Volume 64, Number 9

OBSTETRICAL AND GYNECOLOGICAL SURVEY

Copyright © 2009 by Lippincott Williams & Wilkins
OBSTETRICS
Acupuncture as Pain Relief During
Delivery: A Randomized Controlled Trial
Lissa Borup, Winnie Wurlitzer, Morten Hedegaard, Ulrik S. Kesmodel,
and Lone Hvidman
Department of Obstetrics and Gynecology, Aarhus University Hospital, Skejby, Denmark; Department of
Obstetrics at the Rigshospitalet, Copenhagen, Denmark; Department of Epidemiology at the Institute of
Public Health, University of Aarhus, Aarhus, Denmark; and Department of Obstetrics and Gynecology at the
Aarhus University Hospital, Skejby, Denmark
Birth 2009;36:5–12

ABSTRACT
The effect of acupuncture on labor pain has been poorly documented. The few randomized controlled trials published
after 2002 that evaluated the effectiveness of acupuncture for pain relief during labor, varied in size and quality. This
randomized unblinded study compared the effect of acupuncture on pain relief and relaxation during delivery with the
pain relief provided by transcutaneous electric nerve stimulation (TENS) or traditional analgesics. A total of 607 healthy
laboring women with normal singleton pregnancies at term (37–42 weeks) were randomized to receive either
acupuncture (n ⫽ 314), TENS (n ⫽ 144) or traditional analgesics (n ⫽ 149). The primary outcomes were a need for
pharmacological and invasive treatment for pain relief and a subjective assessment of birth experience and satisfaction
with delivery at 2 months postpartum. The level of pain was determined by a visual analogue scale. Of the 517 women
who completed the treatment, 490 responded to a questionnaire 2 months after delivery, assessing their experience and
satisfaction with the 3 methods.
Significantly fewer women in the acupuncture group received pharmacological or invasive methods for pain relief,
or both, compared with the traditional group (P ⬍ 0.001) or the TENS group (P ⬍ 0.031). No significant difference
between the 3 treatment groups was found in pain scores or in secondary outcomes, including the need for oxytocin,
mode of delivery, duration of labor, mean Apgar score at 5 minutes, or cord pH blood value. When asked if they would
want to use acupuncture again for a future delivery, 53% of the women in the acupuncture group answered positively
and 18% negatively. In contrast, when asked this same question, 18% of the women in the TENS group answered
positively and 66% negatively. The degree of relaxation and control of the situation reported by women in the
acupuncture group tended to be higher compared to the other 2 groups, but no differences were reported for other birth
experience variables.
These findings suggest that the need for pharmacological and invasive methods to achieve pain relief may be
significantly lower during labor with acupuncture compared to use of TENS or traditional analgesics. The investigators
suggest that acupuncture may be a useful supplement to other types of pain relief for women in labor.

EDITORIAL COMMENT
(Since 2001, several randomized trials of acu- ation, but reported equivalent pain intensity to
puncture treatment during labor have been re- women in the conventional analgesia group.
ported. In the first, by Ramnero et al, 90 women Maternal and perinatal outcomes did not differ
were randomized to receive acupuncture during between groups (BJOG 2002;109:637). Skiland
labor (as an adjunct or alternative to conven- et al. randomized 210 healthy parturients in
tional analgesia) or to receive conventional an- spontaneous active labor to either real acu-
algesia alone. Women allocated to acupuncture puncture (insertion of needles into recognized
used epidural analgesia less frequently, 12% acupoints) or false acupuncture (insertion of nee-
versus 22%, and had a higher degree of relax- dles into points not on classical acupuncture me-
www.obgynsurvey.com | 565
566 Obstetrical and Gynecological Survey
ridians). Visual analog pain scores, use of epidural
analgesia (10% vs. 27%), interval from random-
ization to delivery, and the use of oxytocin aug-
mentation were all lower in the real acupuncture
group (Acta Obstet Gynecol Scand 2002;81:943).
In the unblinded trial of Neishem et al. con-
ducted among 204 parturients, use of meperi-
dine (the analgesic of choice on their unit) was
the primary outcome measure, and the rate of
usage by women randomized to acupuncture
was significantly lower, 11% versus 37% (Clin J
Pain 2003;19:187). Hantoushzadeh et al ran-
domized 144 healthy nulliparous women in ac-
tive labor to real or false acupuncture. After
randomization, women in the real acupuncture
group had lower visual analog pain scores, and
active phase labor durations that were 1 hour
shorter, 31⁄2 versus 41⁄2 hours (Aust NZ J Obstet
Gynecol 2007;47:26). In contrast, Ziaei and Ha-
jipour found acupuncture in labor to be without
benefit. They randomized 150 women to real
acupuncture (n ⫽ 90), fake acupuncture (n ⫽
30), or neither (n ⫽ 30) and found no difference
between the groups in pain or relaxation scores,
duration of labor, or cesarean delivery rates
(IJOG 2006;92:71).
The abstracted trial of Borup et al is the largest
to date. It was a major undertaking, requiring the
certification of 50 midwives, all of whom com-
pleted a 5-day acupuncture course and re-
ceived 6 months of clinical training on how to
use acupuncture during labor. Participating
women allocated to acupuncture received it in
one of 34 specified points. The needles were to
be left in for as little as 30 minutes or for as long
as 2 hours, and were managed by a project
midwife at all times.
Although reported as a positive trial, in fact it
was negative. The prespecified primary out-
come was the combined rate of epidural anal-
gesia or meperidine use during labor. In the
acupuncture group this rate was 18.2% and in
the TENS group it was 17.4% (TENS was cho-
sen for one of the control groups because its
analgesic effects are known to be weak, thus
making it a good placebo). Pain scores did not
differ between groups, nor did the duration of
labor or use of oxytocin. Cesarean delivery rates
were comparable across groups.
Based on the above studies, is there a place
for acupuncture in labor management? Maybe,
but that place is not well-defined because the
studies are not consistent. How acupuncture
was actually employed in the studies (how
many needles, where, and for how long) is not
clearly described. Blinding is difficult when
studying acupuncture, and thus biased out-
come assessment is more likely. For example,
patients of caregivers with positive attitudes
about acupuncture might be influenced by
that positive attitude to forgo epidural analge-
sia that they otherwise would have utilized.
That said, if a trained provider is available and
a women in labor requests acupuncture as an
adjunct or alternative to conventional analge-
sia, there is no reason to withhold it.—DJR)
 Management of Labor, Delivery, and the Puerperium 567

Effect of Food Intake During Labor on

Obstetric Outcome: Randomized
Controlled Trial
Geraldine O’Sullivan, Bing Liu, Darren Hart, Paul Seed,
and Andrew Shennan
Department of Anaesthesia, St Thomas’ Hospital, Guy’s and St Thomas’ NHS Foundation Trust, London,
United Kingdom; Maternal and Fetal Research Unit, King’s College London, London, United Kingdom;
and Guy’s and St Thomas’ NHS Foundation Trust London, United Kingdom
BMJ 2009;338:b784

ABSTRACT
There is no clear evidence that routine fasting in labor reduces the incidence of pulmonary aspiration or is associated
with improved labor outcomes. Oral intake during labor is limited in the United States primarily to clear liquids. In
contrast, birth units in several European countries allow access to food and drink. Five previous randomized controlled
trials involving a total of fewer than 1000 women have investigated the effect of food intake on labor outcomes. Three
reported no effect. One showed increased duration of labor and another showed a nearly 3-fold increase in cesarean
deliveries. All 5 studies were limited by relatively small numbers of participants and limited power.
This randomized controlled trial investigated the effect of food intake during labor on the rate of spontaneous delivery
and other obstetric and neonatal outcomes. The participants were 2426 healthy nulliparous singleton women at term with
a vertex presentation and a cervical dilatation less than 6 cm. Study subjects were randomized to either an eating group
(n ⫽ 1219) or a water only group (n ⫽ 1207). Women in the eating group were allowed a low fat, low residue diet during
labor. All participants had free access to water. The primary outcome was the rate of spontaneous vaginal delivery. Other
measured outcomes were the instrumental and cesarean delivery rate, incidence of vomiting, need for augmentation with
oxytocin, duration of labor, and neonatal outcome. Data were analyzed according to the intention to treat principle.
There was no significant difference in the rate of normal vaginal delivery between the 2 groups. The rates were 44%
(533/1219) in the eating group and 44% (534/1207) in the water only group (risk ratio [RR], 0.99; 95% confidence
interval [CI], 0.91–1.09). No significant difference between the eating and water only groups were found in the cesarean
delivery rate (30% vs. 30%; RR, 0.99; 95% CI, 0.87–1.12), the incidence of vomiting (35% vs. 34%; RR, 1.05; 95%
CI, 0.9–1.2), or duration of labor (597 minutes vs. 612 minutes; RR of geometric means, 0.975; 95% CI, 0.93–1.03).
In addition, no differences were noted between the groups in the rate of instrumental vaginal delivery, need for
augmentation with oxytocin, Apgar scores, or admission to neonatal intensive care or special care units.

EDITORIAL COMMENT
(After C. L. Mendelson described the patho-
physiology of aspiration during obstetric anes-
thesia, fasting during labor became the norm
(Am J Obstet 1946;52:191). Today in developed
countries, parturients rarely die from anesthesia-
associated aspiration; in fact, according to the
7th Confidential Enquiries into Maternal Death
Report, in the United Kingdom from 2003 to 2005,
there were no anesthesia-associated aspiration
deaths (London: CEMACH, 2007). Several factors
explain this success: the vast majority of cesare-
ans do not involve tracheal intubation, preopera-
tive antacids are given routinely to de-acidify the
stomach contents, anesthesiologists and nurse
anesthetists are generally well-trained and
highly skilled, and most women in labor are not
allowed solid food.
The American line on oral intake during labor
has been a relatively hard one, but more so for
solids than liquids: “The oral intake of modest
amounts of clear liquids may be allowed for
uncomplicated laboring patients . . . Examples of
clear liquids include, but are not limited to,
water, fruit juices without pulp, carbonated bev-
erages, clear tea, black coffee, and sports dri-
nks . . . However patients with additional risk
factors for aspiration (morbid obesity, diabetes,
difficult airway) or patients at increased risk for
568 Obstetrical and Gynecological Survey
operative delivery (e.g., nonreassuring fetal
heart rate pattern) may have further restrictions
of oral intake, determined on a case by case-
by-case basis.” As for solids “The consultants
and ASA members both agree that the oral in-
take of solids during labor increases maternal
complications” (American Society of Anesthesi-
ologist’s Task Force on Obstetric Anesthesia.
Practice guidelines for obstetric anesthesia. An-
esthesiology 2007;106:843).
Other countries have taken more liberal ap-
proaches. Four-fifths of women in the Nether-
lands and one-third in the United Kingdom are
allowed liquids and solids in labor (Scheepers
HC, et al. Eur J Obstet Gynecol Reprod Biol
1998;78:37; Michael S, et al. Anesthesia 1991;
46:1071). The liberalized approach of these
countries to oral intake during labor has not
been associated with an increased incidence of
aspiration. Indeed, as described above, aspira-
tion, or at least deaths from aspiration, have
declined in the United Kingdom.
The abstracted study of O’Sullivan et al dem-
onstrates that from the standpoint of objective
obstetric and neonatal outcomes, allowing in-
take of solid food during labor makes no differ-
ence. But in a study of healthy, well-nourished
women who labored on average for 10 hours,
were allowed to drink water, and, in two-third of
cases, received more than 500 mL of intrave-
nous fluid, I wouldn’t have expected any differ-
ences in these outcomes. It seems to me that,
for most women, allowing solid food in labor is a
satisfaction issue, and this wasn’t measured.
Interestingly, some women did not enroll in the
study because they planned (and were allowed)
to eat during labor, and 20% of the women who
enrolled and were randomized to only water
partook of solid food.
In the O’Sullivan trial of healthy laboring nul-
liparas, 30% underwent cesarean delivery. Al-
though it is true that most cesareans are
accomplished with regional anesthesia, 10%
are performed under general anesthesia (Bloom
SL, et al. Obstet Gynecol 2005; 106:281), and
presence of particulate matter in the stomach
makes those procedures riskier. I have man-
aged the delivery of 1 patient who aspirated
popcorn and required mechanical ventilation
and I would not prefer to manage such a
situation again. I understand that some pa-
tients want to eat during labor and from a
probability standpoint, they are not being at all
unreasonable. But if we are to emulate the
safety record of the airline industry, we are
going to have to accept some of the dissatis-
faction that goes along with it, and maybe
even go it one better—no peanuts.—DJR)
 Management of Labor, Delivery, and the Puerperium 569

Early Compared With Late Neuraxial

Analgesia in Nulliparous Labor Induction:
A Randomized Controlled Trial
Cynthia A. Wong, Robert J. McCarthy, John T. Sullivan,
Barbara M. Scavone, Susan E. Gerber, and Edward A. Yaghmour
Northwestern University Feinberg School of Medicine, Departments of Anesthesiology and Obstetrics
and Gynecology, Chicago, Illinois
Obstet Gynecol 2009;113:1066–1074

ABSTRACT
In observational studies, the rate of cesarean delivery among women who received epidural analgesia early in labor
at cervical dilation less than 4 cm was twice that of women who first received epidural analgesia at greater cervical
dilations. In contrast to these findings, 2 previous randomized studies, one a study of mixed spontaneous and induced
labor and the other of spontaneous labor, as well as an extensive review of 9 studies reported no such association
between timing of initiation of epidural analgesia and cesarean delivery rates.
This randomized study investigated whether early initiation and maintenance of neuraxial (combined spinal-epidural)
analgesia in nulliparas undergoing induction of labor increased the risk of cesarean delivery in comparison with systemic
opioid analgesia in early labor followed by epidural analgesia in late labor. The final analysis included 806 singleton
term nulliparas who were randomized to either combined spinal-epidural (early, n ⫽ 400) or systemic opioid (late, n ⫽
406) analgesia at first request for analgesia when cervical dilation was ⱕ4 cm. Analgesia was maintained in the early
group with patient-controlled epidural analgesia at the second analgesia request, whereas in the late group epidural
analgesia was not initiated until the third request for analgesia. At the first and second analgesia requests, patients rated
their pain using a verbal 0 to 10 rating score (0 indicating no pain).
There was no difference in the cesarean delivery rate between the groups (early [combined spinal-epidural]: 32.7%
vs. late [systemic]: 31.5%); the 95% confidence interval for the difference in the medians was ⫺3% to 6% (P ⫽ 0.65).
No differences between the groups were found in the mode of vaginal delivery or in Apgar scores at 1 and 5 minutes.
The median pain scores were significantly lower in the early compared to the late group (early: 1 vs. late: 5, P ⬍ 0.001)
and labor duration in the early group was shorter (early: median 528 minutes vs. late: 569 minutes, P ⫽ 0.047). After
analgesia, there was no significant difference between the groups in the rate of reassuring fetal heart rate tracings, or
in the incidence of persistent variable decelerations or late decelerations.
These findings show no difference in the rate of cesarean delivery among a population of nulliparas undergoing
induction of labor who received neuraxial analgesia in early as opposed to later labor.

EDITORIAL COMMENT
(During my residency, which ended in 1992, it
was thought that too-early administration of la-
bor epidural analgesia increased a parturient’s
risk of cesarean delivery for nonprogressive la-
bor. Thus, we typically delayed initiation of epi-
dural analgesia until a cervical dilation of 4 to 5
cm. In fact, in the year 2000, the ACOG Task
Force on Cesarean Delivery Rates recom-
mended that, when feasible, epidural analgesia
should be delayed in nulliparas until 4 to 5 cm.
The basis for this recommendation was obser-
vational data showing an inverse relationship
between cervical dilation at epidural initiation
and risk of cesarean delivery. These data, how-
ever, did not directly address causality, because
women who request or need analgesia earlier in
labor are probably at inherently higher risk of
subsequent cesarean delivery for abnormal la-
bor progress.
The first randomized trial to challenge the pre-
vailing wisdom of delaying epidural analgesia
was conducted by Chestnut et al. They random-
ized 150 laboring women who were receiving
oxytocin to early (ⱖ3 cm) or to late (ⱖ5 cm)
initiation of epidural analgesia. Two-thirds of the
participants had entered labor spontaneously
570 Obstetrical and Gynecological Survey
and one-third had been induced. The actual
median cervical dilations at epidural initiation
were 3.5 cm and 5.0 cm, respectively. The ce-
sarean rates in the 2 groups were 18% and
19%, respectively. A limitation of this trial is that
“early” initiation, at 3.5 cm, was not actually all
that early (Anesthesiology 1994;80:1193).
Subsequently, Wong et al randomized 750
term nulliparas in spontaneous labor who had a
cervical dilation of ⬍4 cm to 1 of 2 groups. In the
early group, a combined spinal-epidural was
placed at the first request for pain relief and
intrathecal fentanyl given, followed by initiation
of epidural analgesia at the next request for pain
relief. In the delayed group, systemic narcotics
were given until 4 cm (or until a third request for
analgesia), at which time epidural analgesia was
initiated. Not only did early initiation of neuraxial
analgesia decrease the cesarean rate—17.8%
early versus 20.7% late—it was associated with
a 1.5 hour shorter duration of labor, and much
better pain scores following the first pain in-
tervention in each group. “Early” in this trial
was more truly early. The median dilation at
initiation of intrathecal analgesia in the early
group was 2 cm (NEJM 2005;352:655). Ohel et
al reported similar results in a trial of 449
nulliparas, approximately two-third of whom
entered labor spontaneously. Women in the
early group had epidural analgesia initiated at
mean cervical dilation of 2.4 cm, whereas
women in the late group had it initiated at a
mean of 4.6 cm. Respective rates of cesarean in
the 2 groups were not significantly different,
13% versus 11%. The randomization to full di-
lation interval was 45 minutes shorter in the
early group, and participants indicated a prefer-
ence for early epidural in their next pregnancy
(Am J Obstet Gynecol 2006;194:600).
The abstracted study of Wong et al is a com-
panion study to her trial in spontaneously labor-
ing nulliparas and has essentially the same
design features. Neuraxial (intrathecal) analge-
sia was initiated at a median cervical dilation of
2 cm in the early group, and epidural analgesia
initiated at a median 4 cm in the delayed group.
Collectively, these studies provide strong ev-
idence that there is no valid reason to withhold
neuraxial analgesia (spinal or epidural) until 4
cm. Simply, this causes suffering for no benefit.
It might make sense to delay initiation of
neuraxial analgesia in some women undergoing
induction who are very remote from delivery, so
as to avoid catheter migration over the course of
many hours. But in general, if a woman is being
induced or is in labor and has no contraindica-
tion, neuraxial analgesia should be initiated
upon request.—DJR)
 Management of Labor, Delivery, and the Puerperium 571

Antibiotic Prophylaxis at Urinary

Catheter Removal Prevents Urinary
Tract Infections. A Prospective
Randomized Trial
Urs Pfefferkorn, Sanlav Lea, Jörg Moldenhauer, Ralph Peterli,
Markus von Flüe, and Christoph Ackermann
Department of Surgery, St. Claraspital, Basel, Switzerland
Ann Surg 2009;249:573–575

ABSTRACT
Urinary tract infections are common after removal of a urinary catheter and antibiotic prophylaxis is used to prevent
them, although there is no supporting evidence that it is beneficial. One previous randomized placebo-controlled trial
reported that antibiotic prophylaxis did not significantly reduce the rate of urinary tract infections following catheter
removal but the number of patients included was small. This randomized controlled study assessed the effect of
antibiotic prophylaxis at urinary catheter removal on the rate of subsequent urinary tract infections. The subjects in the
final analysis, 205 patients undergoing elective abdominal surgery, were randomized to receive at catheter removal
either 3 doses of trimethoprim-sulfamethoxazole (n ⫽ 103) or no prophylaxis (n ⫽ 102). Patients with known allergy
to trimethoprim-sulfamethoxazole were given ciprofloxacin. Urinary cultures were obtained the day before and 3 days
after catheter removal, and pathogenic bacteria were identified using Center of Disease Control definitions. The urinary
catheters were left in place for 6.5 to 7 days in the 2 groups (P ⫽ 0.68). An independent study-blind urologist assessed
subjective symptoms 4 ⫾ 2 days after removal of the urinary catheter. The primary study end point was the occurrence
of a urinary tract infection following catheter removal.
Significantly fewer urinary tract infections (4.9%, 5/103) occurred in patients with antibiotic prophylaxis compared
to those without prophylaxis (21.6%, 22/102) (P ⬍ 0.001). The absolute risk reduction for the presence of a urinary tract
infection in the prophylaxis group was 16.7%, with a 95% confidence interval (CI) of 7.8% to 22.1%. The relative risk
reduction was 77.5% (95% CI: 45.7%–91.9%), and the number needed to treat was 6 (95% CI: 4.5–12.8). There was
also less significant bacteriuria 3 days after catheter removal in the antibiotic prophylaxis group (16.5%, 17/103)
compared with the no prophylaxis group (41.2%, 42/102) (P ⬍ 0.001).
The data suggest to the investigators that antibiotic prophylaxis at urinary catheter removal may reduce the rate of
urinary tract infections. These findings are not consistent with those of a previous study showing no benefit of antibiotic
prophylaxis for prevention of urinary tract infections after catheter removal.

EDITORIAL COMMENT
(Although it is standard to screen all pregnant
women during the prenatal period for asymp-
tomatic bacteriuria (ASB), screening in the post-
partum period is not routine, even although ASB
is relatively common after delivery. In their co-
hort of 667 postpartum women, Harris et al re-
ported an ASB rate of 2.8% among women who
had not been catheterized, and a rate of 6.1%
among those who had. How long the catheter
remained in place influenced the rate of ASB; it
was 3.5% when the duration of catheterization
was under 24 hours, and ⬎25% when the cathe-
ter was indwelling for at least 24 hours (SMJ 1977;
70:1273). In women who delivered vaginally, Rehu
et al reported that a single “in-and-out” catheter-
ization during labor was associated with a dou-
bled risk of postpartum bacteriuria, from 4% to
9%; 2 or more catheterizations were associated
with a 10-fold increased risk—43%. In contrast,
single catheterization before cesarean delivery
was not associated with an increased risk of post-
partum bacteriuria; the rate was 7% in women
who did not undergo catheterization before cesar-
ean, and 7% in those who underwent a single
in-and-out catheterization (Ann Clin Research
1980:112).
572 Obstetrical and Gynecological Survey
In their series of 1438 women who underwent
cesarean, Buchholz et al reported that the rate of
postpartum bacteriuria was 11.7%. All women in
this series had indwelling catheters for an average
of 2.3 days (Eur J Obstet Gynecol Reprod Biol
1994;56:161). Recently, Elram et al reported
that 13% of women with negative urine cultures
on admission for delivery developed bacteriuria
during the labor and delivery process (J Mat Fet
Neonatal Med 2008;21:483).
If anything, the issue of ASB in the postpartum
period is probably growing in importance, in that
urethral catheterization, especially if prolonged,
increases the risk of bacteriuria, and a high per-
centage of parturients now undergo placement
of an in dwelling catheter in conjunction with the
receipt of epidural analgesia. This is why I se-
lected the abstracted article of Pfefferkorn, even
although it did not involve pregnant patients. It
suggests that there are benefits to routine anti-
microbial prophylaxis at the time of catheter
removal. However, on average, the catheters
were left in place for 1 week, which is far longer
than in the average pregnant or postpartum pa-
tient. This lengthy duration of catheterization was
predictably associated with a higher rate of bac-
teriuria (36%) than would be seen in a typical
postpartum patient. Moreover, the trial was not
blinded, and because “symptomatic” urinary tract
infection was the end point, outcome assessment
may have been biased.
Are there data specific to pregnancy that
answer the questions “should we screen
some or even all women for ASB after delivery
and treat those who are positive?” or “should
we treat some women presumptively?” Unfor-
tunately, there are none that are very compel-
ling. Stray-Pedersen et al showed that ASB in
the postpartum period can be fairly success-
fully eradicated with a short course of antibi-
otics, and that (1⁄4) of women with ASB in the
immediate postpartum period, who do not get
treated will have persistent bacteriuria 10
weeks later (Am J Obstet Gynecol 1990;162:
792). But in healthy, nonpregnant women without
urologic abnormalities, ASB rarely progresses to
symptomatic infection. Thus, without a defini-
tive clinical trial that establishes routine screen-
ing for postpartum ASB is worthwhile, or that
routine antimicrobial prophylaxis after catheter-
ization is effective, for most of our patients we
would do better to simply be alert for symptom-
atic urinary tract infection and reserve antimi-
crobial prophylaxis for selected women with
complicated histories or prolonged or repetitive
catheterization.—DJR)
 Newborn Medicine 573

Magnesium Sulfate in Severe Perinatal

Asphyxia: A Randomized,
Placebo-Controlled Trial
Mushtaq Ahmad Bhat, Bashir Ahmad Charoo, Javeed Iqbal Bhat,
Sheikh Mushtaq Ahmad, Syed Wajid Ali, and Masood-ul-Hassan Mufti
Department of Pediatrics, Sheri-Kashmir Institute of Medical Sciences, Srinagar, India
Pediatrics 2009;123;e764–e769

ABSTRACT
Hypoxic-ischemic encephalopathy (HIE) is associated with high morbidity and mortality in the neonatal period and
permanent neurodevelopmental deficits among survivors. Animal studies have shown beneficial effects of magnesium
sulfate in perinatal asphyxia but only 1 uncontrolled human trial has shown beneficial effects of postnatal magnesium
sulfate in neonates with perinatal asphyxia. This longitudinal randomized, placebo-controlled trial tested whether
postnatal magnesium sulfate treatment could improve neurologic outcomes at discharge among term neonates with
severe perinatal asphyxia. The study subjects were 40 term (ⱖ37 weeks) neonates born at a tertiary-care hospital in India
between 2004 and 2006 who were ⬍6 hours of age at the time of admission. Infants were randomly assigned to receive
infusions of either magnesium sulfate (250 mg/kg per dose (1 mL/kg per dose) over 1 hour, with 2 additional doses
repeated at intervals of 24 hours (treatment group, n ⫽ 20) or 3 doses of normal saline (1 mL/kg per dose) 24 hours apart
(placebo group, n ⫽ 20). Supportive care available for both groups when needed included respiratory and presser support.
At admission, the percentage of patients with moderate and severe HIE were similar in the treatment and placebo
groups (treated: 35% 7/20 and 65% 13/20 vs. placebo: 40% 8/20 and 60% 12/20). The mean serum magnesium
concentration remained within the therapeutic range (at or above 1.2 mmol/L) for 72 hours after the first infusion. At
discharge, neurologic abnormalities were present in 22% (4/18) of infants in the treatment group and in 56% (10/18)
of infants in the placebo group. On day 14 of life, neuroimaging (head computed tomography) results were abnormal
among fewer infants in the treatment group (16% 3/18) than in the placebo group (44% 8/18). At discharge, 77% (14/18)
infants in the treatment group were receiving oral feeding through sucking compared with 37% (7/18) infants in the
placebo group. Good short-term outcomes (a composite measure of no neurologic abnormality, normal neuroimaging
and electroencephalographic findings, and oral feeding through sucking) were 77% (14/18) for patients in the treatment
group, compared with 37% (7/18) for patients in the placebo group.
These findings demonstrate that postnatal treatment with magnesium sulfate improves neurologic outcomes at
discharge for term neonates with HIE. This is the first placebo-controlled study to show the efficacy of postnatal
magnesium sulfate treatment in perinatal asphyxia.

EDITORIAL COMMENT
(In developed countries, it is estimated that in
term newborns, hypoxic ischemic encephalop-
athy (HIE) occurs at a rate of 1 to 8 per 1000
births (ACOG 2003. Neonatal Encephalopathy
and Cerebral Palsy). When moderate or severe,
the risk of death is high. Those who survive are
at great risk of neurologic handicap. HIE is a
condition that all obstetricians and parents want
to avoid. Unfortunately, we can not always do
so, but our ability to mitigate the consequences
of HIE has improved.
In a prior SURVEY (2006;3:158), I discussed
the excellent study of whole-body cooling for
HIE by Shankaran et al. The combined rate of
death or severe disability at 18 to 22 months
was significantly less frequent in the cooling
group, 44% versus 62%. The respective rates of
death were 24% and 37%, and of cerebral palsy
19% and 30% (NEJM 2005;353:1574). This
beneficial intervention is now utilized in many
neonatal intensive care units across the country.
In both the Shankaran and the Bhat study,
eligibility depended on the diagnosis of as-
phyxia according to predefined criteria, and a
clinical diagnosis of moderate or severe en-
cephalopathy. The criteria for asphyxia in the 2
574 Obstetrical and Gynecological Survey
studies were not identical; they were more strin-
gent in the Shankaran study. But the criteria for
encephalopathy were identical. In the Bhat
study, 2/3 of infants had severe encephalopa-
thy, whereas in the Shankaran study, 2/3 had
moderate encephalopathy.
The Bhat study was very well done. It was
double-blinded and placebo-controlled, and all
enrolled infants were accounted for. It is limited
principally by its small size (40 total infants vs.
208 in the Shankaran study) and the short du-
ration of follow-up, which, although not explic-
itly stated, seems to have been only to hospital
discharge. The magnesium sulfate infusions
seemed generally safe, although in 2 of 20
cases they caused apnea that required intuba-
tion and mechanical ventilation. Also, more in-
fants in the magnesium sulfate group required
pressor support for hypotension, 7 versus 5, a
difference that was not statistically significant. A
somewhat curious finding of the study was that
although the infants were sicker than in the
Shankaran study, their death rate was lower
(10% vs. 30%), possibly due to the short-term
follow-up or unwillingness to withdraw life sup-
port in infants destined to die without it.
In a randomized but nonblinded study of 37
asphyxiated infants, Ichiba et al also reported
short-term benefits to magnesium sulfate infu-
sion (Ped Int 2002;44:505). That magnesium sul-
fate is neuroprotective when given in the first
few days of life to asphyxiated infants is biolog-
ically plausible. Asphyxia results in glutamate
excess. This excess keeps open N-methyl-D-
aspartate channels, allowing excessive calcium
neuronal influx and permanent neuronal dam-
age. This second wave of damage occurs for up
to 72 hours after the primary asphyxial insult has
occurred and resolved (Lorek A, et al. Pediatr
Res 1994;36:699).
Maternally administered magnesium sulfate
prevents eclamptic convulsions (Altman D, et al.
Lancet 2002;359:1877) and lowers the risk of
the cerebral palsy among the survivors of early
preterm birth (Doyle LW, et al. Cochrane Data-
base of Systematic Reviews 2009:CD004661).
The study of Bhat suggests that it may have
additional perinatal neuroprotective applica-
tions. Perhaps it is time for a trial of magnesium
sulfate ⫹ whole-body cooling versus whole-
body cooling alone for neonatal hypoxic isch-
emic encephalopathy.—DJR)
 Medical Complications of Pregnancy 575

Maternal Cardiovascular Disease Risk in

Relation to the Number of Offspring
Born Small for Gestational Age:
National, Multigenerational Study of 2.7
Million Births
Peter M. Nilsson, Xinjun Li, Jan Sundquist, and Kristina Sundquist
Department of Clinical Sciences, Lund University, University Hospital, Malmö, Sweden; Center for Family
Medicine (CeFAM), Karolinska Institute, Huddinge, Stockholm, Sweden; Center for Primary Health Care
Research, Lund University, Lund, Sweden; and Stanford Prevention Research Center, Stanford University
School of Medicine, Stanford, CA
Acta Pediatrica 2009;98:985–989

ABSTRACT
A number of studies have found an association between low birth weight or fetal growth impairment and an increased
risk of hypertension, type 2 diabetes, and cardiovascular disease (CVD) in adult life. This interaction may be due to
nutritional and lifestyle-related factors during pregnancy or genetic factors. Other studies reported a similar interaction
but were limited by sample size and did not include nonfatal events and adjustment for social background (educational
level). This large, national, observational study assessed the risk of small for gestational age (SGA) births in relation
to maternal history of CVD across 2 generations.
Data for 1.4 million women and 2.7 million offsprings were obtained for the years 1973 to 2004 from the Swedish
national, multigeneration register and analyzed. The primary outcome measure was determination of the risk of being
SGA in relation to maternal history of CVD (n ⫽ 10,436) across 2 generations. Secondary outcomes included
determination of the risk of maternal CVD based on number of SGA births and assessment of the possible contribution
of level of education for these associations.
For both male and female offspring with a positive maternal history of CVD, the risk of being SGA was higher than
for offspring with no family history of CVD (reference); the hazard ratio (HR) for females was 1.11, with a 95%
confidence interval (CI), of 1.09–1.13), and for males it was 1.12 (1.09–1.14). A 2 generation history of CVD (in both
mother and grandmother) was associated with the highest risk (for girls, HR: 1.32, 95% CI: 1.24–1.39 and for boys, 1.43,
1.35–1.51).
Compared to women with no SGA infants, the risk of a maternal history of CVD increased along with the number
of SGA offspring (1 SGA, HR: 1.41, 95% CI: 1.36–1.46; 2 SGAs: HR: 1.74, 95% CI: 1.58–1.93; and 3 or more SGAs:
HR: 1.86, 95% CI: 1.35–2.57). Among the 3 levels of education evaluated, the interaction between maternal CVD and
risk of SGA birth was the strongest in the group with the lowest level of education and increased in a graded manner
with 1, 2, and 3 SGAs.
These data, based on multigenerational observational epidemiological data, indicate that the risk of SGA offspring is
positively related to the occurrence of CVD over 2 generations of mothers, and that risk of maternal CVD is associated
with the number of SGA offsprings.

EDITORIAL COMMENT
(The data presented by Nilsson et al are both
fascinating and frustrating; fascinating because
of the strong association detected between
eventual CVD in women and growth restriction
in their fetuses, and frustrating because the data
do not explain why this is so. Is this link merely
the result of several generations of poor health
habits, such as smoking, drinking, and poor nu-
trition? Or does this represent a true epigenetic
phenomenon, in which some antenatal factor or
factors prompted genetic imprinting of the fe-
tus, resulting in poor growth and small size?
A link between fetal growth restriction and
maternal cardiovascular disease has been re-
576 Obstetrical and Gynecological Survey
ported before, but the nature of the relationship
is incompletely understood. One theory is that,
because both fetal growth restriction and heart
disease are caused by vascular pathology, hav-
ing a growth restricted fetus is an event that
warns of future maternal cardiovascular prob-
lems. This theory is supported by the observa-
tion that both cardiovascular disease and fetal
growth restriction are associated with maternal
hypertension, obesity, and insulin resistance,
and the fact that endothelial dysfunction has
been shown to underlie both fetal growth re-
striction and heart disease. One study that an-
alyzed biochemical and anthropometric data
from 3 generations of Dutch subjects found that,
9 years after a group of 56 women had given
birth to a growth restricted fetus, they were 3
times more likely than control women to have
developed hypertension (26.95% vs. 8.9%, P ⬍
0.01); those who had developed hypertension
already had carotid artery intimal thickening,
and their fathers had significantly higher fasting
glucose levels than the fathers of control women
(Berends AL, et al. Hypertension 2008;51:1034).
For these women, having a growth restricted
fetus was one of the first indications that they
would eventually develop familial hypertension
and possibly insulin resistance.
However, another equally interesting theory is
that something about pregnancy itself predis-
poses to the eventual development of cardio-
vascular disease in certain women. In 1 large
epidemiologic study of how body composition
affects morbidity—the Health, Aging, and Body
Composition (Health ABC) Study—the health
status of women age 70 to 79 was evaluated in
terms of their pregnancy history; 89 nulliparous
women were compared to 321 parous women
who had had only completely normal pregnan-
cies and 130 parous women had had at least 1
pregnancy complicated by hypertension, pre-
eclampsia, low birth weight, preterm birth, or
stillbirth (Catov JM, et al. Ann Epidemiol 2008;
18:873). This fascinating study showed that,
even although the parous women were younger
and were less likely to smoke or to be obese
than the nulliparous women, and even although
adjustments were made for age, race, smoking,
use of statins, and HDL levels, the parous
women were significantly more likely to have
cardiovascular disease than the nulliparous
women. Importantly, although women who had
had 1 or more pregnancy complications were
at highest risk (adjusted OR: 2.67, 95% CI:
1.34–5.33), even parous women who had had
only completely normal pregnancies were
twice as likely to have cardiovascular disease
as their nulliparous colleagues (adjusted OR:
1.95, 1.03–3.68).
As these data illustrate, the origins of cardio-
vascular disease in women are complex and in
many ways different from those in men. Although
cardiovascular disease in women develops, on
average 10 years later than it does in men, a
phenomenon attributed to the effects of female
hormones, it also seems that pregnancy—a time
of soaring female hormone levels—actually in-
creases the risk of developing heart disease. Ad-
verse pregnancy outcomes such as hypertension,
preeclampsia, and fetal growth restriction can
be both the result of existing endothelial dys-
function and the harbinger of its eventual de-
velopment. Much more research is needed to
explain these complex relationships, and thus
to provide the basis of strategies for women to
avoid cardiovascular disease as well as effec-
tive therapies to treat it. In the mean time,
obstetricians should be aware that a woman
who has had a pregnancy complicated by fetal
growth restriction, or by hypertension, pre-
eclampsia or diabetes, is at increased risk of
developing cardiovascular disease. Such
women should be counseled about strategies
to reduce their risk and monitored so that
heart disease will be detected early and
treated appropriately.—KDW)
 Obstetric Complications 577

Preeclampsia in the Second Pregnancy:

Does Previous Outcome Matter?
Sohinee Bhattacharya, Doris M. Campbell, and Norman C. Smith
Dugald Baird Centre for Research on Women’s Health, Aberdeen Maternity Hospital, Aberdeen, Scotland,
UK; and Department of Obstetrics, Aberdeen Maternity Hospital, Aberdeen, Scotland, UK
Eur J Obstet Gynecol Reprod Biol 2009;144:130–134

ABSTRACT
Although it has been observed that the incidence of preeclampsia is lower among women having their second
pregnancy than among women having their first, it is unclear whether protection against preeclampsia results from parity
or gravidity. For example, 1 previous study reported a protective effect in the second pregnancy of a late miscarriage
in the first pregnancy, whereas other studies have found that only deliveries ⬎37 completed weeks have a protective
effect. Another study showed that a previous abortion, either spontaneous or therapeutic, was protective, irrespective of
gestational age. This retrospective observational study was designed to determine whether the reduced risk of
preeclampsia in the second pregnancy is dependent or independent of the first pregnancy’s outcome, including the
gestational age at delivery. Data from 24,500 women with first and second pregnancies between 1986 and 2006 were
obtained from the Aberdeen Maternity and Neonatal Databank in Scotland. The treatment group included all women
who developed preeclampsia in their second pregnancy, whereas the control group included all women with normo-
tensive second pregnancies. Potential independent risk factors for the development of preeclampsia in the second
pregnancy were evaluated, and crude and adjusted odds ratios (OR) were estimated using binary logistic regression.
A total of 903 (3.7%) of the 24,500 women in the study population had preeclampsia in the second pregnancy. Of
these, 167 had a previous history of preeclampsia, for a recurrence rate of 14.2%. The incidence of preeclampsia was
increased in women with interpregnancy intervals of 6 years or more (19.3% vs. 14.7%) and in those with an increase
in BMI (70.8% vs. 63.2%), whereas a change of partner appeared to have a protective effect (3.5% vs. 5.6%). Compared
to women who were normotensive in the first pregnancy, women with a history of preeclampsia in the previous
pregnancy had an adjusted odds ratio of 5.12 (95% confidence interval [CI] of 4.42–6.48) for developing preeclampsia
in the second pregnancy.
Using a model that included only live births, and after controlling for maternal age, BMI, smoking, change of partner,
interpregnancy interval, and previous history of preeclampsia or gestational hypertension, the risk of preeclampsia in the
second pregnancy appeared to decrease with increasing gestational age at delivery of the first pregnancy. Compared to
women who had a previous term delivery, the odds ratio for developing preeclampsia in the second pregnancy was 4.22
(95% CI, 2.54–7.03) for women whose previous pregnancy ended in the second trimester, 2.32 (95% CI, 1.62–3.32) for
women who had a previous preterm birth at 25 to 32 weeks’, and 1.62 (95% CI, 1.46–1.72) for those with a previous
preterm birth at 33 to 36 weeks’. The effect of a previous fetal death was evaluated using a second model that included
all study patients whose previous pregnancy extended beyond 20 weeks. After adjusting for the same variables listed
above, the risk of preeclampsia was the same for women with a previous history of stillbirth as for those with a previous
live birth. Only first deliveries beyond 37 weeks were protective against preeclampsia, independent of outcome, in the
second pregnancy.
These findings suggest that the protective effect of a previous pregnancy against preeclampsia in the second
pregnancy is dependent on the gestational age at delivery of the first pregnancy but not on that pregnancy’s outcome.

EDITORIAL COMMENT

(Factors that predict recurrence of a preg-
nancy complication are important because they
alert the physician to the potential for increased
risk in the next pregnancy. Once we know that
our patient’s first pregnancy was complicated
by preeclampsia, we usually watch her blood
pressure and other signs and symptoms more
closely during her next pregnancy. This is espe-
cially true when the patient has an underlying
health problem such as renal disease or obesity
that independently increases the recurrence
risk. However, factors that predict recurrence are
also important because they provide clues about
the underlying pathophysiology of the complica-
578 Obstetrical and Gynecological Survey
tion. What is it about the experience of a previous
pregnancy that protects against preeclampsia in
subsequent pregnancies, and what does this sug-
gest about the underlying pathophysiology?
Many theories about the cause of preeclamp-
sia have been proposed over the years. My
personal favorite is that preeclampsia is caused
by a helminth infection, a theory that was dis-
proved when the wormlike structures visible in
the peripheral smears of affected women were
shown to be an artifact of the staining process
(Richards FO, et al. JAMA 1983;250:2970). Of
the more rational theories, one is that pre-
eclampsia is related to immune dysfunction
(Dekker GA, Sibai BM. Am J Obstet Gynecol
1998;179:1359). Although the evidence sup-
porting this theory is now fairly complex, it orig-
inated in the observation that the microscopic
placental changes seen in severe preeclampsia
are similar to acute graft rejection (Labarrere C.
Placenta 1988;9:108). According to this theory,
exposure to novel antigens resulting from the
paternal contribution to the fetus’s (and pla-
centa’s) genetic makeup—which is similar to
receiving an organ transplant from anyone but
an identical twin—should prompt the produc-
tion of blocking antibodies that cover the pla-
centa’s antigenic sites. The production of such
antibodies protects the pregnancy; if this im-
mune response is insufficient, “rejection” oc-
curs, in the form of preeclampsia. Antibody
production “immunizes” the woman, and as a
result her next pregnancy is less likely to be
complicated by preeclampsia. If sufficient
blocking antibodies are not produced during the
first pregnancy, or if the pregnancy ends before
the production of blocking antibodies can result
in “immunization,” the woman is still at risk of
having preeclampsia in her next pregnancy.
Another reasonable theory is that inflamma-
tion is the underlying cause of preeclampsia
(Redman CWG, Sargent IL. Placenta 2003;17:
S21). This theory was initially prompted by the
observation that women with preeclampsia fre-
quently have elevated levels of proinflammatory
cytokines and white blood cells, before the clin-
ical manifestations of preeclampsia become evi-
dent (Redman S, et al. Am J Obstet Gynecol
1990;180:499). Chronic inflammation increases
oxidative stress, leading to the production of
reactive oxygen species and free radicals; these
toxic molecules cause endothelial cell injury,
inhibiting the production of prostacyclin and im-
pairing the production of nitric oxide by the en-
dothelium (IBID, Dekker and Sibai). The resulting
endothelial dysfunction results in vasopasm,
abnormal placentation, and the typical signs
and symptoms of preeclampsia. Both theories
also acknowledge the importance of genetic
factors, diet, and other influences on ultimate
pregnancy outcome.
How do Bhattacharya’s data fit with these the-
ories? Using a large dataset that included data
from all deliveries in one area of Scotland over a
50 year period, these investigators found that a
previous term pregnancy is protective against
preeclampsia in the next pregnancy while a pre-
vious preterm pregnancy is not, regardless of
whether the pregnancy resulted in a live birth or a
stillbirth. The data also showed that the risk of
preeclampsia decreases as the gestational age at
the time of preterm delivery increases, regardless
of the survival of the preterm infant. Although the
dataset did not provide some much needed
information—for example, whether the preterm
births were spontaneous or indicated for com-
plications such as severe preeclampsia—as a
whole, the results indicate that the most impor-
tant protective feature of a previous pregnancy
is the gestational age at delivery, not the viability
of the fetus.
These data are consistent with many previous
reports—for example indicating that a previous
miscarriage does not significantly reduce the risk
of developing preeclampsia in the next pregnancy
(Strickland DM et al. Am J Obstet Gynecol 1986;
154:146) and that only previous term deliveries
are protective (Xiong X, et al. Am J Obstet Gy-
necol 2002;187:1013). They are also consistent
with both theories about etiology. The produc-
tion of blocking antibodies takes time and must
continue throughout the pregnancy—hence a
previous term pregnancy would be most protec-
tive against recurrent preeclampsia. On the
other hand, inflammation could occur and recur
regardless of fetal viability, thus the fetal out-
come of the first pregnancy would not influence
the risk of preeclampsia in the second.
 Obstetric Complications 579

One piece of data reported by Bhattacharya
and coworker is puzzling—that is, that having a
pregnancy with a new partner reduced the inci-
dence of preeclampsia. Many other investiga-
tors have found that having a pregnancy with a
new partner increases the risk of preeclampsia—
presumably because it results in first time expo-
sure to another set of novel antigens (Trupin LS,
et al. Epidemiology 1996;7:240; Mostello D, et
al. Am J Obstet Gynecol 2002;187:425). Bhat-
tacharya’s finding might be related to the un-
detected interaction of multiple factors—for
example, if a woman lost weight after her first
pregnancy, her reduced BMI might be protec-
tive against preeclampsia in the next preg-
nancy even although she had a new partner.
The data could also reflect the importance of
genetic and epigenetic factors in the pathogen-
esis of preeclampsia (Chelbi ST, Vaiman D. Mol
Cell Epidemiol 2008;282:120); after a pregnancy
complicated by preeclampsia, a new partner
might result in greater maternal—fetal genetic
compatibility.
There is an empty stone plaque on Chicago’s
Lying in Hospital reserved for the name of the
person who discovers the etiology of pre-
eclampsia. Considering the complexity of the
disease, this space will likely not be enough to
accommodate the names of all the researchers
who have reported pieces of the puzzle.—KDW)

Expectant Management of Severe

Preeclampsia at 27 (0/7) to 33 (6/7)
Weeks’ Gestation: Maternal and
Perinatal Outcomes According to
Gestational Age by Weeks at Onset of
Expectant Management
Annette E. Bombrys, John R. Barton, Mounira Habli, and Baha M. Sibai
Department of Obstetrics and Gynecology, University of Cincinnati, Cincinnati, Ohio; and Central Baptist
Hospital, Lexington, Kentucky
Am J Perinatol 2009;26:441–446

ABSTRACT
The aim of expectant management of severe preeclampsia between 27 (0/7) and 33 (6/7) weeks of gestation is to
improve perinatal outcome without compromising maternal safety. Perinatal outcome is dependent in large part on the
gestational age at the time of disease onset and maternal and fetal status at the time of presentation. This retrospective
study evaluated maternal morbidity and perinatal outcome with expectant management of severe preeclampsia between
27 (0/7) and 33 (6/7) weeks’ gestation, with data stratified by gestational age at the time of onset of expectant
management. The investigators reviewed the medical records of 336 gravid women with severe primary or superim-
posed preeclampsia who delivered before 34 (0/7) weeks at an Ohio medical center, to identify a study population of
66 patients (71 fetuses) between 27 (0/7) and 33 (6/7) weeks. All patients were treated with corticosteroids for fetal lung
maturity.
Expectant management resulted in pregnancies being prolonged by a median of 5 days (range: 3–35). Nineteen (27%)
newborns had a birth weight below the 10th percentile for gestational age, and 6 (8%) had a birth weight below the fifth
percentile. Among patients in whom expectant management was instituted at 27–27 (6/7) weeks, there was 1 neonatal
580 Obstetrical and Gynecological Survey

death, 5 cases of respiratory distress syndrome, 2 cases of bronchopulmonary dysplasia, and 2 cases of necrotizing
enterocolitis, but no cases of grade III or IV intraventricular hemorrhage. Among those whose expectant management
began at ⱖ32 weeks, neonatal morbidity was minimal.
Maternal morbidities including placental abruption, pulmonary edema, HELLP syndrome, renal insufficiency, and
eclampsia were evaluated as individual outcomes and also as a composite outcome. There were 2 cases of transient renal
insufficiency, 5 cases of HELLP syndrome, and 6 cases of pulmonary edema, of which 0, 1, and 3 cases, respectively,
occurred in women in whom expectant management started at ⱖ32 weeks. The composite maternal morbidity was 27%;
the morbidity was 40% at 32–32 (6/7) weeks and 33% at 33–33 (6/7) weeks. Although most maternal complications did
not appear to be related to the gestational age at the institution of expectant management, placental abruption was more
common among pregnancies managed expectantly at ⱕ28 weeks than among those ⬎28 weeks (25% vs. 6%, P ⫽ 0.05).
There were no maternal deaths, cases of eclampsia, or disseminated intravascular coagulopathy.
These data support a role for expectant management of early severe preeclampsia from 27 to 31 (6/7) weeks’.
However, on the basis of these data and previous findings, the investigators recommend that women with severe
preeclampsia at ⱖ32 weeks’ gestation should be delivered following corticosteroid administration.

EDITORIAL COMMENT

(The management dilemma presented by se- by severe preeclampsia be delivered by 32 (0/7)

vere preeclampsia remote from term has been
studied before. In a 2007 review, Sibai and Bar-
ton (Am J Obstet Gynecol 2007;196:514) sum-
marized the findings of 2 randomized trials and
11 observational studies, including a total of
1677 pregnancies between 24 and 34 weeks,
and concluded that certain patients with severe
preeclampsia remote from term can be man-
aged expectantly. The criteria for expectant
management included an estimated gestational
age of at least 23 and preferably 24 weeks and
the absence of: eclampsia, neurologic deficit,
pulmonary edema, disseminated intravascular
coagulation, suspected abruption, evidence of
HELLP (hemolysis, elevated liver enzymes, low
platelets) syndrome, or nonreassuring fetal
heart rate tracing. The investigators stated that
such pregnancies need intensive monitoring of
both the mother and the fetus, and that deliv-
ery should be accomplished if any of the
above conditions develop. In the 2 random-
ized trials, expectantly managed pregnancies
were prolonged by 7 days with no perinatal
deaths and 15 days with a perinatal death rate
of 17%; in the 8 observational studies, preg-
nancies were prolonged by an average of 5 to
14 days, with perinatal death rates of 4% to
14%. Associated morbidities experienced by
the mothers included HELLP syndrome in 4%
to 17%, renal failure in 2% to 17%, pulmonary
edema in 1% to 9%, and eclampsia in 2% to
6%. Based on the data reviewed, the authors
recommended that all pregnancies complicated
weeks.
The gold standard for clinical research is the
prospective randomized trial. Interestingly, one
of the randomized trials included in the above
review was conducted by an author of the
present retrospective study. The data from that
trial (Sibai BM, et al. Am J Obstet Gynecol 1994;
171:818) strongly supported expectant man-
agement of the small group of patients who met
the criteria listed above. Compared to infants
from similar pregnancies who were delivered 48
hours after maternal steroid administration, the
infants of pregnancies managed expectantly
gained on average 7.1 days in utero and were
significantly less likely to be admitted to the
neonatal intensive care unit or the intermediate
care nursery, spent significantly fewer days in
the NICU or intermediate care nursery, and had
significantly less respiratory distress syndrome
and necrotizing enterocolitis. The authors con-
cluded that expectant management of pregnan-
cies remote from term, as long as they met strict
criteria and were monitored closely, was reason-
able. Considering this study and the published
literature, why was the current retrospective study
needed?
The investigators stated that data regarding
the effect of gestational age on outcome, or in
other words, neonatal and maternal outcome
according to the gestational age at which ex-
pectant management was begun, is lacking in
the published literature; in their own prospective
randomized trial, pregnancies were grouped for
 Obstetric Complications
analysis into only 2 groups, expectant versus
aggressive, and the effect of gestational age at
the onset of expectant management was not
assessed. The goal of the current study was
therefore to provide these data for the gesta-
tional age range 27 (0/7) to 33 (6/7) weeks. As
might have been predicted, the incidence of
neonatal respiratory distress and other compli-
cations decreased as gestational age at delivery
increased, and after 32 (0/7) weeks, there was
very little morbidity and neonatal outcomes
were good. Interestingly, the rate of abruption
was high at 27 and 28 weeks but dropped sig-
nificantly after 28 weeks (from 25% to 6%; P ⫽
0.05). Maternal complications, however, did not
follow the same pattern. Although the number of
patients at each week of gestation was too
small to permit calculation of reliable incidence
rates, cases of HELLP syndrome and pulmonary
edema occurred at all gestational ages and the
composite maternal mortality was fairly con-
stant from 27 (0/7) to 33 (6/7) weeks. Thus at
ⱖ32 weeks, a gestational age at which delivery
was associated with a good neonatal outcome,
the maternal risk associated with continued
pregnancy remained high.
One important piece of data provided by this
study, that hasn’t always been presented in
studies of this type, is the number of pregnan-
581
cies in the denominator. In other words, given
the stringent criteria for selecting severe pre-
eclamptic pregnancies that can be managed
expectantly, for what percentage of patients is
this management option a consideration? The
investigators reported that, of 336 patients with
severe preeclampsia at 27 (0/7) to 33 (6/7)
weeks who were evaluated for participation,
only 112 (33%) met the enrollment criteria. In
their prospective trial, 74% (95 of 129 patients)
were eligible, but since patients were not en-
rolled and randomized until after an initial 24
hour observation period, the denominator likely
included additional patients who had to be deliv-
ered before enrollment, making the incidence of
eligible patients lower than reported. In the other
prospective study included in Sibai and Burton’s
review, (Odendaal HJ, et al. Obstet Gynecol 1990;
76:1070), 58 women were deemed eligible for ex-
pectant management but 20 had to be delivered
before randomization—leaving 66% eligible for
participation. Thus, expectant management of
preeclampsia remote from term is possible in a
substantial number of patients, and by all reports
improves neonatal outcome. Because neonatal
mortality drops significantly after 32 (0/7) weeks
while maternal morbidity remains high until deliv-
ery, this study and previous reports support elec-
tive delivery at 32 weeks.—KDW)
582 Obstetrical and Gynecological Survey

Skin Adhesive Versus Subcuticular Suture

for Perineal Skin Repair After Episiotomy:
A Randomized Controlled Trial
Raquel Mota, Fernanda Costa, Alexandra Amaral Fátima Oliveira,
Cristina Costa Santos, and Diogo Ayres-De-Campos
Department of Obstetrics and Gynecology, S. João Hospital, Porto, Portugal; Department of Biostatistics and
Medical Informatics, Centre for Research in Health Technologies and Information Systems, Porto Faculty of
Medicine, University of Porto, Porto, Portugal; and Department of Obstetrics and Gynecology, Porto Faculty
of Medicine, Biomedical Engineering Institute, University of Porto, Porto, Portugal
Acta Obstet Gynecol Scand 2009;88:660–666

ABSTRACT
During the puerperium, surgical repair of perineal lesions is frequently associated with pain and discomfort interfering
with the normal daily activities. The use of skin adhesive glue (octyl-2-cyanoacrilate) is a relatively new method for
repair of cutaneous lesions that appears to be a safe and rapid alternative to a continuous subcuticular suture of the
perineal skin, a traditional method commonly used in many countries. Preliminary studies suggested several advantages
of skin adhesive over other types of suture. However, these studies had small sample sizes, lacked a control group, were
not randomized or did not use subcuticular suture for comparison.
This randomized controlled trial compared the skin adhesive glue with continuous subcuticular suture for perineal skin
repair after episiotomy. A total of 100 women having mediolateral episiotomy at vaginal delivery were randomized to
receive either skin adhesive (n ⫽ 53) or subcuticular suture (n ⫽ 47) for closure of perineal skin. The primary study
outcome was self-assessed perineal pain during the first 30 days after delivery. Secondary outcomes included technical
difficulties and duration of the perineal repair procedure, wound complications at 42 to 68 hours or in the first 30 days
after repair, and the reinitiation of sexual activity by 30 days postpartum. To evaluate these outcome measures, patients
were asked to complete a daily questionnaire.
Forty-nine (92%) of the women in the skin adhesive group and 37 (79%) of those in the subcuticular suture group
returned the questionnaire. No significant differences between the 2 groups were reported in the incidence of pain,
technical difficulties, or failure of the perineal repair procedure. In addition, no significant differences were observed
in the number of wound complications observed at 42 to 68 hours, in reinitiation of sexual activity by the 30th day
postpartum, self-reported measures of pain at 7 and 30 days, or the need to be evaluated at a healthcare facility within
30 days. The only significant difference between the 2 study groups was a 4 minute shorter duration of wound repair
with the skin adhesive.
Compared to continuous subcuticular suture repair of episiotomy, the only advantage of skin adhesive devices appears
to be a slightly shorter duration of surgical repair, which is of questionable clinical significance. The investigators
conclude that these findings provide no evidence to support the use of skin adhesive devices in episiotomy repair.

EDITORIAL
(In this interesting study, Mota et al random-
ized women who had episiotomies, but no vag-
inal or other lacerations, to undergo a standard
perineal repair until the perineal skin was
reached, and then to have either subcuticular
closure of the skin or skin closure using an
adhesive. The investigators stated that they re-
cruited only women with mediolateral episioto-
mies to standardize the type of incision as much
as possible, reasoning that each perineal lacer-
ation is different and including such lacerations
COMMENT
would make it hard to insure that the lacera-
tions in each group were similar. All episioto-
mies were repaired in the standard manner,
with continuous suturing of the vaginal wall
and continuous closure of the perineal mus-
cles and subcutaneous tissue, until the skin
was reached. At that point, the perineal skin
was closed using either a continuous subcu-
ticular suture of polyglactin 910 or skin adhe-
sive applied with the fingers, using Kocher
clamps to appose the skin edges.
 Surgical Technique in Obstetrics
Amazingly, it took them only a year to recruit
100 patients—amazing both because the study
was done on a midwife service, and midwives
usually try to avoid any kind of episiotomy, and
because a mediolateral episiotomy is consid-
ered by many to be fairly invasive and thus
rarely used; a true mediolateral incision is so
deep that it results in exposure of the fat in the
ishiorectal fossa (Hankins GDV, Clark SL Cun-
ningham FG, et al. Operative Obstetrics. Nor-
walk, CT: Appleton and Lange, 1995). However,
because this type of incision can be difficult to
close and has been associated with more post-
operative pain and dyspareunia than midline in-
cisions, the investigators certainly gave the skin
adhesive a challenging test. The investigators
found that both skin closure techniques were
adequate and very similar in terms of pain dur-
ing or after the procedure, wound complica-
tions, and time to initiation of sexual activity. In
fact, the only significant difference was that
when adhesive was used the repair took 4 fewer
minutes than when a subcuticular stitch was
placed.
Interestingly, the investigators noted that the
skin adhesive peeled off early in many cases,
sometimes by the third day, although this did
not appear to affect the result. The reason this is
interesting is that many authorities do not be-
lieve that closure of the perineal skin is even
necessary—thus loss of the adhesive before the
skin had healed would not be expected to have
any effect on ultimate wound healing. The super-
fluousness of perineal skin closure has been
shown by many studies. For example, Gordon et
al randomized 1780 women who had an episiot-
omy or first or second degree perineal laceration
to either a standard 3 stage repair including clo-
sure of the skin with a subcuticular suture or a 2
stage repair with no skin closure (Br J Obstet
Gynecol 1998;105:435). There were no differ-
ences between the groups in terms of perineal
583
pain at 24 to 48 hours or wound breakdown. How-
ever, compared to the stitched group, the group
whose perineal skin had been left open had less
perineal pain at 3 months and a greater number
had resumed intercourse. In a similar study,
Kindberg et al (Br J Obstet Gynecol 2008;115:
472) randomized 395 women with a second de-
gree laceration or an episiotomy to either stan-
dard repair with interrupted stitches using
inverted knots or repair of the vagina and peri-
neum with a continuous suture but the skin left
open. The 2 techniques resulted in similar out-
comes in terms of perineal pain, wound healing,
dyspareunia, need for resuturing, and patient
satisfaction, but the open skin technique was
faster and required less suture material.
Although it is good to know that perineal skin
can be safely closed with skin adhesive rather
than sutures, it seems more useful to know that
skin closure is probably not necessary. As long
as the anal sphincter is intact or has been re-
paired, the vaginal mucosa has been closed,
and the crown stitch placed to reapproximate
the bulbocavernosus muscles and the labia, the
skin can be left open. The skin edge must of
course be hemostatic; as a fellow I had an older
professor who clamped the perineal skin edges
together with Kochers for a few minutes at the
end of the repair to compress any bleeders
along the skin edge, which he would then leave
unsutured. I cringe thinking about that tech-
nique, but of course there are many other less
painful ways to control skin bleeding. Restora-
tion of the underlying perineal anatomy results
in the cut skin edges being reapproximated
without suturing, and the absence of subcutic-
ular stitches (or adhesive) reduces postopera-
tive and long-term pain. Leaving the skin open
also reduces the amount of time needed to
complete the repair, which will no doubt be
appreciated by the mother who is trying to bond
with her infant.—KDW)
584 Obstetrical and Gynecological Survey

A Randomised Controlled Trial of

Amniotomy and Immediate Oxytocin
Infusion Versus Amniotomy and
Delayed Oxytocin Infusion for
Induction of Labor at Term
Dan Selo-Ojeme, Pradnya Pisal, Olalekan Lawal, Cathy Rogers,
Abhijeet Shah, and Smitha Sinha

Department of Obstetrics and Gynaecology, Women’s Health Division, Barnet; and Chase Farm Hospitals
NHS Trust, The Ridgeway, Enfield, United Kingdom
Arch Gynecol Obstet 2009;279:813–820

ABSTRACT
The combination of amniotomy and intravenous oxytocin is believed to be more effective for labor induction than
amniotomy alone. However, the optimal time interval between amniotomy and administration of intravenous oxytocin
has not been determined in well-accepted, prospective randomized studies. This randomized controlled trial compared
2 groups of women 4 hours after amniotomy: (1) women who had an amniotomy followed by immediate oxytocin
infusion (immediate group, n ⫽ 61) and women who underwent an amniotomy with delayed oxytocin infusion (delayed
group, n ⫽ 62). The proportion of women in labor within 4 hours of the start of induction and the proportion who
delivered within 12 hours of amniotomy were compared. Women with planned induction of labor who met eligibility
criteria were enrolled between 2006 and 2007. Participants were asked to complete a questionnaire after delivery that
assessed their satisfaction with the induction process.
Women in the immediate group were significantly more likely to be in established labor 4 hours after amniotomy
than women in the delayed group (relative risk [RR], 12.8; 95% confidence interval [CI], 55.1–111.7, P ⬍ 0.001).
They were also more likely to have a shorter median period from amniotomy to vaginal delivery (P ⬍ 0.001), and
to achieve vaginal delivery within 12 hours of amniotomy (RR, 1.5; 95% CI, 1.0–12.6, P ⫽ 0.015). No difference
was found between the groups in terms of mode of delivery or perinatal outcomes, including incidence of uterine
hyperstimulation, the incidence of abnormal fetal heart rate during labor, or the occurrence of postpartum
hemorrhage. There were no perinatal deaths. Satisfaction questionnaires were returned by 75 (60.9%) women, 39
in the immediate group and 36 from the delayed group. Compared with the delayed group, significantly more
women in the immediate group were satisfied with the induction process (RR, 4.1; 95% CI, 1.1–16.1, P ⫽ 0.03),
their care during labor (RR, 2.1; 95% CI, 1.1–5.2, P ⫽ 0.02), and the duration of labor (RR, 1.8; 95% CI, 1.0–3.3,
P ⫽ 0.02).
These findings indicate that immediate oxytocin infusion after amniotomy results in a greater proportion of women
in established labor at 4 hours and achieving vaginal delivery within 12 hours, and a shorter amniotomy to delivery
interval. Patient satisfaction with induction of labor appears to be closely related to the speed of the induction process.
The apparent lack of effect of the induction regimens on other maternal and neonatal outcomes is inconclusive because
of the small sample size.

EDITORIAL COMMENT

(This study, conducted in London from De-
cember 2006 to September 2007, was de-
signed to determine when oxytocin should be
started during labor induction by amniotomy.
Apparently, amniotomy is a relatively common
way to induce labor in England, and whether and
when to start intravenous oxytocin remains a mat-
ter of debate. The investigators recruited 123 nul-
liparas who had singleton vertex pregnancies at
ⱖ37 weeks’, intact membranes, no regular uterine
 Labor Induction
contractions, and favorable cervices, defined as
Bishop score ⱖ6. Amniotomy was performed in
all women, after which half were given intrave-
nous oxytocin immediately, and half were given
oxytocin 4 hours later, if active labor had not
commenced. The immediate oxytocin group
went into active labor earlier, delivered earlier,
and was more satisfied with their experience
than the delayed oxytocin group.
Several aspects of this trial are confusing—
the most important being that, although the
protocol states that participants received only
amniotomy and oxytocin, one of the tables
indicates that more than half the patients in
each study group also received vaginal pros-
taglandin. Since the dose and timing of this
intervention is not provided, its relationship to
the induction protocol is unclear—but the par-
ticipants most likely got vaginal prostaglandin
sometime before their amniotomy was per-
formed, and possibly within a few hours of
induction. Thus there were really 4 treatment
arms, not 2—prostaglandin and immediate
oxytocin, no prostaglandin and immediate
oxytocin, prostaglandin and delayed oxytocin,
and no prostaglandin and delayed oxytocin—
although the data was not analyzed this way.
Twenty percent of the immediate oxytocin
group and 27% of the delayed group were
delivered by cesarean and 67% and 53% of the
cesareans in each group were for “fetal dis-
tress,” although no data is provided to explain
why. Prostaglandin is associated with a 1% to
5% rate of uterine hyperstimulation, which can
occur as long as 91⁄2 hours after placement of
the vaginal insert (Induction of Labor. ACOG
Practice Bulletin 10, November 1999) Aggres-
sive oxytocin dosing can also cause hyper-
stimulation (the investigator’s oxytocin protocol
is not provided), and early amniotomy can result
in severe variable decelerations. It is, therefore,
frustrating that the investigators did not indicate
which intervention was most closely associated
with “fetal distress.” Importantly, no information
about intrapartum infection or fever during labor
is provided, even although that is one of the
main complications of early amniotomy. Need-
less to say, it is hard to draw any conclusions
from this study.
585
Fortunately, there have been other similar
but better designed studies from which con-
clusions about optimal induction manage-
ment, including the timing of amniotomy, can
be drawn. One very well done study of early
(as soon as it was considered safe) versus late
(when cervical dilation reached ⱖ5 cm) am-
niotomy for labor induction was performed by
Mercer et al (Am J Obstet Gynecol 1995:173:
1321). This study included 209 women with
singleton, vertex, term pregnancies who had
not had cervical ripening or any kind of geni-
tourinary tract infection. The study subjects
were randomly assigned to each group before
initiating oxytocin, and the oxytocin was ad-
ministered according to a standard protocol in
which the dose was doubled either every 30 or
every 60 minutes, up to a maximum of 32
mU/min, until there was a consistent contrac-
tion pattern of 3 to 5 contractions in 10 min-
utes. Compared to the late amniotomy group,
the early group had more severe variable de-
celerations and a significantly higher inci-
dence of chorioamnionitis, mainly in the group
in which oxytocin was increased every 60 min-
utes (22: vs. 6.8%; P ⫽ 0.002). Although there
were a similar number of cesareans in each
group, 6 women in the early amniotomy
group required cesareans for moderate-to-
severe variables associated with fetal dis-
tress, whereas none of the women in the late
amniotomy group required cesarean for this
reason. Women in the early amniotomy group
whose oxytocin was increased at 30 minute
intervals had a significantly shorter time to
delivery (11.2 hours) compared to women who
had late amniotomy or had early amniotomy
with their oxytocin dose increased every 60
minutes (18.0 hours for late amniotomy and 60
minute oxytocin dosing; 17.7 hours for late
amniotomy and 30 minute oxytocin dosing,
and 15.3 hours for early amniotomy with 60
minute oxytocin dosing, not significantly dif-
ferent). The investigators concluded that both
early and late amniotomy had advantages
and disadvantages, which supported a more
individualized approach to labor induction.
For example, growth restricted or otherwise
stressed fetuses might not tolerate repetitive
586 Obstetrical and Gynecological Survey

variable decelerations after early amniotomy
or rapid increases in the oxytocin rate, and
might benefit from later amniotomy and 60
minute oxytocin dosing. In contrast, the
woman with a favorable cervix, no risk factors
for infection, and an AGA fetus might benefit
from early amniotomy and 30 minute oxytocin
dosing.
Although Selo-Ojeme et al did not prove their
hypothesis, there are other published data vali-
dating several different induction protocols,
which support some individualization of induc-
tion management. Selecting the best induction
protocol for a particular patient probably leads
to more patient satisfaction than trying to man-
age every patient the same way.—KDW)