- Email: [email protected]
A multicenter controlled trial of fetal pulse oximetry in the intrapartum management of nonreassuring fetal heart rate patterns
American Journal of Obstetrics and Gynecology Founded in 1920 volume 183 number 5 NOVEMBER 2000
TRANSACTIONS OF THE TWENTIETH ANNUAL MEETING OF THE SOCIETY FOR MATERNAL-FETAL MEDICINE—CONTINUED
A multicenter controlled trial of fetal pulse oximetry in the intrapartum management of nonreassuring fetal heart rate patterns Thomas J. Garite, MD,a Gary A. Dildy, MD,b Helen McNamara, MD, MSc,c Michael P. Nageotte, MD,d Frank H. Boehm, MD,e Eric H. Dellinger, MD,f Robert A. Knuppel, MD,g Richard P. Porreco, MD,h Hugh S. Miller, MD,i Shiraz Sunderji, MD,j Michael W. Varner, MD,k and David B. Swedlow, MDl Orange, Long Beach, and Danville, California, Provo and Salt Lake City, Utah, Nashville, Tennessee, Greenville, South Carolina, New Brunswick, New Jersey, Denver, Colorado, Tucson, Arizona, St Louis, Missouri, and Montreal, Quebec, Canada OBJECTIVE: Recent developments permit the use of pulse oximetry to evaluate fetal oxygenation in labor. We tested the hypothesis that the addition of fetal pulse oximetry in the evaluation of abnormal fetal heart rate patterns in labor improves the accuracy of fetal assessment and allows safe reduction of cesarean deliveries performed because of nonreassuring fetal status. STUDY DESIGN: A randomized, controlled trial was conducted concurrently in 9 centers. The patients had term pregnancies and were in active labor when abnormal fetal heart rate patterns developed. The patients were randomized to electronic fetal heart rate monitoring alone (control group) or to the combination of electronic fetal monitoring and continuous fetal pulse oximetry (study group). The primary outcome was a reduction in cesarean deliveries for nonreassuring fetal status as a measure of improved accuracy of assessment of fetal oxygenation. RESULTS: A total of 1010 patients were randomized, 502 to the control group and 508 to the study group. There was a reduction of >50% in the number of cesarean deliveries performed because of nonreassuring fetal status in the study group (study, 4.5%; vs control, 10.2%; P = .007). However, there was no net difference in overall cesarean delivery rates (study, n = 147 [29%]; vs control, 130 [26%]; P = .49) because of an increase in cesarean deliveries performed because of dystocia in the study group. In a blinded partogram analysis 89% of the study patients and 91% of the control patients who had a cesarean delivery because of From the Department of Obstetrics and Gynecology, University of California Irvine Medical Centera; the Department of Obstetrics and Gynecology, Utah Valley Regional Medical Center, Provob; the Department of Obstetrics and Gynecology, Royal Victoria Hospital, McGill University, Montrealc; the Department of Obstetrics and Gynecology, Long Beach Memorial Medical Centerd; the Department of Obstetrics and Gynecology, Vanderbilt University Medical Center, Nashvillee; the Department of Obstetrics and Gynecology, Greenville Hospital Systemf; the Department of Obstetrics and Gynecology, St Peter’s Medical Center, New Brunswickg; the Department of Obstetrics and Gynecology, Presbyterian St Luke’s Hospital, Denverh; the Department of Obstetrics and Gynecology, University of Arizona Health Science Centeri; the Department of Obstetrics and Gynecology, St John’s Mercy Medical Center, St Louisj; the Department of Obstetrics and
Gynecology, University of Utah Medical Center, Salt Lake Cityk; and the Swedlow Group Medical Technology Consultants, Danville.l Supported by the Nellcor Division of Mallinckrodt Inc. Presented at the Twentieth Annual Meeting of the Society for MaternalFetal Medicine, January 31–February 5, 2000. Reprint requests: Thomas J. Garite, MD, Professor and Chairman, Department of Obstetrics and Gynecology, University of California Irvine Medical Center, PO Box 14091, Orange, CA 92863-1491. E-mail [email protected]. Am J Obstet Gynecol 2000;183:1049-58. Copyright © 2000 by Mosby, Inc. 0002-9378/2000 $12.00 + 0 6/6/110632 doi:10.1067/mob.2000.110632
1049
1050 Garite et al
November 2000 Am J Obstet Gynecol
dystocia met defined criteria for actual dystocia. There was no difference between the 2 groups in adverse maternal or neonatal outcomes. In terms of the operative intervention for nonreassuring fetal status, there was an improvement in both the sensitivity and the specificity for the study group compared with the control group for the end points of metabolic acidosis and need for resuscitation. CONCLUSION: The study confirmed its primary hypothesis of a safe reduction in cesarean deliveries performed because of nonreassuring fetal status. However, the addition of fetal pulse oximetry did not result in an overall reduction in cesarean deliveries. The increase in cesarean deliveries because of dystocia in the study group did appear to result from a well-documented arrest of labor. Fetal pulse oximetry improved the obstetrician’s ability to more appropriately intervene by cesarean or operative vaginal delivery for fetuses who were actually depressed and acidotic. The unexpected increase in operative delivery for dystocia in the study group is of concern and remains to be explained. (Am J Obstet Gynecol 2000;183:1049-58.)
Key words: Fetal pulse oximetry, fetal oxygen saturation monitoring, electronic fetal heart rate monitoring, randomized controlled trials in obstetrics Electronic fetal heart rate (FHR) monitoring is the predominant method used for assessing fetal oxygenation in labor. Whereas electronic FHR monitoring is the best modality available, it is a nonspecific and indirect way to identify hypoxia and acidosis. Electronic FHR monitoring is associated with increased cesarean delivery rates in randomized trials in part because nonreassuring FHR patterns are an inaccurate indicator of fetal hypoxia and acidosis.1-5 A more direct means of evaluating fetal oxygenation is needed. The introduction of adult and pediatric pulse oximetry has resulted in decreases in death rates, lawsuits, and malpractice insurance premiums.6 Technical difficulties have delayed the development of pulse oximetry for the more inaccessible and physiologically different fetus.7, 8 The use of special far red and near-infrared wavelengths and the placement of the sensor transcervically to lodge against the fetal cheek have largely overcome these difficulties.7, 9, 10 The range of normal and abnormal fetal oxygen saturation in labor11 and a critical threshold of fetal oxygen saturation as measured by pulse oximetry (FSpO2) for metabolic acidosis have now been identified. Human and animal studies have now been published with relatively universal agreement that in the fetus, which normally has an oxygen saturation in labor of 35% to 65%, a metabolic acidosis does not develop until the oxygen saturation falls below 30% for at least 10 to 15 minutes.12-15 Given the rarity of adverse outcomes in monitored fetuses, we chose to test for an improvement in fetal assessment by evaluating whether fetal pulse oximetry allows a significant and safe reduction in cesarean deliveries done because of nonreassuring fetal status. If the oximeter has greater specificity than electronic FHR monitoring to determine the absence of significant fetal hypoxia, the clinician might safely avoid unnecessary operative interventions for nonreassuring FHR patterns. Material and methods A multicenter randomized, controlled trial was developed involving 9 centers with diverse patient populations and practice types throughout the United States. The hypothesis to be tested was as follows: In patients in labor
with an FHR pattern generally considered nonreassuring, the addition of fetal pulse oximetry, compared with electronic FHR monitoring alone, meaningfully reduces the rate of cesarean deliveries done because of nonreassuring fetal status without increasing adverse outcomes for the mother, fetus, or newborn. The protocol was approved by the Food and Drug Administration and the institutional review boards of each center. The study was conducted in 3 phases. In the observational phase all centers prospectively coded deliveries to determine the proportion of cesarean deliveries done for each of two indications—those for nonreassuring fetal status alone and those for failure to progress (dystocia) combined with a concern over the FHR pattern (termed fetal intolerance to labor). During the second, or learning, phase each center placed between 15 and 25 sensors and demonstrated the ability to recruit and randomize patients, obtain adequate FSpO2 signals, and execute the research protocol. Patient inclusion criteria for both the learning phase and the randomized phase were the same. In the final, randomized phase the patients gave consent for possible study inclusion on admission but were only randomized and included if one or more of the FHR patterns developed that are listed in Table I. These FHR patterns allowing study inclusion were defined liberally to include patients who might subsequently have more significant nonreassuring FHR patterns. The study was limited to patients at ≥36 weeks 0 days’ gestation and in active labor, with a single fetus in a cephalic presentation with the cervix dilated to at least 2 cm and at the –2 station or below. All patients had ruptured membranes, although amniotomy was permitted. Exclusion criteria included the following: planned cesarean delivery, placenta previa, need for immediate delivery, active genital herpes or known human immunodeficiency virus infection, which precluded internal monitoring, or participation in other studies. Once a patient met the inclusion criteria, she was randomized to the control group or the study group according to a computerized telephone system at a central site, 24 hours a day, 7 days a week. The study design prescribed
Garite et al 1051
Volume 183, Number 5 Am J Obstet Gynecol
Fig 1. Technique for transcervical insertion of intrauterine Nellcor FS14 fetal oxygen sensor.
Table I. FHR patterns allowing study inclusion (≥1 of the following) • Baseline FHR between 100 and 120 beats/min with no accelerations >15 beats/min for >15 s • Baseline FHR <100 beats/min with accelerations • Increased variability >25 beats/min for >30 min • Mild or moderate variable decelerations for >30 min • Late decelerations (≥1 per 30 min) • Persistent late decelerations (>50% of contractions) for >15 min • Decreased variability <5 beats/min for >30 min • Tachycardia >160 beats/min with variability <5 beats/min • Sinusoidal pattern • Variable decelerations with any of the following: A relative drop of ≥70 beats/min or an absolute drop to ≤70 beats/min for >60 s A persistent slow return to baseline Variability <5 beats/min Tachycardia >160 beats/min Recurrent prolonged decelerations (≥2 decelerations of <70 beats/min for >90 s in 15 min)
the presence of the research nurse from randomization to delivery for patients in both groups. All randomized patients underwent FHR monitoring with either Doppler or scalp electrode or both. For patients in the study group a Nellcor FS14 fetal oxygen sensor was placed and connected to a Nellcor N-400 monitor. The FS14 sensor is flexible and can be inserted transcervically until it rests against the fetal cheek (Fig 1). On the sensor’s surface are electrodes that determine sensor contact, 2 photoemitters that alternately emit far red (735 nm) and nearinfrared (890 nm) light and a photodetector. The sensor is connected to the N-400 monitor, which processes the signals. The N-400 monitor is connected to a conventional electronic fetal monitor that continuously prints the fetal oxygen saturation, averaged over 45 seconds, superimposed on the lower contraction channel of the fetal monitor, which is already calibrated at 0 to 100 (Fig 2).
In patients in the control group electronic FHR monitoring alone was used. Management of the patient according to group assignment was as follows. For both groups the FHR was defined as reassuring, nonreassuring, or ominous. In both groups, when the FHR was reassuring, labor was allowed to continue. An ominous FHR pattern, defined as a FHR persistently <70 beats/min for at least 7 minutes, required immediate delivery in both groups. Therefore the difference between management of FHR patterns between the groups was limited to persistently nonreassuring FHR patterns as defined in Table II. In the study group a fetus with a nonreassuring FHR pattern was considered to be normally oxygenated as long as the FSpO2 was >30% at any time during the interval between two contractions. If the FSpO2 remained <30% for the entire interval between two contractions or if a satisfactory FSpO2 tracing could not be
1052 Garite et al
November 2000 Am J Obstet Gynecol
Fig 2. FHR and contraction tracing with FSpO2 continuously superimposed (arrow) on contraction channel.
Table II. Classification of FHR patterns defining study management FHR classification III. Continue labor III. Management between study and control groups differed only in this classification of nonreassuring FHR patterns
III. Immediate delivery
FHR criteria Reassuring group: Any FHR pattern that does not meet criteria for groups II or III Nonreassuring group (any one of the following for >15 min): 1. Persistent late decelerations (>50% of contractions) 2. Sinusoidal pattern 3. Variable decelerations with ≥1 of the following: • Relative drop of ≥70 beats/min or absolute drop to ≤70 beats/min for >60 s • Persistent slow return to baseline • Long-term variability <5 beats/min • Tachycardia >160 beats/min 4. Recurrent prolonged decelerations (≥2 below 70 beats/min for >90 s) 5. Any one of the following for >60 min: a. Tachycardia >160 beats/min with long-term variability <5 beats/min b. Persistent decreased variability (≤5 beats/min for >60 min) Ominous group: Prolonged deceleration to <70 beats/min for >7 min
obtained, the clinician reverted to the FHR and used the same criteria decribed in the subsequent text for the control group for management. For both the control group and the study group, when the FSpO2 was <30% or unobtainable, the management prescribed was as follows: If the FHR was persistently nonreassuring, the clinician had the option of using the presence of spontaneous or induced FHR accelerations or scalp pH to rule out fetal acidosis. If reassurance could not be established and the FHR pattern persisted, intervention by cesarean or operative vaginal delivery was undertaken. In both groups, before such intervention, nonsurgical means of improving fetal oxygenation such as intravenous fluids, oxygen administration, lateral positioning, and amnioinfusion, where appropriate, or a combination of these, were attempted. The remainder of the management was at the discretion of the managing physician.
One author (Michael P. Nageotte, MD), whose institution did not participate in the actual clinical study, reviewed all electronic FHR tracings for inclusion criteria, appropriate nonoperative interventions and evaluation, and indications for operative intervention. We calculated a need for 425 patients in each of the 2 groups using an a priori estimate of 10% for the rate of cesarean delivery for nonreassuring fetal status and a desired reduction of 50% to 5% with α = .05 and a power of .20. This a priori estimate was confirmed in the baseline observational phase of the study. Statistical methods used were as follows, unless otherwise specified. The Pearson χ2 test of homogeneity was used to compare the relative cesarean delivery rates except in 2 × 2 group comparisons, where the Fisher exact test was used. The χ2 and either the Student t test or the Mann-Whitney U test were used for demographic comparisons. For complex rela-
Garite et al 1053
Volume 183, Number 5 Am J Obstet Gynecol
Table III. Demographic comparisons Control (n = 502) Age (y) Parity Race (No.) White Black Hispanic Asian Insurance (No.) Medicaid Other Dilatation on admission (cm, mean) Station on admission (mean) Previous cesarean delivery (No.) Previous cesarean delivery for dystocia (No.) Maternal risk factors* (≥1) (No.) Fetal risk factors† (≥1) (No.) Birth weight (g, mean) Gestational age (d, mean) Birth weight >4000 g (No.) Amniotomy (No.) Meconium (No.) Epidural anesthesia (No.) Epidural anesthesia at dilataion <5 cm (No.) Prostaglandin (No.) Induction of labor (No.)
27.3 0.7
Study (n = 508)
Statistical significance
27.6 0.7
NS NS
303 (60%) 72 (14%) 104 (21%) 20 (4%)
331 (65%) 51 (10%) 103 (20%) 18 (4%)
NS NS NS NS
181 (36%) 321 (64%) 5.5 –0.74 60 (12%) 29 (6%) 331 (66%) 199 (40%) 3311 277 45 (9%) 318 (63%) 98 (20%) 471 (94%) 318 (71%) 123 (25%) 246 (49%)
171 (34) 336 (66%) 5.7 –0.72 64 (13%) 31 (6%) 355 (70%) 215 (42%) 3341 277 43 (8.5%) 340 (67%) 93 (18%) 487 (96%) 340 (73%) 157 (31%) 286 (56%)
NS NS NS NS NS NS NS NS NS NS NS NS NS NS NS P = .02 P = .02
NS, Not significant. *Maternal risk factors consisted of the following: hypertension, obesity, preeclampsia, diabetes, or cardiac or renal disease. †Fetal risk factors consisted of the following: intrauterine growth restriction, polyhydramnios or oligohydramnios, gestational age >42 weeks or <37 weeks, or maternal drug or alcohol abuse.
Table IV. Delivery route and indication Delivery mode and indication Spontaneous vaginal delivery Assisted vaginal delivery For nonreassuring fetal status For all other indications Cesarean deliveries, all indications Nonreassuring fetal status, single indication Fetal intolerance to labor with dystocia, mixed indication Dystocia, single indication Other indication
tionships involving multiple risk factors influencing cesarean delivery rates, the rates of cesarean deliveries for the various indications were modeled by logistic regression to allow adjustments for covariates and interactions between risk factors. All analyses were on an intent-totreat basis and included all randomized patients. Results In the observational phase, 472 patients were included, and 180 patients were included in the learning phase. For the randomized trial 4545 patients were approached, 2996 consented, and 1010 patients met entry criteria and were ultimately enrolled and randomized, 508 in the study and 502 in the control groups. All 1010 patients
FHR alone (n = 502) (No.)
FHR plus FSpO2 (n = 508) (No.)
255 (51%) 117 (23%) 57 (11%) 60 (12%) 130 (26%) 51 (10%) 35 (7%) 43 (9%) 1 (0%)
241 (47%) 120 (24%) 55 (11%) 65 (13%) 147 (29%) 23 (5%) 27 (5%) 94 (19%) 3 (1%)
Statistical significance by χ2 NS NS NS (P = .49) P < .0001 NS P < .0001 NS
were included in the intent-to-treat analyses. Maternal and fetal characteristics before randomization were well matched between the 2 groups (Table III) with the exception of a higher rate of induced labor and of receiving prostaglandin for cervical ripening in the study group. As shown in Table IV, cesarean delivery for nonreassuring fetal status was reduced by >50% with the addition of fetal pulse oximetry. However, the overall cesarean delivery rate remained unchanged because the rate of cesarean delivery as a result of dystocia was significantly higher in the study group. Combining the fetal intolerance to labor group with either the nonreassuring fetal status group or the dystocia group did not change the differences between the study and control groups in either
1054 Garite et al
November 2000 Am J Obstet Gynecol
Table VA. FHR patterns at study entry FHR pattern before randomization
FHR alone (n = 502) (No.)
Baseline FHR between 100 and 110 beats/min with no accelerations >15 beats/min for >15 s Baseline FHR <100 beats/min with accelerations Increased variability >25 bpm for >30 min Mild or moderate variable decelerations for >30 min Late decelerations (at least 1 per 30 minutes) Decreased variability <5 beats/min for >30 min Persistent late decelerations (>50% of contractions) for >15 min Tachycardia >160 beats/min with long-term variability <5 beats/min Sinusoidal pattern Variable decelerations with relative drop in heart rate of ≥70 beats/min or absolute drop to <70 beats/min for 60 s Variable decelerations with persistent slow return to baseline Variable decelerations with long-term variability <5 beats/min Variable decelerations with tachycardia >160 beats/min Recurrent prolonged decelerations (≥2 below 70 beats/min for >90 s in 15 min)
Table VB. FHR patterns 30 minutes before delivery FHR (n = 502) (No.)
Pattern Total nonreassuring FHR pattern Persistent late decelerations Tachycardia >160 beats/min with long-term variabililty Sinusoidal Variable decelerations <70 beats/min for >60 s With persistent slow return to baseline With long-term variability <5 beats/min With tachycardia >160 beats/min Recurrent prolonged decelerations Persistent decreased variability
FSpO2 (n = 508) (No.)
297 (59%) 306 (60%) 96 (19%) 92 (18%) 8 (2%) 19 (4%) 1 (0%)
0 (0%)
85 (17%) 87 (17%) 104 (21%) 114 (23%) 17 (3%)
23 (5%)
39 (8%) 7 (1%) 17 (3%)
44 (9%) 7 (1%) 12 (2%)
Table VC. FHR classification 30 minutes before delivery Randomized control trial FHR characteristic FHR (n = 502) (No.) FSpO2 (n = 508) (No.) Reassuring (I) Nonreassuring (II) Ominous (III)
195 (39%) 297 (59%) 09 (2%)
198 (39%) 306 (60%) 05 (1%)
combination. As FHR patterns allowing study entry were defined liberally, a subanalysis of patients with more severe FHR patterns at entry (persistent late decelerations, nonreassuring variable decelerations, and persistent FHR variability <5 beats/min) was performed. Cesarean delivery for nonreassuring fetal status in this subgroup totaled 14 of 194 (7.2%) in the study group and 27 of 161 (16.7%) in the control group (P = .007). The remainder of the patients with less severe FHR patterns on admission also had a similar reduction in cesarean delivery for nonreassuring
FHR plus FSpO2 (n = 508) (No.)
1
1
2 0 293 184 47 62 10 0 20
1 1 297 189 63 65 5 0 21
21 1 10 10
31 3 11 13
Table VI. Labor interventions and fetal evaluations in each group
Labor interventions and evaluations Repositioning Hydration Correction of hypotension Tocolytics for hypertonic contractions Maternal oxygen Amnioinfusion Correction of oxytocin administration Scalp pH measurement Scalp stimulation Vibroacoustic stimulation External manipulation Other
FHR (control) (n = 502) (No.)
FSpO2 (study) (n = 508) (No.)
467 (93%) 332 (66%) 49 (10%) 53 (11%) 378 (75%) 180 (36%) 254 (51%) 26 (5%) 218 (43%) 47 (9%) 6 (1%) 9 (2%)
479 (94%) 353 (69%) 71 (14%) 49 (10%) 407 (80%) 177 (35%) 277 (55%) 15 (3%) 211 (41%) 34 (7%) 3 (1%) 8 (2%)
fetal status with 9 of 314 (2.8%) in the oximetry group and 24 of 341 (7.0%) in the control group (P = .03). FHR patterns at study entry and at 30 minutes before delivery were similar between the 2 groups whether they were analyzed by study site personnel during the study or by the independent reviewer (Tables VA, VB, and VC). The independent reviewer identified 3 cesarean deliveries in each group done for nonreassuring fetal status in which there were protocol violations. In all other cases of cesarean delivery for nonreassuring fetal status the reviewer agreed with the interpretation of the monitor(s) and agreed that the cases met the criteria for intervention as defined in the protocol. Elimination of these 6 patients with protocol noncompliance would not have changed any of the statistically significant findings for indications for cesarean delivery. The frequencies of nonoperative interventions and evaluations for nonreassuring FHR patterns were also similar between the 2 groups (Table VI).
Garite et al 1055
Volume 183, Number 5 Am J Obstet Gynecol
Table VII. Accuracy of fetal assessment: Agreement between operative intervention* for nonreassuring fetal status and index values in cases of fetal acidosis and depression Sensitivity (%)
Specificity (%)
Immediate neonatal condition
FHR (n = 108) (No.)
SpO2 (n = 78) (No.)
FHR
SpO2
FHR
SpO2
Odds ratio†
Arterial pH <7.05 Arterial base excess ≤–10 Apgar score <7 at 5 min Bag and mask ventilation Tracheal intubation
11 32 18 58 14
8 29 9 73 6
27 34 28 22 21
75 52 33 27 50
78 79 79 79 79
86 87 85 87 85
P = .001 P = .001 P = .17 P = .02 P = .25
A comparison is presented between the study and control groups including only those patients in whom cesarean or operative vaginal delivery was performed because of nonreassuring fetal status. *Includes cesarean and operative vaginal delivery intervention because of nonreassuring fetal status. †Cochran-Mantel-Haenszel test for homogeneity of odds ratio.
To explore the increased rate of cesarean delivery for dystocia in the study group compared with the control group, we performed several post hoc analyses. To determine whether cesarean delivery for dystocia in the study group was really done for nonreassuring fetal status but mislabeled as being done for dystocia, we performed partogram analyses of all cases of cesarean delivery because of dystocia, with the reviewer blinded to the group assignment. We specified arrest of dilation for >3 hours and arrest of descent for >2 hours at complete dilatation as criteria for actual dystocia, and we considered labor inductions that lasted >12 hours in the presence of ruptured membranes but never achieved the active stage of labor (≥4 cm of dilatation) to be the criteria for failed induction. There were nearly identical numbers of patients in each category (actual dystocia: study, 78%; vs control, 80%; failed induction: study and control, 12% in each; did not meet criteria for either: study, 10%; vs control, 8%). To further evaluate the question of mislabeling of the indication for cesarean delivery, a Kaplan-Meier survival analysis of the fraction of patients remaining undelivered over time from randomization to delivery was performed to determine whether patients designated as undergoing cesarean delivery because of dystocia in the study group might have been delivered sooner than the patients with dystocia in the control group, because of a sense of urgency over a nonreassuring FHR pattern. In this analysis there was no difference in the survival curves (P = .2) for patients undergoing cesarean delivery because of dystocia between the study group and the control group. In actuality, there was a trend toward more rapid intervention in the control group. The next analysis was done to determine whether the device itself interfered with labor. A similar Kaplan-Meier survival analysis for the fraction of patients remaining undelivered over time showed that there was no difference in the time from randomization to delivery between the study group and the control group regardless of route of delivery. Multiple logistic regression analysis failed to demonstrate that any imbalance in risk factors for dystocia accounted for the
difference in the cesarean delivery rate between the 2 groups. In particular, the differences in induction and the frequency of preinduction cervical ripening with prostaglandin between the 2 groups at the time of randomization did not account for the difference in cesarean delivery performed because of dystocia. The final analysis focused on whether an FHR pattern that did not result in a cesarean delivery in the study group because of a reassuring FSpO2 ultimately predicted a group of patients in whom true dystocia would develop. The rates of nonreassuring FHR patterns were higher in patients in the study group with cesarean delivery for dystocia than in the control group (control, 17/43; vs study, 67/94; P = .0006). The majority of this difference is caused by the greater frequency of nonreassuring variable decelerations (control, 6/43; vs study, 38/94; P = .003). An analysis was performed to determine whether the addition of fetal pulse oximetry to electronic FHR monitoring provided more accurate selection of patients with operative intervention for nonreassuring fetal status for truly depressed and acidotic fetuses. In an analysis limited to patients in whom intervention by cesarean or operative vaginal delivery was performed for nonreassuring fetal status, we compared the 2 groups to see how well the decision to intervene because of concern over fetal oxygenation compared with immediate newborn outcome. Table VII shows some examples of various cutoff points for these measures, including parameters of newborn depression (Apgar scores and need for resuscitation) and fetal acidosis (umbilical cord arterial pH and base excess). These, as well as most other cutoff points for low pH, demonstrate that the addition of FSpO2 monitoring provided a statistically significant improvement in appropriateness of operative delivery for the acidotic baby and in the need for bag and mask resuscitation. The frequencies of adverse maternal events were similar in the 2 groups, including abruptio placentae (control, 0; study, 1), uterine rupture and dehiscence (2 in each group), postpartum hemorrhage (control, 16; study, 15), wound infection (control, 1; study, 3), and endometritis
1056 Garite et al
November 2000 Am J Obstet Gynecol
Table VIII. Neonatal outcomes Variable 1-min Apgar score <4 (No.) 5-min Apgar score <7 (No.) Apgar score (mean) 1-min 5-min Cord arterial pH Mean <7.15 <7.10 <7.05 <7.0 Cord arterial base excess (mean) Resuscitation (No.) Bag and mask Intubation Neonatal intensive care unit admissions (No.) Suspected sepsis (No.) Respiratory disease (No.) Jaundice (No.) Neonatal length of stay (d)
Control 29 19
Statistical Study significance 26 8
NS P = .05
7.6 8.8
7.5 8.7
NS NS
7.24 56 26 11 4 –3.4
7.24 62 28 8 3 –3.6
NS NS NS NS NS NS
58 14 74
73 6 92
NS NS NS
21 41 59 1.35
25 39 58 1.40
NS NS NS NS
(control, 16; study, 15). The occurrence of intrapartum fever after randomization was also similar (control, 40/502 [8%]; vs study, 48/508 [9%]). Length of maternal postpartum hospital stay did not differ between the groups (median: control, 1.43 days; vs study, 1.47 days). Neonatal outcomes were similar between the 2 groups as shown in Table VIII. There were 5 neonatal deaths, 3 in the study and 2 in the control group. Four of the 5 deaths were caused by complex congenital heart anomalies. The fifth death occurred in a baby from the study group with postnatal asphyxia as a result of bilateral tension pneumothoraces whose 5-minute Apgar score and umbilical cord pH had been normal. In babies who survived, umbilical cord arterial pH was <7.05 in 8 cases in the study group and in 11 cases in the control group; pH was <7.0 in 4 cases in the study group and in 4 cases in the control group. Of these (all with pH values <7.05), cesarean or operative vaginal delivery for nonreassuring fetal status was performed in 6 of 8 in the study group and in 3 of 11 in the control group (P = .07). Neonatal outcomes according to the same parameters defined in Table VIII were also similar between the 2 groups in the subanalysis of patients with the more severe FHR patterns at study entry. Study and clinical staffs successfully placed the fetal oximeter sensor and obtained a signal 95% of the time. The median time that an adequate signal and tracing were obtained was 67%; thus, for an average hour of tracing, approximately 40 minutes of continuous tracing of fetal oxygen saturation was obtained. Comment Whereas fetal pulse oximetry may ultimately improve our ability to avoid adverse perinatal outcome, electronic
FHR monitoring is generally good at detecting hypoxia (ie, sensitive),16, 17 and adverse outcomes caused by intrapartum hypoxia are rare. It would take a prohibitively large sample size for fetal pulse oximetry to demonstrate benefit in reducing fetal or neonatal complications. We therefore chose to study whether fetal pulse oximetry improved the clinician’s accuracy of fetal assessment in deciding on operative intervention because of concern over potential fetal compromise. Therefore this study was designed to test the hypothesis that the addition of fetal pulse oximetry to electronic FHR monitoring would result in a lower rate of cesarean deliveries performed for nonreassuring fetal status without any increase in adverse neonatal outcome. The study confirmed the primary hypothesis, demonstrating a >50% reduction in the rate of cesarean delivery for nonreassuring fetal status. However, the study did not result in an overall decrease in cesarean deliveries. As an alternative method of assessing the accuracy of fetal pulse oximetry, we looked at how the decision for operative intervention for nonreassuring fetal status correlated with various indexes of neonatal hypoxia, acidosis, and depression. Fetal pulse oximetry did significantly improve the sensitivity and specificity of surgical intervention for nonreassuring fetal status. The improved identification and intervention for those babies who required resuscitation and who were acidotic is an important result of this trial. The experience of an obstetrician who rushes to perform cesarean or operative vaginal delivery of the fetus with a concerning FHR pattern, only to deliver a vigorous healthy newborn, is all too common and disconcerting. A device that actually measures the parameter about which the clinician is concerned, fetal oxygenation, and as a result allows more precise prediction and intervention for a more appropriate diagnosis has the potential to result in improved care. Hurried performance of operative delivery, especially when there is no real indication, has the potential to result in suboptimal outcome, with possible increases in infection, hemorrhage, and anesthetic accidents. The additional observation of improved sensitivity in the oximetry group was unexpected. The generally accepted premise that electronic FHR monitoring is a sensitive modality16, 17 was not borne out by the control arm of this study, because many depressed and acidotic babies managed with electronic FHR monitoring alone did not have operative intervention for nonreassuring fetal status (Table VI). Perhaps the discrepancy is because of the fact that previous studies correlated the FHR before delivery with outcome but did not use the behavior of operative intervention as the independent predictor. One could certainly argue whether the FHR patterns allowing entry into this study are appropriate and could also assert that they are arbitrary and not absolutely defined (eg, how many variable decelerations or how long
Volume 183, Number 5 Am J Obstet Gynecol
they should persist before the patient is enrolled). The subanalyses comparing patients with FHR patterns with more and less concern at study entry showed similar rates of reduction in cesarean delivery for nonreassuring fetal status and neonatal safety, suggesting that more stringent entry criteria would not have altered the study results and that the liberal study entry criteria were appropriate. One could also question whether the FHR criteria used for intervention were consistent with common practice. Because FHR pattern interpretation can vary and given substantial differences in management with nonreassuring FHR patterns, what was critical to the success of this study was the creation of a uniform protocol for monitor interpretation and management to which 9 centers would agree and to have an investigator who was not a participant in the clinical study review every tracing. The fact that this was accomplished and that the independent reviewer found in excess of 95% compliance with the protocol was as much as we could hope to accomplish with a study of this magnitude involving electronic FHR monitoring. The increase in cesarean delivery for dystocia in the study group is an interesting and unanticipated outcome. Four possible explanations for this observation were considered. Multiple logistic regression failed to demonstrate any difference in cesarean deliveries for dystocia as a result of an imbalance in randomization. In particular, the difference in rates of dystocia remained after logistic regression analysis even when we corrected for the differences noted between the 2 groups in rates of induction and prostaglandin use. The second possibility was that the clinicians were actually performing cesarean deliveries for nonreassuring fetal status in the study group but labeled those cases as dystocia. This is the most concerning possibility, because in an unblinded study there is a potential for bias by simple mislabeling of the indication for the operative delivery. Partogram analysis demonstrated similar and high frequencies of documented criteria for failure to progress in labor for the 2 groups. A decision for operative intervention because of dystocia in the oximetry group did not occur sooner than the same decision in the control group. Therefore, whether these patients did or did not continue to have nonreassuring FHR patterns, it does appear that they had actual failure to progress. What cannot be determined, because this was not anticipated in the study design, was whether all efforts at augmenting labor were really similar between the 2 groups. Thus these analyses did not find any evidence of investigator bias as the explanation for the higher rate of cesarean delivery for dystocia in the oximetry group. There was no evidence that the device itself slowed or interfered with labor. Finally, the question was asked whether the same patients who avoided cesarean delivery because of nonreassuring fetal status actually went on to have a cesarean delivery because of dystocia.
Garite et al 1057
Analysis of the worst FHR pattern, confirmed by the independent reviewer, in patients undergoing cesarean delivery for dystocia demonstrated that patients in the study group had a higher frequency of abnormal FHR patterns that would likely have resulted in cesarean delivery for nonreassuring fetal status in the absence of pulse oximetry. This last analysis suggests that FHR patterns commonly interpreted as nonreassuring, especially nonreassuring variable decelerations, may actually be indicators for an underlying risk of dystocia. A normal FSpO2 permits the obstetrician to allow labor to proceed through periods of nonreassuring FHR patterns. Ultimately, the patients in the study group with true dystocia were delivered by cesarean because of dystocia, whereas patients in the electronic FHR monitoring group more often underwent cesarean delivery because of nonreassuring fetal status. It is interesting that at least 2 previous randomized studies of electronic FRH monitoring1, 18 showed an increase not only in cesarean delivery for nonreassuring fetal status in the electronically monitored group but also in cesarean delivery for dystocia in that group as well. Future studies investigating this issue of dystocia are warranted. Perhaps the most important single result of this study was that allowing labor to continue in the presence of a nonreassuring FHR but a reassuring FSpO2 with the consequent reduction in cesarean delivery for nonreassuring fetal status did not result in an apparent increase in adverse neonatal outcome. This study did not have the statistical power to analyze small increases in adverse outcome. Nevertheless, it is unlikely that any single randomized study of this complexity can achieve a substantially larger number. In conclusion, this randomized study confirmed the primary study hypothesis that the addition of fetal pulse oximetry to electronic FHR monitoring in fetuses with concerning FHR patterns can lower the cesarean delivery rate for nonreassuring fetal status. This was accomplished without a demonstrable increase in adverse fetal or neonatal outcome. However, the addition of fetal pulse oximetry did not result in a decrease in the overall cesarean delivery rate because of an increase in cesarean delivery for dystocia in the study group. The failure to reduce the overall cesarean delivery rate may cause some clinicians to question either the validity of the study or the value of the reduction in cesarean deliveries for nonreassuring fetal status. A detailed analysis allowed us to provide some reassurance that the increase in cesarean deliveries for dystocia in the study group was apparently not the result of investigator bias, but this cannot be completely ruled out. The improvement in sensitivity and specificity of intervention by cesarean delivery for nonreassuring fetal status demonstrates that fetal pulse oximetry provides additional confidence that the clinician will have the capability to more accurately assess the well-
1058 Garite et al
being of the fetus in labor. The improved sensitivity for detecting true hypoxia and acidosis is an observation that may give hope that some additional babies in distress may be detected by this new modality. It is also possible that with improvement in the technology of fetal pulse oximetry the accuracy of fetal assessment can be improved even further. We should proceed cautiously with any new technology, but this large, multicenter, prospective, randomized study suggests that fetal pulse oximetry offers an opportunity to more accurately assess fetal oxygenation in labor and may allow us to act more appropriately for the fetus truly in need of intervention. We gratefully acknowledge the following primary research nurses and the study sites involved, as well as the many attending and resident physicians, nurses, and patients who participated in and contributed to the successful completion of this study. Presbyterian St Luke’s Hospital, Denver, Colo (Leslie Gardner, RNC) St John’s Mercy Hospital, St Louis, Mo (Kathleen Simpson, PhD, Pat Flynn, RN, BSN) Greenville Hospital System, Greenville, SC (Karen Nichols, RN) St Peter’s Medical Center, New Brunswick, NJ (Kate McGuire, RNC, Sherrill Sorensen, RNC) University of Arizona Health Sciences Center, Tucson, Ariz (Karen Hallbauer, RN, Melissa Goldsmith RN) University of California Irvine Medical Center, Orange, Calif (Klm Peters-Phair, RN) University of Utah Medical Center, Salt Lake City, Utah (Chris Monson, RN) Utah Valley Regional Medical Center, Provo, Utah (Carol Loucks, RNC) Vanderbilt University Medical Center, Nashville, Tenn (Nancy Townsend, MSN) We also express gratitude to the following individuals who were instrumental in the development and completion of this trial: Debbie Reisenthel, Simon Thomas, Marti Letko-Porter, Mary Kenney, Susan Nixon, and Pamela Rumney.
November 2000 Am J Obstet Gynecol
REFERENCES
1. Haverkamp AD, Thompson HE, McFee JG, et al. The evaluation of continuous fetal heart rate monitoring in high-risk pregnancy. Am J Obstet Gynecol 1976;125:310-7. 2. MacDonald D, Grant A, Sheridan-Pereira M, et al. The Dublin randomized control trial of intrapartum fetal heart rate monitoring. Am J Obstet Gynecol 1985;152:524-39. 3. Leveno KJ, Cunningham FG, Nelson S, et al. A prospective comparison of selective and universal electronic fetal monitoring in 34,995 pregnancies. N Engl J Med 1986;315:615-9. 4. Renou P, Chang A, Anderson I, et al. Controlled trials of fetal intensive care. Am J Obstet Gynecol 1976;126:470-6. 5. Kelso IM, Parsons RJ, Lawrence GF, et al. An assessment of continuous fetal heart rate monitoring in labor; a randomized trial. Am J Obstet Gynecol 1978;131:526-32. 6. Johnson N. Development and potential of fetal pulse oximetry. Contemp Rev Obstet Gynecol 1991;3:193-200. 7. Mannheimer PD, Fein ME, Casciani JR. Physio-optical considerations for the design of fetal pulse oximetry sensors. Eur J Obstet Gynecol Reprod Biol 1997;1(Suppl):S9-19. 8. Mannheimer PD, Casciani JR, Fein ME. Wavelength selection for fetal pulse oximetry selection. IEEE Trans Biomed Eng 1997;44:148-58. 9. Dildy GA, Clark SL, Loucks CA. Intrapartum fetal pulse oximetry: past, present, and future. Am J Obstet Gynecol 1996;175:1-9. 10. McNamara H, Chung DC, Lilford R, et al. Do fetal pulse oximetry readings at delivery correlate with cord blood oxygenation and acidaemia? Br J Obstet Gynaecol 1999;99:735-8. 11. Dildy GA, van den Berg PP, Katz M, et al. Intrapartum fetal pulse oximetry: fetal oxygen saturation trends during labor and relation to delivery outcome. Am J Obstet Gynecol 1994;171:679-84. 12. Kuhnert M, Seelbach-Gobel B, Butterwegge M. Predictive agreement between the fetal arterial oxygen saturation and fetal scalp pH: results of the German multicenter study. Am J Obstet Gynecol 1998;178:330-5. 13. Seelbach-Gobel M, Huepel M, Kuhned M, et al. The prediction of fetal acidosis by means of intrapartum fetal pulse oximetry. Am J Obstet Gynecol 1999;180:73-81. 14. Carbonne B, Langer B, Goffinet F, et al. Multicenter study on the clinical value of fetal pulse oximetry. II. Compared predictive values of pulse oximetry and fetal blood analysis. Am J Obstet Gynecol 1997;17:593-8. 15. Kubli FW, Hon EH, Kazin AE, et al. Observations on heart rate and pH in the human fetus during labor. Am J Obstet Gynecol 1969;104:1190-206. 16. Schifrin BS, Dame L. Fetal heart rate patterns. Prediction of Apgar score. JAMA 1972;219:1322-5. 17. Bissonnette JM. Relationship between continuous FHR patterns and Apgar score in the newborn. Br J Obstet Gynaecol 1975; 82:24-8. 18. Haverkamp AD, Orleans M, Langendoerfer S, et al. A controlled trial of the differential effects of continuous fetal heart rate monitoring in labor: a randomized trial. Am J Obstet Gynecol 1979;134:399-412.
TRANSACTIONS OF THE TWENTIETH ANNUAL MEETING OF THE SOCIETY FOR MATERNAL-FETAL MEDICINE—CONTINUED
A multicenter controlled trial of fetal pulse oximetry in the intrapartum management of nonreassuring fetal heart rate patterns Thomas J. Garite, MD,a Gary A. Dildy, MD,b Helen McNamara, MD, MSc,c Michael P. Nageotte, MD,d Frank H. Boehm, MD,e Eric H. Dellinger, MD,f Robert A. Knuppel, MD,g Richard P. Porreco, MD,h Hugh S. Miller, MD,i Shiraz Sunderji, MD,j Michael W. Varner, MD,k and David B. Swedlow, MDl Orange, Long Beach, and Danville, California, Provo and Salt Lake City, Utah, Nashville, Tennessee, Greenville, South Carolina, New Brunswick, New Jersey, Denver, Colorado, Tucson, Arizona, St Louis, Missouri, and Montreal, Quebec, Canada OBJECTIVE: Recent developments permit the use of pulse oximetry to evaluate fetal oxygenation in labor. We tested the hypothesis that the addition of fetal pulse oximetry in the evaluation of abnormal fetal heart rate patterns in labor improves the accuracy of fetal assessment and allows safe reduction of cesarean deliveries performed because of nonreassuring fetal status. STUDY DESIGN: A randomized, controlled trial was conducted concurrently in 9 centers. The patients had term pregnancies and were in active labor when abnormal fetal heart rate patterns developed. The patients were randomized to electronic fetal heart rate monitoring alone (control group) or to the combination of electronic fetal monitoring and continuous fetal pulse oximetry (study group). The primary outcome was a reduction in cesarean deliveries for nonreassuring fetal status as a measure of improved accuracy of assessment of fetal oxygenation. RESULTS: A total of 1010 patients were randomized, 502 to the control group and 508 to the study group. There was a reduction of >50% in the number of cesarean deliveries performed because of nonreassuring fetal status in the study group (study, 4.5%; vs control, 10.2%; P = .007). However, there was no net difference in overall cesarean delivery rates (study, n = 147 [29%]; vs control, 130 [26%]; P = .49) because of an increase in cesarean deliveries performed because of dystocia in the study group. In a blinded partogram analysis 89% of the study patients and 91% of the control patients who had a cesarean delivery because of From the Department of Obstetrics and Gynecology, University of California Irvine Medical Centera; the Department of Obstetrics and Gynecology, Utah Valley Regional Medical Center, Provob; the Department of Obstetrics and Gynecology, Royal Victoria Hospital, McGill University, Montrealc; the Department of Obstetrics and Gynecology, Long Beach Memorial Medical Centerd; the Department of Obstetrics and Gynecology, Vanderbilt University Medical Center, Nashvillee; the Department of Obstetrics and Gynecology, Greenville Hospital Systemf; the Department of Obstetrics and Gynecology, St Peter’s Medical Center, New Brunswickg; the Department of Obstetrics and Gynecology, Presbyterian St Luke’s Hospital, Denverh; the Department of Obstetrics and Gynecology, University of Arizona Health Science Centeri; the Department of Obstetrics and Gynecology, St John’s Mercy Medical Center, St Louisj; the Department of Obstetrics and
Gynecology, University of Utah Medical Center, Salt Lake Cityk; and the Swedlow Group Medical Technology Consultants, Danville.l Supported by the Nellcor Division of Mallinckrodt Inc. Presented at the Twentieth Annual Meeting of the Society for MaternalFetal Medicine, January 31–February 5, 2000. Reprint requests: Thomas J. Garite, MD, Professor and Chairman, Department of Obstetrics and Gynecology, University of California Irvine Medical Center, PO Box 14091, Orange, CA 92863-1491. E-mail [email protected]. Am J Obstet Gynecol 2000;183:1049-58. Copyright © 2000 by Mosby, Inc. 0002-9378/2000 $12.00 + 0 6/6/110632 doi:10.1067/mob.2000.110632
1049
1050 Garite et al
November 2000 Am J Obstet Gynecol
dystocia met defined criteria for actual dystocia. There was no difference between the 2 groups in adverse maternal or neonatal outcomes. In terms of the operative intervention for nonreassuring fetal status, there was an improvement in both the sensitivity and the specificity for the study group compared with the control group for the end points of metabolic acidosis and need for resuscitation. CONCLUSION: The study confirmed its primary hypothesis of a safe reduction in cesarean deliveries performed because of nonreassuring fetal status. However, the addition of fetal pulse oximetry did not result in an overall reduction in cesarean deliveries. The increase in cesarean deliveries because of dystocia in the study group did appear to result from a well-documented arrest of labor. Fetal pulse oximetry improved the obstetrician’s ability to more appropriately intervene by cesarean or operative vaginal delivery for fetuses who were actually depressed and acidotic. The unexpected increase in operative delivery for dystocia in the study group is of concern and remains to be explained. (Am J Obstet Gynecol 2000;183:1049-58.)
Key words: Fetal pulse oximetry, fetal oxygen saturation monitoring, electronic fetal heart rate monitoring, randomized controlled trials in obstetrics Electronic fetal heart rate (FHR) monitoring is the predominant method used for assessing fetal oxygenation in labor. Whereas electronic FHR monitoring is the best modality available, it is a nonspecific and indirect way to identify hypoxia and acidosis. Electronic FHR monitoring is associated with increased cesarean delivery rates in randomized trials in part because nonreassuring FHR patterns are an inaccurate indicator of fetal hypoxia and acidosis.1-5 A more direct means of evaluating fetal oxygenation is needed. The introduction of adult and pediatric pulse oximetry has resulted in decreases in death rates, lawsuits, and malpractice insurance premiums.6 Technical difficulties have delayed the development of pulse oximetry for the more inaccessible and physiologically different fetus.7, 8 The use of special far red and near-infrared wavelengths and the placement of the sensor transcervically to lodge against the fetal cheek have largely overcome these difficulties.7, 9, 10 The range of normal and abnormal fetal oxygen saturation in labor11 and a critical threshold of fetal oxygen saturation as measured by pulse oximetry (FSpO2) for metabolic acidosis have now been identified. Human and animal studies have now been published with relatively universal agreement that in the fetus, which normally has an oxygen saturation in labor of 35% to 65%, a metabolic acidosis does not develop until the oxygen saturation falls below 30% for at least 10 to 15 minutes.12-15 Given the rarity of adverse outcomes in monitored fetuses, we chose to test for an improvement in fetal assessment by evaluating whether fetal pulse oximetry allows a significant and safe reduction in cesarean deliveries done because of nonreassuring fetal status. If the oximeter has greater specificity than electronic FHR monitoring to determine the absence of significant fetal hypoxia, the clinician might safely avoid unnecessary operative interventions for nonreassuring FHR patterns. Material and methods A multicenter randomized, controlled trial was developed involving 9 centers with diverse patient populations and practice types throughout the United States. The hypothesis to be tested was as follows: In patients in labor
with an FHR pattern generally considered nonreassuring, the addition of fetal pulse oximetry, compared with electronic FHR monitoring alone, meaningfully reduces the rate of cesarean deliveries done because of nonreassuring fetal status without increasing adverse outcomes for the mother, fetus, or newborn. The protocol was approved by the Food and Drug Administration and the institutional review boards of each center. The study was conducted in 3 phases. In the observational phase all centers prospectively coded deliveries to determine the proportion of cesarean deliveries done for each of two indications—those for nonreassuring fetal status alone and those for failure to progress (dystocia) combined with a concern over the FHR pattern (termed fetal intolerance to labor). During the second, or learning, phase each center placed between 15 and 25 sensors and demonstrated the ability to recruit and randomize patients, obtain adequate FSpO2 signals, and execute the research protocol. Patient inclusion criteria for both the learning phase and the randomized phase were the same. In the final, randomized phase the patients gave consent for possible study inclusion on admission but were only randomized and included if one or more of the FHR patterns developed that are listed in Table I. These FHR patterns allowing study inclusion were defined liberally to include patients who might subsequently have more significant nonreassuring FHR patterns. The study was limited to patients at ≥36 weeks 0 days’ gestation and in active labor, with a single fetus in a cephalic presentation with the cervix dilated to at least 2 cm and at the –2 station or below. All patients had ruptured membranes, although amniotomy was permitted. Exclusion criteria included the following: planned cesarean delivery, placenta previa, need for immediate delivery, active genital herpes or known human immunodeficiency virus infection, which precluded internal monitoring, or participation in other studies. Once a patient met the inclusion criteria, she was randomized to the control group or the study group according to a computerized telephone system at a central site, 24 hours a day, 7 days a week. The study design prescribed
Garite et al 1051
Volume 183, Number 5 Am J Obstet Gynecol
Fig 1. Technique for transcervical insertion of intrauterine Nellcor FS14 fetal oxygen sensor.
Table I. FHR patterns allowing study inclusion (≥1 of the following) • Baseline FHR between 100 and 120 beats/min with no accelerations >15 beats/min for >15 s • Baseline FHR <100 beats/min with accelerations • Increased variability >25 beats/min for >30 min • Mild or moderate variable decelerations for >30 min • Late decelerations (≥1 per 30 min) • Persistent late decelerations (>50% of contractions) for >15 min • Decreased variability <5 beats/min for >30 min • Tachycardia >160 beats/min with variability <5 beats/min • Sinusoidal pattern • Variable decelerations with any of the following: A relative drop of ≥70 beats/min or an absolute drop to ≤70 beats/min for >60 s A persistent slow return to baseline Variability <5 beats/min Tachycardia >160 beats/min Recurrent prolonged decelerations (≥2 decelerations of <70 beats/min for >90 s in 15 min)
the presence of the research nurse from randomization to delivery for patients in both groups. All randomized patients underwent FHR monitoring with either Doppler or scalp electrode or both. For patients in the study group a Nellcor FS14 fetal oxygen sensor was placed and connected to a Nellcor N-400 monitor. The FS14 sensor is flexible and can be inserted transcervically until it rests against the fetal cheek (Fig 1). On the sensor’s surface are electrodes that determine sensor contact, 2 photoemitters that alternately emit far red (735 nm) and nearinfrared (890 nm) light and a photodetector. The sensor is connected to the N-400 monitor, which processes the signals. The N-400 monitor is connected to a conventional electronic fetal monitor that continuously prints the fetal oxygen saturation, averaged over 45 seconds, superimposed on the lower contraction channel of the fetal monitor, which is already calibrated at 0 to 100 (Fig 2).
In patients in the control group electronic FHR monitoring alone was used. Management of the patient according to group assignment was as follows. For both groups the FHR was defined as reassuring, nonreassuring, or ominous. In both groups, when the FHR was reassuring, labor was allowed to continue. An ominous FHR pattern, defined as a FHR persistently <70 beats/min for at least 7 minutes, required immediate delivery in both groups. Therefore the difference between management of FHR patterns between the groups was limited to persistently nonreassuring FHR patterns as defined in Table II. In the study group a fetus with a nonreassuring FHR pattern was considered to be normally oxygenated as long as the FSpO2 was >30% at any time during the interval between two contractions. If the FSpO2 remained <30% for the entire interval between two contractions or if a satisfactory FSpO2 tracing could not be
1052 Garite et al
November 2000 Am J Obstet Gynecol
Fig 2. FHR and contraction tracing with FSpO2 continuously superimposed (arrow) on contraction channel.
Table II. Classification of FHR patterns defining study management FHR classification III. Continue labor III. Management between study and control groups differed only in this classification of nonreassuring FHR patterns
III. Immediate delivery
FHR criteria Reassuring group: Any FHR pattern that does not meet criteria for groups II or III Nonreassuring group (any one of the following for >15 min): 1. Persistent late decelerations (>50% of contractions) 2. Sinusoidal pattern 3. Variable decelerations with ≥1 of the following: • Relative drop of ≥70 beats/min or absolute drop to ≤70 beats/min for >60 s • Persistent slow return to baseline • Long-term variability <5 beats/min • Tachycardia >160 beats/min 4. Recurrent prolonged decelerations (≥2 below 70 beats/min for >90 s) 5. Any one of the following for >60 min: a. Tachycardia >160 beats/min with long-term variability <5 beats/min b. Persistent decreased variability (≤5 beats/min for >60 min) Ominous group: Prolonged deceleration to <70 beats/min for >7 min
obtained, the clinician reverted to the FHR and used the same criteria decribed in the subsequent text for the control group for management. For both the control group and the study group, when the FSpO2 was <30% or unobtainable, the management prescribed was as follows: If the FHR was persistently nonreassuring, the clinician had the option of using the presence of spontaneous or induced FHR accelerations or scalp pH to rule out fetal acidosis. If reassurance could not be established and the FHR pattern persisted, intervention by cesarean or operative vaginal delivery was undertaken. In both groups, before such intervention, nonsurgical means of improving fetal oxygenation such as intravenous fluids, oxygen administration, lateral positioning, and amnioinfusion, where appropriate, or a combination of these, were attempted. The remainder of the management was at the discretion of the managing physician.
One author (Michael P. Nageotte, MD), whose institution did not participate in the actual clinical study, reviewed all electronic FHR tracings for inclusion criteria, appropriate nonoperative interventions and evaluation, and indications for operative intervention. We calculated a need for 425 patients in each of the 2 groups using an a priori estimate of 10% for the rate of cesarean delivery for nonreassuring fetal status and a desired reduction of 50% to 5% with α = .05 and a power of .20. This a priori estimate was confirmed in the baseline observational phase of the study. Statistical methods used were as follows, unless otherwise specified. The Pearson χ2 test of homogeneity was used to compare the relative cesarean delivery rates except in 2 × 2 group comparisons, where the Fisher exact test was used. The χ2 and either the Student t test or the Mann-Whitney U test were used for demographic comparisons. For complex rela-
Garite et al 1053
Volume 183, Number 5 Am J Obstet Gynecol
Table III. Demographic comparisons Control (n = 502) Age (y) Parity Race (No.) White Black Hispanic Asian Insurance (No.) Medicaid Other Dilatation on admission (cm, mean) Station on admission (mean) Previous cesarean delivery (No.) Previous cesarean delivery for dystocia (No.) Maternal risk factors* (≥1) (No.) Fetal risk factors† (≥1) (No.) Birth weight (g, mean) Gestational age (d, mean) Birth weight >4000 g (No.) Amniotomy (No.) Meconium (No.) Epidural anesthesia (No.) Epidural anesthesia at dilataion <5 cm (No.) Prostaglandin (No.) Induction of labor (No.)
27.3 0.7
Study (n = 508)
Statistical significance
27.6 0.7
NS NS
303 (60%) 72 (14%) 104 (21%) 20 (4%)
331 (65%) 51 (10%) 103 (20%) 18 (4%)
NS NS NS NS
181 (36%) 321 (64%) 5.5 –0.74 60 (12%) 29 (6%) 331 (66%) 199 (40%) 3311 277 45 (9%) 318 (63%) 98 (20%) 471 (94%) 318 (71%) 123 (25%) 246 (49%)
171 (34) 336 (66%) 5.7 –0.72 64 (13%) 31 (6%) 355 (70%) 215 (42%) 3341 277 43 (8.5%) 340 (67%) 93 (18%) 487 (96%) 340 (73%) 157 (31%) 286 (56%)
NS NS NS NS NS NS NS NS NS NS NS NS NS NS NS P = .02 P = .02
NS, Not significant. *Maternal risk factors consisted of the following: hypertension, obesity, preeclampsia, diabetes, or cardiac or renal disease. †Fetal risk factors consisted of the following: intrauterine growth restriction, polyhydramnios or oligohydramnios, gestational age >42 weeks or <37 weeks, or maternal drug or alcohol abuse.
Table IV. Delivery route and indication Delivery mode and indication Spontaneous vaginal delivery Assisted vaginal delivery For nonreassuring fetal status For all other indications Cesarean deliveries, all indications Nonreassuring fetal status, single indication Fetal intolerance to labor with dystocia, mixed indication Dystocia, single indication Other indication
tionships involving multiple risk factors influencing cesarean delivery rates, the rates of cesarean deliveries for the various indications were modeled by logistic regression to allow adjustments for covariates and interactions between risk factors. All analyses were on an intent-totreat basis and included all randomized patients. Results In the observational phase, 472 patients were included, and 180 patients were included in the learning phase. For the randomized trial 4545 patients were approached, 2996 consented, and 1010 patients met entry criteria and were ultimately enrolled and randomized, 508 in the study and 502 in the control groups. All 1010 patients
FHR alone (n = 502) (No.)
FHR plus FSpO2 (n = 508) (No.)
255 (51%) 117 (23%) 57 (11%) 60 (12%) 130 (26%) 51 (10%) 35 (7%) 43 (9%) 1 (0%)
241 (47%) 120 (24%) 55 (11%) 65 (13%) 147 (29%) 23 (5%) 27 (5%) 94 (19%) 3 (1%)
Statistical significance by χ2 NS NS NS (P = .49) P < .0001 NS P < .0001 NS
were included in the intent-to-treat analyses. Maternal and fetal characteristics before randomization were well matched between the 2 groups (Table III) with the exception of a higher rate of induced labor and of receiving prostaglandin for cervical ripening in the study group. As shown in Table IV, cesarean delivery for nonreassuring fetal status was reduced by >50% with the addition of fetal pulse oximetry. However, the overall cesarean delivery rate remained unchanged because the rate of cesarean delivery as a result of dystocia was significantly higher in the study group. Combining the fetal intolerance to labor group with either the nonreassuring fetal status group or the dystocia group did not change the differences between the study and control groups in either
1054 Garite et al
November 2000 Am J Obstet Gynecol
Table VA. FHR patterns at study entry FHR pattern before randomization
FHR alone (n = 502) (No.)
Baseline FHR between 100 and 110 beats/min with no accelerations >15 beats/min for >15 s Baseline FHR <100 beats/min with accelerations Increased variability >25 bpm for >30 min Mild or moderate variable decelerations for >30 min Late decelerations (at least 1 per 30 minutes) Decreased variability <5 beats/min for >30 min Persistent late decelerations (>50% of contractions) for >15 min Tachycardia >160 beats/min with long-term variability <5 beats/min Sinusoidal pattern Variable decelerations with relative drop in heart rate of ≥70 beats/min or absolute drop to <70 beats/min for 60 s Variable decelerations with persistent slow return to baseline Variable decelerations with long-term variability <5 beats/min Variable decelerations with tachycardia >160 beats/min Recurrent prolonged decelerations (≥2 below 70 beats/min for >90 s in 15 min)
Table VB. FHR patterns 30 minutes before delivery FHR (n = 502) (No.)
Pattern Total nonreassuring FHR pattern Persistent late decelerations Tachycardia >160 beats/min with long-term variabililty Sinusoidal Variable decelerations <70 beats/min for >60 s With persistent slow return to baseline With long-term variability <5 beats/min With tachycardia >160 beats/min Recurrent prolonged decelerations Persistent decreased variability
FSpO2 (n = 508) (No.)
297 (59%) 306 (60%) 96 (19%) 92 (18%) 8 (2%) 19 (4%) 1 (0%)
0 (0%)
85 (17%) 87 (17%) 104 (21%) 114 (23%) 17 (3%)
23 (5%)
39 (8%) 7 (1%) 17 (3%)
44 (9%) 7 (1%) 12 (2%)
Table VC. FHR classification 30 minutes before delivery Randomized control trial FHR characteristic FHR (n = 502) (No.) FSpO2 (n = 508) (No.) Reassuring (I) Nonreassuring (II) Ominous (III)
195 (39%) 297 (59%) 09 (2%)
198 (39%) 306 (60%) 05 (1%)
combination. As FHR patterns allowing study entry were defined liberally, a subanalysis of patients with more severe FHR patterns at entry (persistent late decelerations, nonreassuring variable decelerations, and persistent FHR variability <5 beats/min) was performed. Cesarean delivery for nonreassuring fetal status in this subgroup totaled 14 of 194 (7.2%) in the study group and 27 of 161 (16.7%) in the control group (P = .007). The remainder of the patients with less severe FHR patterns on admission also had a similar reduction in cesarean delivery for nonreassuring
FHR plus FSpO2 (n = 508) (No.)
1
1
2 0 293 184 47 62 10 0 20
1 1 297 189 63 65 5 0 21
21 1 10 10
31 3 11 13
Table VI. Labor interventions and fetal evaluations in each group
Labor interventions and evaluations Repositioning Hydration Correction of hypotension Tocolytics for hypertonic contractions Maternal oxygen Amnioinfusion Correction of oxytocin administration Scalp pH measurement Scalp stimulation Vibroacoustic stimulation External manipulation Other
FHR (control) (n = 502) (No.)
FSpO2 (study) (n = 508) (No.)
467 (93%) 332 (66%) 49 (10%) 53 (11%) 378 (75%) 180 (36%) 254 (51%) 26 (5%) 218 (43%) 47 (9%) 6 (1%) 9 (2%)
479 (94%) 353 (69%) 71 (14%) 49 (10%) 407 (80%) 177 (35%) 277 (55%) 15 (3%) 211 (41%) 34 (7%) 3 (1%) 8 (2%)
fetal status with 9 of 314 (2.8%) in the oximetry group and 24 of 341 (7.0%) in the control group (P = .03). FHR patterns at study entry and at 30 minutes before delivery were similar between the 2 groups whether they were analyzed by study site personnel during the study or by the independent reviewer (Tables VA, VB, and VC). The independent reviewer identified 3 cesarean deliveries in each group done for nonreassuring fetal status in which there were protocol violations. In all other cases of cesarean delivery for nonreassuring fetal status the reviewer agreed with the interpretation of the monitor(s) and agreed that the cases met the criteria for intervention as defined in the protocol. Elimination of these 6 patients with protocol noncompliance would not have changed any of the statistically significant findings for indications for cesarean delivery. The frequencies of nonoperative interventions and evaluations for nonreassuring FHR patterns were also similar between the 2 groups (Table VI).
Garite et al 1055
Volume 183, Number 5 Am J Obstet Gynecol
Table VII. Accuracy of fetal assessment: Agreement between operative intervention* for nonreassuring fetal status and index values in cases of fetal acidosis and depression Sensitivity (%)
Specificity (%)
Immediate neonatal condition
FHR (n = 108) (No.)
SpO2 (n = 78) (No.)
FHR
SpO2
FHR
SpO2
Odds ratio†
Arterial pH <7.05 Arterial base excess ≤–10 Apgar score <7 at 5 min Bag and mask ventilation Tracheal intubation
11 32 18 58 14
8 29 9 73 6
27 34 28 22 21
75 52 33 27 50
78 79 79 79 79
86 87 85 87 85
P = .001 P = .001 P = .17 P = .02 P = .25
A comparison is presented between the study and control groups including only those patients in whom cesarean or operative vaginal delivery was performed because of nonreassuring fetal status. *Includes cesarean and operative vaginal delivery intervention because of nonreassuring fetal status. †Cochran-Mantel-Haenszel test for homogeneity of odds ratio.
To explore the increased rate of cesarean delivery for dystocia in the study group compared with the control group, we performed several post hoc analyses. To determine whether cesarean delivery for dystocia in the study group was really done for nonreassuring fetal status but mislabeled as being done for dystocia, we performed partogram analyses of all cases of cesarean delivery because of dystocia, with the reviewer blinded to the group assignment. We specified arrest of dilation for >3 hours and arrest of descent for >2 hours at complete dilatation as criteria for actual dystocia, and we considered labor inductions that lasted >12 hours in the presence of ruptured membranes but never achieved the active stage of labor (≥4 cm of dilatation) to be the criteria for failed induction. There were nearly identical numbers of patients in each category (actual dystocia: study, 78%; vs control, 80%; failed induction: study and control, 12% in each; did not meet criteria for either: study, 10%; vs control, 8%). To further evaluate the question of mislabeling of the indication for cesarean delivery, a Kaplan-Meier survival analysis of the fraction of patients remaining undelivered over time from randomization to delivery was performed to determine whether patients designated as undergoing cesarean delivery because of dystocia in the study group might have been delivered sooner than the patients with dystocia in the control group, because of a sense of urgency over a nonreassuring FHR pattern. In this analysis there was no difference in the survival curves (P = .2) for patients undergoing cesarean delivery because of dystocia between the study group and the control group. In actuality, there was a trend toward more rapid intervention in the control group. The next analysis was done to determine whether the device itself interfered with labor. A similar Kaplan-Meier survival analysis for the fraction of patients remaining undelivered over time showed that there was no difference in the time from randomization to delivery between the study group and the control group regardless of route of delivery. Multiple logistic regression analysis failed to demonstrate that any imbalance in risk factors for dystocia accounted for the
difference in the cesarean delivery rate between the 2 groups. In particular, the differences in induction and the frequency of preinduction cervical ripening with prostaglandin between the 2 groups at the time of randomization did not account for the difference in cesarean delivery performed because of dystocia. The final analysis focused on whether an FHR pattern that did not result in a cesarean delivery in the study group because of a reassuring FSpO2 ultimately predicted a group of patients in whom true dystocia would develop. The rates of nonreassuring FHR patterns were higher in patients in the study group with cesarean delivery for dystocia than in the control group (control, 17/43; vs study, 67/94; P = .0006). The majority of this difference is caused by the greater frequency of nonreassuring variable decelerations (control, 6/43; vs study, 38/94; P = .003). An analysis was performed to determine whether the addition of fetal pulse oximetry to electronic FHR monitoring provided more accurate selection of patients with operative intervention for nonreassuring fetal status for truly depressed and acidotic fetuses. In an analysis limited to patients in whom intervention by cesarean or operative vaginal delivery was performed for nonreassuring fetal status, we compared the 2 groups to see how well the decision to intervene because of concern over fetal oxygenation compared with immediate newborn outcome. Table VII shows some examples of various cutoff points for these measures, including parameters of newborn depression (Apgar scores and need for resuscitation) and fetal acidosis (umbilical cord arterial pH and base excess). These, as well as most other cutoff points for low pH, demonstrate that the addition of FSpO2 monitoring provided a statistically significant improvement in appropriateness of operative delivery for the acidotic baby and in the need for bag and mask resuscitation. The frequencies of adverse maternal events were similar in the 2 groups, including abruptio placentae (control, 0; study, 1), uterine rupture and dehiscence (2 in each group), postpartum hemorrhage (control, 16; study, 15), wound infection (control, 1; study, 3), and endometritis
1056 Garite et al
November 2000 Am J Obstet Gynecol
Table VIII. Neonatal outcomes Variable 1-min Apgar score <4 (No.) 5-min Apgar score <7 (No.) Apgar score (mean) 1-min 5-min Cord arterial pH Mean <7.15 <7.10 <7.05 <7.0 Cord arterial base excess (mean) Resuscitation (No.) Bag and mask Intubation Neonatal intensive care unit admissions (No.) Suspected sepsis (No.) Respiratory disease (No.) Jaundice (No.) Neonatal length of stay (d)
Control 29 19
Statistical Study significance 26 8
NS P = .05
7.6 8.8
7.5 8.7
NS NS
7.24 56 26 11 4 –3.4
7.24 62 28 8 3 –3.6
NS NS NS NS NS NS
58 14 74
73 6 92
NS NS NS
21 41 59 1.35
25 39 58 1.40
NS NS NS NS
(control, 16; study, 15). The occurrence of intrapartum fever after randomization was also similar (control, 40/502 [8%]; vs study, 48/508 [9%]). Length of maternal postpartum hospital stay did not differ between the groups (median: control, 1.43 days; vs study, 1.47 days). Neonatal outcomes were similar between the 2 groups as shown in Table VIII. There were 5 neonatal deaths, 3 in the study and 2 in the control group. Four of the 5 deaths were caused by complex congenital heart anomalies. The fifth death occurred in a baby from the study group with postnatal asphyxia as a result of bilateral tension pneumothoraces whose 5-minute Apgar score and umbilical cord pH had been normal. In babies who survived, umbilical cord arterial pH was <7.05 in 8 cases in the study group and in 11 cases in the control group; pH was <7.0 in 4 cases in the study group and in 4 cases in the control group. Of these (all with pH values <7.05), cesarean or operative vaginal delivery for nonreassuring fetal status was performed in 6 of 8 in the study group and in 3 of 11 in the control group (P = .07). Neonatal outcomes according to the same parameters defined in Table VIII were also similar between the 2 groups in the subanalysis of patients with the more severe FHR patterns at study entry. Study and clinical staffs successfully placed the fetal oximeter sensor and obtained a signal 95% of the time. The median time that an adequate signal and tracing were obtained was 67%; thus, for an average hour of tracing, approximately 40 minutes of continuous tracing of fetal oxygen saturation was obtained. Comment Whereas fetal pulse oximetry may ultimately improve our ability to avoid adverse perinatal outcome, electronic
FHR monitoring is generally good at detecting hypoxia (ie, sensitive),16, 17 and adverse outcomes caused by intrapartum hypoxia are rare. It would take a prohibitively large sample size for fetal pulse oximetry to demonstrate benefit in reducing fetal or neonatal complications. We therefore chose to study whether fetal pulse oximetry improved the clinician’s accuracy of fetal assessment in deciding on operative intervention because of concern over potential fetal compromise. Therefore this study was designed to test the hypothesis that the addition of fetal pulse oximetry to electronic FHR monitoring would result in a lower rate of cesarean deliveries performed for nonreassuring fetal status without any increase in adverse neonatal outcome. The study confirmed the primary hypothesis, demonstrating a >50% reduction in the rate of cesarean delivery for nonreassuring fetal status. However, the study did not result in an overall decrease in cesarean deliveries. As an alternative method of assessing the accuracy of fetal pulse oximetry, we looked at how the decision for operative intervention for nonreassuring fetal status correlated with various indexes of neonatal hypoxia, acidosis, and depression. Fetal pulse oximetry did significantly improve the sensitivity and specificity of surgical intervention for nonreassuring fetal status. The improved identification and intervention for those babies who required resuscitation and who were acidotic is an important result of this trial. The experience of an obstetrician who rushes to perform cesarean or operative vaginal delivery of the fetus with a concerning FHR pattern, only to deliver a vigorous healthy newborn, is all too common and disconcerting. A device that actually measures the parameter about which the clinician is concerned, fetal oxygenation, and as a result allows more precise prediction and intervention for a more appropriate diagnosis has the potential to result in improved care. Hurried performance of operative delivery, especially when there is no real indication, has the potential to result in suboptimal outcome, with possible increases in infection, hemorrhage, and anesthetic accidents. The additional observation of improved sensitivity in the oximetry group was unexpected. The generally accepted premise that electronic FHR monitoring is a sensitive modality16, 17 was not borne out by the control arm of this study, because many depressed and acidotic babies managed with electronic FHR monitoring alone did not have operative intervention for nonreassuring fetal status (Table VI). Perhaps the discrepancy is because of the fact that previous studies correlated the FHR before delivery with outcome but did not use the behavior of operative intervention as the independent predictor. One could certainly argue whether the FHR patterns allowing entry into this study are appropriate and could also assert that they are arbitrary and not absolutely defined (eg, how many variable decelerations or how long
Volume 183, Number 5 Am J Obstet Gynecol
they should persist before the patient is enrolled). The subanalyses comparing patients with FHR patterns with more and less concern at study entry showed similar rates of reduction in cesarean delivery for nonreassuring fetal status and neonatal safety, suggesting that more stringent entry criteria would not have altered the study results and that the liberal study entry criteria were appropriate. One could also question whether the FHR criteria used for intervention were consistent with common practice. Because FHR pattern interpretation can vary and given substantial differences in management with nonreassuring FHR patterns, what was critical to the success of this study was the creation of a uniform protocol for monitor interpretation and management to which 9 centers would agree and to have an investigator who was not a participant in the clinical study review every tracing. The fact that this was accomplished and that the independent reviewer found in excess of 95% compliance with the protocol was as much as we could hope to accomplish with a study of this magnitude involving electronic FHR monitoring. The increase in cesarean delivery for dystocia in the study group is an interesting and unanticipated outcome. Four possible explanations for this observation were considered. Multiple logistic regression failed to demonstrate any difference in cesarean deliveries for dystocia as a result of an imbalance in randomization. In particular, the difference in rates of dystocia remained after logistic regression analysis even when we corrected for the differences noted between the 2 groups in rates of induction and prostaglandin use. The second possibility was that the clinicians were actually performing cesarean deliveries for nonreassuring fetal status in the study group but labeled those cases as dystocia. This is the most concerning possibility, because in an unblinded study there is a potential for bias by simple mislabeling of the indication for the operative delivery. Partogram analysis demonstrated similar and high frequencies of documented criteria for failure to progress in labor for the 2 groups. A decision for operative intervention because of dystocia in the oximetry group did not occur sooner than the same decision in the control group. Therefore, whether these patients did or did not continue to have nonreassuring FHR patterns, it does appear that they had actual failure to progress. What cannot be determined, because this was not anticipated in the study design, was whether all efforts at augmenting labor were really similar between the 2 groups. Thus these analyses did not find any evidence of investigator bias as the explanation for the higher rate of cesarean delivery for dystocia in the oximetry group. There was no evidence that the device itself slowed or interfered with labor. Finally, the question was asked whether the same patients who avoided cesarean delivery because of nonreassuring fetal status actually went on to have a cesarean delivery because of dystocia.
Garite et al 1057
Analysis of the worst FHR pattern, confirmed by the independent reviewer, in patients undergoing cesarean delivery for dystocia demonstrated that patients in the study group had a higher frequency of abnormal FHR patterns that would likely have resulted in cesarean delivery for nonreassuring fetal status in the absence of pulse oximetry. This last analysis suggests that FHR patterns commonly interpreted as nonreassuring, especially nonreassuring variable decelerations, may actually be indicators for an underlying risk of dystocia. A normal FSpO2 permits the obstetrician to allow labor to proceed through periods of nonreassuring FHR patterns. Ultimately, the patients in the study group with true dystocia were delivered by cesarean because of dystocia, whereas patients in the electronic FHR monitoring group more often underwent cesarean delivery because of nonreassuring fetal status. It is interesting that at least 2 previous randomized studies of electronic FRH monitoring1, 18 showed an increase not only in cesarean delivery for nonreassuring fetal status in the electronically monitored group but also in cesarean delivery for dystocia in that group as well. Future studies investigating this issue of dystocia are warranted. Perhaps the most important single result of this study was that allowing labor to continue in the presence of a nonreassuring FHR but a reassuring FSpO2 with the consequent reduction in cesarean delivery for nonreassuring fetal status did not result in an apparent increase in adverse neonatal outcome. This study did not have the statistical power to analyze small increases in adverse outcome. Nevertheless, it is unlikely that any single randomized study of this complexity can achieve a substantially larger number. In conclusion, this randomized study confirmed the primary study hypothesis that the addition of fetal pulse oximetry to electronic FHR monitoring in fetuses with concerning FHR patterns can lower the cesarean delivery rate for nonreassuring fetal status. This was accomplished without a demonstrable increase in adverse fetal or neonatal outcome. However, the addition of fetal pulse oximetry did not result in a decrease in the overall cesarean delivery rate because of an increase in cesarean delivery for dystocia in the study group. The failure to reduce the overall cesarean delivery rate may cause some clinicians to question either the validity of the study or the value of the reduction in cesarean deliveries for nonreassuring fetal status. A detailed analysis allowed us to provide some reassurance that the increase in cesarean deliveries for dystocia in the study group was apparently not the result of investigator bias, but this cannot be completely ruled out. The improvement in sensitivity and specificity of intervention by cesarean delivery for nonreassuring fetal status demonstrates that fetal pulse oximetry provides additional confidence that the clinician will have the capability to more accurately assess the well-
1058 Garite et al
being of the fetus in labor. The improved sensitivity for detecting true hypoxia and acidosis is an observation that may give hope that some additional babies in distress may be detected by this new modality. It is also possible that with improvement in the technology of fetal pulse oximetry the accuracy of fetal assessment can be improved even further. We should proceed cautiously with any new technology, but this large, multicenter, prospective, randomized study suggests that fetal pulse oximetry offers an opportunity to more accurately assess fetal oxygenation in labor and may allow us to act more appropriately for the fetus truly in need of intervention. We gratefully acknowledge the following primary research nurses and the study sites involved, as well as the many attending and resident physicians, nurses, and patients who participated in and contributed to the successful completion of this study. Presbyterian St Luke’s Hospital, Denver, Colo (Leslie Gardner, RNC) St John’s Mercy Hospital, St Louis, Mo (Kathleen Simpson, PhD, Pat Flynn, RN, BSN) Greenville Hospital System, Greenville, SC (Karen Nichols, RN) St Peter’s Medical Center, New Brunswick, NJ (Kate McGuire, RNC, Sherrill Sorensen, RNC) University of Arizona Health Sciences Center, Tucson, Ariz (Karen Hallbauer, RN, Melissa Goldsmith RN) University of California Irvine Medical Center, Orange, Calif (Klm Peters-Phair, RN) University of Utah Medical Center, Salt Lake City, Utah (Chris Monson, RN) Utah Valley Regional Medical Center, Provo, Utah (Carol Loucks, RNC) Vanderbilt University Medical Center, Nashville, Tenn (Nancy Townsend, MSN) We also express gratitude to the following individuals who were instrumental in the development and completion of this trial: Debbie Reisenthel, Simon Thomas, Marti Letko-Porter, Mary Kenney, Susan Nixon, and Pamela Rumney.
November 2000 Am J Obstet Gynecol
REFERENCES
1. Haverkamp AD, Thompson HE, McFee JG, et al. The evaluation of continuous fetal heart rate monitoring in high-risk pregnancy. Am J Obstet Gynecol 1976;125:310-7. 2. MacDonald D, Grant A, Sheridan-Pereira M, et al. The Dublin randomized control trial of intrapartum fetal heart rate monitoring. Am J Obstet Gynecol 1985;152:524-39. 3. Leveno KJ, Cunningham FG, Nelson S, et al. A prospective comparison of selective and universal electronic fetal monitoring in 34,995 pregnancies. N Engl J Med 1986;315:615-9. 4. Renou P, Chang A, Anderson I, et al. Controlled trials of fetal intensive care. Am J Obstet Gynecol 1976;126:470-6. 5. Kelso IM, Parsons RJ, Lawrence GF, et al. An assessment of continuous fetal heart rate monitoring in labor; a randomized trial. Am J Obstet Gynecol 1978;131:526-32. 6. Johnson N. Development and potential of fetal pulse oximetry. Contemp Rev Obstet Gynecol 1991;3:193-200. 7. Mannheimer PD, Fein ME, Casciani JR. Physio-optical considerations for the design of fetal pulse oximetry sensors. Eur J Obstet Gynecol Reprod Biol 1997;1(Suppl):S9-19. 8. Mannheimer PD, Casciani JR, Fein ME. Wavelength selection for fetal pulse oximetry selection. IEEE Trans Biomed Eng 1997;44:148-58. 9. Dildy GA, Clark SL, Loucks CA. Intrapartum fetal pulse oximetry: past, present, and future. Am J Obstet Gynecol 1996;175:1-9. 10. McNamara H, Chung DC, Lilford R, et al. Do fetal pulse oximetry readings at delivery correlate with cord blood oxygenation and acidaemia? Br J Obstet Gynaecol 1999;99:735-8. 11. Dildy GA, van den Berg PP, Katz M, et al. Intrapartum fetal pulse oximetry: fetal oxygen saturation trends during labor and relation to delivery outcome. Am J Obstet Gynecol 1994;171:679-84. 12. Kuhnert M, Seelbach-Gobel B, Butterwegge M. Predictive agreement between the fetal arterial oxygen saturation and fetal scalp pH: results of the German multicenter study. Am J Obstet Gynecol 1998;178:330-5. 13. Seelbach-Gobel M, Huepel M, Kuhned M, et al. The prediction of fetal acidosis by means of intrapartum fetal pulse oximetry. Am J Obstet Gynecol 1999;180:73-81. 14. Carbonne B, Langer B, Goffinet F, et al. Multicenter study on the clinical value of fetal pulse oximetry. II. Compared predictive values of pulse oximetry and fetal blood analysis. Am J Obstet Gynecol 1997;17:593-8. 15. Kubli FW, Hon EH, Kazin AE, et al. Observations on heart rate and pH in the human fetus during labor. Am J Obstet Gynecol 1969;104:1190-206. 16. Schifrin BS, Dame L. Fetal heart rate patterns. Prediction of Apgar score. JAMA 1972;219:1322-5. 17. Bissonnette JM. Relationship between continuous FHR patterns and Apgar score in the newborn. Br J Obstet Gynaecol 1975; 82:24-8. 18. Haverkamp AD, Orleans M, Langendoerfer S, et al. A controlled trial of the differential effects of continuous fetal heart rate monitoring in labor: a randomized trial. Am J Obstet Gynecol 1979;134:399-412.