Neonatal outcome after preterm delivery for preeclampsia

Steven A. Friedman, MD, Eyal Schifl~ MD, Lu Kao, RN, and Baha M. Sibai, MD
Memphis, Tennessee

OBJECTIVE: Our purpose was to determine whether maternal preeclampsia per se has a beneficial
effect on neonatal outcome after delivery before 35 weeks.
STUDY DESIGN: A matched cohort study design was used. Two hundred twenty-three infants of strictly
defined preeclamptic women were matched for gestational age, race, gender, and mode of delivery with
infants of normotensive women with preterm labor and delivery. Pregnancies with multiple gestation,
premature rupture of membranes, known fetal anomalies, diabetes, or maternal medical disease were
excluded. Information was obtained by review of maternal and neonatal charts. Paired categoric and
continuous data were compared by McNemar's test and the Wilcoxon signed-rank test, respectively.
RESULTS: There was no difference in the incidence of neonatal death (4.5% vs 4.5%, p = 0.82),
respiratory distress syndrome (22.0% vs 22.0%, p = 0.88), grades 3 and 4 intraventricular hemorrhage
(2.2% vs 2.2%, p = 0.72), grades 2 and 3 necrotizing enterocolitis (5.8% vs 4.0%, p = 0.48), and
culture-proved sepsis (9.0% vs 9.0%, p = 0.85). Results were similar when analysis was limited to infants
born at _<32 weeks, infants born to mothers with severe preeclampsia, and infants with intrauterine
growth restriction.
CONCLUSION: Maternal preeclampsia per se does not have a beneficial effect on the postnatal course
of infants born at 24 to 35 weeks' gestation. (AM J OBSTETGYNECOL1995;172:1785-92.)

Key words: Preeclampsia, neonatal outcome, respiratory distress syndrome, intraventricular

hemorrhage, necrotizing enterocolitis, sepsis

Preeclampsia accounts for approximately 5% to 10%
of all p r e t e r m births? T h e decision to terminate a
pregnancy complicated by preeclampsia remote from
term is often determined by the physician's assessment
of the fetal-neonatal prognosis. It is a commonly held
view that fetuses of preeclamptic mothers are subjected
to "stress" in utero and therefore have accelerated
maturation and a better prognosis than other preterm
infants, e T h e r e are conflicting data on the incidence of
respiratory distress syndrome (RDS) in infants of pre-
eclamptic mothers. Data regarding other important
neonatal outcome variables, such as intraventricular
hemorrhage, necrotizing enterocolitis, and sepsis, are
sparse. Therefore we sought to determine whether
maternal preeclampsia has a protective effect on neo-
natal outcome after delivery before 35 weeks.
Material and m e t h o d s
A matched cohort study design was used. On the
basis of a desired power of 0.8, an c~ of 0.05, and an
From the Division of Maternal-Fetal Medicine and the Newborn
Center, Department of Obstetrics and Gynecology, University of
Tennessee, Memphis.
Presented at the Sixty-second Annual Meeting of The Central Asso-
ciation of Obstetricians and Gynecologists, Memphis, Tennessee,
October 13-15, 1994.
Reprint requests: Steven A. Friedman, MD, Department of Obstetrics
and Gynecology, University of Tennessee, Memphis, 853 Jefferson
Ave., Suite El02, Memphis, TN 38103.
Copyright © 1995 by Mosby-Year Book, Inc.
0002-9378/95 $3.00 + 0 6/6/64181
incidence of an outcome variable of 10%, 199 patients
would be required to detect a twofold difference be-
tween the groups. For outcome variables with an ex-
pected incidence of 20% and the same power and ct
values, 82 patients would be n e e d e d to detect a twofold
difference.
We reviewed the charts of 404 hypertensive patients
who were delivered of preterm infants between January
1, 1989, and April 30, 1994. Patients were diagnosed as
preeclamptic if they met all of the following criteria:
systolic blood pressure -> 140 m m H g or diastolic blood
pressure -> 90 m m H g on two occasions 6 hours apart,
proteinuria -> 300 mg p e r 24 hours, and serum uric
acid concentration -> 5.0 mg/dl. Patients were included
in this study only if gestational age was supported by at
least one ultrasonographic examination before 24
weeks' gestation. Exclusion criteria included multifetal
pregnancy, pre~tature rupture of membranes, known
fetal anomalies, diabetes, or maternal medical disease.
A total of 223 patients fulfilled these criteria and con-
stituted our preeclamptic study group. Severe pre-
eclampsia was defined as systolic blood pressure > 160
m m H g or diastolic blood pressure > 110 m m H g
measured at least twice 6 hours apart, proteinuria
>5000 mg per 24 hours, eclampsia, platelet count
persistently < 100,000/mm 3, serum transaminases more
than twice normal, or persistent severe headache. In-
trauterine growth restriction (IUGR) was diagnosed if
birth weight was < 10th percentile for gestational a g e ?
Two h u n d r e d twenty-three normotensive patients

1785
1786 Friedman et al. June 1995
Am J Obstet Gynecol

T a b l e I. Clinical data

Preeclampsia (n = 223) Control (n = 223) ] Significance

Age (yr) 21 (17, 32) 21 (17, 30) p = 0.82
Nulliparity (%) 71.7 27.8 p < 0.0001
Gestational age (wk) 32.7 (27.9, 34.7) 32.6 (27.7, 34.6) *
Race (%)
Black 71.3 71.3 *
White 28.7 28.7 *
Glucocorticoid therapy (%) 55.6 36.8 p < 0.0001
Female infant (%) 57.4 57.4 *
Birth weight (gm) 1515 (822, 2205) 1875 (997, 2426) p---0.0001
Cesarean delivery (%) 63.7 63.7 *

Data reported as median, 10th percentile, and 90th percentile or incidence.

*Matched variable.

T a b l e II. Neonatal outcomes of m a t c h e d pregnancies delivered at < 35 weeks

Preeclampsia (n = 223) Control (n = 223) RR and 95% C1 Significance
Neonatal death (%) 4.5 4.5 1.00 (0.42-2.36) p = 0.82
Intensive or intermediate care (%) 78.0 63.7 1.23 (1.09-1.38) p = 0.001
Intensive or intermediate care (days) 12 (0, 48) 8 (0, 45) p = 0.0009
Mechanical ventilation (%) 30.5 29.1 1.05 (0.79-1.39) p = 0.78
Mechanical ventilation (days) 0.0 (0.0, 6.2) 0.0 (0.0, 7.2) p = 0.93
RDS (%) 22.0 22.0 1.00 (0.70-1.42) p = 0.88
IVH grades 3 and 4 (%) 2.2 2.2 1.00 (0.29-3.41) p = 0.72
NEC grades 2 and 3 (%) 5.8 4.0 1.44 (0.63-3.31) p = 0.48
Sepsis (%) 9.0 9.0 1.00 (0.55-1.81) p = 0.85
5 rain Apgar score < 6 (%) 17.9 17.9 1.00 (0.67-1.49) p = 0.89

Data reported as median, 10th percentile, and 90th percentile or incidence. RR, Relative risk; CI, confidence interval; IVH,
intraventricular hemorrhage; NEC, necrotizing enterocolitis.

T a b l e I I I . Neonatal outcomes of m a t c h e d pregnancies delivered at -< 32 weeks

Preeclampsia (n = I00) ] Control (n = 100) I RR and 95% CI [ Significance
Death (%) 10.0 9.0 1.11 (0.47-2.62) p = 1.00
RDS (%) 48.0 43.0 1.12 (0.82-1.51) p = 0.53
IVH grades 3 and 4 (%) 5.0 5.0 1.00 (0.30-3.35) p = 0.72
NEC grades 2 and 3 (%) 10.0 8.0 1.25 (0.51-3.04) p = 0.79
Sepsis (%) 14.0 19~0 0.74 (0.39-1.39) p = 0.40
5 min Apgar score < 6 (%) 25.0 26.0 0.96 (0.60-1.54) p = 1.00

Abbreviations as for Table II.

who were delivered d u r i n g the same p e r i o d because of
refractory p r e t e r m labor were m a t c h e d to the pre-
eclamptic patients by gestational age, race, infant gen-
der, a n d m o d e of delivery. All 223 m a t c h e d controls
likewise h a d gestational age based o n u l t r a s o n o g r a p h y
before 24 weeks, absence of the exclusion criteria used
for the study group, a n d ~tbsence of hypertension a n d
p r o t e i n u r i a at any time.
Neonatal charts were reviewed for several outcome
variables. RDS was defined as a n oxygen r e q u i r e m e n t at
24 hours of age with Fio 2 > 0.4, a c c o m p a n i e d by char-
acteristic x-ray findings. Intraventricular h e m o r r h a g e
was graded according to traditional criteria? Necrotiz-
ing enterocolitis was diagnosed w h e n there was radio-
logic evidence of p n e u m a t o s i s intestinalis (grade 2) or
bowel perforation (grade 3). Neonatal sepsis was de-
fined as culture-proved bacteremia along with charac-
teristic clinical signs.
Paired categoric a n d c o n t i n u o u s data were c o m p a r e d
by McNemar's test a n d the Wilcoxon signed-rank test,
respectively. A p value < 0 . 0 5 was considered statisti-
cally significant.
Results
Clinical data o n the 223 m a t c h e d pairs of p r e g n a n -
cies are p r e s e n t e d in Table I. As expected, the inci-
dence of nulliparity a n d birth weight differed between
the groups. O f note, m o r e patients in the preeclampsia
group received a full course of glucocorticoids (be-
t a m e t h a s o n e 12 m g intramuscularly, two doses 12 to 24
hours apart).
T a b l e II presents major n e o n a t a l outcome variables
Volume 172, Number 6 Friedman ot al. 1787
Am J Obstet Gynecol

Table IV. Neonatal outcomes of pregnancies with severe preeclampsia delivered at -< 35 weeks
[ Preeclampsia(n=160) I Control(n=160) ] RR and 95% CI [ Significance
Neonatal death (%) 5.6 5.6 1.00 (0.41-2.45) p = 0.75
RDS (%) 27.5 27.5 1.00 (0.70-1.43) p = 0.87
IVH grades 3 and 4 (%) 3.1 3.1 1.00 (0.30-3.39) p = 0.72
NEC grades 2 and 3 (%) 7.5 5.0 1.50 (0.63-3.57) p = 0.45
Sepsis (%) 10.0 11.9 0.84 (0.45-1.58) p = 0.68
5 rain Apgar score -<6 (%) 18.1 18.8 0.97 (0.61-1.53) p = 1.00

Abbreviations as for Table II.

Table V. Neonatal outcomes of pregnancies with both preeclampsia and intrauterine growth restriction
delivered at -< 35 weeks
Preeclampsia (n = 40) Control (n = 40) I RR and 95% CI Significance
Neonatal death (%) 2.5 2.5 1.00 (0.06-15.44) p = 1.00
RDS (%) 12.5 7.5 1.67 (0.43-6.51) p = 0.68
IVH grades 3 and 4 (%) 0.0 0.0 Undefined
NEC grades 2 and 3 (%) 7.5 5.0 1.50 (0.26-8.50) p = 1.00
Sepsis (%) 12.5 5.0 2.50 (0.90-54.32) p = 0.25
5 min Apgar score -<6 (%) 22.5 12.5 1.80 (0.66-4.90) p = 0.34

Abbreviations as for Table II.

Table VI. Neonatal outcomes of pregnancies matched additionally for antenatal glucocorticoid exposure
I Preeclampsia(n=119) Control (n=119) t RR and 95% C1 I Significance
I I I

Neonatal death (%) 3.4 3.4 1.00 (0.26-3.91) p = 0.73

RDS (%) 18.5 19.3 0.96 (0.57-1.62) p = 1.00
IVH grades 3 and 4 (%) 2.5 1.7 1.50 (0.26-8.82) p = 1.00
NEC grades 2 and 3 (%) 5.9 2.5 2.33 (0.62-8.81) p = 0.29
Sepsis (%) 5.9 9.2 0.64 (0.26-1.59) p = 0.42
5 min Apgar score -<6 (%) 16.8 18.5 0.91 (0.52-1.58) p = 0.86

Abbreviations as for Table II.

for all 446 pregnancies studied. The only outcomes that
were significantly different between the groups were
incidence and duration of admission to the newborn
center (intensive and intermediate care). The inci-
dences of neonatal death, RDS, intraventricular hemor-
rhage, necrotizing enterocolitis, and sepsis were similar
between the two groups.
Outcomes for neonates delivered at -< 32.0 weeks are
described in Table III. There were no significant differ-
ences between the groups. Table IV contains data on
pregnancies with severe preeclampsia and their
matched controls. In Table V outcomes from 40 preg-
nancies with IUGR are compared with their matched
controls. Table VI contains data on the 119 pregnancies
matched additionally for antenatal glucocorticoid expo-
sure (68 did not receive a full course of glucocorticoids;
51 did receive a full course).
Comment
Although the belief persists that preeclampsia sub-
jects fetuses to "stress" in utero and thereby enhances
their maturation, 2 surprisingly little has been written to
support or refute this claim. We have reported that lung
maturation is not accelerated in fetuses of preeclamptic
women, as evidenced by biochemical tests of lung ma-
turity from amniocentesis specimens. ~ In addition,
three recent studies have reported an increased inci-
dence of RDS in the neonates of preeclamptic
women. 68 O n the other hand, earlier studies demon-
strated biochemical evidence of accelerated lung matu-
ration 9, 10 or reduced rates of RDS TM 12 in the infants of
preeclamptic women. In the current study we found no
evidence that preeclampsia accelerates fetal lung matu-
ration and reduces the incidence of RDS either in the
entire group of 223 matched pairs or in the subgroups
limited to patients delivered at -<32 weeks, patients
with severe preeclampsia, or fetuses with IUGR.
Little has been written about other important out-
comes after premature delivery, including intraventricu-
lar hemorrhage, necrotizing enterocolitis, and sepsis.
Kuban et al. 13 reported a decreased incidence of ger-
minal matrix hemorrhage (grade 1 intraventricular
hemorrhage) in the infants of preeclamptic women, but
their findings must be viewed with caution. First, 96% of
1788 Friedman et al.
the infants with intraventricular hemorrhage had grade
1 or 2 disease rather than the prognostically more
important grades 3 and 4. Second, their definition of
preeclampsia was questionable at best. Third, they pro-
vided no information about how gestational age, a
critical determinant of outcome, was assigned in indi-
vidual pregnancies. In our carefully matched popula-
tion we found no evidence of a reduced rate of grades
3 and 4 intraventricular hemorrhage in infants of
preeclamptic women, although with the current sample
size we have limited statistical power to conclude that
there is no difference. Banias et al., 6 whose control
group was matched only for gestational age, r e p o r t e d
no difference in the incidence of intraventricular hem-
orrhage and necrotizing enterocolitis but a statistically
significant increase in the perinatal death rate in the
infants of preeclamptic women. Neonatal death rates in
our study were virtually identical in the two groups.
We must emphasize that our control group, matched
for gestational age, race, gender, and mode of delivery,
comprises infants delivered because of idiopathic pre-
term labor. These infants were chosen as controls be-
cause, fortunately, it is very rare to find absolutely
normal infants delivered preterm (for example, an
inadvertent preterm delivery of a patient requesting an
elective repeat cesarean section). Nevertheless, it must
be r e m e m b e r e d that infants whose mothers had idio-
pathic preterm delivery may have yet-unrecognized
abnormalities of their own, which could bias the data.
Finally, a larger percentage of infants in the pre-
eclampsia group were treated with glucocorticoids in
utero. Such a difference would be expected to bias the
data toward better outcomes in the preeclampsia
group. Nevertheless, we could find no evidence of a
protective effect of preeclampsia on neonatal outcome
in spite of this advantage. In addition, Table VI con-
tains data on those pairs concordant for antenatal
glucocorticoid exposure, and, again, no significant dif-
ferences were found between the two groups. Thus the
hypothesis that preeclampsia has a protective effect on
neonatal outcome is further weakened.
In conclusion, we have found no evidence to support
the view that fetuses of preeclamptic women have ac-
celerated maturation and improved neonatal outcome
consequent to "stress" in utero. When the clinician is
faced with the possibility of delivering a severely pre-
eclamptic patient remote from term, it should be as-
sumed that the infant will have the same prognosis as
other preterm infants taken from the same population.
REFERENCES
1. Main DM, Gabbe SG, Richardson D, Strong S. Can pre-
term deliveries be prevented? AMJ OBsxExGVNECOL1985;
151:892-8.
June 1995
Am J Obstet Gynecol
2. Cunningham FG, MacDonald PC, Gant NE Leveno KJ,
Gilstrap LC III. Williams' obstetrics. 19th ed. Norwalk,
Connecticut: Appleton & Lange, 1993:154.
3. Williams RL, Creasy RK, Cunningham GC, Hawes WE,
Norris FD, Tashiro M. Fetal growth and perinatal viability
in California. Obstet Gynecol 1982;59:624-32.
4. Papile L-A, Burstein J, Burstein R, Koffler H. Incidence
and evolution of subependymal and intraventricular
hemorrhage: a study of infants with birth weights less than
1,500 gm. J Pediatr 1978;92:529-34.
5. Schiff E, Friedman SA, Mercer BM, Sibai BM. Fetal lung
maturity is not accelerated in preeclamptic pregnancies.
AMJ OBSTETGYNECOL1993;169:1096-101.
6. Banias BB, Devoe LD, Nolan TE. Severe preeclampsia in
preterm pregnancy between 26 and 32 weeks' gestation.
Am J Perinatol 1992;9:357-60.
7. Tubman TRJ, Rollins MD, Patterson C, Halliday HL.
Increased incidence of respiratory distress syndrome in
babies of hypertensive mothers. Arch Dis Child 1991;66:
52-4.
8. Bowen JR, Leslie GI, Arnold JD, Jones ME Gallery EDM.
Increased incidence of respiratory distress syndrome in
infants following pregnancies complicated by hyperten-
sion. Aust N Z J Obstet Gynaecol 1988;28:109-12.
9. Gluck L, Kulovich MV. Lecithin/sphingomyelin ratios in
amniotic fluid in normal and abnormal pregnancy. AMJ
OBSTETGYNECOL1973; 115:539-46.
10. Kulovich MV, Gluck L. The lung profile. II. Complicated
pregnancy. AMJ OBSTETGVNECOL1979;135:64-70.
11. Yoon Jj, Kohl S, Harper RG. The relationship between
maternal hypertensive disease of pregnancy and the inci-
dence of idiopathic respiratory distress syndrome. Pediat-
rics 1980;65:735-9.
12. Chiswick ML. Prolonged rupture of membranes, pre-
eclamptic toxaemia, and respiratory distress syndrome.
Arch Dis Child 1976;51:674-9.
13. Kuban KCK, Leviton A, Pagano M, Fenton T, Strassfeld R,
Wolff M. Maternal toxemia is associated with reduced
incidence of germinal matrix hemorrhage in premature
babies. J Child Neurol 1992;7:70-6.
Discussion
DR. EROL AMON, St. Louis, Missouri. The pathophysi-
ologic features of preeclampsia can affect every major
maternal organ system, resulting in severe, protean
manifestations. Modern comprehensive maternal-fetal
medicine practice mandates establishment of secondary
diagnoses affecting each of these end organs, including
the individual components of the fetal-placental and
amniotic fluid end-organ system. Timing and mode of
delivery for preterm preeclamptic women are based on
specific maternal or fetal indications. T h e cesarean
route is generally based on unfavorable Bishop scores,
fetal malpresentation, fetal distress, and hemorrhage,
not on preeclampsia per se. On the basis of our collec-
tive clinical experience, we know that preeclamptic
women and their fetuses are at increased risk for unto-
ward outcome. However, do such neonates do better,
worse, or no different than norrnotensive controls?
The authors, known for their work in preeclampsia,
should be c o m m e n d e d for their prior contributions and
their current effort attempting to answer this important
question. They conclude by suggesting that clinicians
should generally assume that neonatal outcome after
preterm preeclamptic pregnancy is similar and that
pulmonary maturation is not enhanced. Knowing that
Volume 172, Number 6
Am J Obstet Gynecol
Table I. RDS
Length of gestation
(wk) Preeclampsia
24-35 49/223 (22%)
-<32 48/100 (48%)
>32 1/123 (0.8%)
No difference overall in 223 matched patients. No difference in 100 patients < 32 weeks. Is there a true difference in rates > 32
weeks?
preterm preeclampsia is a stressful situation for mother
and fetus, how can we make sense of the authors'
findings? My answer will focus on the nature of the
control group and the incidence of RDS.
First, all patients were selected from a larger obstetric
population who were delivered at the University of
Tennessee, City of Memphis Hospital. These patients
were primarily of low socioeconomic status with associ-
ated additional clinical characteristics. It is generally
agreed that RDS rates are much lower in this particular
obstetric population than in middle to u p p e r social
classes.l' ~ Thus it is not surprising that in the current
study only 49 patients p e r group had RDS.
Second, the control group selected did not comprise
normal unstressed control patients undergoing elective
cesarean section. They all had refractory preterm labor
with intact membranes, which in itself may be associated
with accelerated maturation.
Third, 60% of these control patients underwent ce-
sarean section as required by the matching process in
their study design. For 60% of preterm labor patients to
u n d e r g o cesarean section rather than vaginal delivery is
an unusual occurrence in any population and implies
that these controls had additional complications, some
of which may be associated with accelerated maturation.
Concerning RDS, rates were not truly different in the
overall group nor in the patient subset at -< 32 weeks'
gestation. However, rates were about six times higher in
infants delivered because of refractory preterm labor
after 32 weeks' gestation, six of 123 (4.9%) versus one of
123 (0.8%) (Table I). Is this observed difference in the
RDS rate a true difference or not? Because the sample
size was reduced to 123 matched patients in the subset
beyond 32 weeks' gestation, the power to make a valid
scientific statement was also reduced. Nonetheless, ac-
cording to my calculations, the power to detect a true
sixfold difference, with an ct value of 0.05 in patients
> 32 weeks' gestation is 88% and the p value is 0.03. 3
To summarize, the authors m o u n t e d an excellent
effort, unmatched by other clinical investigators, to
answer a very pertinent clinical question. However,
because of the nature of the population studied, selec-
tion of control group, and improved RDS rates beyond
32 weeks, their study results and conclusions are not
persuasive. This alternative view is consistent with the
theory that when fetuses at 32 to 35 weeks are poised to
produce endogenous surfactant, preeclampsia appears
to preferentially stimulate this process.
Friodman et al. 1789
Preterm labor Significance
49/223 (22%) p = 1.00
43/100 (37%) p --- 0.53
6/123 (4.9%) ?
I have the following questions. If available, would the
authors provide us with fetal data including compari-
sons of amniotic fluid tests results, cocaine usage, oli-
gohydramnios, and fetal distreSs as measured by an-
tepartum testing, intrapartum fetal heart rate and um-
bilical cord gases, and neonatal treatment with artificial
surfactant?
REFERENCES
1. Robertson PA, Sniderrnan SH, Laros RKJr, et al. Neonatal
morbidity according to gestational age and birth weight
from five tertiary care centers in the United States, 1983
through 1986. A~J OBSTETGYNECOL1992;166:1629-45.
2. Chiswick MI. Prolonged rupture of membranes, pre-
eclamptic toxaemia, and respiratory distress syndrome.
Arch Dis Child 1976;51:674-9.
3. Rosner B. Fundamentals of biostatistics. 3rd ed. Boston:
Duxbury Press.
DR. Yov.aM SOROKIN,Detroit, Michigan. The authors
continue to address important clinical issues in hyper-
tensive pregnancies. The current study is a follow-up to
their recent publication suggesting lack of support to
the contention that amniotic fluid fetal lung maturity is
accelerated in pregnancies "stressed" with preeclamp-
sia/ In this current study neonatal outcomes of 223
preterm infants of preeclamptic pregnancies were not
different from neonatal outcomes of 223 infants of
normotensive pregnancies.
The authors carried out a well-designed study. There
are several points I would like to raise. T h e authors
controled for the most important variables: gestational
age, race, infant gender, and mode of delivery. Early
and accurate dating is one of the strengths of this study;
relatively less accurate dating and inclusion of possibly
"older" pregnancies in the hypertensive group is sug-
gested as an explanation for results of earlier studies
during the 1970s that found enhanced pulmonary ma-
turity and decreased RDS in hypertensive pregnancies.
Other factors have been r e p o r t e d to affect the rate of
neonatal mortality, RDS, necrotizing enterocolitis, intra-
ventricular hemorrhage, and sepsis (e.g., antenatal
hemorrhage, presence of labor before delivery, amniotic
fluid maturity tests, use of tocolysis, and birth weight
percentile). Neonatal m a n a g e m e n t (e.g., use of surfac-
tant) may affect neonatal morbidity and mortality. Sev-
eral other publications in the recent literature did not
r e p o r t on these variables in their analysis of the inci-
dence of RDS in hypertensive and nonhypertensive
1790 Friedman et al.
pregnancies. Do you have any information on these
factors?
The authors noted in the Comment section that the
matched control group included pregnancies with re-
fractory "idiopathic" preterm labor and that they may
have as-yet unrecognized abnormalities. I agree; the
control group is not optimal. Recent literature suggests
that nearly all pregnancies with "idiopathic" preterm
labor or premature rupture of membranes that leads to
preterm delivery have pathologic processes. 2-4 Preterm
labor is a syndrome; it is a pathologic event. Preterm
labor is a disease process, not term labor before its time.
Evidence of heterogeneity of preterm labor comes from
cytologic, biochemical, immunologic, pathologic, mi-
crobiologic, and clinical data. 2 Pathologic insults to the
mother or fetoplacental unit may lead to preterm labor
and delivery. Infection, ischemia, uterine anomaly, al-
lergic phenomena, and abnormal allogeneic response
are the basic mechanisms of disease found to be in-
volved as causes of preterm labor. These processes can
affect neonatal morbidity and mortality and can there-
fore change the "control" group and bias the data in
both directions. These processes may produce "stress"
and acceleration of fetal lung maturity. Although the
control group is not optimal, the authors did their best
to find the best possible control group. It is not ethical
to have an experimental group of preterm deliveries. It
is difficult to choose a better control group; I do not
have a suggestion for a better one. The authors appro-
priately acknowledged this problem in the Comment
section. Do you have any data on the placentas of
pregnancies in both groups? Did the control group
include patients with premature rupture of membranes
in labor or patients with premature rupture of mem-
branes that were later delivered prematurely? Do you
have information on amniotic fluid tests (for fetal lung
maturity or to rule out intraamniotic infection) in both
hypertensive and control groups?
Neonatal morbidity and mortality of preterm deliv-
eries have markedly decreased in the past 20 years. 5' 6
With such low morbidity and mortality rates, studies
require very large numbers of pregnancies to show lack
of effects. Corrected neonatal mortality rates of 1147
neonates weighing I000 to 2499 gm born during 1990
to 1991 were 10%, 3%, 0%, and 0% for birth weight
groups 1000 to 1299, 1300 to 1599, 1600 to 1899, and
1900 to 2199 gm, respectively. 5 A study of neonatal
morbidities of 20,680 deliveries during 1983 to 1986
found very low rates (e.g., intraventricular hemorrhage
(grades 3 to 4) of 3.6%, 2.8%, 1.9%, 2.0%, 0.9%, 0%,
0%, and 0% for neonates at 28, 29, 30, 31, 32, 33, 34,
and 35 weeks, respectively. 6 T h e r e are wide variations
in morbidity and mortality rates among gestational ages
or birth weights. 5' 6
T h e spectrum of gestational ages (median 32.6
weeks, range 27 to 35 weeks) and birth weights (medi-
ans 1515 and 1875 gin, range 800 to 2500 gm) in the
current study are usually too wide for such comparisons
of neonatal mortality and morbidities. T h e separate
June 1995
Am J Obstet Gynecol
analysis of neonatal outcome for pregnancies delivered
at -< 32 weeks (n = 100 in each group) tries to respond
to such criticism. The use of McNemar's nonparametric
test by the authors helps to combat this problem to a
large extent.
Neonatal mortality and intraventricular hemorrhage
and necrotizing enterocolitis rates in hypertensive and
control groups in this study are very low, 2% to 9% for
both pregnancies delivered at < 3 5 or < 3 2 weeks.
Sepsis rates were 9% for < 35 weeks and 14% and 15%
for the two groups at -<32 weeks. As stated in the
Comment section, there is a limitation of statistical
power to make a conclusion concerning lack of effect on
neonatal mortality, intraventricular hemorrhage, necro-
tizing enterocolitis, and sepsis. RDS rates are higher
(22% at < 35 weeks' gestation and 48% and 43% at -< 32
weeks), and there were no significant differences be-
tween the hypertensive and control groups.
In conclusion, the authors are to be congratulated on
a well-designed study. The study controlled for the
most important variables affecting neonatal outcome. I
have a question concerning the possible effects of other
uncontrolled confounding variables. T h e authors ac-
knowledged the possible bias from a nonoptimal con-
trol group of pregnancies with "idiopathic" preterm
labor that probably have subclinical pathologic features.
I would like to know whether the authors have data on
the placentas, the presence of premature rupture of
membranes in labor and before labor, and information
on amniotic fluid tests. In a period of very low morbid-
ity and mortality rates, much larger sample sizes are
necessary. The study has limitations of statistical power
to conclude lack of effect on neonatal mortality, intra-
ventricular hemorrhage, necrotizing enterocolitis, and
sepsis of preeclampsia in pregnancy. The primary neo-
natal outcome variable, RDS was not reduced in pa-
tients with preeclampsia at < 35 weeks, or -< 32 weeks,
or in patients with severe preeclampsia.
REFERENCES
1. Schiff E, Friedman SA, Mercer BM, Sibai BM. Fetal lung
maturity is not accelerated in preeclamptic pregnancies. AM
J OBSTETGYNECOL1993;169:1096-101.
2. Romero R, Sepulveda W, Baumann P, et al. The preterm
labor syndrome: biochemical, cytologic immunologic,
pathologic, microbiologic, and clinical evidence that pre-
term labor is a heterogeneous disease [Abstract]. AM J
OBsTrx GYNECOL1993; 168:288.
3. Arias F, Rodriguez L, Rayne SC, Kraus FT. Maternal pla-
cental vasculopathy and infection: two distinct subgroups
among patients with preterm labor and preterm ruptured
membranes. AMJ OBSTETGVNECOL1993;168:585-91.
4. Lettieri L, Vintzileos AM, Rodis JF, Albini SM, Salafia CM.
Does "idiopathic" preterm labor resulting in preterm birth
exist? AMJ OBSTETGVNECOL1993;18:1480-5.
5. DePalma RT, Leveno KJ, Kelly MA, Sherman ML, Carmody
TJ. Birth weight threshold for postponing preterm birth.
AMJ OBSTETGVNECOL1992;167:1145-9.
6. Robertson PA, Sniderman SH, Laros RKJr, et al. Neonatal
morbidity according to gestational age and birth weight
from five tertiary care centers in the United States, 1983
through 1986. AMJ OBSTETGYNECOL1992;166:1629-45.
Volume 172, Number 6
Am J Obstet Gynecol
DR. ROBERT CARPENTER,Houston, Texas. Are there
any recurrent patients in either of the groups from the
standpoint of statistical analysis?
Second, we know that at least 160 pairs of the
patients were in the hospital for at least 24 hours. What
was the interval from admission to delivery in both sets
of patients, and how did that influence outcome?
T h e r e appears in some centers to be a more marked
effort to keep some patients with severe preeclampsia
p r e g n a n t longer and longer, walking them out to a
point where they are very sick. How rational is that in
view of this study?
Last, if you have an incidence of RDS of 24% to 48%
in the population, it is far greater than for the standard
gestational ages. So what, if anything, are we doing with
antenatal steroid therapy in trying to obtain 24 hours in
some of these patients?
DR. JosEPH DEWANE, Memphis, Tennessee. With the
proved ability of steroids in decreasing the incidence of
RDS and intraventricular hemorrhage, why was their
use so low in this population, both in the preectamptic
and control groups, particularly in the control group
where the need for delivery would have been less?
Second, did you compare the groups of patients who
received steroids with those who did not?
DR. FRIEDMAN (Closing). In response to Drs. Amon
and Sorokin, we agree with your observations on the
difficulties with the control group, and we too thought
that a group of patients who had preterm labor without
premature rupture of membranes, diabetes, or other
medical problems would be the best control group. As
Dr. Sorokin pointed out, it wouldn't be ethical to have
a true control group of normal patients who underwent
cesarean section or vaginal delivery for no indications at
all at premature gestations.
Because our primary study question dealt with neo-
natal outcome as a consequence of maternal disease, we
did not collect extensive fetal data, which may or may
not have played an important role mechanistically.
Instead, we looked at maternal disease and looked at
the final outcome of the neonates, because that is the
more objective data. Therefore I do not have with me
complete data regarding oligohydramnios, antenatal
testing, umbilical cord gases, and amniocentesis results.
Although we do not have drug screens on all patients,
the incidence of a positive drug screen in our popula-
tion of preterm labor patients has been in the past
around 5%.
Surfactant therapy was used in 35 of the 223 pre-
eclamptic patients and 24 of the 223 control patients,
but it is difficult to interpret data on surfactant therapy.
At our institution surfactant therapy, which began about
1990, h a s been used only for rescue, not for prophy-
laxis. Therefore the use of this therapy is not only a
determinant of pulmonary outcome but also a reflec-
tion of its severity in the first several hours of life,
because the severity of disease was used to determine
whether an infant would receive surfactant therapy.
If anything, the increased use of surfactant therapy in
Friedman el al. 1791
the infants of preeclamptic women would indicate more
severe RDS in that group.
I would also like to point out that the data in this
study are consistent with the data that we published in
1993, as Dr. Sorokin mentioned, demonstrating that in
anmiocentesis specimens biochemical evidence of ma-
turity is not accelerated at any gestational age in pre-
eclamptic patients from 29 to 37 weeks.
Dr. Amon pointed out that there is a trend toward
increased pulmonary maturity in the preeclamptic
group from 32 to 35 weeks, and he attributed this
finding to enhanced pulmonary maturation in this
group of patients.
We believe that a more logical explanation, which is
much more consistent with our previous amniocentesis
data, is selection bias in this group. A considerable
number of patients delivered in the 32 to 35-week
gestational age range had mild preeclampsia; because
their disease was mild, they could have been managed
expectantly at our institution. Nevertheless, they were
delivered in many cases because of known pulmonary
maturity. That may have stacked the deck a little bit in
favor of the preeclamptics of 32 to 35 weeks. Therefore
this group was probably somewhat enriched in patients
with a low risk for RDS after delivery.
We were asked to comment on tocolysis. At our
institution, magnesium sulfate has been the tocolytic of
choice; and the patients who had preeclampsia also
almost uniformly received magnesium sulfate before
delivery. So there was overwhelming use of magnesium
sulfate in both groups.
The question was asked about why the use of gluco-
corticoids was so low, especially in the control group. At
our institution we use glucocorticoids very generously.
We treat almost every preterm patient with any kind of
complication with glucocorticoids. What is in our statis-
tics are the number of pregnancies in which the full 48
hours was achieved from the beginning of the course of
glucocorticoids. So the question is, why is glucocorticoid
usage particularly low in the control population? The
answer is that many patients in the preterm labor group
had refractory preterm labor that we were unable to
stop. So, although there was no indication for their
delivery before completing their course of steroids, they
took the question out of our hands.
Regarding the question about whether we c o m p a r e d
the groups of patients who received steroids with those
who didn't in both the preeclamptic and control
groups, I've done that preliminarily. I don't have final
statistics, but there were very few differences.
There was a question about whether there were any
patients in our study with more than one pregnancy.
There were three women in the preeclamptic group
and 14 women in the control group who had data on
two pregnancies included in this study.
In response to the final questions, I do not have the
data immediately available on the admission-to-delivery
interval, because that was one of the antepartum vari-
ables on which we really did not tabulate the data.
 Fisher et al. June 1995
Am J Obstet Gynecol

In response to the question on how rational is con-
servative m a n a g e m e n t of severe preeclampsia, I refer
you to our recent publications on the subject. 1' 2
REFERENCES
1. Sibai BM, Mercer BM, Schiff E, Friedman SA. Aggressive
versus expectant management of severe preeclampsia at 28
to 32 weeks' gestation: a randomized controlled trial. AMJ
O~STE'r GVNECOL1994; 171:818-22.
2. Schiff E, Friedman SA, Sibai BM. Conservative manage-
ment of severe preeclampsia remote from term. Obstet
Gynecol 1994;84:626-30.

The expert witness: Real issues and suggestions

Charles W. Fisher, JD," Mitchell P. Dombrowski, MD~ b Stanislaw E. Jaszczak, MD~ b
Cheryl D. Cook, BArb and Robert J. Sokol, MD b
Detroit, Michigan

Medical malpractice lawsuits generally require expert testimony. Defendants and plaintiffs deserve expert
testimony that is exacting, accurate, and consistent. A study of four frequently testifying experts was
undertaken with review of depositions, reports, and trial transcripts of those experts, Contradictions in
claimed medical principles from one case to the next were found and examples were cited for each
expert. The review suggested that expert testimony re~arding the standard of care may be neither reliable
nor accurate for the purposes of judging physician conduct is lawsuits. Presently, no peer review or
sanction process has been implemented to ensure accuracy and reliability of expert testimony used in
medical malpractice lawsuits. We recommend changes that would include independent court-appointed
experts, central filing of opinion letters by exp~rts with authoritative text citations, and a sanction process
by courts and/or authorized boards for testimony that is deemed inaccurate, false, or contradictory to
the standard of care. (AM J OBSTETGYNECOL1995;172:1792-800.)

Key words: Expert witness, peer review, medical malpractice

Although the legal system purports to mete out
justice for civil liability through impaneled juries, it has
serious shortcomings in terms of being an objective
arbiter. The current malpractice (liability) crisis, is often
ascribed to "dishonest lawyers" and "ignorant juries."
However, the most significant shortcoming may very
well be the expert witness. The majority of expert
witnesses are honest and motivated by a duty to be fair
and just. However, a small but significant minority (i.e.,
the "professional expert witness") is motivated by finan-
cial incentive rather than fairness or justice. 1-s Such
professional witnesses, especially physicians who testify
in malpractice cases, may earn most or all of their
From Kitch, Drutches, Wagner & Kenney, P. C.,~and the Department
of Obstetrics and Gynecology, Wayne State University/Hutzel Hos-
pital?
Presented at the Sixty-second Annual Meeting of The Central Asso-
ciation of Obstetricians and Gynecologists, Memphis, Tennessee,
October 13-15, 1994.
Reprint requests: Charles W. Fisher, Kitch, Drutches, Wagner &
Kenney, P.C., One WoodwardAve., Tenth Floor, Detroit, MI 48226.
Copyright © 1995 by Mosby-Year Book, Inc.
0002-9378/95 $3.00 + 0 6/6/63635
income from fees changed for medical-legal reviews,
depositions, or trim testimony.
The generic problem of scientific p r o o f becomes
highly amplified in a medical malpractice case when
experts are called to testify. The court allows medical
experts to testify to standards of care without substan-
tive basis or written proof. In nearly every medical
malpractice case the plaintiff must identify an expert
witness to define a particular standard of care and to
testify that the defendant violated that standard. 4 With-
out such expert testimony the lay jury will have no basis
on which to make a finding as to the culpability or
negligence of the defendant2 At the heart of the medi-
cal malpractice case lie the experts for the plaintiff and
the experts for the defendant but not necessarily the
objective facts.
If there are national standards of care, then the
objective determination of such should be attainable.
However, the vast majority of cases have experts on
opposing sides in complete disagreement about the
particular issues. It is therefore unfair and absurd to
expect a lay jury, in the face of diverse expert testimony,

1792