Keywords
implementation - fetal resuscitation - category II fetal heart tracing - maternal oxygenation - fetal acidemia - de-implementation
Fetal heart rate (FHR) monitoring during labor is the most common obstetrical procedure in the United States aimed at determining fetal acid–base status at the time of observation.[1] Intrapartum FHR tracings are classified into one of three categories based on baseline rate, beat variability, and deceleration events.[2] Category I tracings are strongly predictive of normal fetal acid–base status while category III are strongly predictive of abnormal fetal acid–base status with subsequent risk for neonatal encephalopathy and cerebral palsy.[3] Category II tracings are classified as indeterminate, and while not predictive of fetal acid–base status, clinical guidance suggests consideration of intrauterine resuscitative measures to promote fetal oxygenation.[2]
[3]
Routine management of category II tracings has historically included supplemental maternal oxygen based on the hypothesis that increased maternal oxygenation can subsequently improve fetal respiratory status.[3] Maternal hyperoxygenation has been demonstrated alone or as part of a resuscitative bundle to improve FHR tracings.[4]
[5]
[6]
[7] However, emerging data challenge this assumption, showing routine maternal oxygenation having no effect on neonatal acidemia, intensive care admission, or other complications.[8]
[9]
[10] In a recent randomized trial among patients with category II tracings, room oxygen was noninferior to supplemental oxygen to improve FHR tracings[9] or prevent neonatal acidosis.[8]
Due to the coronavirus disease 2019 (COVID-19) pandemic, there were concerns about noninvasive oxygen supplementation facilitating iatrogenic transmission of the SARS-CoV-2 virus (severe acute respiratory syndrome-coronavirus-2).[11] Therefore, many organizations, including our institution, affirmed the decision to de-implement the practice of routine maternal oxygen supplementation for nonreassuring FHR tracings.[12]
[13]
[14] While the American College of Obstetricians and Gynecologists (ACOG) has recently updated guidance against routine oxygen supplemental oxygen in the context of Category II or III tracings,[15] the Association of Women's Health, Obstetric, and Neonatal Nurses continues to recommend supplemental oxygen for nonreassuring FHR tracings.[16]
We leveraged our institution's system-wide change to validate withholding routine maternal supplemental oxygen for nonreassuring FHR in a population-based observational study. We hypothesized that the policy change did not increase the risk of neonatal acidemia.
Materials and Methods
This is a population-based, retrospective cohort study performed at a tertiary care hospital in New Orleans, LA, serving a diverse urban population. Our community was affected by one of the earliest COVID-19 spikes in the United States; noninvasive oxygen supplementation facilitating iatrogenic transmission of the SARS-CoV-2 virus was a major concern.[11] With questions regarding the benefits of routine oxygen supplementation for fetal resuscitation having been sufficiently raised,[8] our department de-implemented the routine use of routine oxygen supplementation for category II tracings at the end of March 2020. In addition to email messages and person-to-person communication of the change, high-flow oxygen masks were removed from labor rooms. The global and local climate of rapidly changing practices at the time facilitated the de-implementation of the long-standing bedrock of fetal resuscitation at our institution, as did the perceived reduction of risk to health care personnel. This study was approved by our hospital's institutional review board and was conducted in accordance with the Helsinki Declaration.
All singleton or twin infants with continuous intrapartum electronic fetal monitoring delivered between 23 and 42 weeks estimated gestational age from January 2019 to July 2021 were evaluated. Infants delivered after April 1, 2020, were included in the postdeimplementation cohort, which was compared to historical controls delivered prior to that date. Those undergoing labor with persistent intrapartum category II FHR tracings were included. In our system, intrapartum FHR tracings are evaluated by the immediate care team, as well as through a centralized remote fetal monitoring system staffed by experienced nurses certified in electronic fetal monitoring interpretation.[17] Persistently nonreassuring tracings are flagged and documented, and from this record, we identified persistent category II tracings. Persistently concerning tracings were defined as those meeting ACOG's three-tiered Category II criteria for 30 minutes, or sooner if the monitoring nurse determines that findings (such as recurrent late decelerations) raise clinical concern warranting earlier intervention. Records were excluded if there were known major fetal anomalies, a contraindication to labor (e.g., prior uterine incision other than low transverse hysterotomy or breech presentation), and maternal indications for oxygen supplementation, including active respiratory infection such as COVID-19 ([Fig. 1]).
Fig. 1 Flow diagram for identification of deliveries with persistent category II fetal heart rate tracings by time period.
Category II labor tracings were identified and pooled for data extraction from electronic medical records, with comorbidities defined by International Classification of Disease, 10th Revision (ICD-10) codes. Baseline maternal data included maternal age, gestational age at delivery, pregravid and admission body mass index (BMI), gravidity, parity, Black race, Hispanic ethnicity, pregestational diabetes, chronic hypertension, asthma/chronic respiratory disorders, current smoking status, and illicit drug use during pregnancy. Advanced maternal age was defined as greater than 35 years while obesity was defined as pregravid BMI greater than 30 m/kg2. Pregnancy complications included gestational diabetes, COVID-19 during pregnancy, COVID-19 at admission, fetal growth restriction (FGR), and hypertensive disorders of pregnancy. Labor characteristics evaluated included induction of labor, augmentation of labor, and epidural analgesia. The primary outcome was defined as neonatal acidemia as determined by an umbilical artery or vein pH less than 7.20, the most common categorical outcome for fetal acidemia reported in randomized trials for maternal oxygenation.[18] Secondary neonatal outcomes included median umbilical pH, median umbilical base excess, moderate neonatal acidemia (pH < 7.10), severe neonatal acidemia (pH < 7.00), 5-minute Apgar score <4, and neonatal intensive care admission. Mode of delivery, including the failed trial of labor after cesarean and emergency cesarean, was likewise evaluated. Umbilical artery and venous cord gas assessments are not performed universally for deliveries at our institution, though guidelines specify their use in the case of nonreassuring category II and III heart tracings. The decision to draw umbilical blood gases was dependent on the discretion of the delivering provider based on suspected fetal status at time of delivery. In cases where umbilical artery assessment was not able to be performed secondary to insufficient sampling, umbilical vein parameters were reported; when both were available arterial sample was reported over venous.
Statistical Analysis
Univariable statistics were used to compare baseline characteristics. Categorical variables were compared with the chi-square test while medians of continuous variables were compared with the Wilcoxon ranked-sum test. Outcomes were compared via logistic regression generating adjusted odds ratios (aORs) with 95% confidence intervals (CIs), controlling for prespecified confounders known to be associated with neonatal acidemia that included pregestational diabetes, chronic hypertension, advanced maternal age, obesity, and antenatally diagnosed FGR. Nonnormalized continuous outcome medians were compared via analysis of variance. Missing paired umbilical gases were assumed not to be acidemic in accordance with clinician concern and standard operating practice outlined above. Functional markers of fetal compromise such as low Apgar score and admission to the neonatal intensive care unit were available for the entire cohort.
The two-sided alpha was set at 0.05. Estimating 40% of laboring patients would have category II tracings, we assumed a conservative 4% frequency of neonatal acidemia in our historical controls. Equivalence power calculation for sample size was applied to this study since we hypothesized there to be no difference between the two cohorts, with a beta of 0.10 correlating with a power of 90% to discover a difference if present. To detect a 10% frequency of acidemia in the postdeimplementation group, we estimated the need for 616 patients in each group and findings are reported in accordance with the STROBE Statement ([Supplementary Material S1], available in the online version).
Results
Among 9,088 deliveries between 23 and 42 weeks gestation from January 2019 to June 2021, 1,232 (13.6%) were identified as persistent category II by centralized remote fetal monitoring ([Fig. 1]). After excluding deliveries that had a contraindication to a trial of labor, major known fetal anomalies, and a maternal indication for oxygenation, a total of 1,162 records were included in analysis. The historical cohort of 481 infants delivered between January 2019 and March 2020, with the postdeimplementation cohort consisting of 681 deliveries that occurred between April 2020 and June 2021. There were no significant differences in baseline characteristics between historical controls and the postdeimplementation cohort, including maternal age, baseline BMI, race/ethnicity, and comorbidities ([Table 1]). Pregnancy complications and labor characteristics were likewise similar save for active or historic COVID-19, which as expected was more prevalent after April 2020.
Table 1
Baseline and pregnancy characteristics by historical controls versus postdeimplementation of routine maternal oxygenation for category II intrapartum fetal heart tracings
|
Historical controls
N = 481
Mean (±SD) or N (%)
|
Postdeimplementation
N = 681
Mean (±SD) or N (%)
|
p-Value
|
|
Maternal age (y)
|
29.6 (± 5.4)
|
29.8 (± 5.7)
|
0.436
|
|
Black race, %
|
201 (42.0%)
|
293 (43.0%)
|
0.716
|
|
Hispanic ethnicity, %
|
28 (6.0%)
|
32 (5.0%)
|
0.422
|
|
Baseline BMI, kg/m2
|
28.3 (± 7.7)
|
29.5 (± 8.0)
|
0.009
|
|
Chronic hypertension, %
|
36 (8.0%)
|
62 (9.0%)
|
0.337
|
|
Pregestational DM, %
|
20 (4.0%)
|
20 (3.0%)
|
0.327
|
|
Asthma, %
|
43 (9.0%)
|
65 (9.0%)
|
0.759
|
|
Nulliparous, %
|
304 (63.0%)
|
406 (60.0%)
|
0.220
|
|
Gestational DM, %
|
19 (4.0%)
|
22 (3.0%)
|
0.524
|
|
Pregnancy-associated hypertension, %
|
68 (14.0%)
|
112 (16.0%)
|
0.286
|
|
Gestational age, weeks
|
38.6 (± 16.4)
|
38.6 (± 16.6)
|
0.591
|
|
Fetal growth restriction, %
|
30 (6.0%)
|
31 (5.0%)
|
0.230
|
|
Labor induction, %
|
254 (53.0%)
|
390 (57.0%)
|
0.135
|
|
Labor augmentation, %
|
313 (65.0%)
|
455 (67.0%)
|
0.571
|
|
Epidural anesthesia, %
|
433 (90.0%)
|
620 (91.0%)
|
0.610
|
Abbreviations: BMI, body mass index; DM, diabetes mellitus; SD, standard deviation.
Among 481 historical controls, 181 (38%) had recorded umbilical blood gas results, comparable to the 232 (34%, p = 0.235) recorded postdeimplementation. The prevalence of umbilical blood acidemia was similar between the postdeimplementation cohort (13.8%) and historic controls (15.4%, aOR = 0.87, 95% CI: 0.62, 1.22; [Table 2]). There was no difference in the median umbilical pH between the postdeimplementation cohort (7.2, interquartile range [IQR]: 7.2–7.3; [Table 3]) and historical controls (7.2, IQR: 7.2–7.3, p = 0.882); median base excess was likewise similar (postdeimplementation −6, IQR: −9 to −4 vs. historic controls −7, IQR: −9 to −5, p = 0.7402). No difference was observed for moderate or severe neonatal acidemia ([Table 4]). Clinical outcomes including 5-minute Apgar <4 (postdeimplementation 2.1% vs. historic controls 1.1%, aOR = 1.88, 95% CI: 0.70, 5.07) and neonatal intensive care admission (19.3 vs. 16.2%, OR = 0.80, CI: 0.59, 1.09) were similar.
Table 2
Outcomes by historical controls versus postdeimplementation of routine maternal oxygenation for category II intrapartum fetal heart tracings
|
Historical controls
N = 481
N (%)
|
Postdeimplementation
N = 681
N (%)
|
aOR (95% CI)
|
|
Neonatal acidemia (umbilical pH < 7.2)
|
74 (15.4%)
|
95 (13.8%)
|
0.87 (0.62, 1.22)
|
|
Moderate neonatal acidemia (umbilical pH < 7.1)
|
21 (4.4%)
|
31 (4.6%)
|
1.04 (0.59, 1.82)
|
|
Severe neonatal acidemia (umbilical pH < 7.0)
|
6 (1.3%)
|
12 (1.8%)
|
1.37 (0.53, 3.55)
|
|
5-minute Apgar score <4
|
5 (1.1%)
|
14 (2.1%)
|
1.88 (0.70, 5.07)
|
|
Neonatal intensive care
|
93 (19.3%)
|
110 (16.2%)
|
0.80 (0.59, 1.09)
|
Abbreviations: aOR, adjusted odds ratio; CI, confidence interval.
Table 3
Umbilical artery blood gas analysis by historical controls versus postdeimplementation of routine maternal oxygenation for category II intrapartum fetal heart tracings
|
Historical controls
N = 481
Median (IQR)
|
Postdeimplementation
N = 681
Median (IQR)
|
p-Value
|
|
Umbilical artery pH
|
7.2 (7.2, 7.3)
|
7.2 (7.2, 7.3)
|
0.880
|
|
Umbilical artery base excess
|
−7.0 (−9.0, −5.0)
|
−6.0 (−9.0, −4.0)
|
0.740
|
Abbreviation: IQR, interquartile range.
Table 4
Mode of delivery by historical controls versus postdeimplementation of routine maternal oxygenation for category II intrapartum fetal heart tracings
|
Historical controls
N = 481
N (%)
|
Postdeimplementation
N = 681
N (%)
|
OR (95% CI)
|
|
All deliveries
|
|
Spontaneous vaginal delivery
|
310 (64.5%)
|
415 (60.9%)
|
0.86 (0.68, 1.10)
|
|
Operative vaginal delivery
|
28 (5.8%)
|
31 (4.6%)
|
0.77 (0.46, 1.3)
|
|
Cesarean section
|
143 (29.7%)
|
235 (34.5%)
|
1.25 (0.97, 1.6)
|
|
Trial of labor after cesarean
|
24 (5.0%)
|
52 (7.6%)
|
1.57 (0.96, 2.59)
|
|
VBAC
|
20 (83.3%)
|
33 (63.4%)
|
0.35 (0.10, 1.17)
|
|
Failed TOLAC
|
4 (16.6%)
|
19 (36.5%)
|
2.88 (0.86, 9.68)
|
|
Emergency cesarean
|
16 (3.3%)
|
54 (7.9%)
|
2.5 (1.42, 4.45)
|
Abbreviations: CI, confidence interval; OR, odds ratio; TOLAC, trial of labor after cesarean; VBAC, vaginal birth after cesarean section.
Mode of delivery was similar between groups, with 34.5% (N = 235) of postdeimplementation neonates delivered by cesarean compared to 29.7% (N = 143) of historical controls (p = 0.179, [Table 4]). Both vaginal birth after cesarean and failed trial of labor after cesarean were similar between groups. The proportion of emergency cesareans was higher in the postdeimplementation group (7.9 vs. 3.3%, OR = 2.5, 95% CI: 1.42, 4.43).
Principal Findings
Among 681 infants with category II FHR tracings delivered after the de-implementation of routine maternal oxygen supplementation for fetal resuscitation in a single center, we found no increased prevalence of umbilical blood acidemia compared to predeimplementation controls. At-risk fetal cardiac tracings, identified through a previously described centralized remote fetal monitoring system,[17] demonstrated a similar median umbilical pH and base excess with comparable prevalence of moderate and severe neonatal acidemia. While umbilical cord gas assessments were not universally drawn on the population, clinical outcomes including low Apgar and neonatal intensive care were likewise similar. While the mode of delivery did not differ between groups, emergency cesarean was more common following de-implementation.
Discussion
Interpretation and management of category II FHR tracings has long posed a challenge. Characterized as indeterminate, the tracings can neither confirm nor rule out fetal acid–base perturbations and warrant continued surveillance along with targeted corrective measures based on the clinical scenario. Absence of clearly identifiable upstream conditions such as maternal hypotension, supine maternal positioning, or tachysystole, care providers have few interventions that have been demonstrated to improve FHR tracings. Category II tracings are likewise exceedingly common, observed in over 80% of laboring patients.[13] Despite the widespread use of maternal oxygenation,[10] evidence suggesting no benefit to neonatal outcomes continues to mount in controlled settings.[8]
[9]
[18]
[19] Our findings align with emerging data to reinforce that discontinuation of routine maternal oxygenation for fetal resuscitation does not increase the risk for adverse perinatal outcomes.
Lags between the publication of scientific breakthroughs and widespread implementation are notoriously long, and obstetrics is no exception.[20]
[21] De-implementation of ineffective or detrimental current practices similarly poses a challenge. A wide range of obstacles slows the diffusion of evidence into routine practice at the patient, provider, organization, and policy levels. The COVID-19 pandemic was a disruption event that introduced radical changes at each of these levels. Accordingly, despite relatively recent questioning of the well-established practice, rapid de-implementation of routine maternal oxygenation was able to be accomplished at our institution, facilitated through the perceived potential for the practice to facilitate the spread of the SARS-CoV-2 virus. Here, we demonstrate no adverse perinatal outcomes related to the rapid de-implementation of a long-standing practice.
Interestingly, emergency cesarean was observed to increase in the postimplementation group. While our observational dataset is limited in its ability to assess causality, prospective studies do not suggest a difference in the persistence of category II tracings related to maternal oxygen exposure.[9] The effect of the COVID-19 pandemic on staffing, workflow, and provider decision-making is not assessable here, though such differences may have contributed to the observed difference. Alternatively, we hypothesize that presurgical interventions, even those that do not improve fetal heart tracings (FHT), may function as a stalling mechanism that allows for improvements in the tracing. With fewer interventions available to the care team, the decision for cesarean delivery may have been accelerated. Reassuringly, no difference in cesarean delivery overall was observed.
While our findings support the de-implementation of routine maternal oxygen supplementation for fetal resuscitation at our institution, important questions remain. Limited to a 90% power to determine a clinically meaningful increase in fetal acidemia, we call on other centers that implemented a similar policy change to replicate our methods at their institutions. Should evidence of even a small increase in adverse perinatal outcomes come to light, consideration of de-implementation of routine maternal oxygen supplementation for fetal resuscitation should be revisited.
Strengths and Limitations
Strengths and Limitations
This single-center retrospective study has a range of limitations. Importantly, the application of maternal oxygen supplementation was inconsistently recorded in the electronic medical record (EMR), allowing us only to test outcomes by time period (before and after the policy change) as opposed to documented exposure to oxygen specifically for fetal resuscitation. While patients with documented exposure to supplemental oxygen were excluded from the postdeimplementation group, we cannot validate that none in that group were exposed, potentially biasing our findings towards the null hypothesis if policy adherence was inconsistent. Removal of oxygen masks from patient rooms likely modulated this risk. We identified only 481 of the estimated 616 preimplementation patients required for 90% power, exposing the findings to potential type II error. Umbilical blood gases were collected in only around 36% of all identified persistent category II tracings, likewise limiting the power of the primary outcome; presumably, these gases were not collected based on low clinician suspicion for fetal compromise, hopefully limiting the number of fetuses with acidemia that were not identified. Reassuringly, acidemia-related clinical outcomes including low Apgar scores and neonatal intensive care admission were similar between groups.
Additionally, differences between the two cohorts related to COVID-19 may exist that could confound our findings. While patients with documented COVID-19 were excluded, differences at the personnel or hospital level may impede internal validity. Social impacts of long-term isolation and decisions regarding labor management may have influenced outcomes, potentially contributing to the difference seen in emergency cesareans between the two cohorts.
The project likewise has several strengths. Our institution's well-established, previously described centralized remote fetal monitoring program identifies persistently nonreassuring tracings.[17] De-implementation is a notoriously difficult system-level intervention, though our process was facilitated by the disruption event that was the COVID-19 pandemic, facilitating broad behavioral changes at the personal, provider, and hospital levels. We characterized perinatal compromise not only through umbilical blood assessment but also through clinical outcomes. In presenting real-world clinical findings, we follow basic science,[22]
[23] observational data,[5]
[24] and randomized trial outcomes[8] through to an evidenced-based conclusion.
Conclusion
Here we demonstrate clinical validity and generalizability to mounting evidence that routine intrapartum maternal oxygenation for category II tracings does not improve outcomes for fetuses at risk for acidemia. Our findings validate updated guidelines recommending against routine maternal oxygenation for fetal resuscitation.
