Vous êtes sur la page 1sur 10


Defining failed induction of labor

William A. Grobman, MD, MBA; Jennifer Bailit, MD, MPH; Yinglei Lai, PhD; Uma M. Reddy, MD, MPH;
Ronald J. Wapner, MD; Michael W. Varner, MD; John M. Thorp Jr, MD; Kenneth J. Leveno, MD; Steve N. Caritis, MD;
Mona Prasad, DO; Alan T. N. Tita, MD, PhD; George Saade, MD; Yoram Sorokin, MD; Dwight J. Rouse, MD;
Sean C. Blackwell, MD; Jorge E. Tolosa, MD, MSCE; for the Eunice Kennedy Shriver National Institute of Child Health
and Human Development Maternal-Fetal Medicine Units Network

BACKGROUND: While there are well-accepted standards for the RESULTS: A total of 10,677 women were available for analysis. In the
diagnosis of arrested active-phase labor, the definition of a “failed” vast majority (96.4%) of women, the active phase had been reached by
induction of labor remains less certain. One approach to diagnosing a 15 hours. The longer the duration of a woman’s latent phase,
failed induction is based on the duration of the latent phase. However, a the greater her chance of ultimately undergoing a cesarean delivery
standard for the minimum duration that the latent phase of a labor (P < .001, for time both as a continuous and categorical independent
induction should continue, absent acute maternal or fetal indications for variable), although >40% of women whose latent phase lasted≤18
cesarean delivery, remains lacking. hours still had a vaginal delivery. Several maternal morbidities, such as
OBJECTIVE: The objective of this study was to determine the frequency postpartum hemorrhage (P < .001) and chorioamnionitis (P < .001),
of adverse maternal and perinatal outcomes as a function of the duration increased in frequency as the length of latent phase increased.
of the latent phase among nulliparous women undergoing labor induction. Conversely, the frequencies of most adverse perinatal outcomes were
STUDY DESIGN: This study is based on data from an obstetric cohort statistically stable over time.
of women delivering at 25 US hospitals from 2008 through 2011. CONCLUSION: The large majority of women undergoing labor in-
Nulliparous women who had a term singleton gestation in the cephalic duction will have entered the active phase by 15 hours after oxytocin has
presentation were eligible for this analysis if they underwent a labor started and rupture of membranes has occurred. Maternal adverse out-
induction. Consistent with prior studies, the latent phase was determined comes become statistically more frequent with greater time in the latent
to begin once cervical ripening had ended, oxytocin was initiated, and phase, although the absolute increase in frequency is relatively small.
rupture of membranes had occurred, and was determined to end once These data suggest that cesarean delivery should not be undertaken
5-cm dilation was achieved. The frequencies of cesarean delivery, as well during the latent phase prior to at least 15 hours after oxytocin and rupture
as of adverse maternal (eg, postpartum hemorrhage, chorioamnionitis) of membranes have occurred. The decision to continue labor beyond this
and perinatal (eg, a composite frequency of seizures, sepsis, bone or nerve point should be individualized, and may take into account factors such as
injury, encephalopathy, or death) outcomes, were compared as a function other evidence of labor progress.
of the duration of the latent phase (analyzed with time both as a continuous
measure and categorized in 3-hour increments). Key words: labor induction, latent phase, outcomes

Introduction Obstetric outcomes were then studied as

Induction of labor has become an a function of the length of the latent
increasingly utilized obstetric interven- cesarean delivery occurs in the latent phase in induced labors. They concluded
tion. Over the last 2 decades, its use has phase of a labor induction, the indica- that the latent phase could be allowed to
more than doubled, and at present, tion is sometimes labeled as “failed.” extend to at least 12 hours without excess
approximately 1 in 4 pregnant women However, there has not been consensus obstetric morbidity. However, their
have their labor induced.1 One conun- regarding the criterion for this indica- study population was relatively small
drum faced by clinicians who are caring tion, and as a result, the approach to and from a single site, and they could not
for women undergoing labor induction obstetric management in the latent adequately assess durations of the latent
is whether the benefits outweigh the phase for women undergoing labor in- phase >12 hours. Three other studies
risks of continuing labor when a woman duction varies among providers and were performed that approached the
remains in the latent phase for an institutions.2 diagnosis of a failed induction from this
extended period of time. When a Rouse et al3 formulated one approach perspective, and to varying degrees had
to defining a failed induction.3 They similar methodological limitations.4-6
Determining a standard and evidence-
based criterion for a cesarean that is
Cite this article as: Grobman WA, Bailit J, Lai Y, et al. defined the latent phase as beginning performed in the latent phase for the sole
Defining failed induction of labor. Am J Obstet Gynecol when both oxytocin had been initiated
reason that the patient has not entered
and rupture of membranes (ROM) had the active phase is important if unnec-
0002-9378/free occurred, and ending at either 4-cm essary cesarean deliveries are to be
ª 2017 Elsevier Inc. All rights reserved. dilation and 90% effacement or 5-cm
dilation regardless of effacement.
122.e1 American Journal of Obstetrics & Gynecology JANUARY 2018
Original Research ajog.org

JANUARY 2018 American Journal of Obstetrics & Gynecology 122.e2

Original Research OBSTETRICS ajog.org

Development Maternal-Fetal Medicine

Units Network performed an observa-
Flowchart illustrating composition of study population
tional study (ie, the APEX study). In
this study, patient characteristics, intra-
partum events, and pregnancy outcomes
were collected on all women of at least 23
0/7 weeks with a live fetus on admission
and delivered on randomly selected days
representing one third of deliveries over
a 3-year period at 25 participating hos-
pitals. Trained and certified research
personnel abstracted all charts. All cen-
ters obtained institutional review
board approval and a waiver of informed
consent. Full details of the technique
of data collection were described
Women were considered eligible for
this analysis if they were nulliparous;
had a singleton, cephalic gestation at
≤37 weeks; and underwent labor in-
duction. The duration of the latent
phase was defined in a similar fashion
to that first elaborated by Rouse et al3
and subsequently used by others in
their analyses of the latent phase during
labor induction. Specifically, the latent
phase of labor in the setting of induc-
tion was defined to begin once any
cervical ripening had been completed
(ie, when it was no longer used),
oxytocin had begun, and ROM (either
spontaneously or artificially) had
occurred. Latent phase labor was
defined to end once at least 5-cm dila-
tion had been reached (or if cesarean
occurred before that dilation). Women
were excluded from the primary anal-
ysis if any of the times needed to
calculate the length of the latent phase
(eg, time at ROM, time at oxytocin
initiation, time at least 5 cm was
reached) were not available in the chart
and, correspondingly, the length of the
Flowchart illustrating composition of study population of nulliparous women at term with non- latent phase could not be determined.
anomalous vertex singleton gestations undergoing labor induction. Patient outcomes, including the fre-
Grobman et al. Defining failed induction of labor. Am J Obstet Gynecol 2018. quency of cesarean delivery, adverse
maternal outcomes (clinically diagnosed
chorioamnionitis, postpartum hemor-
minimized and interinstitutional com- outcomes associated with the length of rhage, hysterectomy), and adverse
parisons of care are to be possible.7 Thus, the latent phase of labor. neonatal outcomes were compared as a
the purpose of this analysis was to function of the duration of the latent
determine, among a large and Materials and Methods phase. The primary adverse neonatal
geographically varied population of From 2008 through 2011, investigators outcome was a composite that was
nulliparous women undergoing labor at the Eunice Kennedy Shriver National defined to occur when a neonate had any
induction, the maternal and neonatal Institute of Child Health and Human of the following: seizures, culture-proven

122.e3 American Journal of Obstetrics & Gynecology JANUARY 2018

ajog.org OBSTETRICS Original Research

9.4; SAS Institute Inc, Cary, NC) was

used for the analyses. All tests were 2-
Characteristics of study Proportion of women no longer in
sided and P < .05 was used to define
population latent phase after initiation of
statistical significance with no adjust-
labor induction
Characteristic N ¼ 10,677 ment for multiple comparisons.
Maternal age, y 26.4 T 6.1 Latent Cumulative
phase, h N % % Results
Body mass index, kg/m2 32.2 T 6.7 A total of 10,677 women met inclusion
0e2.9 3523 33.0 33.0
Gestational age, wk 39.8 T 1.3 criteria and were available for analysis
3e5.9 3470 32.5 65.5
Race/ethnicity (Figure), 1725 (16.2%) of whom un-
6e8.9 1997 18.7 84.2 derwent induction for PROM and 5582
White 5724 (53.6)
9e11.9 921 8.6 92.8 (52.3%) of whom underwent cervical
Black 2153 (20.2) ripening (Table 1). For women who did
12e14.9 380 3.6 96.4
Hispanic 1665 (15.6) not present with PROM, the median
15e17.9 192 1.8 98.2
Asian 570 (5.3) duration between initiation of oxytocin
≤18 194 1.8 100.0 and ROM was 215 minutes (inter-
Other 565 (5.3)
Women at least 5 cm (or who had cesarean within given quartile range [IQR] 75-418 minutes)
Reason for labor time interval) after cervical ripening had been completed,
oxytocin had begun, and rupture of membranes (either for women who underwent cervical
induction spontaneously or artificially) had occurred. ripening and 180 minutes (IQR 65-332
Maternal medical 2562 (24.0) Grobman et al. Defining failed induction of labor.
minutes) for women who did not un-
conditiona Am J Obstet Gynecol 2018.
dergo cervical ripening. The median
Late or postterm 3004 (28.1) duration from having had both oxytocin
Fetal statusb 1636 (15.3) 3-hour increments. A relative risk for started and ROM (defined in this anal-
PROM 1725 (16.2) each 3-hour interval was estimated with ysis as the start of the latent phase) to
the generalized linear model utilizing the active labor (or cesarean delivery if active
Elective 1468 (13.8)
midpoint of the time interval as the value labor was not reached) was 262 minutes
Other 282 (2.6) for the independent variable in the (IQR 141-435 minutes). By 6 hours
Cervical ripening 5582 (52.3) equation. The proportion of cesareans almost two thirds of women had pro-
Epidural use that occurred in the latent phase, in the gressed from the start of the latent phase
10,038 (95.0)
active phase, and in the second stage, to active labor, and in the vast majority
Birthweight, g 3369 T 477 with the primary indications of non- (96.4%) of women, the active phase had
Data presented as mean T SD or N (%). reassuring fetal status (which included been reached by 15 hours (Table 2).
PROM, premature rupture of membranes. nonreassuring fetal heart tracing, cord The longer the duration of a woman’s
Includes maternal comorbidities, such as hypertensive prolapse, or abruption) or dystocia, was latent phase, the greater her chance of
disease and diabetes mellitus, that are indications for
labor induction; b Includes fetal growth restriction, calculated for each 3-hour time interval ultimately undergoing a cesarean
oligohydramnios, nonreassuring antepartum as well. delivery (P < .001 for time both as a
Grobman et al. Defining failed induction of labor.
To assess the robustness of our find- continuous and categorical independent
Am J Obstet Gynecol 2018. ings, several sensitivity analyses were variable) (Table 3). Nevertheless, >40%
performed. In one sensitivity analysis, the of women whose latent phase lasted for
results of the analysis were reestimated ≤ 18 hours delivered vaginally. The in-
sepsis, bone or nerve injury, after the number of missing time values dications for cesarean, stratified by phase
encephalopathy, or death. These analyses was reduced. In this analysis, the latent and stage of labor, are presented in
were performed with time expressed both phase starting time was assigned ac- Table 4. As is illustrated, the majority of
as a continuous variable and as a cate- cording to the time at ROM or oxytocin, the cesareans at any of the time
gorical variable in 3-hour increments. when only 1 of those times was available. intervalseand in particular at the earlier
Generalized linear models for binary In further sensitivity analyses, general- time intervals that were <15 hoursewere
outcome with log-link function were ized linear models with the log-link not performed in the latent phase. Several
used to estimate relative risks with 95% function, with time as a continuous in- maternal morbidities (ie, chorioamnio-
confidence intervals for the association dependent variable, were rerun after nitis, postpartum hemorrhage, and blood
of latent phase duration (expressed as a adjusting for each of the following 3 transfusion) also increased in frequency
continuous variable in hours) with ob- factors: cervical ripening used (yes/no), as the length of the latent phase increased.
stetric outcomes. The Cochran- elective induction (yes/no), or premature Conversely, the frequency of most
Armitage or exact test9 for trend was ROM (PROM) (yes/no). Adjustment in neonatal adverse outcomes did not differ
used to assess whether the frequency of these models included the main effect of as a function of time. Specifically, there
obstetric outcomes changed as a func- the factor as well as the interaction of the was no statistical increase in the fre-
tion of the duration of the latent phase in factor with time. Software (SAS, Version quency of the primary neonatal adverse

JANUARY 2018 American Journal of Obstetrics & Gynecology 122.e4

Original Research OBSTETRICS ajog.org

outcome, or of outcomes such as low

Apgar score, acidemia, or shoulder


Hysterectomy, Relative risk





(95% CI) dystocia. The frequency of neonatal

intensive care unit (NICU) admission







increased with duration of the latent
phase (Table 5).
Lastly, for sensitivity analyses, we
examined whether results differed after
6 (0.06)

2 (0.53)

1 (0.52)
2 (0.06)
1 (0.03)
N (%)

the number of missing values for latent

phase duration were reduced (leaving

only 746 latent phase durations
transfusion, Relative risk






missing), or after adjustment for
(95% CI)


whether cervical ripening had been used,







the induction was undertaken without a
medical indication (ie, an elective in-
duction), or PROM had occurred. In all
175 (1.6)

36 (1.8)

16 (1.7)

13 (3.4)

10 (5.2)
7 (3.7)
39 (1.1)
54 (1.6)

N (%)

sensitivity analyses, the associations be-
tween duration of the latent phase and
outcomes remained similar to those
degree, Relative risk






of the primary analysis (data not shown).
N (%) (95% CI)

182 (5.2) Referent

572 (5.4) 0.985

199 (5.7) 0.96

119 (6.0) 0.91

42 (4.6) 0.87

13 (3.4) 0.84

10 (5.2) 0.80

7 (3.6) 0.76

This study described several aspects of


the latent phase in the setting of labor

induction that may be helpful when
considering recommendations regarding
(1.044e 1.092)
Relative risk

management of labor. First, a majority of






(95% CI)

women (ie, >96%) will enter the active

66 (1.9) Referent

<.001 <.001
268 (2.6) 1.068

79 (2.4) 1.22

51 (2.6) 1.49

26 (2.9) 1.81

23 (6.2) 2.21

11 (5.8) 2.69

12 (6.3) 3.28

phase within 15 hours of the completion

of cervical ripening (if any is needed),
the initiation of oxytocin, and ROM. The
Frequency of maternal outcomes stratified by duration of latent phase

N (%)

women who do are more likely than not

to have a vaginal delivery and be free of
Chorioamnionitis, Relative risk

maternal and perinatal morbidity. These






(95% CI)

patterns were extant regardless of









whether the induction was without

medical indication, was after PROM, or
was after cervical ripening. Also, there is
Grobman et al. Defining failed induction of labor. Am J Obstet Gynecol 2018.

no one time at which complications

251 (12.6)

84 (22.1)

47 (24.5)

32 (16.5)
1109 (10.4)

160 (17.4)

suddenly arise, although there is an

241 (6.8)
294 (8.5)

incremental increase in the frequency of

N (%)


several maternal complications, and of

NICU admission, as time progresses.
CI, confidence interval; PPH, postpartum hemorrhage.
Time as categorical variable, 3-h increments
Cesarean, Relative risk

These findings extend the findings of






(95% CI)

3523 932 (26.5) Referent

other investigators who have performed

10,677 3608 (33.8) 1.047

787 (39.4) 1.32

433 (47.0) 1.52

207 (54.5) 1.74

115 (59.9) 2.00

109 (56.2) 2.30

3470 1025 (29.5) 1.15

similar analyses.3-6 Studies such as those

Cochran-Armitage or Exact trend test.

of Chelmow et al10 and Maghoma and

Time as continuous variable, h

Buchmann,11 for example, demon-


strated that prolonged latent phases were

N (%)

associated with more frequent maternal

and neonatal complications, although




these studies neither were restricted to


labor induction nor used a single



P value


consensus definition of “prolonged.”



Rouse et al3 performed the first study


that used a similar approach to that of

122.e5 American Journal of Obstetrics & Gynecology JANUARY 2018

ajog.org OBSTETRICS Original Research

Frequency of indications for cesarean delivery, stratified by phase and stage of labor
Latent phase Active phase Second stage
Latent phase, h NRFS Dystocia NRFS Dystocia NRFS Dystocia
0e2.9 [N ¼ 932] 144 (15.5) 41 (4.4) 178 (19.1) 369 (40.1) 50 (5.4) 148 (15.9)
3e5.9 [N ¼ 1025] 114 (11.1) 144 (14.0) 129 (12.6) 385 (37.6) 46 (4.5) 206 (20.1)
6e8.9 [N ¼ 787] 77 (9.8) 234 (29.7) 74 (9.4) 271 (34.4) 19 (2.4) 109 (13.9)
9e11.9 [N ¼ 433] 39 (9.0) 139 (32.1) 38 (8.8) 151 (34.9) 9 (2.1) 57 (13.2)
12e14.9 [N ¼ 207] 25 (12.1) 71 (34.3) 16 (7.7) 65 (31.4) 4 (1.9) 24 (11.6)
15e17.9 [N ¼ 115] 9 (7.8) 55 (47.8) 9 (7.8) 31 (27.0) 3 (2.6) 8 (7.0)
≤18 [N ¼ 109] 9 (8.3) 48 (44.0) 5 (4.6) 33 (30.3) 2 (1.8) 12 (11.0)
Data presented as N (%).
NRFS, nonreassuring fetal status.
Grobman et al. Defining failed induction of labor. Am J Obstet Gynecol 2018.

the present analysis to try to determine which 6 cm was used to define the end of If this missingness is not random, but
the association between duration of the latent labor, admission to the NICU (but related systematically to the outcome
latent phase and obstetric complications not mechanical ventilation or sepsis) was and exposure, bias may be introduced.
in the setting of labor induction. Their statistically more frequent (8.7% at 12 However, in a sensitivity analysis in
study included 509 women of mixed hours vs 6.7% at 9 hours) once 12 hours which the number of missing times was
parity and demonstrated that of the latent phase had been reached.6 reduced by >75%, the results remained
“continued labor induction allowed While that study used 6 cm to define unchanged. Also, these findings are for
some women to have vaginal deliveries,” the end of the latent phase, the results are nulliparous women and cannot be
that chorioamnionitis rose with longer similar to our study (in which 5 cm was generalized to parous women. Never-
times of the latent phase, and that major used as the terminal dilation for the theless, we believe nulliparous women
maternal and perinatal complications latent phase). are the population most in need of study,
were uncommon. They did not, howev- There are several strengths of this given their much greater chance of pro-
er, have sufficient sample size to compare study, including the size of the popula- longed labor and cesarean delivery.
more uncommon neonatal complica- tion, the quality of the data (which were Because this study was concerned with
tions, such as umbilical artery pH <7.0, abstracted by trained research personnel the length of the latent phase during
according to latent phase duration or from each chart), and the diversity of a induction, it provides no insight into the
reliably estimate the outcomes after≤12 nationwide cohort. Women were latent phase during spontaneous labor.
hours of the latent phase. included with a variety of indications for Although this study occurred at many
Subsequent analyses by Simon and labor induction and with differing needs institutions, most were academic centers
Grobman4 (n ¼397) and Rouse et al5 for cervical ripening, and the associa- with training programs, and thus
(n ¼ 1347) included more nulliparous tions observed remained present even generalizability to community hospitals
women with latent phase durations >12 after taking these factors into account. cannot be certain. Yet, it is not evident
hours and concluded that even after 12 Conversely, factors such as maternal age why the presence of associations between
hours in the latent phase, vaginal de- or body mass index were not adjusted duration of a phase of labor and obstetric
livery occurred with reasonable fre- for, given that these may not be mere outcomes should differ based on com-
quency and complications remained covariates but causally related to both munity or academic setting. And, one
uncommon. For example, in the analysis exposure and outcome.12 could consider the many institutions and
by Simon and Grobman,4 67% of Despite these strengths, its limitations the lack of a single protocol for induction
women who had a latent phase of 12-18 should be acknowledged. Because of its or labor management as a strength of the
hours after the completion of any cer- observational nature, the associations study, as these characteristics increase
vical ripening, oxytocin initiation, and observed cannot be known to imply the applicability of the findings to other
ROM had a vaginal delivery without a causality. Even with good data quality institutions, which similarly lack a single
discernible increase in perinatal com- and control processes, particular types of standard for all aspects of labor man-
plications. In a recent analysis of data data (eg, times for multiple events agement. Other studies of labor stan-
from 9763 nulliparous women in the throughout labor) may be missing in the dards, such as those from the
Consortium of Safe Labor study, in chart and thus not able to be abstracted. Consortium of Safe Labor, also have

JANUARY 2018 American Journal of Obstetrics & Gynecology 122.e6

Original Research OBSTETRICS ajog.org

benefited from analysis of labor man-

agement as it occurs in actual clinical

1.034 (1.021e1.047)
settings and not after imposition of a

1.10 (1.06e1.15)
199 (10.0) 1.22 (1.13e1.31)
98 (10.6) 1.35 (1.20e1.51)
44 (11.6) 1.49 (1.28e1.73)
29 (15.1) 1.64 (1.36e1.98)
33 (17.0) 1.81 (1.45e2.27)
standard study protocol.13 Nevertheless,
Relative risk
we cannot know with certainty whether
NICU, N (%) (95% CI)


these findings are generalizable to all
health care settings.
This study presents one, but certainly
1.032 (0.995e1.072) 1004 (9.4)
299 (8.5)
302 (8.7)
not the only approach, to assessing the

extent to which the latent phase should
continue, in the absence of acute indi-
cation for delivery, during labor induc-
tion. Other approaches, for example,
1.10 (0.98e1.23)
1.21 (0.97e1.52)
1.33 (0.95e1.87)
1.47 (0.94e2.30)
1.61 (0.92e2.83)
1.78 (0.91e3.48)
include determination of inflection
Relative risk

points on labor curves as well as cate-

ANC, N (%) (95% CI)


gorizing abnormality based on

ANC, adverse neonatal composite (seizures, sepsis, bone or nerve injury, encephalopathy, death); CI, confidence interval; NICU, neonatal intensive care unit admission; UA, umbilical artery.
population-level percentiles.13-16 This
presently applied approach, however,
has several advantages, including that it
1.034 (0.975e1.097) 243 (2.2) 0.985 (0.955e1.016) 125 (1.2)
38 (1.1)
37 (1.1)
20 (1.0)
16 (1.8)
4 (1.1)
8 (4.2)
2 (1.0)
standardizes the duration as a function
of several aspects of management to
establish a common “clock” and directly
assesses the relationship between dura-
86 (2.5) 0.96 (0.87e1.05)
46 (2.3) 0.91 (0.76e1.10)
16 (1.7) 0.87 (0.66e1.15)
7 (1.9) 0.83 (0.58e1.21)
4 (2.1) 0.80 (0.50e1.26)
3 (1.6) 0.76 (0.44e1.32)

tion and obstetric outcome. Indeed,

dystocia, Relative risk

some investigators have stressed the

(95% CI)

81 (2.3) Referent

importance of determining labor defi-


nitions and utility of different manage-

ment approaches in the context of

maternal and neonatal outcomes.17,18

N (%)

Regardless of the specific approach, a

standard for the minimum duration of
Frequency of neonatal outcomes stratified by duration of latent phase

the latent phase that should be employed

1.11 (0.93e1.32)
1.22 (0.86e1.75)
1.35 (0.80e2.31)
1.50 (0.74e3.05)
1.66 (0.68e4.03)
1.83 (0.63e5.32)

in the setting of labor induction is sorely

needed. Obstetric providers are
pH <7.0,a Relative risk
(95% CI)

routinely faced with the question, for



women who have not entered the active

phase, of whether the benefits of allow-
Grobman et al. Defining failed induction of labor. Am J Obstet Gynecol 2018.
Data are missing for 6666 neonates; b Cochran-Armitage or exact trend test.

ing an induction to continue outweigh

1.072 (0.995e1.155) 62 (1.5)
15 (1.2)
10 (0.8)
9 (1.2)
3 (0.8)
2 (1.1)
2 (2.3)
3 (3.3)

the risks. Lacking a standard, inter-

N (%)

patient and interinstitutional variability

result. As our data demonstrate, even
3.51 (0.92e13.44)

though cesarean for dystocia or failed

1.23 (0.99e1.54)
1.52 (0.97e2.38)
1.87 (0.96e3.67)
2.31 (0.94e5.65)
2.85 (0.93e8.72)

induction was not frequent, it not only

Apgar <4, Relative risk

was cited as an indication but occurred at

(95% CI)


a variety of durations throughout the


latent phase.
Converting these data into a discrete
Time as a continuous variable

clinical recommendation is challenging

10,677 25 (0.2)
6 (0.2)
8 (0.2)
5 (0.3)
921 4 (0.4)

194 2 (1.0)
N (%)

given there is no single time interval at


which the complications suddenly arise,


at which the marginal increase in the

frequency of complications dwarfs
phase, h N

P value e

the marginal increase during antecedent


intervals, or at which there is no


longer a balance of benefit (eg, avoidance


of additional cesareans) with risk.

122.e7 American Journal of Obstetrics & Gynecology JANUARY 2018

ajog.org OBSTETRICS Original Research

However, given that the vast majority of Campbell, C. Collins, N. Jackson, M. Dinsmoor Eunice Kennedy Shriver National Institute of
women will progress to the active phase (NorthShore University HealthSystem), J. Senka Child Health and Human Development,
(NorthShore University HealthSystem), K. Pay- Bethesda, MDeC. Spong, S. Tolivaisa.
by 15 hours, that most women who do chek (NorthShore University HealthSystem), A. Maternal-Fetal Medicine Units Network
will progress to a vaginal delivery, and Peaceman. Steering Committee Chair (Medical University of
that relatively few will have adverse Columbia University, New York, NYeM. South Carolina, Charleston, SC)eJ. P. Van
outcomes, we believe that our results are Talucci, M. Zylfijaj, Z. Reid (Drexel University), Dorsten, MD.
consistent with prior recommenda- R. Leed (Drexel University), J. Benson (Chris-
tions.3-5,7 Specifically, among women tiana Hospital), S. Forester (Christiana Hospital),
C. Kitto (Christiana Hospital), S. Davis (St Peter’s References
undergoing labor induction, when University Hospital), M. Falk (St Peter’s Univer-
1. Martin JA, Hamilton BE, Osterman MJ,
maternal and fetal maternal and fetal sity Hospital), C. Perez (St Peter’s University
Curtin SC, Mathews TJ. Births: final data for
conditions permit, cesarean delivery Hospital).
2013. Natl Vital Stat Rep 2015;64:1-65.
should not be undertaken in the latent University of Utah Health Sciences Center,
2. Baños N, Migliorelli F, Posadas E, Ferreri J,
Salt Lake City, UTeK. Hill, A. Sowles, J. Postma
phase prior to at least 15 hours after (LDS Hospital), S. Alexander (LDS Hospital),
Palacio M. Definition of failed induction of labor
ROM has occurred and oxytocin has and its predictive factors: two unsolved issues of
G. Andersen (LDS Hospital), V. Scott (McKay-
an everyday clinical situation. Fetal Diagn Ther
been started. The decision to continue Dee), V. Morby (McKay-Dee), K. Jolley
labor beyond this point should be indi- (Utah Valley Regional Medical Center), J. Miller
3. Rouse DJ, Owen J, Hauth JC. Criteria for
vidualized, and may take into account (Utah Valley Regional Medical Center), B. Berg
failed labor induction: prospective evaluation of a
(Utah Valley Regional Medical Center).
factors such as other evidence of labor University of North Carolina at Chapel Hill,
standardized protocol. Obstet Gynecol
progress. 2000;96:671-7.
Chapel Hill, NCeK. Dorman, J. Mitchell, E. 4. Simon CE, Grobman WA. When has an in-
The advantages of such an approach Kaluta, K. Clark (WakeMed), K. Spicer duction failed? Obstet Gynecol 2005;105:
are illustrated by 1 study that demon- (WakeMed), S. Timlin (Rex), K. Wilson (Rex).
strated that introduction of a protocol University of Texas Southwestern Medical
5. Rouse DJ, Weiner SJ, Bloom SL, et al. Failed
Center, Dallas, TXeL. Moseley, M. Santillan, J.
that specified a minimum amount of labor induction: toward an objective diagnosis.
Price, K. Buentipo, V. Bludau, T. Thomas, L. Fay,
time before a failed induction was Obstet Gynecol 2011;117:267-72.
C. Melton, J. Kingsbery, R. Benezue.
6. Kawakita T, Reddy UM, Iqbal SN, et al.
diagnosed (in that case, 12 hours after University of Pittsburgh, Pittsburgh, PAeH. Duration of oxytocin and rupture of the mem-
ripening, oxytocin, and ROM) was Simhan, M. Bickus, D. Fischer, T. Kamon branes before diagnosing a failed induction of
associated with a significantly lower (deceased), D. DeAngelis. labor. Obstet Gynecol 2016;128:373-80.
Case Western Reserve University-
frequency of cesarean and no greater MetroHealth Medical Center, Cleveland,
7. Caughey AB, Cahill AG, Guise JM, Rouse DJ.
frequency of adverse outcomes.19 And, Safe prevention of the primary cesarean delivery.
OHeB. Mercer, C. Milluzzi, W. Dalton, T. Dot- Am J Obstet Gynecol 2014;210:179-93.
based on our data, the public health son, P. McDonald, C. Brezine, A. McGrail. 8. Bailit JL, Grobman WA, Rice MM, et al. Risk-
ramifications of adherence to a stan- Ohio State University, Columbus, OHeC. adjusted models for adverse obstetric outcomes
dard definition for “failed induction” Latimer, L. Guzzo (St Ann’s), F. Johnson, L. and variation in risk-adjusted outcomes across
Gerwig (St Ann’s), S. Fyffe, D. Loux (St Ann’s), S.
can be estimated. If induction were to hospitals. Am J Obstet Gynecol 2013;209:446.
Frantz, D. Cline, S. Wylie, J. Iams.
be designated as “failed” if a woman University of Alabama at Birmingham, Bir-
9. Agresti A. A survey of exact inference for
continued to be in the latent phase at mingham, ALeM. Wallace, A. Northen, J. Grant, contingency tables. Stat Sci 1992;7:131-77.
15 as opposed to 6 hours after the C. Colquitt, D. Rouse, W. Andrews. 10. Chelmow D, Kilpatrick SJ, Laros RK Jr.
initiation of oxytocin and ROM, University of Texas Medical Branch, Galveston, Maternal and neonatal outcomes after pro-
approximately 70,000 additional cesar- TXeJ. Moss, A. Salazar, A. Acosta, G. Hankins. longed latent phase. Obstet Gynecol 1993;81:
Wayne State University, Detroit, MIeN. Hauff, 486-91.
ean deliveries could be avoided among L. Palmer, P. Lockhart, D. Driscoll, L. Wynn, C. 11. Maghoma J, Buchmann EJ. Maternal and
the 400,000 nulliparous women who Sudz, D. Dengate, C. Girard, S. Field. fetal risks associated with prolonged latent
undergo induction in the United States Brown University, Providence, RIeP. Breault, phase of labor. J Obstet Gynaecol 2002;22:
every year. n F. Smith, N. Annunziata, D. Allard, J. Silva, M. 16-9.
Gamage, J. Hunt, J. Tillinghast, N. Corcoran, M. Ananth CV, Schisterman EF. Confounding,
12. Jimenez. causality, and confusion: the role of intermediate
Acknowledgment University of Texas Health Science Center at variables in interpreting observational studies in
The authors thank Cynthia Milluzzi, RN, and Houston-Children’s Memorial Hermann Hospi- obstetrics. Am J Obstet Gynecol 2017;217:
Joan Moss, RNC, MSN, for protocol develop- tal, Houston, TXeF. Ortiz, P. Givens, B. Rech, C. 167-75.
ment and coordination between clinical research Moran, M. Hutchinson, Z. Spears, C. Carreno, 13. Zhang J, Landy H, Branch DW, et al.
centers; and Elizabeth Thom, PhD, Madeline M. B. Heaps, G. Zamora. Contemporary patterns of spontaneous labor
Rice, PhD, Brian M. Mercer, MD, and Catherine Oregon Health and Science University, Port- with normal neonatal outcomes. Am J Obstet
Y. Spong, MD, for protocol development and land, OReJ. Seguin, M. Rincon, J. Snyder, Gynecol 2010;116:1281-7.
oversight. C. Farrar, E. Lairson, C. Bonino, W. Smith 14. Friedman EA. The graphic analysis of labor.
In addition to the authors, other members of (Kaiser Permanente), K. Beach (Kaiser Perma- Am J Obstet Gynecol 1954;68:1568-75.
the Eunice Kennedy Shriver National Institute of nente), S. Van Dyke (Kaiser Permanente), 15. Friedman EA. Primigravid labor: a graph-
Child Health and Human Development S. Butcher (Kaiser Permanente). icostatistical analysis. Obstet Gynecol 1955;6:
Maternal-Fetal Medicine Units Network are as George Washington University Biostatistics 567-89.
follows: Center, Washington, DCeE. Thom, M. Rice, 16. Zhang J, Troendle J, Yancey MK. Reas-
Northwestern University, Chicago, ILeG. Y. Zhao, P. McGee, V. Momirova, R. Palugod, sessing the labor curve in nulliparous women.
Mallett, M. Ramos-Brinson, A. Roy, L. Stein, P. B. Reamer, M. Larsen. Am J Obstet Gynecol 2002;187:824-8.

JANUARY 2018 American Journal of Obstetrics & Gynecology 122.e8

Original Research OBSTETRICS ajog.org

17. Hanley GE, Munro S, Greyson D, et al. MetroHealth Medical Center-Case Western Reserve Hospital, Houston, TX (Dr Blackwell); and Oregon Health
Diagnosing onset of labor: a systematic review of University, Cleveland, OH (Dr Bailit); George Washington and Science University, Portland, OR (Dr Tolosa).
definitions in the research literature. BMC Preg- University Biostatistics Center, Washington, DC (Dr Lai); Received Oct. 2, 2017; revised Oct. 27, 2017;
nancy Childbirth 2016;16:71. Eunice Kennedy Shriver National Institute of Child Health accepted Nov. 6, 2017.
18. Lavender T, Hart A, Smyth R. Effect of par- and Human Development, Bethesda, MD (Dr Reddy); The project described was supported by grants from
togram use on outcomes for women in spon- Columbia University, New York, NY (Dr Wapner); Univer- the Eunice Kennedy Shriver National Institute of Child
taneous labor at term. Cochrane Database Syst sity of Utah Health Sciences Center, Salt Lake City, UT Health and Human Development (HD21410, HD27869,
Rev 2013;7:CD005461. (Dr Varner); University of North Carolina at Chapel Hill, HD27915, HD27917, HD34116, HD34208, HD36801,
19. Rhinehart-Ventura J, Eppes C, Sangi- Chapel Hill, NC (Dr Thorp); University of Texas South- HD40500, HD40512, HD40544, HD40545, HD40560,
Haghpeykar H, et al. Evaluation of outcomes western Medical Center, Dallas, TX (Dr Leveno); University HD40485, HD53097, HD53118) and the National Center
after implementation of an induction-of-labor of Pittsburgh, Pittsburgh, PA (Dr Caritis); Ohio State Uni- for Research Resources (UL1 RR024989; 5UL1
protocol. Am J Obstet Gynecol 2014;211:301. versity, Columbus, OH (Dr Prasad); University of Alabama RR025764). Comments and views of the authors do not
e1-7. at Birmingham, Birmingham, AL (Dr Tita); University of necessarily represent views of the National Institutes of
Texas Medical Branch, Galveston, TX (Dr Saade); Wayne Health.
Author and article information State University, Detroit, MI (Dr Sorokin); Brown University, The authors report no conflict of interest.
From the Departments of Obstetrics and Gynecology of Providence, RI (Dr Rouse); University of Texas Health Corresponding author: William A. Grobman, MD,
Northwestern University, Chicago, IL (Dr Grobman); Science Center at Houston-Children’s Memorial Hermann MBA. w-grobman@northwestern.edu

122.e9 American Journal of Obstetrics & Gynecology JANUARY 2018

ajog.org OBSTETRICS Original Research

JANUARY 2018 American Journal of Obstetrics & Gynecology