- Email: [email protected]
A clinical trial of induction of labor versus expectant management in postterm pregnancy
A clinical trial of induction of labor versus expectant in management postterm pregnancy The National Institute of Child Health and Human Development Network of Maternal-Fetal Medicine Units OBJECTIVE:Managementof the uncomplicated pregnancy prolonged beyond the estimated date of confinementIs controversial,particularlywhen the cervix Is unfavorablefor Induction. The benefit of reducing potential fetal risk with Induction of labor must be balanced against the morbidity associated with this procedure. The objective of this study was to compare two strategiesfor managing postterm pregnancy (i.e., ImmediateInduction and expectant mangement). STUDY DESIGN: Four hundred forty patients with uncomplicated pregnanciesat 41 weeks' gestation were randomizedto either immediate Induction of labor (n = 265) or expectant management (n = 175). Patientswith expectant managementunderwent nonstresstesting and amnloic fluid volume assesment twice per week. PatientsIn the induction group underwent Induction within 24 hours of randomization.To evaluatethe efficacy of Intracervicallprostaglandin E2 gel, patients In the Induction group were randomized In a 2: 1 scheme to receive either 0.5 mg prostaglandin E2 gel or placebo gel intracervically 12 hours before Induction of labor with oxytocin. RESULTS:The Incidence of adverse perinatal outcome (neonatalseizures, Intracranialhemorrhage,the need for mechanicalventilation,or nerve Injury) was 1.5% In the Induction group and 1% in the expectant managementgroup (p > 0.05). There were no fetal deaths In either group. There were no differences In mean birth weight or the frequency of macrosomia(birth weight ;-- 4000 gm) between the two groups (a > 0.05). Regardlessof parity, prostaglandin E2 Intracervicalgel was not more effectivethan placebo in ripening the cervix. The cesarean delivery rate was not significantlydifferent In the expectant (18%), prostaglandin E2 gel (23%).or placebo gel (18%) groups. , CONCLUSIONS:Adverse perinatal outcome In otherwise uncomplicated pregnanciesof ý!41 weeks Is very low with either of the managementschemes described. Thus from the perspective of perinatall 1994;170:716-23.) morbidity or mortality either managementscheme is acceptable. (AmJ OesTuGYNECOL
Key words: Postterm pregnancy,prostaglandin E2 gC' The postterm.pregnancy is one that extends beyond the forty-second week of amenorrhea. The management of such a pregnancy that is otherwise uncomplicated is controversial.' Central to this controversy is whether the fetus is at increasingrisk of deterioration as the pregnancyadvances.Two managementschemesare often used to manage the postterm pregnancy. In one, pregnancy is allowed to progress to 42 weeks and beyond. Labor is induced only if the cervix is well effaced or dilated, or both, or if fetal compromise occurs. 'Me fetal condition is evaluated regularly by Members of thestudygroupandparticipatinginstitutionsarelistedat the endof thearticle. Supportedby grants HD 21410, HD 21434, HD 21366, HD 21380, HD 19897, HD 21414, HD 21386, HD 21366, and HD 21363 ftom the National Institute of Child Health and Human Development. for publicationApril 13,1993; revisedAugust 4,1993; Received October27,1993. accepted Reprintrequests: DonaldMcNellis,MD, NationalInstituteof Child Health and Human Development, National Institutesof Health, EPN Building, Room643, Bethesda, MD 20892. 611152458 716
various techniques. In the second scheme, labor is aggressivelyinduced at 42 weeks or earlier. Cervical ripening agents such as prostaglandins' are used to prepare the cervix and, if necessary,oxytocin and amniotomy are also used. Recently,severalprospective randomized trials comparing these two plans of management have yielded contradictory results." 'Me purpose of the current study was to compared thesetwo managementschemes in a subset of women whose pregnancies reached 41 weeksbut were otherwiseuncomplicated. We also compared the cervical ripening effects of 0.5 mg prostaglandin E2 (PGE2) and placebo gels placed into the cervix. This comparison is pertinent to the management of postterm pregnancy. Methods The trial was designed and implemented by the participants of the Maternal-Fetal Medicine Units Network under the direction and sponsorhsip of the National Institute of Child Health and Human Development. An investigationalnew drug (IND) exemption for
Volume 170, Number 3 Am I Obstet Gynecol
the use of PGE2 gel was obtained from the Food and Drug Administration. Informed consent was obtained from each participating subject. Patient selection. Pregnant patients identified as having a gestational age of at least 287 days from the last menstrual period were screened for eligibility. Patients were excluded from the trial if they had any medical or obstetric complications requiring induction of labor, cesarean section, or frequent monitoring of maternal or fetal condition. Additionally, patients with by 4500 fetal clinigm, an estimated weight exceeding cal or ultrasonographic examination, were also excluded. The presence of any one of the following five criteria was accepted as evidence of accurate gestational dating: (1) a clearly defined menstrual period and audible fetal heartbeat documented for at least 21 weeks by fetoscope or at least 30 weeks by a Doppler device, (2) a clearly defined menstrual period and uterine size estibe by to 524 mated weeks physical examination at : compatible with the duration of amenorrhea, (3) a clearly defined menstrual period and a positive blood in enough or urine pregnancy test result obtained early 41 exceeded pregnancy to assure that the gestation 32 least for documented (4) feal heartbeat at weeks, a weeks by Doppler device if the date of the last menstrual period was uncertain, or (5) an ultrasonographic estimate of gestational age obtained before 26 completed weeks' gestation. Only patients considered to have a gestational age of 287 days but <301 days (41 to 43 completed weeks' gestation) were further evaluated for enrollment into the trial. Randomization. Patients meeting these inclusion criteria and willing to participate in the trial underwent a prerandomization evaluation that included cervical examination to determine a Bishop score' (maximum score of 13), a nonstress test, and an estimation of amniotic fluid volume by ultrasonography. ' Patients were excluded from the trial if the modified Bishop score was _-7, if the nonstress test result was not reassuring (nonreactive or showing decelerations), or if the largest pocket of amniotic fluid was < 2.0 cm. Eligible patients were then randomized to one of three treatment groups according to a computer-assigned randomization scheme arranged by the data coordinating center. Randomization was stratified by clinical site and week of gestation (two strata). Randomization to treatment group was done according to a 2: 2: 1 ratio: expectant management, cervical priming with PGE2 gel followed 12 hours later with oxytocin, or cervical priming with placebo gel followed 12 hours later with oxytocin. Management protocol. Women randomized to the expectant management group were evaluated weekly
NICHHD Network of Maternal-Fetal Medicine Units
717
with a cervical examination and twice weekly with a nonstress test and an ultrasonographic estimation of labor Spontaneous fluid wasawaited, volume. amnionic but induction of labor could be undertaken if any of the following situations occurred: the Bishop score exceeded6, the estimatedfetal weight exceeded4500 gm, delivery developed, indication for obstetric or medical a the largest pocket of amniotic fluid was < 2.0 cm, or an abnormal nonstress test result occurred (nonreactive fetal heart rate or a fetal heart rate with mild variable decelerations exceeding 15 beats/min and lasting 15 by followed a positive contraction stress test seconds) result. If the contraction stresstest wasnegativeand the fetal heart rate remained persistently nonreactive,testing wasrepeated in 24 hours. Patientsin the expectant by 308 comundelivered group who were management pleted days (44 weeks)were releasedfrom the protocol and managed by the method appropriate for the clinical situation. Subjectsin the induction group underwent induction within 24 hours after randomization with instillation into the cervix of either PGE, gel or placebo gel. The investigatorswere unawareof and unable to determine whether the gel contained placeboor active drug. Both PGE2(0.5 mg) and placebo gels were prepackaged by The Upjohn Company for researchuse in a sterile unit dose syringewith a modified catheter and safetydevice to limit the depth of insertion into the intracervical canal to I to 2 cm and to prevent extraamniotic application of gel. Before gel insertion, a modified Bishop scorewas performed. The fetal heart rate and uterine contractions were monitored continuously by noninvasive meansfor at least 4 hours after get administration. If labor did not ensue within 12 hours, a cervical examination was performed and the Bishop score was determined. We used only a single application of PGE2 gel becausethere is no apparent benefit of repeated applications." When clinically feasible,amniotomy was performed before the administration of oxytocin. If amniotomy was not feasible,oxytocin infusion was initiated according to a uniform protocol. If the patient had not entered the active phaseof labor after 24 hours of oxytocin administration, a cesareansection was performed or induction of labor wascontinued for a longer time. For all patients randomized in the trial, decisions regarding the need for cesareandelivey and the use of antibiotics for infection (urinary tract infections,chorioby the made were amnionitis, or endomyometritis) physicianswho were taking care of these patients and by However, dictated the the protocol. not study were indications for cesarean delivery and for the use of antibiotics were recorded. Continuous electronic recording of the fetal heart The in labor during each patient. performed was rate
718 NICHHDNetwork of Maternal-FetalMedicine Units
fetal heart rate tracings were evaluated in terms of baseline heart rate, beat-to-beat variability, and the presenceof decelerations.Variable decelerationswere defined as severeif the heart rate decreasedbelow 60 beats/min and lasted a 60 seconds.The fetus was considered to be in "distress" if repeated late or severe variable decelerationsoccurred. In each patient the presenceof meconium wasnoted either at the time of amniotomy or subsequentlyduring labor. At delivery, every infant with thick meconium in the amniotic fluid had immediate suctioning of the oropharynx on delivery of the head, Outcome variables. For the comparisonof expectant management and immediate induction, the primary outcomevariablesincluded perinatal or maternal death or perinatal morbidity (a composite variable including neonatal seizures, intracranial hemorrhage, the need for mechanicalventilation, Erb's palsy,or other brachial plexus or facial nerve injury). Secondaryoutcome variablesincluded the cesareandelivery rate, the incidence of maternal infection, the need for maternal blood transfusion, the presence of severe variable or late decelerationsduring labor, the incidence of a 5-minute Apgar score <4, the presence of neonatal meconium aspiration, and the neonatal birth weight. For the comparison of PGE2 and placebo gels, the primary outcome variables were the. gel insertion-delivery interval, the change in Bishop score for the 12 hours after gel insertion, and the incidence of failed induction (i.e., the failure to enter the active phase of labor). Secondaryoutcome variables included the incidencesof hyperstimulation or cesareandelivery for fetal "distress" after gel insertion and nonclosure of the ductus arteriosus. Statistical analysis. For comparison of induction and expectant management,we anticipated an incidence of 2% for the composite primary outcome variablesin the group of patients managed expectantly. A sample size of 2800 patients would be required to demonstrate a 50% reduction in the incidence of the primary outcome variablesin the group undergoing immediate induction of labor. This provided a type I error of 0.05 (onesided) and a power of 0.80. To determine sample size for the comparison of PGE2and placebo gels, we assumedthat the incidence of failed induction would be the variable least likely to be reduced by PGE2 gel treatment so we calculated sample size on the basisof this variable. We assumeda failed induction rate of 25%.A sample sizeof 750 (500 receiving PGE2 and 250 receiving placebo gel) would have a 90% power (a = 0.10,0 = 0.05) to detect a 40% reduction in the rate of failed induction by PGE2 gel, Statistical comparisons were performed for ordinal valued variables (Bishop score, gel insertion-delivery interval, and latent phase duration) with the Kruskal-
March 1994 Am j Obstet Gynecol
Wallis rank sum test. Categoric measurementswere compared with the y,' test. All primary outcome analyseswere basedon the total cohort of patients randomized into the trial. Patients were included in their randomly assigned treatment group, and treatment group assignmentwas not altered on the basis of the patient's adherenceto the assignedtreatment regimen. We planned an interim analysisto reassessthe incidence of adverse outcomes and to recalculate study power on the basis of the observed incidence rates. After 440 patients were enrolled in 18 months, the incidence of adverseoutcome wasonly 1.1%.A sample size in excessof 5600 patients would thus be required to adequately test the hypothesis proposed. On the basis of the low incidence of adverse effects and the anticipated annual rate of recruitment, the Data and Safety Monitoring Committee recommended that the study be stopped. Results BetweenDecember 1987 and July 1989we screened 4566 pregnant women who had reached a gestation of ;ý41 weeks.Of these,4126 were ineligible for the study: 2428 (59%) becauseof uncertain gestational age, 650 (16%) becauseof a medical or obstetric complication, 549 (13%) becauseof a Bishop score of Z!7,72 (2%) becauseof decreasedamniotic fluid volume, and 100 (2%) becauseof a nonreactive nonstresstest. An additional 323 patients (8%), although eligible, refused to participate in the study. Finally, in four patients no reason for exclusion was available.Of the 440 patients randomized, 175 were randomized to the expectant management arm and 265 to the induction arm. Of those in the induction arm, 91 received placebo gel whereas 174 received PGE2gel into the cervix. Table I comparesselectedcharacteristicsof the study population at the time of randomization. When comparison by treatment was done, there were no differencesin any of the demographic variableslisted. Additional variables compared at randomization and not found to differ significantly by treatment group include incidence of previous stillbirth, neonatal death, elective abortion, prepregnancy weight and height, estimated fetal weight by ultrasonography, and largest amniotic fluid pocket. The vast majority of patients (80%) were randomized at a gestationalage of 41 to 42 weeks(287 to 293 days). The cervix in these patients was not favorable for induction as evidencedby the low Bishop score. Table II summarizesthe primary outcome variables for the induction and expectantly managed groups. There were no perinatal or maternal deaths, and no infant had an intracranial hemorrhage. In the entire study group only 5 infants (1.1%) experienced one or more of the outcomes listed in Table 11.In the PGE2
Volume 170, Number 3 Arn] Obstet Gynecol
NICHHD Network of Maternal-Fetal Medicine Units
719
Table 1. Characteristics of study population at randomization Induction of labor
Age (mean and SD) Parity (No. ) 0 >1 Missing Previous postterm birth (No. ) Race (No. ) Black White Not available Gestational age at entry (No. ) 287-293 days 295-301 days Bishop score at randomization Mean and SD Median and range
PGE2 (n = 174)
Placebo (n = 91)
25.4 (5.7)
25.4 (5.3)
105 66 3 16
54 37 0 4
(60%) (38%) (2%) (9%)
(59%) (41%) (0%) (4%)
Expectant management (n = 175) 26.1 (5.8) 94 79 2 12
(54%) (45%) (1%) (7%)
56 (32%) 117 (67%) 1 (1%)
34 (37%) 57 (63%) 0 (0%)
67 (38%) 105 (60%) 3 (2%)
141 (81%) 33 (19%)
72 (79%) 19 (21%)
139 (79%) 36 (21%)
4.0 (1.4) 4 (0-6)
3.8 (1.4) 4 (1-6)
3.9 (1.5) 4 (0-6)
Table 11. Incidence of primary outcome variables Induchon of labor PGE, Patients (No. ) Mechanical ventilation (No. ) Nerve injury (No. ) Seizures (No. ) Babies with (No. ) I adverse outcome -ý
174 011 100 021 I (I %)
group one infant of 41 weeks' gestation and weighing 3647 gm had a brachial plexus injury. In the placebo group two babies (one of 41 weeks' and another of 42 weeks' gestation) had seizures only and another baby of 41 weeks 3 days' gestation required mechanical ventilation only. None of these infants was asphyxiated at delivery. Neither of the two infants with seizures required antiseizure medication. In the expectant management group, one infant of 41 weeks' gestation had seizures and required antiseizure medication. This baby also required mechanical ventilation. The baby was unasphyxiated at delivery but had a hypoplastic right ventricle. Maternal and perinatal outcome variables are sumniarized in Tables IIIA and 11113.The randomizationdelivery interval was significantly greater in the group managed expectantly than in either induction group (85 vs 35 and 36 hours). The range of values for the induction-delivery interval is similar in all three groups because in this analysis we did not exclude those patients in whom the protocol was violated. If such patients are excluded, the induction-delivery interval is still significantly greater in the expectantly managed group. Gestational age at delivery was significantly greater in the expectantly managed patients. Sixty-one percent of
Placebo
Expectant management
91
175
3 (3%)
women in the expectantly managed group was delivered at 294 days (42 weeks) or later. Only 35% of women in the two induction groups were delivered at 42 weeks or later. Uterine hyperactivity occurred in three women; one received placebo gel and two received PGE2 gel. The incidence of cesarean delivery was similar in each of the three treatment groups (i. e., 23%, 18%, and 18%, respectively, in the PGE21 placebo, and expectant groups). The combination of indications generally considered to represent cephalopelvic disproportion (i. e., arrest of dilatation, arrest of descent, and protracted descent) was the most commonly listed indication for cesarean delivery, accounting for 56%, 38%, and 56%, respectively, of cesarean deliveries in the PGE21placebo, and expectant management groups. The cumulative frequency of these three indications did not differ among the treatment groups. The frequency of failed induction leading to cesarean section was also similar in the three treatment groups (i. e., 15%, 19%, and 11%, respectively). Table IIIA demonstrates that of the 175 women in the expectant management group 55 (3 1%) underwent induction of labor whereas 118 (69%) labored spontaneously. One woman underwent cesarean delivery with-
720 NICHHDNetwork of Maternal-FetalMedicine Units
Table IIIA.
March 1994 Am j Obstet Gynecol
Matemal outcome Induction of labor
Randomization-delivery interval (hr, median and range) Gestational age at delivery* 287-293 days 294-301 days > 302 days Not available Matemal infection (No. ) Need for transfusion (No. ) Uterine hyperactivity (No. ) Route of delivery (No. ) Vaginal Cesarean Not available
1
PGE2 (n = 174)
Placebo (n = 91)
36 (6-492)
35 (7-487)
85 (5-538)*
60 29 2 0 13 0 1
67 83 24 1 25 3 0
111 60 2 1 33 2 2
(64%) (34%) (1%) (1%) (19%) (1%) (1%)
(66%) (32%) (2%) (0%) (14%) (0%) (1%)
75 (82%) 16 (18%) 0 (00/c)
134 (77%) 39 (22%) 1 (1%)
Expectant management (n = 175)
(38%) (47%) (14%) (1%) (14%) (2%) (0%)
I 142 (81%) 32 (18%) 1 (1%)
< 0.001 by Kruskal-Wallis test.
Table 11111.Neonatal outcome Inductionof labor PGE2 Babies (No. ) Apgar score <4 at 5 min (No. ) Late decelerations in labor (No. ) Meconium (No. ) None 'Min '17hick Not available Meconium aspiration pneumonia (No. ) Birth weight (gm, mean and SD) ;! 4000 gm (No. ) &4500 gm (No. )
174 0 1 131 31 10 2 1 3607 27 1
out labor, and in one woman data regarding the onset of labor were not available.Of thoseundergoing induction of labor the indications were as follows: favorable cervix (n = 18), ruptured membranes (n = 6), nonreassuring nonstress test result or low amniotic fluid volume (n = 23), and "other" indications (n = 8). Table IIIB summarizesneonatal outcome variables. The incidence of asphyxia was low in all groups as evidenced by the infrequent occurrence of a low 5minute Apgar score or late decelerations of the fetal heart rate. Meconium was detected in amnionic fluid with similar frequency in all treatment groups, and in 9% the meconium was describedas thick. Interestingly, of the four infants in whom meconium aspiration pneumonia developed, all had thick meconium detected during labor. None of the 365 infants with thin or absent meconium had meconium aspiration pneumonia. The mean birth weight and the percentage of infants with a birth weight >4000 gm or >4500 gm were similar in each group.
(75%) (18%) (6%) (1%) (382) (16%) (1%)
Placebo
Expectant management
91 0 0 (0%)
175 1 (1) 3 (2%)
71 12 7 1 1 3532 16 3
(78%) (13%) (8%) (1%) (464) (18%) (3%)
109 42 20 4 2 3606 31 6
(62%) (24%) (12%) (2%) (440) (18%) (4%)
Table IV summarizesthe effect of placebo and PGE2 gels on the onset of persistent contractions and on parameters of cervical ripening in the two induction groups. This subgroup analysisincludes only patients for whom the induction protocol was followed. In nulliparous women only, placebo gel was significantly less effective than PGE2in starting persistent contractions. PGE2gel and placebogel, however,were comparablein all other parametersmeasured.The increasein Bishop score, the gel insertion delivery interval, the duration of the latent phaseand the number of women delivered within 12 hours of gel instillation were not significantly different among treatment groups regardlessof parity. Comment This study was stopped becausethe incidence of the primary outcome variable (perinatal mortality or morbidity) was only 1%. Thus a sample size in excessof 5600 would be required to adequatelycompare the two schemesdescribed. Becauseadverseperinatal outcome
Volume 170, Number 3 Am j Obstet Gynecol
NICHHDNetwork of Maternal-FetalMedicine Units 721
Table IV. Effect of placebo and PGE, on contractions and cervical ripening Nulliparouswomen Placebo (n = 44) Cases with persistent contractions* (No. ) Bishop scoreT (median and range) At gel insertion After 12 hr Increase in Bishop scorej (No. ) 0-1 2-3 a4 Not available Gel insertion-delivery interval (hr, median and range) Latent-phase duration (hr, median and range) Delivered within 12 hr (No. )
8 (19)t
4 (1-6) 5 (1-10) 23 10 9 2 33
(52%) (23%) (20%) (5%) (7-90)
9 (< 1-66) 0 (0%)
PGE2 (n = 93) 45 (48)
4 (1-7) 6 (0-12) 47 14 32 0 34
(51%) (15%) (34%) (0%) (7-79)
10 (< 1-57) 3 (3%)
MuRiparouswomen Placebo (n = 34) 14 (41)
4 (1-6) 5 (1-10) 18 6 9 1 27
(53%) (18%) (26%) (3%) (9-70)
7 (< 1-54) 3 (9%)
T
PGE2 (n = 60) 23 (38)
4 (0-6) 6 (0-10) 27 11 20 2 25
(45%) (18%) (33%) (3%) (3-58)
5 (< 1-41) 10 (17%)
Three PGE2 subjects were excluded from the table because parity was undetermined. Thirty-one subjects (13 PGE21 18 placebo) were excluded because the induction protocol was not followed. Of these, 24 did not receive gel or placebo and 7 went home undelivered. *After gel administration but before the start of oxytocin. tSignificant difference when compared with nulliparous women receiving PGE2. TWomen who were delivered vaginally within 12 hours of gel instillation were considered to have a Bishop score of 10 and a change in Bishop score >4.
is so uncommon in postterm pregnancy, the cost of such a trial is not justified. Furthermore, the low prevalence of postterm pregnancy in a well-dated population that is eligible for expectant management (i. e., an unripe cervix, a normal nonstress test result, and a normal amniotic fluid volume) makes such a trial impractical. Thus from the clinician's perspective either expectant management or immediate induction is an acceptable option for the otherwise uncomplicated postterm pregnancy. Postterm pregnancy has historically been considered .. Before a risk factor for adverse perinatal outcome .... the introduction of fetal surveillance techniques, prolonged pregnancy had been associated with a twofold to tenfold increase in perinatal mortality and a twofold to fourfold increase in the incidence of fetal distress. " Induction of labor emerged as a means of reducing perinatal risks in the prolonged pregnancy. The development and application of modern techniques of fetal assessment have been associated with a reduction in perinatal risk in the prolonged pregnancy. In 13 studies betwen 1978 and 1987 in which antenatal fetal surveillance was used for follow-up of the postterm pregnancy, the risk of perinatal mortality was similar to that of pregnancies delivered at term. ' Such reports demonstrated that expectant management of the postterm pregnancy was an acceptable alternative to the induction of labor. However, several recent studies have suggested that in spite of modern monitoring tech-
niques the postterm fetus remains at risk for certain perinatal morbidities such as meconium aspiration, fetal distress in labor, and macrosomatiawith its attendant complications,'"' These findings have rekindled the controversy surrounding the optimal management of the prolonged pregnancy. In response,severalprospective randomized trials using contemporary management schemeshave compared induction and expectant management in the prolonged pregnancy."' These trials have yielded conflicting results that have been attributed to differences in patient selection (ripe vs unripe cervix at entry), methods of labor induction (PGE2with or without oxytocin, amniotomy, or stripping of the membranes),and techniquesof anteparturn fetal surveillance.' These conflicting results leaveunresolvedthe issueof management of the prolonged pregnancy. Therefore we undertook this trial to assessthe role of labor induction with or without PGE2gel in the management of prolonged pregnancy. Patients were enrolled at 287 days (41 weeks)becauserecent data indicate an ý-_ increased risk of perinatal complications beyond this gestational length. " The study was limited to patients with an unfavorablecervix (modified Bishop score !56) and a normally grown, uncompromised fetus because the management of this group of women is the most vexing one when pregnancy is prolonged. We found no difference in neonatal mortality or morbidityý cesarean delivery rates, or maternal morbidity between the two
722 NICHHDNetwork of Maternal-FetalMedicine Units
managementschemes.Specifically,no perinatal deaths occurred and neonatal morbidity was extremely low (1%) in both groups. The absenceof any perinatal death in our study is typical of results of other recent prospective randomized trials of postterm pregnancy." Among the 723 pregnancies reported in these trials, only two fetal deaths occurred among infants without anomaliesfor a corrected perinatal mortality rate of 2.2 per 1000. The risk of perinatal morbidity in the postterm pregnancy is much less clearly defined becauseof wide variation in morbidity rates from center to center. For example, in a recent review of postterm pregnancy trials, the risk of meconium aspiration ranged from 13%to 46%, the risk of a 5-minute Apgar score of :r.6 ranged from I% to 8%, and the risk of fetal distress ranged from 6% to 38V This wide variation in risk is a major cause of controversyin the managementof postterm pregnancy. All but one"of the recent prospectiverandomized trials of postterm pregnancyhave demonstrateda similar risk of perinatal morbidity in pregnanciesmanaged expectantly and those managed by induction. In that study (Dyson et aP) the risk of severalperinatal morbidities greatly exceededthose in the current study (e.g., meconium aspiration [4% vs 1.2%] and late decelerationsof the fetal heart rate [6.7% vs 1.0%]). Those trials of postterm pregnancywith low rates of perinatal morbidity (including ours) have not demonstrated a benefit of inducing labor routinely as opposed to an expectant approach. The conclusion of Dyson et al.,' that induction of labor in the prolonged pregnancy is superior to expectant management,would not apply if the rate of perinatal morbidity in the group of women managed expectantly was similar to that in our study and in the other trials studying prolonged pregnancy. One possiWe explanation for the high rate of perinatal morbidity in the study of Dyson et al.-' is that reduced amniotic fluid volume was defined as a pocket of :r. I cm rather 2 cm. Perinatal risk is substantially than a pocket of _-5 greater when the former definition is used.' The high rate of perinatal morbidity in the study of Dyson et al.-' may therefore be due in part to the definition of "normal" amniotic fluid volume. Most randomized trials of postterm pregnancy indicate that the uncompromised, normally grown fetus at term is at low risk of mortality or serious morbidity when modern surveillancetechniques are used."" Surveillance of the fetal condition by nonstresstesting and measurementof amniotic fluid volume appears to be sufficiently sensitive to detect fetal compromise in the small percentageof infants who do undergo deterioration with advancing gestation. Tberefore in the otherwise uncomplicated postterm pregnancy there is little difference in perinatal outcome with expectant management or with immediate induction.
March 1994 Am j Obstet Gynecol
Our data indicate that induction of labor in the postterm pregnancy (with or without PGE2) is not associatedwith an increase in cesareandelivery rate. Among other studies of postterm pregnancies,the cesareandelivery rate was either increased,decreased,or unchanged in the group of women with induction compared with those managed expectantly." The fact that various means of induction were used in these studies (i.e., oxytocin alone or combined with PGE2) limits direct comparisons.Meta-analysisof studies examining the effect of a single application of endocervical PGE2for induction of labor (not limited to postterm pregnancies) indicates that the administration of this agent is not associatedwith a change in the cesarean section rate (typical odds ratio 0.87,95% confidence interval 0.69 to 1.08). It is noteworthy that the effect of PGE2 administration in our trial differs from that reported in other studies."" A single administration of PGE2 into the cervix has been associatedwith an increase in the Bishop score, uterine contracility, and onset of labor and a decreasein the requirement for formal induction of labor when comparedwith placeboor no treatment.' In our study the only difference found between PGE2 and placebo gel was a higher incidence of persistent contractions in nulliparous patients. The discrepancy betweenour results and those reported in the literature biases be to observer or the method of may attributable drug administration. 'nie first concern was adequately addressedin our study design by blinding the drug to both investigators and patients. The second issue relates to the specific method of administration and the dose of PGE2. In our study placebo and active drug were administered via catheter with a guard used to limit the depth of insertion. This guard wasrequired by the Food and Drug Administration before granting permission for the investigatonal use of the drug. This drug delivery systemwasnot used in other studies.It is possiblethat extraamniotic placementof the PGE2may haveresulted in a higher incidenceof induction of labor in other studies, which would account for the differencesobserved.The dose of PGE2used in our study is unlikely to account for its lack of effectiveness,because most clinical trials in which a biologic effect has been documented used a similar dose of the drug. In conclusion,with al sample sizeof 440 subjects,we could not demonstrate a difference in maternal or perinatal outcome when an uncomplicated pregnancy of L-41 weeks'gestation is managed either expectantly or by immediate induction of labor. We havealso shown that a single intracervical administration of 0.5 mg of PGE2 gel is more effective than placebo in initiating persistent contractions in nulliparous patients only, but PGE2gel does not reduce the cesareansection rate or the induction-delivery interval.
Volume 170, Number 3 Am Obstet Gynecol .1
The National Institute of Child Health and Human Development Maternal-Fetal Medicine Units Network was established by the National Institute of Child Health and Human Development in 1986. Participating institutions include Columbia University, " George Washington University Biostatistics Center, ' Johns Hopkins University, ' Thomas Jefferson University, ' University of North Carolina, ' Magee-Womens Hospital, University of Pittsburgh, ' University of Southern h California at Los Angeles, 9 University of Tennessee,
Yale University, ' and the National Institute of Child Health and Human inDevelopment. ' Investigators clude Donald McNellis, MD (Program Officer, ' Arnold L. Medearis, MD, 11 Sarah Fowler, Phl), b Roberto Romero, MD, ' Baha M. Sibai, MD, ` Steve N. Caritis, MD, 'Richard H. Paul, MD, 9 Richard Depp, MD, ' Frank Witter, MD, 'John C. Hobbins, MD, 'Janet Horenstein, MD, 1 Robert C. Cefalo, MD, ' Tavia Gordon, ' Sumner Yaffe, MD, 1 Mark Klebanoff, MD, 1 Heinz Berendes, ' Charlotte Catz, 1Catherine Walla, RN, 11Peggy Cotroneo, RN, 'Susan Tannenbaum, RN d Eileen Bray, RN, ' Ginny Sabo, RN, 'Laura Rocco, RN, MS, ' Angela Portale, RN, ' Lynda Green, MS, b and John D. Green. '
Addendum After submission of our manuscript, the Candian Multicenter Postterm Pregnancy Trial Group also reported the results of a study of postterm pregnancy. Several differences in the current study and the Canadian study limit a direct comparison of most outcome variables. For example, the Canadian investigators used PGE, intracervical gel repeatedly, if needed. We administered the agent only once. Further, the Canadian investigators defined a cervical dilatation of 3 cm as an entry criterion whereas we used a Bishop score of :56. Even with these limitations, the major conclusion of the two studies was the same (i. e., perinatal outcome in postterm pregnancy is similar whether the pregnancy is managed by immediate induction or expectantly). A secondary finding of the Canadian study was that the cesarean section rate was higher with expectant management than with induction at 41 weeks' gestation (24.5% vs 21.2%, respectively). The higher cesarean section rate in the expectantly managed group was primarily due to a higher incidence of cesearan delivery for fetal distress in that group. The criteria for the diagnosis of fetal distress were not controlled. Consequently, the Canadian investigators suggested that ". differences delivery the in the cesarean of rates .. may have been due to differences in the interpretation
NICHHD Network of Maternal-Fetal Medicine Units
723
of the fetal heart rate tracings. " We did not observe a difference in cesarean section rates or in the diagnosis of fetal distress in either of the induction groups or the expectantly managed group.
REFERENCES 1. Dyson DC. Fetal surveillance vs. labor induction at 42 weeks in postterm gestation. j Reprod Med 1988-,33: 2627o. 2. Kerise MJNC, Van Oppen ACC. Preparing the cervix for induction of labour. In: Chalmers E, Keirse M, eds. Effective care in pregnancy and childbirth. London: Oxford University Press, 1989. 3. Katz Z, Yemini M, Lancet M, Mogilner BM, Ben-Hur 11, Caspi B. Non-aggressive management of post-date pregnancies. Eurj Obstet Gynecol Reprod Biol 1983; 15:71-9. 4. Cardozo L, Fysh J, Pearce. 1M. Prolonged pregnancy: the management debate. BMJ 1986; 293: 1059-63. 5. Dyson DC, Miller PD, Armstrong MA. Management of prolonged pregnancy: induction of labor versus anteparturn fetal testing. Am j Oiis-rEl GYNECOL1987; 156:928-34. 6. Augensen K, Bergsjo P, Eikeland T, Askvik K, Carlsen J. Randomised comparison of early versus late induction of labour in post-term pregnancy. BMJ 1987;294: 1192-5. 7. Bergsjo P, Huang G-d, Yu S-q, Gao Z-z, Bakketeig L. Comparison of induced versus non-induced labor in post-term pregnancy. Acta Obstet Gynecol Scand 1989;68: 683-7. 8. Bishop EH. Pelvic scoring for elective induction. Obstet Gynecol 1964; 24: 266-8. 9. Chamberlain PF, Manning FA, Morrison 1, Harman CR, Lange IR. Ultrasound evaluation of amniotic fluid volume. Am j OBgrri GYNECOL 1984; 150: 245-9.
10. Prins RP, Neilson DRjr, Bolton RN, Mark C 111,Watson P. Preinduction cervical ripening with sequential use of prostaglandin E2 gel. Am. 1 ORsrY.i GYNKC01,1986;154: 1275-9. 11. McClure-Browne Postmaturity. Am j ORsi F. -r GYNECOL 1962;85: 573-82. .1C. 12. Vorherr H. Placental insufficiency in relation to postterm pregnancy and fetal postmaturity. Eur j Obstet Gynecol Reprod Biol 1974; 123:67-100. 13. Arias F. Predictability of complications associated with prolongation of pregnancy. Obstet Gynecol 1987,70: 101-6. 14. DeVoe LD, Sholl JS. Postdates pregnancy. Reprod Med .) 1983; 28: 576-80. 15. Freeman RK, Garite Tj, Modanlou H, Dorchester W. Rommal C, Devaney M. Postdate pregnancy: utilization of contraction stress testing for primary fetal surveillance. Am j ClBs-rt'. 'i GYNECOL1981; 140: 128-35.
16. Bernstein
P, Leyland N, Gurland P, Gare D. Cervical E2 gel; a ripening and labor induction with prostaglandin placebo-controlled study. Am OBSTEF GYNECOL 1987; 156: .1 336-40.
17. Noah M L, DeCoster j M, Fraser Tj, Orr j D. Preinduction cervical softening with enclovercial PGE2 gel. Acta Obstet Gynecol Scand 1987; 66: 3-7. 18. Yonekura ML, Songster G, Smith-Wallace T. Preinduction cervical priming with PGE2 intracervical gel. Am j Perinatol 1985;2: 30.5-10.
Key words: Postterm pregnancy,prostaglandin E2 gC' The postterm.pregnancy is one that extends beyond the forty-second week of amenorrhea. The management of such a pregnancy that is otherwise uncomplicated is controversial.' Central to this controversy is whether the fetus is at increasingrisk of deterioration as the pregnancyadvances.Two managementschemesare often used to manage the postterm pregnancy. In one, pregnancy is allowed to progress to 42 weeks and beyond. Labor is induced only if the cervix is well effaced or dilated, or both, or if fetal compromise occurs. 'Me fetal condition is evaluated regularly by Members of thestudygroupandparticipatinginstitutionsarelistedat the endof thearticle. Supportedby grants HD 21410, HD 21434, HD 21366, HD 21380, HD 19897, HD 21414, HD 21386, HD 21366, and HD 21363 ftom the National Institute of Child Health and Human Development. for publicationApril 13,1993; revisedAugust 4,1993; Received October27,1993. accepted Reprintrequests: DonaldMcNellis,MD, NationalInstituteof Child Health and Human Development, National Institutesof Health, EPN Building, Room643, Bethesda, MD 20892. 611152458 716
various techniques. In the second scheme, labor is aggressivelyinduced at 42 weeks or earlier. Cervical ripening agents such as prostaglandins' are used to prepare the cervix and, if necessary,oxytocin and amniotomy are also used. Recently,severalprospective randomized trials comparing these two plans of management have yielded contradictory results." 'Me purpose of the current study was to compared thesetwo managementschemes in a subset of women whose pregnancies reached 41 weeksbut were otherwiseuncomplicated. We also compared the cervical ripening effects of 0.5 mg prostaglandin E2 (PGE2) and placebo gels placed into the cervix. This comparison is pertinent to the management of postterm pregnancy. Methods The trial was designed and implemented by the participants of the Maternal-Fetal Medicine Units Network under the direction and sponsorhsip of the National Institute of Child Health and Human Development. An investigationalnew drug (IND) exemption for
Volume 170, Number 3 Am I Obstet Gynecol
the use of PGE2 gel was obtained from the Food and Drug Administration. Informed consent was obtained from each participating subject. Patient selection. Pregnant patients identified as having a gestational age of at least 287 days from the last menstrual period were screened for eligibility. Patients were excluded from the trial if they had any medical or obstetric complications requiring induction of labor, cesarean section, or frequent monitoring of maternal or fetal condition. Additionally, patients with by 4500 fetal clinigm, an estimated weight exceeding cal or ultrasonographic examination, were also excluded. The presence of any one of the following five criteria was accepted as evidence of accurate gestational dating: (1) a clearly defined menstrual period and audible fetal heartbeat documented for at least 21 weeks by fetoscope or at least 30 weeks by a Doppler device, (2) a clearly defined menstrual period and uterine size estibe by to 524 mated weeks physical examination at : compatible with the duration of amenorrhea, (3) a clearly defined menstrual period and a positive blood in enough or urine pregnancy test result obtained early 41 exceeded pregnancy to assure that the gestation 32 least for documented (4) feal heartbeat at weeks, a weeks by Doppler device if the date of the last menstrual period was uncertain, or (5) an ultrasonographic estimate of gestational age obtained before 26 completed weeks' gestation. Only patients considered to have a gestational age of 287 days but <301 days (41 to 43 completed weeks' gestation) were further evaluated for enrollment into the trial. Randomization. Patients meeting these inclusion criteria and willing to participate in the trial underwent a prerandomization evaluation that included cervical examination to determine a Bishop score' (maximum score of 13), a nonstress test, and an estimation of amniotic fluid volume by ultrasonography. ' Patients were excluded from the trial if the modified Bishop score was _-7, if the nonstress test result was not reassuring (nonreactive or showing decelerations), or if the largest pocket of amniotic fluid was < 2.0 cm. Eligible patients were then randomized to one of three treatment groups according to a computer-assigned randomization scheme arranged by the data coordinating center. Randomization was stratified by clinical site and week of gestation (two strata). Randomization to treatment group was done according to a 2: 2: 1 ratio: expectant management, cervical priming with PGE2 gel followed 12 hours later with oxytocin, or cervical priming with placebo gel followed 12 hours later with oxytocin. Management protocol. Women randomized to the expectant management group were evaluated weekly
NICHHD Network of Maternal-Fetal Medicine Units
717
with a cervical examination and twice weekly with a nonstress test and an ultrasonographic estimation of labor Spontaneous fluid wasawaited, volume. amnionic but induction of labor could be undertaken if any of the following situations occurred: the Bishop score exceeded6, the estimatedfetal weight exceeded4500 gm, delivery developed, indication for obstetric or medical a the largest pocket of amniotic fluid was < 2.0 cm, or an abnormal nonstress test result occurred (nonreactive fetal heart rate or a fetal heart rate with mild variable decelerations exceeding 15 beats/min and lasting 15 by followed a positive contraction stress test seconds) result. If the contraction stresstest wasnegativeand the fetal heart rate remained persistently nonreactive,testing wasrepeated in 24 hours. Patientsin the expectant by 308 comundelivered group who were management pleted days (44 weeks)were releasedfrom the protocol and managed by the method appropriate for the clinical situation. Subjectsin the induction group underwent induction within 24 hours after randomization with instillation into the cervix of either PGE, gel or placebo gel. The investigatorswere unawareof and unable to determine whether the gel contained placeboor active drug. Both PGE2(0.5 mg) and placebo gels were prepackaged by The Upjohn Company for researchuse in a sterile unit dose syringewith a modified catheter and safetydevice to limit the depth of insertion into the intracervical canal to I to 2 cm and to prevent extraamniotic application of gel. Before gel insertion, a modified Bishop scorewas performed. The fetal heart rate and uterine contractions were monitored continuously by noninvasive meansfor at least 4 hours after get administration. If labor did not ensue within 12 hours, a cervical examination was performed and the Bishop score was determined. We used only a single application of PGE2 gel becausethere is no apparent benefit of repeated applications." When clinically feasible,amniotomy was performed before the administration of oxytocin. If amniotomy was not feasible,oxytocin infusion was initiated according to a uniform protocol. If the patient had not entered the active phaseof labor after 24 hours of oxytocin administration, a cesareansection was performed or induction of labor wascontinued for a longer time. For all patients randomized in the trial, decisions regarding the need for cesareandelivey and the use of antibiotics for infection (urinary tract infections,chorioby the made were amnionitis, or endomyometritis) physicianswho were taking care of these patients and by However, dictated the the protocol. not study were indications for cesarean delivery and for the use of antibiotics were recorded. Continuous electronic recording of the fetal heart The in labor during each patient. performed was rate
718 NICHHDNetwork of Maternal-FetalMedicine Units
fetal heart rate tracings were evaluated in terms of baseline heart rate, beat-to-beat variability, and the presenceof decelerations.Variable decelerationswere defined as severeif the heart rate decreasedbelow 60 beats/min and lasted a 60 seconds.The fetus was considered to be in "distress" if repeated late or severe variable decelerationsoccurred. In each patient the presenceof meconium wasnoted either at the time of amniotomy or subsequentlyduring labor. At delivery, every infant with thick meconium in the amniotic fluid had immediate suctioning of the oropharynx on delivery of the head, Outcome variables. For the comparisonof expectant management and immediate induction, the primary outcomevariablesincluded perinatal or maternal death or perinatal morbidity (a composite variable including neonatal seizures, intracranial hemorrhage, the need for mechanicalventilation, Erb's palsy,or other brachial plexus or facial nerve injury). Secondaryoutcome variablesincluded the cesareandelivery rate, the incidence of maternal infection, the need for maternal blood transfusion, the presence of severe variable or late decelerationsduring labor, the incidence of a 5-minute Apgar score <4, the presence of neonatal meconium aspiration, and the neonatal birth weight. For the comparison of PGE2 and placebo gels, the primary outcome variables were the. gel insertion-delivery interval, the change in Bishop score for the 12 hours after gel insertion, and the incidence of failed induction (i.e., the failure to enter the active phase of labor). Secondaryoutcome variables included the incidencesof hyperstimulation or cesareandelivery for fetal "distress" after gel insertion and nonclosure of the ductus arteriosus. Statistical analysis. For comparison of induction and expectant management,we anticipated an incidence of 2% for the composite primary outcome variablesin the group of patients managed expectantly. A sample size of 2800 patients would be required to demonstrate a 50% reduction in the incidence of the primary outcome variablesin the group undergoing immediate induction of labor. This provided a type I error of 0.05 (onesided) and a power of 0.80. To determine sample size for the comparison of PGE2and placebo gels, we assumedthat the incidence of failed induction would be the variable least likely to be reduced by PGE2 gel treatment so we calculated sample size on the basisof this variable. We assumeda failed induction rate of 25%.A sample sizeof 750 (500 receiving PGE2 and 250 receiving placebo gel) would have a 90% power (a = 0.10,0 = 0.05) to detect a 40% reduction in the rate of failed induction by PGE2 gel, Statistical comparisons were performed for ordinal valued variables (Bishop score, gel insertion-delivery interval, and latent phase duration) with the Kruskal-
March 1994 Am j Obstet Gynecol
Wallis rank sum test. Categoric measurementswere compared with the y,' test. All primary outcome analyseswere basedon the total cohort of patients randomized into the trial. Patients were included in their randomly assigned treatment group, and treatment group assignmentwas not altered on the basis of the patient's adherenceto the assignedtreatment regimen. We planned an interim analysisto reassessthe incidence of adverse outcomes and to recalculate study power on the basis of the observed incidence rates. After 440 patients were enrolled in 18 months, the incidence of adverseoutcome wasonly 1.1%.A sample size in excessof 5600 patients would thus be required to adequately test the hypothesis proposed. On the basis of the low incidence of adverse effects and the anticipated annual rate of recruitment, the Data and Safety Monitoring Committee recommended that the study be stopped. Results BetweenDecember 1987 and July 1989we screened 4566 pregnant women who had reached a gestation of ;ý41 weeks.Of these,4126 were ineligible for the study: 2428 (59%) becauseof uncertain gestational age, 650 (16%) becauseof a medical or obstetric complication, 549 (13%) becauseof a Bishop score of Z!7,72 (2%) becauseof decreasedamniotic fluid volume, and 100 (2%) becauseof a nonreactive nonstresstest. An additional 323 patients (8%), although eligible, refused to participate in the study. Finally, in four patients no reason for exclusion was available.Of the 440 patients randomized, 175 were randomized to the expectant management arm and 265 to the induction arm. Of those in the induction arm, 91 received placebo gel whereas 174 received PGE2gel into the cervix. Table I comparesselectedcharacteristicsof the study population at the time of randomization. When comparison by treatment was done, there were no differencesin any of the demographic variableslisted. Additional variables compared at randomization and not found to differ significantly by treatment group include incidence of previous stillbirth, neonatal death, elective abortion, prepregnancy weight and height, estimated fetal weight by ultrasonography, and largest amniotic fluid pocket. The vast majority of patients (80%) were randomized at a gestationalage of 41 to 42 weeks(287 to 293 days). The cervix in these patients was not favorable for induction as evidencedby the low Bishop score. Table II summarizesthe primary outcome variables for the induction and expectantly managed groups. There were no perinatal or maternal deaths, and no infant had an intracranial hemorrhage. In the entire study group only 5 infants (1.1%) experienced one or more of the outcomes listed in Table 11.In the PGE2
Volume 170, Number 3 Arn] Obstet Gynecol
NICHHD Network of Maternal-Fetal Medicine Units
719
Table 1. Characteristics of study population at randomization Induction of labor
Age (mean and SD) Parity (No. ) 0 >1 Missing Previous postterm birth (No. ) Race (No. ) Black White Not available Gestational age at entry (No. ) 287-293 days 295-301 days Bishop score at randomization Mean and SD Median and range
PGE2 (n = 174)
Placebo (n = 91)
25.4 (5.7)
25.4 (5.3)
105 66 3 16
54 37 0 4
(60%) (38%) (2%) (9%)
(59%) (41%) (0%) (4%)
Expectant management (n = 175) 26.1 (5.8) 94 79 2 12
(54%) (45%) (1%) (7%)
56 (32%) 117 (67%) 1 (1%)
34 (37%) 57 (63%) 0 (0%)
67 (38%) 105 (60%) 3 (2%)
141 (81%) 33 (19%)
72 (79%) 19 (21%)
139 (79%) 36 (21%)
4.0 (1.4) 4 (0-6)
3.8 (1.4) 4 (1-6)
3.9 (1.5) 4 (0-6)
Table 11. Incidence of primary outcome variables Induchon of labor PGE, Patients (No. ) Mechanical ventilation (No. ) Nerve injury (No. ) Seizures (No. ) Babies with (No. ) I adverse outcome -ý
174 011 100 021 I (I %)
group one infant of 41 weeks' gestation and weighing 3647 gm had a brachial plexus injury. In the placebo group two babies (one of 41 weeks' and another of 42 weeks' gestation) had seizures only and another baby of 41 weeks 3 days' gestation required mechanical ventilation only. None of these infants was asphyxiated at delivery. Neither of the two infants with seizures required antiseizure medication. In the expectant management group, one infant of 41 weeks' gestation had seizures and required antiseizure medication. This baby also required mechanical ventilation. The baby was unasphyxiated at delivery but had a hypoplastic right ventricle. Maternal and perinatal outcome variables are sumniarized in Tables IIIA and 11113.The randomizationdelivery interval was significantly greater in the group managed expectantly than in either induction group (85 vs 35 and 36 hours). The range of values for the induction-delivery interval is similar in all three groups because in this analysis we did not exclude those patients in whom the protocol was violated. If such patients are excluded, the induction-delivery interval is still significantly greater in the expectantly managed group. Gestational age at delivery was significantly greater in the expectantly managed patients. Sixty-one percent of
Placebo
Expectant management
91
175
3 (3%)
women in the expectantly managed group was delivered at 294 days (42 weeks) or later. Only 35% of women in the two induction groups were delivered at 42 weeks or later. Uterine hyperactivity occurred in three women; one received placebo gel and two received PGE2 gel. The incidence of cesarean delivery was similar in each of the three treatment groups (i. e., 23%, 18%, and 18%, respectively, in the PGE21 placebo, and expectant groups). The combination of indications generally considered to represent cephalopelvic disproportion (i. e., arrest of dilatation, arrest of descent, and protracted descent) was the most commonly listed indication for cesarean delivery, accounting for 56%, 38%, and 56%, respectively, of cesarean deliveries in the PGE21placebo, and expectant management groups. The cumulative frequency of these three indications did not differ among the treatment groups. The frequency of failed induction leading to cesarean section was also similar in the three treatment groups (i. e., 15%, 19%, and 11%, respectively). Table IIIA demonstrates that of the 175 women in the expectant management group 55 (3 1%) underwent induction of labor whereas 118 (69%) labored spontaneously. One woman underwent cesarean delivery with-
720 NICHHDNetwork of Maternal-FetalMedicine Units
Table IIIA.
March 1994 Am j Obstet Gynecol
Matemal outcome Induction of labor
Randomization-delivery interval (hr, median and range) Gestational age at delivery* 287-293 days 294-301 days > 302 days Not available Matemal infection (No. ) Need for transfusion (No. ) Uterine hyperactivity (No. ) Route of delivery (No. ) Vaginal Cesarean Not available
1
PGE2 (n = 174)
Placebo (n = 91)
36 (6-492)
35 (7-487)
85 (5-538)*
60 29 2 0 13 0 1
67 83 24 1 25 3 0
111 60 2 1 33 2 2
(64%) (34%) (1%) (1%) (19%) (1%) (1%)
(66%) (32%) (2%) (0%) (14%) (0%) (1%)
75 (82%) 16 (18%) 0 (00/c)
134 (77%) 39 (22%) 1 (1%)
Expectant management (n = 175)
(38%) (47%) (14%) (1%) (14%) (2%) (0%)
I 142 (81%) 32 (18%) 1 (1%)
< 0.001 by Kruskal-Wallis test.
Table 11111.Neonatal outcome Inductionof labor PGE2 Babies (No. ) Apgar score <4 at 5 min (No. ) Late decelerations in labor (No. ) Meconium (No. ) None 'Min '17hick Not available Meconium aspiration pneumonia (No. ) Birth weight (gm, mean and SD) ;! 4000 gm (No. ) &4500 gm (No. )
174 0 1 131 31 10 2 1 3607 27 1
out labor, and in one woman data regarding the onset of labor were not available.Of thoseundergoing induction of labor the indications were as follows: favorable cervix (n = 18), ruptured membranes (n = 6), nonreassuring nonstress test result or low amniotic fluid volume (n = 23), and "other" indications (n = 8). Table IIIB summarizesneonatal outcome variables. The incidence of asphyxia was low in all groups as evidenced by the infrequent occurrence of a low 5minute Apgar score or late decelerations of the fetal heart rate. Meconium was detected in amnionic fluid with similar frequency in all treatment groups, and in 9% the meconium was describedas thick. Interestingly, of the four infants in whom meconium aspiration pneumonia developed, all had thick meconium detected during labor. None of the 365 infants with thin or absent meconium had meconium aspiration pneumonia. The mean birth weight and the percentage of infants with a birth weight >4000 gm or >4500 gm were similar in each group.
(75%) (18%) (6%) (1%) (382) (16%) (1%)
Placebo
Expectant management
91 0 0 (0%)
175 1 (1) 3 (2%)
71 12 7 1 1 3532 16 3
(78%) (13%) (8%) (1%) (464) (18%) (3%)
109 42 20 4 2 3606 31 6
(62%) (24%) (12%) (2%) (440) (18%) (4%)
Table IV summarizesthe effect of placebo and PGE2 gels on the onset of persistent contractions and on parameters of cervical ripening in the two induction groups. This subgroup analysisincludes only patients for whom the induction protocol was followed. In nulliparous women only, placebo gel was significantly less effective than PGE2in starting persistent contractions. PGE2gel and placebogel, however,were comparablein all other parametersmeasured.The increasein Bishop score, the gel insertion delivery interval, the duration of the latent phaseand the number of women delivered within 12 hours of gel instillation were not significantly different among treatment groups regardlessof parity. Comment This study was stopped becausethe incidence of the primary outcome variable (perinatal mortality or morbidity) was only 1%. Thus a sample size in excessof 5600 would be required to adequatelycompare the two schemesdescribed. Becauseadverseperinatal outcome
Volume 170, Number 3 Am j Obstet Gynecol
NICHHDNetwork of Maternal-FetalMedicine Units 721
Table IV. Effect of placebo and PGE, on contractions and cervical ripening Nulliparouswomen Placebo (n = 44) Cases with persistent contractions* (No. ) Bishop scoreT (median and range) At gel insertion After 12 hr Increase in Bishop scorej (No. ) 0-1 2-3 a4 Not available Gel insertion-delivery interval (hr, median and range) Latent-phase duration (hr, median and range) Delivered within 12 hr (No. )
8 (19)t
4 (1-6) 5 (1-10) 23 10 9 2 33
(52%) (23%) (20%) (5%) (7-90)
9 (< 1-66) 0 (0%)
PGE2 (n = 93) 45 (48)
4 (1-7) 6 (0-12) 47 14 32 0 34
(51%) (15%) (34%) (0%) (7-79)
10 (< 1-57) 3 (3%)
MuRiparouswomen Placebo (n = 34) 14 (41)
4 (1-6) 5 (1-10) 18 6 9 1 27
(53%) (18%) (26%) (3%) (9-70)
7 (< 1-54) 3 (9%)
T
PGE2 (n = 60) 23 (38)
4 (0-6) 6 (0-10) 27 11 20 2 25
(45%) (18%) (33%) (3%) (3-58)
5 (< 1-41) 10 (17%)
Three PGE2 subjects were excluded from the table because parity was undetermined. Thirty-one subjects (13 PGE21 18 placebo) were excluded because the induction protocol was not followed. Of these, 24 did not receive gel or placebo and 7 went home undelivered. *After gel administration but before the start of oxytocin. tSignificant difference when compared with nulliparous women receiving PGE2. TWomen who were delivered vaginally within 12 hours of gel instillation were considered to have a Bishop score of 10 and a change in Bishop score >4.
is so uncommon in postterm pregnancy, the cost of such a trial is not justified. Furthermore, the low prevalence of postterm pregnancy in a well-dated population that is eligible for expectant management (i. e., an unripe cervix, a normal nonstress test result, and a normal amniotic fluid volume) makes such a trial impractical. Thus from the clinician's perspective either expectant management or immediate induction is an acceptable option for the otherwise uncomplicated postterm pregnancy. Postterm pregnancy has historically been considered .. Before a risk factor for adverse perinatal outcome .... the introduction of fetal surveillance techniques, prolonged pregnancy had been associated with a twofold to tenfold increase in perinatal mortality and a twofold to fourfold increase in the incidence of fetal distress. " Induction of labor emerged as a means of reducing perinatal risks in the prolonged pregnancy. The development and application of modern techniques of fetal assessment have been associated with a reduction in perinatal risk in the prolonged pregnancy. In 13 studies betwen 1978 and 1987 in which antenatal fetal surveillance was used for follow-up of the postterm pregnancy, the risk of perinatal mortality was similar to that of pregnancies delivered at term. ' Such reports demonstrated that expectant management of the postterm pregnancy was an acceptable alternative to the induction of labor. However, several recent studies have suggested that in spite of modern monitoring tech-
niques the postterm fetus remains at risk for certain perinatal morbidities such as meconium aspiration, fetal distress in labor, and macrosomatiawith its attendant complications,'"' These findings have rekindled the controversy surrounding the optimal management of the prolonged pregnancy. In response,severalprospective randomized trials using contemporary management schemeshave compared induction and expectant management in the prolonged pregnancy."' These trials have yielded conflicting results that have been attributed to differences in patient selection (ripe vs unripe cervix at entry), methods of labor induction (PGE2with or without oxytocin, amniotomy, or stripping of the membranes),and techniquesof anteparturn fetal surveillance.' These conflicting results leaveunresolvedthe issueof management of the prolonged pregnancy. Therefore we undertook this trial to assessthe role of labor induction with or without PGE2gel in the management of prolonged pregnancy. Patients were enrolled at 287 days (41 weeks)becauserecent data indicate an ý-_ increased risk of perinatal complications beyond this gestational length. " The study was limited to patients with an unfavorablecervix (modified Bishop score !56) and a normally grown, uncompromised fetus because the management of this group of women is the most vexing one when pregnancy is prolonged. We found no difference in neonatal mortality or morbidityý cesarean delivery rates, or maternal morbidity between the two
722 NICHHDNetwork of Maternal-FetalMedicine Units
managementschemes.Specifically,no perinatal deaths occurred and neonatal morbidity was extremely low (1%) in both groups. The absenceof any perinatal death in our study is typical of results of other recent prospective randomized trials of postterm pregnancy." Among the 723 pregnancies reported in these trials, only two fetal deaths occurred among infants without anomaliesfor a corrected perinatal mortality rate of 2.2 per 1000. The risk of perinatal morbidity in the postterm pregnancy is much less clearly defined becauseof wide variation in morbidity rates from center to center. For example, in a recent review of postterm pregnancy trials, the risk of meconium aspiration ranged from 13%to 46%, the risk of a 5-minute Apgar score of :r.6 ranged from I% to 8%, and the risk of fetal distress ranged from 6% to 38V This wide variation in risk is a major cause of controversyin the managementof postterm pregnancy. All but one"of the recent prospectiverandomized trials of postterm pregnancyhave demonstrateda similar risk of perinatal morbidity in pregnanciesmanaged expectantly and those managed by induction. In that study (Dyson et aP) the risk of severalperinatal morbidities greatly exceededthose in the current study (e.g., meconium aspiration [4% vs 1.2%] and late decelerationsof the fetal heart rate [6.7% vs 1.0%]). Those trials of postterm pregnancywith low rates of perinatal morbidity (including ours) have not demonstrated a benefit of inducing labor routinely as opposed to an expectant approach. The conclusion of Dyson et al.,' that induction of labor in the prolonged pregnancy is superior to expectant management,would not apply if the rate of perinatal morbidity in the group of women managed expectantly was similar to that in our study and in the other trials studying prolonged pregnancy. One possiWe explanation for the high rate of perinatal morbidity in the study of Dyson et al.-' is that reduced amniotic fluid volume was defined as a pocket of :r. I cm rather 2 cm. Perinatal risk is substantially than a pocket of _-5 greater when the former definition is used.' The high rate of perinatal morbidity in the study of Dyson et al.-' may therefore be due in part to the definition of "normal" amniotic fluid volume. Most randomized trials of postterm pregnancy indicate that the uncompromised, normally grown fetus at term is at low risk of mortality or serious morbidity when modern surveillancetechniques are used."" Surveillance of the fetal condition by nonstresstesting and measurementof amniotic fluid volume appears to be sufficiently sensitive to detect fetal compromise in the small percentageof infants who do undergo deterioration with advancing gestation. Tberefore in the otherwise uncomplicated postterm pregnancy there is little difference in perinatal outcome with expectant management or with immediate induction.
March 1994 Am j Obstet Gynecol
Our data indicate that induction of labor in the postterm pregnancy (with or without PGE2) is not associatedwith an increase in cesareandelivery rate. Among other studies of postterm pregnancies,the cesareandelivery rate was either increased,decreased,or unchanged in the group of women with induction compared with those managed expectantly." The fact that various means of induction were used in these studies (i.e., oxytocin alone or combined with PGE2) limits direct comparisons.Meta-analysisof studies examining the effect of a single application of endocervical PGE2for induction of labor (not limited to postterm pregnancies) indicates that the administration of this agent is not associatedwith a change in the cesarean section rate (typical odds ratio 0.87,95% confidence interval 0.69 to 1.08). It is noteworthy that the effect of PGE2 administration in our trial differs from that reported in other studies."" A single administration of PGE2 into the cervix has been associatedwith an increase in the Bishop score, uterine contracility, and onset of labor and a decreasein the requirement for formal induction of labor when comparedwith placeboor no treatment.' In our study the only difference found between PGE2 and placebo gel was a higher incidence of persistent contractions in nulliparous patients. The discrepancy betweenour results and those reported in the literature biases be to observer or the method of may attributable drug administration. 'nie first concern was adequately addressedin our study design by blinding the drug to both investigators and patients. The second issue relates to the specific method of administration and the dose of PGE2. In our study placebo and active drug were administered via catheter with a guard used to limit the depth of insertion. This guard wasrequired by the Food and Drug Administration before granting permission for the investigatonal use of the drug. This drug delivery systemwasnot used in other studies.It is possiblethat extraamniotic placementof the PGE2may haveresulted in a higher incidenceof induction of labor in other studies, which would account for the differencesobserved.The dose of PGE2used in our study is unlikely to account for its lack of effectiveness,because most clinical trials in which a biologic effect has been documented used a similar dose of the drug. In conclusion,with al sample sizeof 440 subjects,we could not demonstrate a difference in maternal or perinatal outcome when an uncomplicated pregnancy of L-41 weeks'gestation is managed either expectantly or by immediate induction of labor. We havealso shown that a single intracervical administration of 0.5 mg of PGE2 gel is more effective than placebo in initiating persistent contractions in nulliparous patients only, but PGE2gel does not reduce the cesareansection rate or the induction-delivery interval.
Volume 170, Number 3 Am Obstet Gynecol .1
The National Institute of Child Health and Human Development Maternal-Fetal Medicine Units Network was established by the National Institute of Child Health and Human Development in 1986. Participating institutions include Columbia University, " George Washington University Biostatistics Center, ' Johns Hopkins University, ' Thomas Jefferson University, ' University of North Carolina, ' Magee-Womens Hospital, University of Pittsburgh, ' University of Southern h California at Los Angeles, 9 University of Tennessee,
Yale University, ' and the National Institute of Child Health and Human inDevelopment. ' Investigators clude Donald McNellis, MD (Program Officer, ' Arnold L. Medearis, MD, 11 Sarah Fowler, Phl), b Roberto Romero, MD, ' Baha M. Sibai, MD, ` Steve N. Caritis, MD, 'Richard H. Paul, MD, 9 Richard Depp, MD, ' Frank Witter, MD, 'John C. Hobbins, MD, 'Janet Horenstein, MD, 1 Robert C. Cefalo, MD, ' Tavia Gordon, ' Sumner Yaffe, MD, 1 Mark Klebanoff, MD, 1 Heinz Berendes, ' Charlotte Catz, 1Catherine Walla, RN, 11Peggy Cotroneo, RN, 'Susan Tannenbaum, RN d Eileen Bray, RN, ' Ginny Sabo, RN, 'Laura Rocco, RN, MS, ' Angela Portale, RN, ' Lynda Green, MS, b and John D. Green. '
Addendum After submission of our manuscript, the Candian Multicenter Postterm Pregnancy Trial Group also reported the results of a study of postterm pregnancy. Several differences in the current study and the Canadian study limit a direct comparison of most outcome variables. For example, the Canadian investigators used PGE, intracervical gel repeatedly, if needed. We administered the agent only once. Further, the Canadian investigators defined a cervical dilatation of 3 cm as an entry criterion whereas we used a Bishop score of :56. Even with these limitations, the major conclusion of the two studies was the same (i. e., perinatal outcome in postterm pregnancy is similar whether the pregnancy is managed by immediate induction or expectantly). A secondary finding of the Canadian study was that the cesarean section rate was higher with expectant management than with induction at 41 weeks' gestation (24.5% vs 21.2%, respectively). The higher cesarean section rate in the expectantly managed group was primarily due to a higher incidence of cesearan delivery for fetal distress in that group. The criteria for the diagnosis of fetal distress were not controlled. Consequently, the Canadian investigators suggested that ". differences delivery the in the cesarean of rates .. may have been due to differences in the interpretation
NICHHD Network of Maternal-Fetal Medicine Units
723
of the fetal heart rate tracings. " We did not observe a difference in cesarean section rates or in the diagnosis of fetal distress in either of the induction groups or the expectantly managed group.
REFERENCES 1. Dyson DC. Fetal surveillance vs. labor induction at 42 weeks in postterm gestation. j Reprod Med 1988-,33: 2627o. 2. Kerise MJNC, Van Oppen ACC. Preparing the cervix for induction of labour. In: Chalmers E, Keirse M, eds. Effective care in pregnancy and childbirth. London: Oxford University Press, 1989. 3. Katz Z, Yemini M, Lancet M, Mogilner BM, Ben-Hur 11, Caspi B. Non-aggressive management of post-date pregnancies. Eurj Obstet Gynecol Reprod Biol 1983; 15:71-9. 4. Cardozo L, Fysh J, Pearce. 1M. Prolonged pregnancy: the management debate. BMJ 1986; 293: 1059-63. 5. Dyson DC, Miller PD, Armstrong MA. Management of prolonged pregnancy: induction of labor versus anteparturn fetal testing. Am j Oiis-rEl GYNECOL1987; 156:928-34. 6. Augensen K, Bergsjo P, Eikeland T, Askvik K, Carlsen J. Randomised comparison of early versus late induction of labour in post-term pregnancy. BMJ 1987;294: 1192-5. 7. Bergsjo P, Huang G-d, Yu S-q, Gao Z-z, Bakketeig L. Comparison of induced versus non-induced labor in post-term pregnancy. Acta Obstet Gynecol Scand 1989;68: 683-7. 8. Bishop EH. Pelvic scoring for elective induction. Obstet Gynecol 1964; 24: 266-8. 9. Chamberlain PF, Manning FA, Morrison 1, Harman CR, Lange IR. Ultrasound evaluation of amniotic fluid volume. Am j OBgrri GYNECOL 1984; 150: 245-9.
10. Prins RP, Neilson DRjr, Bolton RN, Mark C 111,Watson P. Preinduction cervical ripening with sequential use of prostaglandin E2 gel. Am. 1 ORsrY.i GYNKC01,1986;154: 1275-9. 11. McClure-Browne Postmaturity. Am j ORsi F. -r GYNECOL 1962;85: 573-82. .1C. 12. Vorherr H. Placental insufficiency in relation to postterm pregnancy and fetal postmaturity. Eur j Obstet Gynecol Reprod Biol 1974; 123:67-100. 13. Arias F. Predictability of complications associated with prolongation of pregnancy. Obstet Gynecol 1987,70: 101-6. 14. DeVoe LD, Sholl JS. Postdates pregnancy. Reprod Med .) 1983; 28: 576-80. 15. Freeman RK, Garite Tj, Modanlou H, Dorchester W. Rommal C, Devaney M. Postdate pregnancy: utilization of contraction stress testing for primary fetal surveillance. Am j ClBs-rt'. 'i GYNECOL1981; 140: 128-35.
16. Bernstein
P, Leyland N, Gurland P, Gare D. Cervical E2 gel; a ripening and labor induction with prostaglandin placebo-controlled study. Am OBSTEF GYNECOL 1987; 156: .1 336-40.
17. Noah M L, DeCoster j M, Fraser Tj, Orr j D. Preinduction cervical softening with enclovercial PGE2 gel. Acta Obstet Gynecol Scand 1987; 66: 3-7. 18. Yonekura ML, Songster G, Smith-Wallace T. Preinduction cervical priming with PGE2 intracervical gel. Am j Perinatol 1985;2: 30.5-10.