medicine. Subcommittee on Nutritional Status Gain during Pregnancy. Nutrition during pregnancy.
I. Weight gain. Washington, DC: National Academy 17. Butman M. Prenatal nutrition: A clinical manual. Massachusetts Department of Health, 1982:26.
do J and Part
Press, 1990. Boston, MA:
18. Robinson HP, Fleming JEE. A critical evaluation of sonar “crownrump length” measurements. Br J Obstet Gynaecol 1975;82:702-10. 19. Daya S. Accuracy of gestational age estimation by means of fetal crown-rump pt 1):903-8.
length
Hediger
measurement.
et al
Am J Obstet
Fetal
Growth
Gynecol1993;168(3
and Delizwy
181
20. Kurtz AB, Wapner RJ, Kurt2 RJ, et al. Analysis of biparietal diameter as an accurate indicator ot gestational age. JCU 1980;8: 31Y -26. 21. Brenner WE, Edelman DA, Hendricks CH. A standard of fetal growth for the United States of America. Am J Obstet Gynecol 1976;126:555-64 22. Greasy RK. Preterm labor and delivery. In: Greasy RK, Resnik R, eds. Maternal-fetal medicine: Principles and practice. Philadelphia: W. B. Saunders Company, 1994:494-520. 23. American Institute of Ultrasound in Medicine. Guidelines for performance of the antepartum obstetrical ultrasound examination. J Ultrasound Med 1991;10:577-8. 24. Campbell S, Thorns A. Ultrasound measurement of the fetal head to abdomen circumference in the assessment of growth retardation. Br J Obstet Gynaecol 1977;84:765-74. VA, Berkowitz RL, Warsof SL, Hobbins JC. 25. Shepard MJ, Richards An evaluation of two equations for predicting fetal weight by ultrasound. Am J Obstet Gynecol 1982;142:47-54. 26. Kleinbaum DG, Kupper LL, Muller KE. Applied regression analysis and other multivariable methods. Boston: PWS-KENT I’ublishing Company, 1988. 27. Scholl TO, llediger ML, Huang J, Johnson FE, Smith W, Antes IG. Young maternal age and parity: Influences on pregnancy outcome. Ann Epidemiol lYY2;2:565-75. 28. Main DM, Gabbe SG, Richardson D, Strong S. Can preterm deliveries be prevented? Am J Obstet Gynecol 1985;151:892-8. 29. Meis PJ, Ernest JM, Moore ML. Causes of low birth weight births in public and private patients. Am J Obstet Gynecol lY87;156: 1165-8. 30. Guihard-Costa AM, Droulle I’, Larroche JC. Growth velocity of the biparietal diameter, abdominal transverse diameter and femur length in the fetal period. Early Hum Dev 1991;27:93-102. 31. Deter RL, Harrist RB. Growth standards for anatomic measurements and growth rates derived from longitudinal studies of normal fetal growth. JCU 1992;20:381-8.
182 Hediger
et al
Fetal
Grozvth
and
De/ivenj
32. Allen SR. Epidemiology of premature rupture of the fetal membranes. Clin Obstet Gynecol 1991;34:685-93. 33. Scholl TO, Hediger ML, Fischer RL, Shearer JW. Anemia vs iron deficiency: Increased risk of preterm delivery in a prospective study. Am J Clin Nutr 1992;55:985-8. 34. Scholl TO, Hediger ML, Schall JI, Fischer RL, Khoo CS. Low zinc intake during pregnancy: Its association with preterm and very preterm delivery. Am J Epidemiol 1993;137:1115-24. 35. Berkowitz GS. Clinical and obstetric risk factors for preterm delivery. Mt Sinai J Med 1985;52:239-47. 36. Hediger ML, Scholl TO, Belsky DH, Antes IG, Salmon RW. Patterns of weight gain in adolescent pregnancy: Effects on birth weight and preterm delivery. Obstet Gynecol 1989;74:6-12. 37. Abrams B, Newman V, Key T, Parker J, Maternal weight gain and preterm delivery. Obstet Gynecol 1989;74:577-83. 38. Collaborative Group on Preterm Birth Prevention. Multicenter randomized, controlled trial of a preterm birth prevention program. Am J Obstet Gynecol 1993;169:352-66.
Address
reprint
Mary
L. Hediger, Department of
requests to: PkD
Obstetrics rind Gynecology
UMDNJ-SOM 401 Haddon Avenue Camden, NJ 08103
Receiord Received Accepted
August 5, 1994. ill revised form October October 13, 1994.
Copyright 0 Gynecologists.
1995 by The
American
3, 1994.
College
of Obstetricians
and
Obstetrics 6 Gynecology
and the Etiology
of Preterm
J
MARY L. HEDIGER, PhD, THERESA 0. SCHOLL, PhD, JOAN I. SCHALL, PhD, LAURIE W. MILLER, BS, RDMS, AND RICHARD L. FISCHER, MD Objective:
To
confirm
with fetal growth various etiologies
that
restriction of preterm
preterm
same degree and type of FGR. Methods: Two hundred ninety gravidas who had routine initial also had subsequent ultrasound gestation. Fetal tween preterm deliveries, obstetric (PROM),
delivery
(FGR), delivery
is
associated
and to determine if the are associated with the young, primarily minority ultrasound examinations examinations at 32 weeks’
growth characteristics (less than 37 weeks’
were compared gestation) and
and among preterm deliveries indications, premature rupture and spontaneous preterm labor.
with of
beterm
medical membranes
or
Results: Forty-six infants (15.9%) were born preterm. At 32 weeks’ gestation, all fetuses later delivered preterm were already smaller than fetuses later delivered at term (P < .05) for all dimensions: circumference (AC), length (FL). However,
head circumference biparietal diameter after stratifying
(HC), (BPD), by cause
abdominal and femur of preterm
delivery for those fetuses later delivered for medical or obstetric indications, we found that only AC was decreased (P < .Ol) and that the HC-AC ratio was elevated (asymmetric FGR). Neonates delivered after unsuccessfully PROM or preterm labor were symmetrically smaller characteristics (HC, AC, BPD, and FL).
treated in all
Conclusion: By 32 weeks’ gestation, fetuses later delivered preterm are already significantly smaller than fetuses later delivered at term. However, when stratified by the etiology of preterm obstetric
delivery, indications
infants delivered had asymmetric
preterm growth
for medical or patterns, which
suggests a growth failure late in pregnancy. Infants delivered preterm after PROM or after failed or no tocolysis for spontaneous preterm labor were proportionately smaller, implying an overall slowing of growth that early in pregnancy and possibly demonstrate stress. (Obstet Gynecol 1995;85:175-82)
From
may originate a more chronic
the Departments of Obstetrics and Gynecology, The University and Dentistry of New Jersey-School of Osteopathic Medicine and Robert Wood Johnson Medical School, Camden, New Jersey. Supported by grant no. HD18269 fr om the National lnstitute of Child Health and Human Development.
of Medicine
VOL.
85,NO.Z,FEBRUARY
1995
Preterm delivery is a major contributor to neonatal and infant mortality and morbidity, but the etiology and risk factors for preterm delivery are not well established. Some of the risk factors associated with preterm delivery are urogenital infection, poor placentation or vasculopathy, multiple gestation, cigarette smoking, substance abuse, black ethnicity, and, possibly, poor diet and nutrition.’ However, identification of risk factors has been complicated by the different etiologic pathways culminating in preterm delivery: spontaneous preterm labor, preterm premature rupture of membranes (PROM), and numerous medical or obstetric indications and complications (eg, severe preeclampsia, placental abruption, or chorioamnionitis).‘,’ The risk factors that may be associated separately with each of the various pathways, especially idiopathic preterm labor and PROM, are also not well known. It has been postulated that, whatever the cause, preterm delivery is characterized by suboptimal fetal growth.” It follows, then, that birth weight for gestational age standards generated from birth weights at preterm gestations may underestimate the size of fetuses that will subsequently be delivered at term, thereby underestimating the number of preterm infants who are born with fetal growth restriction (FGR) and who are at risk for major morbidity. Even given the widespread use of obstetric ultrasonography during pregnancy, only a few studies3-s have examined the relation between preterm delivery and fetal growth. Although previous studies3-s have evaluated the association between preterm delivery and fetal growth, they pose a number of methodologic concerns. In several, growth measurements were determined by ultrasound scan either just preceding delivery., after PROM, or at admission for preterm labor. The problem with cranial measurements obtained after rupture of membranes is that fetuses may have artifactually decreased biparietal diameters (BPD) (dolichocephaly) and may not truly be growth restricted.5’9 Furthermore,
0029-7844/95/$9.50 0029-7844(94)00365-K
175
none of the studies included controls who had no risk factors for preterm delivery and who later delivered at term. In only one of the studies of preterm delivery and fetal growth’ was there any control for other factors known to affect fetal growth,“‘,” such as maternal age, parity, ethnicity, maternal pregravid body mass index (BMI), maternal height, cigarette smoking, or fetal sex. Although there are different proximate causes of preterm delivery, it is not known if each is associated with the same degree and type of FGR.” The objective of this study was to confirm, in a controlled prospective cohort study of primarily young, minority gravidas, that suboptimal fetal growth is characteristic of preterm delivery, and to determine if the various etiologies of preterm delivery are associated with the same degree and type of FGR.
Materials and Methods Thirty percent of gravidas enrolled from October 1990 through November 1993, in a larger, ongoing study of nutrition and growth in adolescent (younger than 19 years) and older gravidas (between 19-29 years), were also recruited to have a subsequent ultrasound examination at 32 weeks’ gestation. All subjects were initially enrolled at entry to prenatal care under the same protocol from an urban clinic in Camden, New Jersey. Gravidas with verified chronic or metabolic disease that could affect nutrition or fetal outcome-such as congenital heart disease, tuberculosis, chronic hypertension, insulin-dependent diabetes mellitus, hemoglobinopathies, cancer, lupus, seizures, drug or alcohol abuse, and psychiatric disorders-were not enrolled in the larger study. Every third subject (290 of 977) enrolled in the larger study during this time period was recruited within a month of entry to prenatal care to have a real-time and Doppler ultrasound scan for research purposes at 32 weeks’ gestation. Funding was available only to perform ultrasound scans at 32 weeks on this proportion of the larger cohort. At about 7 months’ gestation (28 weeks), the subjects identified previously were scheduled for obstetric ultrasound examinations at 32 weeks. Only the results of the real-time ultrasound scans for fetal growth are presented here. The protocol met the guidelines of the Institutional Review Board of the University of Medicine and Dentistry of New Jersey. There were no differences in ethnic distribution between those recruited for an ultrasound scan at 32 weeks’ gestation and the cohort as a whole. However, the 290 subjects enrolled for later ultrasound were slightly, but significantly, younger (17.4 I 2.4 versus 17.9 2 2.6 years, P < .Ol), and consequently there was a higher percentage of primiparas among those enrolled
176
Hediger
et al
Fetal
Growth
and Delioery
for later ultrasound scans (61.4 versus 49.9% primiparas, P < .OOl). The gravidas were interviewed, and anthropometric data, substance use details, and sociodemographic information were obtained at entry to prenatal care and up dated at 28-week and 36-week prenatal examinations. Maternal weight (kg) was measured at each prenatal visit on a beam balance scale, and height (cm) was measured at entry to care using standard anthropometric techniques.” Pregravid weight was obtained by recall at entry to prenatal care. Several studies’-%I5 have addressed the issue of the reliability of a self-reported weight in adolescents and adults and found that the weight reported is sufficiently reliable (Y > 0.90) for research purposes, except for a tendency for very overweight individuals to underestimate their true weight. From the recalled pregravid weight and measured height, we computed pregravid BMI as weight (in kg) divided by height squared (m*). Categories of low pregravid BMI or underweight (BMI less than 19.8) and overweight (BMI greater than 26.0) were based on the BMI criteria proposed by the Institute of Medicine.‘” Adequacy of total gestational weight gain was determined using a published schedule of weight gain for gestational age.” For example, using this schedule, total gestational weight gains less than 7.4 kg for a delivery at 32 weeks’ gestation, 9.0 kg at 36 weeks, 9.7 kg at 38 weeks, and 9.9 kg at 40 weeks were all classified as inadequate. Information on current and past pregnancy outcomes and complications of pregnancy were abstracted from the prenatal record, delivery record, delivery room logbooks, and the infants’ charts. Length of gestation was estimated from the mother’s last menstrual period (LMP) and confirmed or modified by a routine first- or second-trimester ultrasound examination, which was usually performed within a week of entry to prenatal care. For initial ultrasound scans to confirm or establish dates at or before 12 weeks’ gestation, crown-rump length was measured; BPD measurements were the basis for dating for second-trimester ultrasound examinations. If the crown-rump length measurement1a*i9 was within 7 days of menstrual age or the BPD measurement*” was within 10 days of menstrual age, then the estimated date of delivery was based on the LMP. If these limits were exceeded, then the estimated date of delivery was based on the crown-rump length or BPD. Eighty gravidas (27.6%) had their initial ultrasound scans at or before 12 weeks’ gestation; of these, 27.5% (22 of 80) had their estimated dates of delivery modified based on the initial ultrasound. Of the remainder with second-trimester ultrasound examinations, 32.9% (52 of 158) had their estimated dates of delivery modified based on ultrasound scans between 13-20 weeks’ ges-
Obstetrics
& Gynecology
tation, and 32.7% (17 of 52) had modifications when measured after 20 weeks’ gestation. There was no statistically significant difference (,$ = 0.77, P = .68) in the percentage of dates modified based on the timing of the initial ultrasound. The pregnancy outcomes we examined included two estimates of preterm delivery (before 37 completed weeks’ gestation): preterm based on the ultrasoundmodified LMP and preterm from the LMP unmodified by ultrasound. Low birth weight (LBW) was defined as less than 2500 g, and small for gestational age (SCA) status was defined as below the tenth percentile for standards. Small for gestational age infants were categorized using birth weight for gestational age standards, adjusting birth weight for black ethnicity, maternal parity, and fetal sex.*l Preterm deliveries, based on the ultrasound-modified LMP, were also categorized by etiologic pathway.* Medical indications included those delivered for complications that could potentially compromise maternalfetal outcome (eg, placental abruption, severe preeclampsia, nonreassuring fetal heart rate patterns, and a diagnosis of chorioamnionitis, oligohydramnios, or FGR). Premature rupture of membranes was defined as rupture of the chorioamniotic membranes before onset of labor, regardless of gestational age. Preterm labor was diagnosed based on the persistence of uterine contractions (four in 20 minutes or eight in 60 minutes), with either documented cervical changes or effacement of 80% or dilatation of 2 cm or more.22 Tocolysis to arrest preterm labor was attempted, except when cervical dilatation exceeded 4 cm or there were other relative contraindications for tocolysis, such as rupture of membranes. Usual treatment for preterm labor included hospitalization, intravenous hydration, and administration of magnesium sulfate to arrest labor, then maintenance on oral terbutaline. Tocolytic therapy was discontinued at 36-37 weeks’ gestation, and labor was allowed to proceed to delivery any time after termination of therapy. The obstetric ultrasound scan to measure fetal growth was scheduled for 32 weeks’ gestation based on the ultrasound-modified LMP. Fetal growth indices were measured by ultrasound with the Acuson 128XPlO (Acuson Corporation, Mountain View, CA) or General Electric Model RT3200S (General Electric Medical Systems, Milwaukee, WI). Following the guidelines of the American Institute of Ultrasound in Medicine for thirdtrimester sonography,23 the measurements obtained consisted of head circumference (HC), abdominal circumference (AC), BPD, and femur length (FL). Femur length was measured as the length of the femoral diaphysis. From the measurements of HC and AC, when available, we derived the HC-AC ratio, which is
VOL.
85, NO.
2, FEBRUARY
1995
used primarily to indicate asymmetric FGR.24 Estimated fetal weight (EFW) was calculated using the equation of Shepard et alz5: log (EFW) = -1.7492 + 0.166 (BPD) + 0.046 (AC) - 2.646 (AC X BPD)/lOOO. We performed several different kinds of statistical analyses on the collected data. Simple univariate comparisons between groups based on timing of delivery (preterm compared with term) were calculated by t test for continuous variables (age, pregravid BMI, maternal height, weeks of gestation at ultrasound) and by 2 or Fisher exact test for percentages (ethnicity, Medicaid status, smoking, inadequate weight gain, and pregnancy outcomes). The individual fetal measurements (HC, AC, BPD, and FL), HC-AC ratio, and EFW were analyzed in two ways. The biometric indices and the rates of growth (mm/week) from the initial or subsequent ultrasound scan (16-28 weeks’ gestation) to the 32-week scan were compared between the preterm and term groups using analysis of covariance, adjusting the measurements or rates for factors known to affect fetal growth (gestational age at ultrasound, maternal age, black ethnicity, low pregravid BMI, maternal height, tocolysis for preterm labor, parity, the number of cigarettes smoked per day, and fetal sex).2h The results of the analyses of covariance are presented as adjusted least square means + standard error of the mean. Multiple regression analyses predicting the individual fetal growth characteristics were constructed by using both continuous variables (gestational age at ultrasound, maternal age, maternal height, and number of cigarettes smoked per day) and indicator variables (preterm delivery, cause of preterm delivery, black ethnicity, low pregravid BMI, tocolysis for preterm labor, parity, and fetal sex). We did this to allow direct estimation of the magnitude of the indicator effects (such as preterm delivery and cause of preterm delivery) on the growth characteristics (in millimeters) relative to the controls (infants born at term) and model interactions.2h We performed a separate analysis using each growth characteristic (HC, AC, HC-AC ratio, BPD, FL, and EFW) as the dependent variable, and the entire sample was used in all analyses. The results from the multiple regression analyses are presented as the regression coefficient -t standard error associated with preterm delivery (preterm in relation to term controls) or the cause of preterm delivery (separate indicator variables for each cause in a combined model in relation to term controls). Regression coefficients for preterm delivery were considered significant if they differed from zero at an alpha level of .05, using a two-tailed test. Because the expectation based on the overall regressions was for diminished growth with preterm
Hediger
et al
Fetal GYCJZJ.#-I and Delivery
177
Table
1. Sample
Characteristics
by Timing
of Delivery
Preterm No. of patients Age (y) Body mass index (kg/& Height (cm) Initial ultrasound (wk) 32.week ultrasound (wk) Ethnicity (%) her& Rican Black White Medicaid recipient (% ) Primipara (%I Smokers (%) Inadequate weight gain (%I
Table
46 (15.9%) 17.1 5 2.3 22.9 2 5.4 160.7 2 6.6 16.7 -t 5.3 32.0 + 1.1
244 (84.1%) 17.5 2 2.4 23.9 k 5.5 161.1 k 6.1 16.0 t 5.0 32.0 -t 0.8
43.5 50.0 6.5 97.8 63.0 21.7 26.1
38.5 53.5 7.0 95.9 60.7 20.5 20.9
Data are presented as N, mean 2 standard deviation, or % There are no significant differences for any of the variables between the two groups.
delivery, subgroup
we used a one-tailed test of significance analyses (cause of preterm delivery).
for the
Results Forty-six of the 290 gravidas (15.9%) delivered preterm (Table 1). The average age of the cohort was young (17.4 t 2.4 years); 22.8% were between 13-15 years old at entry to care, 50.0% were between 16-18, and 27.2% were between 19-29. There were no differences in average gestational age at entry, pregravid BMI (kg/m2), and maternal height between gravidas delivering preterm and those delivering at term. There were no differences in the gestational age when the initial ultrasound examination was performed, nor in the timing of the 32-week ultrasound. Reflecting the clinic catchment area, the sample primarily represented minority women (Table 1). Overall, 39.3% were Hispanic (Puerto Rican), 53.8% black, and 6.9% white. There were no significant differences in the percentages of Medicaid recipients, primiparas, smokers, or inadequate weight gain between those delivering preterm and those delivering at term. The percentages of delivery indications and adverse outcomes differed between the preterm and term groups (Table 2). Well over half (31 of 46, 67.4%) of the preterm deliveries were preceded at some time by preterm labor; only 11.5% of the term deliveries were complicated by preterm labor. Low birth weight (less than 2500 g) was more common among infants who were delivered preterm, and SGA was about three times more likely among preterm infants. The growth measurements determined by ultrasound at 32 weeks’ gestation by the timing of delivery showed that all were significantly reduced with preterm deliv-
178
Hediger
et al
Fetal
Growth
and Delivery
Outcomes
2. Pregnancy
Term
by Timing
of Delivery Term
Preterm No. of patients Time to delivery from 32.mreek ultrasound (wk) Gestation (ruk) Birth weight (g) Male (%I Spontaneous preterm labor (%) Premature rupture of membranes (‘~1 LOXV birth tveight (%) Small for gestational age (W)
P
46 (15.9%) 3.1 t 1.4
244 (84.1%) 7.0 2 1.4
34.9 2 1.3 2401.4 ? 503.6 60.9 67.4
39.0 2 1.3 3223.0 i 461.2 45.1 11.5
.049 1.0001
39.1
3.3
< .OOOl
52.2 13.0
5.7 4.5
i .OOOl ,036
Data are presented as mean 2 standard deviation or %. Comparisons between percentages are by ,$ analysis or Fisher exact test.
ery compared with those who delivered at term (Table 3). However, the HC-AC ratio was not elevated, and was at the mean for the norm (1.06) at 32 weeks‘ gestation for both groups.21 The discrepancy in fetal growth, attributable to preterm delivery, was determined by regression analyses at 32 weeks’ gestation using both the ultrasoundmodified estimate of gestation (LMP confirmed or modified by initial ultrasound) and gestation estimated from the LMP unmodified by ultrasound (Table 4). There was good concordance using the two methods; 93.5% of the deliveries were classified identically as preterm or term. In nine cases (3.1%), a delivery was preterm from the ultrasound-modified estimate and term from the LMP, and in ten cases (3.4%), the reverse was true. Compared with the term infants, fetuses later delivered preterm, based on the ultrasound-modified LMP, were already significantly smaller in all dimensions at 32 weeks’ gestation. A slightly higher percentage of infants
Table
3. Growth Timing
Characteristics of Delivery
289.6 272.8 1.07 80.0 60.6 1833.7
? +? 2 t ?
Gestation
by
Term
Preterm HC (mm) AC (mm) HC-AC ratio BPD (mm) FL (mm) EFW (g)
at 32 Weeks’
1.8 2.5 0.008 0.5 0.4 41.8
294.8 279.3 1.06 81.6 61.7 1954.7
k 2 2 2 k 2
P 0.8 1.0 0.004 0.2 0.2 17.8
,008 ,016 NS ,006 ,013 ,009
HC = head circumference; AC = abdominal circumference; NS = not significant; BPD = biparietal diameter; FL = femur length; EFW = estimated fetal weight. Data are presented as least square means ? stand error of the mean from models adjusting for gestational age at ultrasound, maternal age, black ethnicity, low’ pregravid body mass index, maternal height, tocolysis for preterm labor, parity, cigarettes smoked per day, and fetal sex.
Obstetrics
G Gynecology
were preterm from the LMP unmodified by ultrasound examination. Although the discrepancy in fetal growth with preterm delivery from the LMP unmodified by ultrasound was smaller because of misclassification (clearly term infants were reassigned to the preterm group), the same pattern of results was found. Head circumference and BPD were significantly smaller (P < .05). Furthermore, the exclusion of the measurements from the 17 fetuses whose estimated dates of delivery may have been biased because they were modified from the LMP based on initial ultrasound scans obtained after 20 weeks’ gestation did not change the finding of reduced fetal size with later preterm delivery. Head circumference (-5.28 ? 2.15 mm, P = .014), AC (-6.17 -C 2.78 mm, P = .027), BPD (-1.63 -C 0.63 mm, P = .Oll), and FL (-1.15 -t 0.45 mm, P = .Oll) were all still significantly smaller among those fetuses later delivering preterm. A total of 54.5% (158 of 290) of the sample had ultrasonography performed between 16-28 weeks’ gestation, either to establish dates (136 of 158,86.1%) or for diagnostic purposes after an earlier ultrasound (22 of 158,13.9%). The sample with ultrasound scans between 16-28 weeks was representative of the total sample. There were no differences in background characteristics (maternal age, pregravid BMI, ethnicity, parity, smoking, and Medicaid status) between those who had an ultrasound examination at 16-28 weeks and those who only had an ultrasound to confirm dates before 16 weeks. Between the ultrasound examinations at 16-28 weeks and later at 32 weeks, there were no differences in the length of the interval to indicate whether the infant was subsequently born preterm or at term (Table 5). There were no differences in AC, BPD, or FL between the two groups, although there was a tendency for FL to be
Table
4.
Regression Coefficients Preterm Delivery
at 32 Weeks’ Gestation Estimate
Ultrasound-modified LMP No. of preterm HC (mm)
deliveries
46 (15.9%)
of gestational LMP
for
age unmodified
by ultrasound
AC (mm) BI’D (mm)
~5.27 -6.51 ~1.63
k 1.99* -c 2.69’ k 0.59’
47 (16.2%) -4.28 2 1.96” ~0.68 2 2.67 -1.15 i 0.58’
FL (mm)
-1.04
2 0.42’
~0.61
? 0.41
LMP = last menstrual period; all other abbreviations as in Table 3. Data are presented as p 2 standard error, adjusting for gestational age at ultrasound, maternal age, black ethnicity, low pregravid BMI, maternal height, tocolysis for preterm labor, parity, cigarettes smoked per day, and fetal sex. *P < .Ol. ’ Significantly different from zero at P < .05.
VOL.
85,
NO.
2, FEBRUARY
1995
Table
5.
Rates of Fetal Growth
bv Timing
of Delivery P
Preterm
Term
No. of patients
30 (19.0%)
128 (81.0%)
Initial
ultrasound
21.0 i 4.0
19.9 2 3.2
NS
(wk)* Interval (wk)* Initial ultrasound
11.1 2 4.4
12.1 i 3.3
NS
156.4 2 1.9
156.6 -t 0.9
NS
48.6 Z 0.5 32.9 2 0.5
48.4 t 0.3 33.7 t 0.2
NS NS
9.1 i 0.4 2.5 2 0.07
10.2 2 0.2 2.8 i 0.03
,008 .0004
2.2 i 0.06
2.3 i 0.03
characteristic AC BPD FL Rates of fetal (mm/wk)* AC
(mm)+
growth
BPD FL
.05
NS = not significant; all other abbreviations as in Table 3. * Data are presented as mean i standard deviation. Comparisons are by t test. ’ Data are presented as least square mean ? standard error of the mean, adjusting for gestational age at initial ultrasound, maternal age, black ethnicity, low pregravid body mass index, maternal height, parity, cigarettes smoked per day, and fetal sex. Instead of gestational age at initial ultrasound, the rates of fetal growth are adjusted for the interval between the initial and 32.week ultrasound examinations.
shorter among those later delivered preterm. However, for AC and BPD, the rates of fetal growth from 20-32 weeks (mm/week) were significantly lower (P < .Ol) for the preterm group, and the difference in the rates for FL was also significant (P = .05), indicating a generally slower rate of overall growth during this interval for infants later delivered preterm. Multiple regression analyses were used to estimate the effects of different causes of preterm delivery on individual characteristics. Of the 46 preterm deliveries, eight (17.4%) were delivered for medical or obstetric indications (two with placental abruption, one with severe preeclampsia, three with FGR and oligohydramnios, and two with chorioamnionitis), 16 (34.8%) delivered after PROM with no evidence of infection, and 22 (47.8%) after spontaneous preterm labor. This latter group was further divided into the 14 who delivered after failed or no tocolysis and the eight who delivered at 36 weeks’ gestation after earlier preterm labor (31-32 weeks’ gestation) had been treated successfully. At 32 weeks’ gestation, both the HC and FL of those who later delivered for medical or obstetric indications did not differ significantly from the term controls, but AC was significantly smaller (P < .Ol), and the HC-AC ratio was elevated (Table 6). This is the pattern seen typically with asymmetric or disproportional types of FGR (head-sparing). However, fetuses that were later delivered after PROM or spontaneous preterm labor and unsuccessful tocolysis were smaller in all dimensions, particularly HC and FL, and the HC-AC ratio was not elevated. This pattern implies a more proportional
Hediger
et al
Fetal
Growth
and Delivery
179
Table
6. Regression Coefficients for Preterm Delivery for Fetal Growth Parameters at 32 Weeks’ Gestation by Cause of Preterm Delivery
Medical/ obstetric indications (N = 8) HC (mm) AC (mm) HC-AC ratio BPD (mm) FL (mm) EFW (g)
Preterm labor failed tocolysis (N = 14)
Premature rupture (N = 16)
-2.06 2 4.31 -11.28 -15.21 ? 5.89* -7.42 0.06 t 0.02* -0.01
2 3.16* 2 4.31+ k 0.01
-2.80 + 1.28’ ~2.71 PO.30 2 0.91 -2.01 -259.5 5 99.5) ~168.9
% 0.93* ~1.64 + 0.67* -1.30 5 72.8+ ~137.3
Preterm labor after tocolysis (N = 8)
~6.30 i 3.30' 5.51 i 4.32 ~6.59 5 4.51 4.58 2 5.90 0.003 2 0.02 0.001 f 0.02 -+ 0.98’ 1.79 5 7.28 2 0.70+ 0.62 2 0.91 Z 76.2+ 152.5 t- 99.8
All abbreviations as in Table 3. Data are presented as 0 t standard error, adjusting for age at ultrasound, black ethnicity, low pregravid body maternal height, tocolysis for preterm labor, maternal cigarettes smoked per day, and fetal sex. *P < .Ol, one-tailed test. + Significantly different from zero at P < .05, one-tailed
gestational mass index, age, parity,
test.
or symmetric slowing of fetal growth (stunting). The neonates born preterm after successful tocolysis for earlier preterm labor did not differ from the controls in regard to their biometric characteristics.
Discussion Consistent with our previous study27 with similar young, primarily minority cohorts, the percentage of preterm deliveries was quite high (nearly 16%). We found that infants who were delivered preterm were already significantly smaller in all fetal growth dimensions by 32 weeks’ gestation. The diminished fetal growth was attributable to slower rates of growth from about 16-32 weeks’ gestation. Fetal growth was diminished with preterm delivery, either estimating length of gestation from the LMP confirmed or modified by an initial ultrasound scan or from the LMP unmodified by ultrasound when the dates were discordant. The findings are also consistent with and expand on the results from previous reports of fetal growth and preterm delivery in older maternal cohorts (approximately 25 years old). In one of the earliest studies using ultrasound, Tamura et al4 evaluated prospectively 148 fetuses at risk for preterm delivery, the majority of which (63%) were measured within 7 days of delivery. They found that an increased percentage had AC and BPD below the tenth percentile. This percentage was more than three times the expected amount. Two studies examined altered growth with preterm labor. Westgren et al6 examined the FL-AC ratio, which is presumed to be independent of gestational age, in 82 fetuses of gravidas hospitalized for preterm labor. Thirty-
180
Hediger
et al
Fetal
Growth
and Delivery
nine infants delivered preterm after failed tocolysis, whereas 43 went on to deliver at term. They found that the FL-AC ratio was significantly increased among those with failed tocolysis, implying that the growth of the fetuses delivering preterm was already suboptimal at the time when preterm labor was diagnosed. In a similar study, MacGregor et al’ performed ultrasound scans on admission for preterm labor in 78 pregnancies. They found that at 31 weeks’ gestation, significant percentages of nearly all growth measurements (BPD, HC, AC, and FL) were below the 25th and tenth percentiles (compared with standards) among the 48 who delivered before 36 weeks’ gestation compared with the 30 who delivered after 36 completed weeks, their cutoff for preterm delivery. Weiner et al5 used extant charts to estimate fetal weights from ultrasound scans and compared EFW at various gestational ages to published birth weight for gestational age norms.” They found that the birth weights of preterm infants were significantly smaller than the EFW of same-age infants who delivered at term. Secher et al7 and Ott3 drew the same conclusion based on similarly designed studies, comparing EFW derived from ultrasound scans to birth weights or sonographically derived standards. We know of no previous studies that have considered the type of FGR in relation to the cause of preterm delivery. In our sample, 17.4% of the infants were delivered principally for medical or obstetric indications, 34.8% for PROM, and 47.8% for preterm labor. These percentages are similar to those reported in studies including minority women.2,28,29 However, we recognize that the causes of preterm delivery may not be mutually exclusive in many individual cases and, likewise, the type of FGR may be relative. Nevertheless, when stratified by principal cause of preterm delivery, infants delivered preterm for medical or obstetric indications tended to have asymmetric FGR at 32 weeks’ gestation (ie, only AC was smaller). This implies a growth failure at a time late in the second and into the third trimester, when the growth velocities for most bone dimensions (eg, HC, BPD, and FL) are decreasing and fetal weight is still increasing.30J31 Severe preeclampsia, chorioamnionitis, and other medical or obstetric complications, which may result in reduced uterine blood flow and placental perfusion with onset late in pregnancy, seem to be associated with a relatively asymmetric FGR as early as 32 weeks‘ gestation. On the other hand, infants delivered preterm after PROM or after failed or no tocolysis for spontaneous preterm labor tended to be proportionately or symmetrically smaller. Both bone (HC, BPD, and FL) and soft-tissue dimensions (AC) were diminished relative to the controls, which suggests a different pattern of
Obstetrics
t3 Gynecology
altered fetal growth (ie, an overall slowing of growth that may originate earlier in pregnancy or possibly reflect a more chronic stress). Based on EFW at 32 weeks’ gestation, the infants later born preterm with PROM or spontaneous preterm labor were not as severely growth restricted as those delivered with medical or obstetric indications, but their growth was suboptimal relative to the controls. However, because our study design precluded very early preterm deliveries, it is possible that the most severely growth restricted fetuses are born preterm with PROM or spontaneous preterm labor at very early gestational ages (less than 33 completed weeks). Factors associated with slowed fetal growth might also be associated with preterm labor or PROM. One such factor may be poor maternal nutrition (ie, diets low in energy and/or essential vitamins and minerals). Early poor nutrition may affect placental development and vascularization, thereby restricting fetal growth by affecting the integrity of the chorioamniotic membranes (leading to PROM)“’ or by increasing susceptibility to infection. We have found in previous observational studies”““4 of both young and older gravidas that low intakes of iron and zinc are related to an increased risk of preterm delivery. For zinc, the risk of preterm delivery with low dietary intake was particularly strong (threefold increased risk) for those whose rupture of membranes preceded labor.“4 Slowed fetal growth in association with preterm delivery is also consistent with findings among both adolescent and adult gravidas that the risk of preterm delivery is increased with low rates of gestational weight gain in the last half of pregnan3537 The low rates of gestational gain may reflect a CY. slowed fetal growth and thus be a marker for impending preterm birth. These results suggest that current birth weight for gestational age standards and fetal growth curves based on the birth weights of delivered infants probably underestimate the growth of infants who go to term and underestimate the proportion of preterm infants who are actually growth restricted. The factors culminating in spontaneous preterm labor or PROM appear to be associated with a proportional slowing of fetal growth; although not necessarily as clinically severe as late asymmetric restriction, this is nevertheless characterized by slower rates of growth. This discrepancy in growth, associated with spontaneous preterm labor or PROM, is already significant by the third trimester of pregnancy. Consequently, research into the etiology of preterm delivery and interventions designed to prevent preterm delivery, especially that resulting from spontaneous preterm labor or PROM, may have to focus on factors (such as periconceptional nutrition) that can be modified early in pregnancy. Interventions3’ aimed
VOL.
85, NO.
2, FEBRUARY
1995
later in pregnancy may affect only those pregnancies in which there has not already been an early insult characterized by a long-term slowing of fetal growth, accounting for ineffectiveness in decreasing the incidence of preterm birth.
References 1
Berkowitz GS, Papiernik miol Rev 1993;15:414&43.
2
Savitz DA, Blackmore CA, Thorp JM. Epidemiologic tics of preterm delivery: Etiologic heterogeneity.
characterisAm J Obstet
Gynecol 1991;164:467-71. Ott WJ. Intrauterine growth retardation J Obstet Gynecol 1993;168:1710-7.
delivery.
3 4
5
E. Epidemiology
Tamura RK, Sabbagha RE, Depp ML. Diminished growth in fetuses labor or rupture of membranes.
of preterm
birth.
and preterm
Depp R. A hypothetical growth in infants deliv-
ered preterm. Obstet Gynecol 1985;65:323-6. 6. Westgren M, Beall M, Divan M, Platt L. Fetal femur abdominal circumference ratio in preterm labor patients
7
successful NJ,
retardation 20. 8. MacGregor
9
Kern
tocolytic
therapy.
Hansen
in preterm
J Ultrasound
P, Thomsen
infants.
SK, Sabbagha
Am
R, Vaisrub N, Dooley SL, Socol born preterm after spontaneous Am J Obstet Gynecol 1984;148:
1105-10. Weiner CP, Sabbagha RE, Vaisrub N, model suggesting suboptimal intrauterine
without 243-5. Secher
Epide-
BL,
Br J Obstet
RE, Tamura
Med
Keiding Gynaecol
RK, Pielet
length/ with and
N.
1986;5: Growth
1987;Y4:115-
SN, Feigenbaum
SL. Differing fetal growth patterns in pregnancies preterm labor. Obstet Gynecol 1988;72:834-7. Wolfson RN, Zador IE, Halvorsen P, Andrews
complicated B, Sokol
by RJ.
Biparietal diameter in premature rupture of membranes: Errors in estimating gestational age. JCU 1983;11:371-4. CR, Goldenberg RL, Hoffman HJ, Cliver SP, 10. Davis RO, Cutter Brumfield CG. Fetal biparietal, head circumference, abdominal circumference and femur length: A comparison by race and sex. J Reprod Med 1993;38:201-6. 11. Kramer MS, Olivier M, McLean FH, Dougherty Usher RH. Determinants of fetal growth and body Pediatrics 12. Cameron F, Tanner Plenum 13. Stewart
GE, Wilhs DM, proportionality.
1990;86:18-26.
N. The methods JM,
eds.
Human
of auxological anthropology. growth. 3. Methodology.
In: Falkner New, York:
and validity
weight
Press, 1986:3-46. AL. The reliability
of self-reported
and
height. J Chron Dis 1982;35:295-309. 14. Palta M, Prineas RJ, Berman R, Hannan I’. Comparison of selfreported and measured height and weight. Am J Epidemiol 1982;115:223-30. 15. Stevens-Simon C, McAnarney pregnant adolescents estimate Adolesc 16. Institute Weight
ER, Coulter MI’. How accurately their weight prior to pregnancy?
Health Care 1986;7:250&4. of Medicine. Subcommittee on Nutritional Status Gain during Pregnancy. Nutrition during pregnancy.
I. Weight gain. Washington, DC: National Academy 17. Butman M. Prenatal nutrition: A clinical manual. Massachusetts Department of Health, 1982:26.
do J and Part
Press, 1990. Boston, MA:
18. Robinson HP, Fleming JEE. A critical evaluation of sonar “crownrump length” measurements. Br J Obstet Gynaecol 1975;82:702-10. 19. Daya S. Accuracy of gestational age estimation by means of fetal crown-rump pt 1):903-8.
length
Hediger
measurement.
et al
Am J Obstet
Fetal
Growth
Gynecol1993;168(3
and Delizwy
181
20. Kurtz AB, Wapner RJ, Kurt2 RJ, et al. Analysis of biparietal diameter as an accurate indicator ot gestational age. JCU 1980;8: 31Y -26. 21. Brenner WE, Edelman DA, Hendricks CH. A standard of fetal growth for the United States of America. Am J Obstet Gynecol 1976;126:555-64 22. Greasy RK. Preterm labor and delivery. In: Greasy RK, Resnik R, eds. Maternal-fetal medicine: Principles and practice. Philadelphia: W. B. Saunders Company, 1994:494-520. 23. American Institute of Ultrasound in Medicine. Guidelines for performance of the antepartum obstetrical ultrasound examination. J Ultrasound Med 1991;10:577-8. 24. Campbell S, Thorns A. Ultrasound measurement of the fetal head to abdomen circumference in the assessment of growth retardation. Br J Obstet Gynaecol 1977;84:765-74. VA, Berkowitz RL, Warsof SL, Hobbins JC. 25. Shepard MJ, Richards An evaluation of two equations for predicting fetal weight by ultrasound. Am J Obstet Gynecol 1982;142:47-54. 26. Kleinbaum DG, Kupper LL, Muller KE. Applied regression analysis and other multivariable methods. Boston: PWS-KENT I’ublishing Company, 1988. 27. Scholl TO, llediger ML, Huang J, Johnson FE, Smith W, Antes IG. Young maternal age and parity: Influences on pregnancy outcome. Ann Epidemiol lYY2;2:565-75. 28. Main DM, Gabbe SG, Richardson D, Strong S. Can preterm deliveries be prevented? Am J Obstet Gynecol 1985;151:892-8. 29. Meis PJ, Ernest JM, Moore ML. Causes of low birth weight births in public and private patients. Am J Obstet Gynecol lY87;156: 1165-8. 30. Guihard-Costa AM, Droulle I’, Larroche JC. Growth velocity of the biparietal diameter, abdominal transverse diameter and femur length in the fetal period. Early Hum Dev 1991;27:93-102. 31. Deter RL, Harrist RB. Growth standards for anatomic measurements and growth rates derived from longitudinal studies of normal fetal growth. JCU 1992;20:381-8.
182 Hediger
et al
Fetal
Grozvth
and
De/ivenj
32. Allen SR. Epidemiology of premature rupture of the fetal membranes. Clin Obstet Gynecol 1991;34:685-93. 33. Scholl TO, Hediger ML, Fischer RL, Shearer JW. Anemia vs iron deficiency: Increased risk of preterm delivery in a prospective study. Am J Clin Nutr 1992;55:985-8. 34. Scholl TO, Hediger ML, Schall JI, Fischer RL, Khoo CS. Low zinc intake during pregnancy: Its association with preterm and very preterm delivery. Am J Epidemiol 1993;137:1115-24. 35. Berkowitz GS. Clinical and obstetric risk factors for preterm delivery. Mt Sinai J Med 1985;52:239-47. 36. Hediger ML, Scholl TO, Belsky DH, Antes IG, Salmon RW. Patterns of weight gain in adolescent pregnancy: Effects on birth weight and preterm delivery. Obstet Gynecol 1989;74:6-12. 37. Abrams B, Newman V, Key T, Parker J, Maternal weight gain and preterm delivery. Obstet Gynecol 1989;74:577-83. 38. Collaborative Group on Preterm Birth Prevention. Multicenter randomized, controlled trial of a preterm birth prevention program. Am J Obstet Gynecol 1993;169:352-66.
Address
reprint
Mary
L. Hediger, Department of
requests to: PkD
Obstetrics rind Gynecology
UMDNJ-SOM 401 Haddon Avenue Camden, NJ 08103
Receiord Received Accepted
August 5, 1994. ill revised form October October 13, 1994.
Copyright 0 Gynecologists.
1995 by The
American
3, 1994.
College
of Obstetricians
and
Obstetrics 6 Gynecology