Nationwide assessment of labor induction at full-term for low-risk pregnancy in the United States from 2018 to 2022

Abstract

Role of labor induction requiring cervical preparation for full-term, low-risk pregnant patients is under active investigation. The objective of this study was to describe nationwide temporal trends, characteristics, and outcomes associated with labor induction requiring cervical preparation at full-term gestation for low-risk pregnant patients in the United States. This cross-sectional study queried the Agency for Healthcare Research and Quality mechanism’s Healthcare Cost and Utilization Project National Inpatient Sample. Study population included 5,003,348 hospital deliveries at 39–40^6/7 weeks’ gestation for singleton, cephalic low-risk pregnancy from 2018 to 2022, excluding historically known presumed pre-labor pregnancy risk factors. Exposure was labor induction status, defined as the performance of cervical preparation (ripening or mechanical dilation), with or without adjunctive intervention (artificial rupture of membranes or oxytocin administration). Main outcome measures were set as cesarean delivery as the primary outcome; induction failure, chorioamnionitis, umbilical cord prolapse, in-labor uterine rupture, postpartum hemorrhage, severe maternal morbidity and mortality as the secondary outcomes, assessed with multivariable generalized linear model. A total of 618,785 (12.4%) deliveries had labor induction requiring cervical preparation. The rate increased from 8.8 to 14.4% between QT1/2018-QT2/2020 (P-trend < .001), at which the rate started gradually decreasing when the COVID-19 pandemic occurred (14.4% at QT2/2020 to 13.4% at QT4/2022, P-trend = .04). The majority of cervical preparation was ripening alone (69.9%), followed by mechanical dilation alone (20.7%) and both ripening and mechanical dilation (9.4%). During the 5-year study period, cervical preparation with mechanical dilation alone (14.4% to 26.9%) and both ripening and mechanical dilation (4.6% to 11.4%) increased (both, P-trend < .001); adjunctive use of both oxytocin and artificial rupture of membranes (18.7% to 27.0%) and oxytocin alone (15.0% to 18.8%) increased while artificial rupture of membranes alone decreased (29.5% to 23.0%) (all, P-trend < .001). When compared to 4,384,563 deliveries without labor induction cervical preparation, induction was associated with cesarean delivery (17.1% vs 9.2%, adjusted-incidence rate ratio [aIR] 1.75, 95% confidence interval [CI] 1.74–1.76), chorioamnionitis (3.2% vs 1.9%, aIR 1.67, 95%CI 1.64–1.69), umbilical cord prolapse (0.2% vs 0.1%, aIR 1.67, 95%CI 1.58–1.77), in-labor uterine rupture (21.8 vs 7.8 per 100,000, aIR 2.66, 95%CI 2.17–3.25), postpartum hemorrhage (5.7% vs 3.9%, aIR 1.42, 95%CI 1.40–1.44), severe maternal morbidity (0.4% vs 0.2%, aIR 1.61, 95%CI 1.54–1.68), and maternal mortality (4.8 vs 0.7 per 100,000, aIR 7.34, 95%CI 4.37–12.33). In conclusion, this nationwide cross-sectional assessment suggests the increasing trend of full-term labor induction for study-defined low-risk pregnancy may have stopped during the COVID-19 pandemic in the United States. The observed association with cesarean delivery and adverse outcomes related to labor induction requiring cervical preparation warrant further investigation.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-42904-2.

Introduction

One third of all births in the United States are following induction of labor.¹ There are numerous maternal and fetal indications for which induction of labor is recommended to optimize maternal and/or neonatal outcomes.² However, many inductions are elective, performed in the absence of a clear obstetric indication among low-risk patients.^2,3 While elective induction prior to 39 weeks is avoided due to increased risk of neonatal morbidity, the optimal management upon reaching 39 weeks is still under active investigation.^4,5

A Randomized Trial of Induction Versus Expectant Management (ARRIVE), published in 2018, demonstrated a decreased rate of cesarean delivery and development of hypertensive disorders without an increase in adverse perinatal outcomes for low-risk nulliparous women undergoing induction at term.⁶ The results of this study challenged previously accepted wisdom regarding the risks of induction among nulliparous patients.⁷ While the dramatic rise of inductions in the United States (elective and those with obstetric indications) predates the ARRIVE trial, the trial catalyzed higher rates of elective inductions.^2,3,8

More recent studies have since evaluated the impact of this trial with mixed results. While studies generally affirm that there is not an increased risk of cesarean section, one study reported that a post-ARRIVE cohort experienced increased risk of chorioamnionitis, blood transfusion, maternal intensive care unit admission, and neonatal respiratory morbidity.^5,8–11

Based on a patient’s initial cervical exam or Bishop score,¹² providers may select different methods for induction of labor. The American College of Obstetricians and Gynecologists (ACOG) suggest a variety of methods that are available to utilize during this process.¹³ Researchers and experts in the field have aimed to find the optimal method for induction of labor to achieve the highest likelihood of successful vaginal delivery.¹⁴

Given the on-going rise of low-risk inductions despite areas of uncertainty, there is a need to better understand the risks and benefits and the frequency at which the various induction agents are being used. The objective of this study was to describe temporal trends, characteristics, and outcomes associated with labor induction requiring cervical preparation at full-term for low-risk pregnant patients.

Materials and methods

Data source

This cross-sectional study utilized the National Inpatient Sample.¹⁵ The database is the Healthcare Cost and Utilization Project’s national-level inpatient database. The Healthcare Cost and Utilization Project is the United States health service data platform that is supported by the Agency for Healthcare Research and Quality, one of the twelve federal agencies within the United States Department of Health and Human Service.

The National Inpatient Sample approximates a stratified sample of 20% of discharges in each center from all the participating hospitals across 48 States and the District of Columbia.¹⁵ Every year the dataset captures nearly seven million inpatient admissions. In 2022, more than 4,500 hospitals participated in the program. When weighted for national survey estimates, the National Inpatient Sample represents more than 97% of U.S. population. The program captures a maximum of 40 diagnoses and 25 procedures for the index admission in each encounter. This study used the National Inpatient Sample due to these robust data collecting mechanisms that capture a nationwide overview of full-term labor inductions requiring cervical preparation in the United States. This study was deemed exempt and patient informed consent was waived by the University of Southern California Institutional Review Board due to the use of publicly available, deidentified secondary data (registration number: HS-16-00481). All methods were carried out in accordance with the institutional review board and related regulations.

Study population

The study population was low-risk pregnant patients, as defined by study-specific administrative codes, who had hospital deliveries at full-term from 2018 to 2022. Hospital deliveries refer to either vaginal or cesarean deliveries, identified by the Diagnosis-Related Group codes or the World Health Organization’s International Classification of Disease tenth revision Clinical Modification or Procedure Classification Schema codes (Table S1).^16–19 Full-term gestation was based on the American College of Obstetrics and Gynecology definition, including 39 weeks 0 days to 40 weeks 6 days of gestation.²⁰ Maternal age followed prior investigations to maximize the catchment of administrative codes used.^16,21 Study starting point was chosen due to the availability of information in the study exposure assignment as shown below.

As there is no society-based specific criteria, the definition of low-risk pregnancy in this study was adopted and modified from a prior study (Fig. 1).⁶ Pregnant patients with historically known presumed pre-labor risk factors were excluded, including placental, fetal, maternal, pregnancy, and uterine characteristics (Fig. 1). (i) Placental factors included placenta previa, placenta accreta spectrum, placenta abruption, and vasa previa. (ii) Fetal factors included multi-fetal gestation, growth restriction, in-utero demise, anomaly, malpresentation, large for gestational age, oligohydramnios, polyhydramnios, and pre-labor abnormal fetal heart rate. (iii) Pregnancy factors included prior cesarean delivery, pregnancy loss, and premature rupture of membranes. (iv) Maternal factors included hypertensive disorders of pregnancy (pre-gestational, gestational, pre-eclampsia including eclampsia, and HELLP syndrome), diabetes mellitus (pre-gestational and gestational), infectious conditions including human immunodeficiency virus and/or acquired immunodeficiency syndrome, genital herpes, and coronavirus 2019, malignancy, organ transplant status, systemic lupus erythematosus, and chronic kidney disease. (v) Uterine factors included myoma, anomaly, endometriosis including adenomyosis, and pre-labor uterine rupture. Identifications of these factors followed prior investigations (Table S1).^{16–19,22–25}

Fig. 1.

Study selection schema. *Antenatal diagnosis. **Including human immunodeficiency virus and/or acquired immunodeficiency syndrome, genital herpes, and coronavirus 2019. Abbreviations: NIS, National Inpatient Sample; IUGR, in-utero growth restriction; IUFD, in-utero fetal demise; LGA, large for gestational age; FHR, fetal heart rate; PROM, premature rupture of membrane; SLE, systemic lupus erythematosus; CKD, chronic kidney disease.

Exposure

Exposures defined

The exposure was the performance of labor inductions requiring cervical preparation. This study followed the California Maternal Quality Care Collaborative and the National Perinatal Information Center definitions for labor induction administrative codes (Table S1). Cervical preparation was performed via ripening and/or mechanical dilation per the California Maternal Quality Care Collaborative coding definitions. Cervical ripening for labor induction refers to the use of cervical inserts or tablets with prostaglandins. Cervical dilation refers to the use of mechanical dilation methods such as balloons. ^26,27 In the current study, to ensure that these administrative codes were used for labor induction rather than labor augmentation, performance of cervical ripening and/or mechanical dilation was set as the minimum requirement of labor induction.

Exposure limitations

The exposures, clinical characteristics, and clinical outcomes are subject to many of the limitations intrinsic to large national retrospective database studies. The National Inpatient Sample consists of data derived from administrative coding that occurs during hospitalization. This may result in coding errors due to human error, misclassification of induction of labor versus augmentation of labor, or differences in coding practices across hospitals.

Patient characteristics and outcomes are further limited by reliance on specific International Classification of Disease codes to identify diagnoses, which results in a lack of granular clinical data within the dataset. Due to the absence of bishop scores or cervical exams, the performance of artificial rupture of membranes and/or oxytocin administration was set as the adjunctive intervention for labor augmentation. This lack of detailed clinical data also limits the ability to determine the actual indication for labor induction (obstetric, medical, or elective),²⁸ the shared decision-making process between obstetrician and patient, the type of cervical dilation used (e.g., Foley ballon)¹⁴ and whether it was done in the inpatient or outpatient setting, patient Bishop score, and parity. These limitations reduce the ability to perform a true intention-to-treat level analysis. The absence of parity data is particularly critical given the recent practice-changing data which was published during this study which may have influenced clinical management patterns.^6,29

Outcome measures

The primary outcome measure was cesarean delivery. This was selected as cesarean delivery has been the main area of research interest in labor induction for low-risk pregnancy.⁶ The secondary outcome measures were induction failure including obstructed labor, chorioamnionitis, abnormal fetal heart rate, umbilical cord prolapse, in-labor uterine rupture, operative delivery (vacuum or forceps), severe perineal laceration (third, fourth, and anal sphincter), postpartum hemorrhage including blood transfusion, and prolonged length of hospitalization. These were selected in relevance of labor induction in general. Identification of these outcome measures followed prior investigations (Table S1).^{16–19,22–25,30,31}

Co-secondary outcome measures included severe maternal morbidity and mortality. Identifications of these outcome measures followed prior investigations (Table S1). This study followed the Centers for Disease Control and Prevention definition and coding schema to identify severe maternal morbidity (a total of 19 indicators):³² acute myocardial infarction, aneurysm, acute renal failure, adult respiratory distress syndrome, amniotic fluid embolism, cardiac arrest / ventricular fibrillation, cardiac rhythm conversion, disseminated intravascular coagulation, heart failure / arrest during surgery or procedure, puerperal cerebrovascular disorders, pulmonary edema / acute heart failure, severe anesthesia complications, sepsis, shock, sickle cell disease with crisis, air and thrombotic embolism, hysterectomy, temporary tracheostomy, and ventilation. The National Inpatient Sample captures maternal mortality event during the index hospitalization. Neonatal outcomes were not included due to the lack of associated data in the National Inpatient Sample.

Study covariates

Study covariates other than study inclusion / exclusion, exposure, and outcome measures included clinical demographics with maternal age (< 25, 25–29, 30–34, 35–39, and ≥ 40 years), year of delivery (every one-year), race and ethnicity (Asian, Black, Hispanic, Native American, Other, and White) determined per the Healthcare Cost and Utilization Project, census-level median household income (every quarter), primary payer (Medicaid, private insurance including HMO, self-pay, and other), U.S. census division (New England, Mid-Atlantic, East North Central, West North Central, South Atlantic, East South Central, West South Central, Mountain, and Pacific; Table S2), hospital relative bed capacity (small, mid, and large) determined per the Healthcare Cost and Utilization Project, and obesity. Cases with unknown status were grouped as one category in each study covariates.

Statistical analysis

The first-step of the analysis was to evaluate the temporal trends of labor induction (performance, modality, and adjunctive intervention) and relevant outcome measures (cesarean delivery, induction failure, and chorioamnionitis). Linear segmental regression model with log-transformation was used for statistical analysis. Time factor was assessed with year-quarter increments to maximize the granularity in temporality. Inflection point was determined in an automated fashion.

The second-step of the analysis was to evaluate the characteristics associated with labor induction and cesarean delivery. A multivariable generalized linear model was created by fitting all the measured clinical factors presumably known prior to labor induction. These included maternal age, year, race and ethnicity, primary payer, census-level median household income, U.S. census division, hospital relative bed capacity, obesity, and gestational age at delivery. The effect size for labor induction was expressed with adjusted-rate ratio (aRR) and a corresponding 95% confidence interval (CI). The second multivariable model was created to assess the characteristics associated with cesarean delivery among those who had labor induction. Additional study covariates included in this model were in-labor factors: cervical preparation modality, additional induction intervention, induction failure, abnormal fetal heart rate, chorioamnionitis, umbilical cord prolapse, and in-labor uterine rupture.

The last-step of the analysis was to assess the association between the exposure and outcome measures. Specifically, generalized linear model was used to assess the association of labor induction and the targeted outcome measures (cesarean delivery, chorioamnionitis, umbilical cord prolapse, in-labor uterine rupture, and severe maternal morbidity and mortality), adjusted for presumed pre-induction characteristics (maternal age, year, race / ethnicity, census-level median household income, primary payer, U.S. census division, hospital relative bed capacity, obesity, and gestational age). Magnitude of statistical strength was expressed with adjusted-incidence rate ratio (aIR) and a corresponding 95%CI.

Several sensitivity analyses were conducted to assess the robustness of study findings. First, temporal trends, characteristics, and outcomes were evaluated per the U.S. census division. The rationale of this assessment was that the regional disparity for full-term labor induction has been relatively understudied. Second, a classification-tree was constructed to visualize the pattern of clinical characteristics associated with cesarean delivery among the patients who had labor induction. All the presumed pre-induction factors were considered in the modeling. Recursive partitioning analysis was used, and the chi-square automatic interaction detector method was employed with a stopping rule of a maximum three layers.³³ In the determined patterns, the incidence rate of cesarean delivery was computed. Third, the exposure-outcome association was assessed by restricting to age < 35 years and non-obesity. Fourth, outcome measures were assessed among whole hospital deliveries during the study period. In an exploratory, hypothesis-generating fashion, individual indicator of severe maternal morbidity was evaluated.

All statistical analyses were based on two-tailed hypotheses, and a probability value of less than 0.05 was considered statistically significant. The weighted values for national estimates provided by the program were used for the analysis. Statistical Package for Social Sciences (IBM SPSS, version 29.0, Armonk, NY, USA) was used for the analysis. The Strengthening the Reporting of Observational Studies in Epidemiology guidelines were followed to outline and report the performance of this study.³⁴

Results

Cohort characteristics

Study selection schema is shown in Fig. 1. A total 17,524,470 hospital deliveries from 2018 to 2022 were screened. Serial exclusions were executed for 6 layers: (i) non-full-term delivery (n = 7,790,443), (ii) placental factors (n = 53,300), (iii) fetal factors (n = 1,390,639), (iv) pregnancy factors (n = 2,068,160), (v) maternal factors (n = 1,160,545), and (vi) uterine factors (n = 58,035). Following these exclusions, a total of 5,003,348 deliveries at 39–40 weeks’ gestation for singleton, cephalic, low-risk pregnancy met the study inclusion criteria.

Cohort-level characteristics are shown in Table 1. The median age was 28 (interquartile range 24–32) years. The majority were White individuals (54.1%), privately insured (53.7%), and delivered at 39 weeks’ gestation.

Table 1.

Study population.

No	N = 5,003,348
Age (y)	28 (24–32)
< 25	1,350,955 (27.0)
25–29	1,510,130 (30.2)
30–34	1,418,055 (28.3)
35–39	614,395 (12.3)
≥ 40	109,815 (2.2)
Year*
Before QT2/2020	2,419,954 (48.4)
QT2/2020 or later	2,583,274 (51.6)
Race / ethnicity
White	2,705,324 (54.1)
Black	583,370 (11.7)
Hispanic	1,021,790 (20.4)
Asian	274,210 (5.5)
Native American	31,520 (0.6)
Other	211,415 (4.2)
Unknown	175,720 (3.5)
Primary payer
Medicaid	2,004,584 (40.1)
Private including HMO	2,684,664 (53.7)
Self-pay	134,585 (2.7)
Other	173,795 (3.5)
Unknown	5,720 (0.1)
Census-level median household income
QT1 (lowest)	1,311,685 (26.2)
QT2	1,267,695 (25.3)
QT3	1,245,685 (24.9)
QT4 (highest)	1,140,010 (22.8)
Unknown	38,275 (0.8)
Census division
Division 1 (New England)	185,450 (3.7)
Division 2 (Mid-Atlantic)	565,255 (11.3)
Division 3 (East North Central)	740,336 (14.8)
Division 4 (West North Central)	380,269 (7.6)
Division 5 (South Atlantic)	926,665 (18.5)
Division 6 (East South Central)	317,995 (6.4)
Division 7 (West South Central)	695,630 (13.9)
Division 8 (Mountain)	416,220 (8.3)
Division 9 (Pacific)	775,530 (15.5)
Hospital relative bed capacity
Small	1,095,498 (21.9)
Mid	1,508,619 (30.2)
Large	2,399,231 (48.0)
Obesity
No	4,580,789 (91.6)
Yes	422,560 (8.4)
Gestational age at delivery
39 weeks	3,074,274 (61.4)
40 weeks	1,929,074 (38.6)

Median (interquartile range) or number (percentage per column) is shown. Abbreviation: QT, quartile.

Labor induction modality

A total of 618,785 (12.4%) deliveries had labor induction requiring cervical preparation, corresponding to approximately one in 8 deliveries. The remaining 4,384,563 (87.6%) deliveries did not have labor induction. The majority of cervical preparation was ripening alone (69.9%), followed by mechanical dilation alone (20.7%) and both ripening and mechanical dilation (9.4%). Adjunctive intervention for labor induction included artificial rupture of membranes (50.0%) and oxytocin administration (42.0%). Combination patterns of these labor induction modalities are shown in Fig. 2.

Fig. 2.

Labor induction modality. The utilization patterns are shown in descending order. Abbreviation: AROM, artificial rupture of membrane.

There was heterogeneity without one single dominant approach. Among 12 patterns of labor induction modality, cervical ripening alone was the most frequent approach (25.7%), followed by cervical ripening and artificial rupture of membranes (17.7%).

Trends of labor induction

At the cohort-level evaluation of 5,004,163 deliveries (Fig. 3), the rate of labor induction increased by 63.6% from 8.8% in the first year-quarter of 2018 to 14.4% in the second year-quarter of 2020 when the COVID-19 global pandemic started (P-trend < 0.001). This increase was followed by a gradual decrease from 14.4% in the second year-quarter of 2020 to 13.4% in the fourth year-quarter of 2022 (P-trend = 0.04).

Fig. 3.

Temporal trend of labor induction for full-term, low-risk pregnancy. Labor induction rate was examined among 5,004,163 hospital deliveries for singleton, cephalic full-term, low-risk pregnancy from 2018 to 2022. Line indicates modeled value. Inflection point was determined in automated fashion. Dots indicate observed values, and bars indicate standard error. (↑) indicates statistically significant increase. (↓) indicates statistically significant increase. Vertical dashed line indicates the beginning of COVID-19 pandemic. The vertical axis is truncated to 0–20% for visibility. Abbreviation: QT, year-quarter; COVID-19, coronavirus 2019.

Temporal trends of cervical preparation and adjunctive intervention were shown in Fig. 4. During the 5-year study period, cervical preparation with mechanical dilation alone (14.4% to 26.9%; 86.8% relative-increase) and both ripening and mechanical dilation (4.6% to 11.4%; 149% relative-increase) increased (both, P-trend < 0.001; Fig. 4A). Adjunctive use of both oxytocin and artificial rupture of membranes (18.7% to 27.0%; 44.4% relative-increase) and oxytocin alone (15.0% to 18.8%; 25.3% relative-increase) increased while artificial rupture of membranes alone decreased (29.5% to 23.0%; 22.0% relative-decrease) (all, P-trend < 0.001; Fig. 4B).

Fig. 4.

Temporal trends of labor induction by method of induction. Among 618,785 hospital deliveries following labor induction from 2018 to 2022, temporal trends of (A) proportional distributions of cervical preparation modality (ripening alone, mechanical dilation alone, or both ripening and mechanical dilation) and (B) performance rate of adjunctive intervention (AROM and/or oxytocin administration) were evaluated. (↑) indicates statistically significant increase. (↓) indicates statistically significant increase. Abbreviations: QT, year-quarter; AROM, artificial rupture of membranes.

Pre-induction characteristics

Results of multivariable analysis are shown in Table 2. Either older or younger maternal age, COVID-19 pandemic era, private insurance, obesity, and 40 weeks’ gestation were associated with increased rate of labor induction. Of those, maternal age ≥ 40 years compared to age 30–34 years had the largest effect size for labor induction (20.8% vs 11.2%, aRR 1.98, 95%CI 1.95–2.00), followed by obesity (19.0% vs 11.8%, aRR 1.60, 95%CI 1.59–1.62).

Table 2.

Multivariable analysis for labor induction.

Characteristic	Labor induction rate	aRR (95%CI)
Age (y)
< 25	13.3%	1.27 (1.26–1.28)
25–29	11.9%	1.09 (1.08–1.10)
30–34	11.2%	1.00 (reference)
35–39	12.8%	1.17 (1.16–1.18)
≥ 40	20.8%	1.98 (1.95–2.00)
Year*
Before QT2/2020	10.9%	1.00 (reference)
QT2/2020 or later	13.7%	1.25 (1.24–1.26)
Race / ethnicity
White	13.4%	1.20 (1.18–1.21)
Black	12.8%	1.12 (1.11–1.14)
Hispanic	10.2%	1.00 (0.99–1.01)
Asian	9.9%	1.00 (reference)
Native American	10.6%	1.02 (0.98–1.05)
Other	11.1%	1.04 (1.02–1.06)
Unknown	13.8%	1.26 (1.24–1.28)
Primary payer
Medicaid	11.4%	1.00 (reference)
Private including HMO	13.1%	1.15 (1.15–1.16)
Self-pay	9.3%	0.86 (0.85–0.88)
Other	13.4%	1.19 (1.17–1.20)
Unknown	11.5%	0.98 (0.92–1.07)
Census-level median household income
QT1 (lowest)	12.5%	1.00 (reference)
QT2	12.4%	0.98 (0.97–0.98)
QT3	12.3%	0.98 (0.98–0.99)
QT4 (highest)	12.3%	1.00 (0.99–1.01)
Unknown	10.8%	1.00 (0.97–1.03)
Census division
Division 1 (New England)	11.5%	1.00 (reference)
Division 2 (Mid-Atlantic)	12.9%	1.13 (1.11–1.15)
Division 3 (East North Central)	15.9%	1.37 (1.35–1.40)
Division 4 (West North Central)	12.8%	1.09 (1.07–1.11)
Division 5 (South Atlantic)	13.1%	1.15 (1.13–1.17)
Division 6 (East South Central)	12.9%	1.13 (1.11–1.15)
Division 7 (West South Central)	11.0%	1.03 (1.02–1.05)
Division 8 (Mountain)	10.9%	0.98 (0.97–0.99)
Division 9 (Pacific)	9.5%	0.87 (0.86–0.88)
Hospital relative bed capacity
Small	12.5%	1.04 (1.03–1.04)
Mid	11.7%	1.00 (reference)
Large	12.7%	1.07 (1.07–1.08)
Obesity
No	11.8%	1.00 (reference)
Yes	19.0%	1.60 (1.59–1.62)
Gestational age at delivery
39 weeks	11.3%	1.00 (reference)
40 weeks	14.0%	1.26 (1.25–1.27)

Generalized linear model for multivariable analysis. All the listed covariates were entered in the model. *Study period was divided to reflect the inflection point of labor induction. Abbreviations: aRR, adjusted-rate ratio; CI, confidence interval; QT, quartile.

Primary outcome: cesarean delivery

When cesarean delivery rates were compared between 618,785 deliveries with labor induction and 4,384,563 deliveries without labor induction (Table 3), the cesarean delivery rate was 75% higher in the labor induction group compared on the non-induction group (17.1% [105,640 of 618,785] vs 9.2% [404,905 of 4384,563], aIR 1.75, 95%CI 1.74–1.76). This association remained among those aged < 35 years and non-obesity (Table 3). The cesarean delivery rate was 32.4% for any hospital deliveries from 2018 to 2022.

Table3.

Association between labor induction and outcome measures.

	Base cohort			Age < 35 / no obesity disorder
Outcome measure	Labor induction (-)	Labor induction ( +)	aIR (95%CI) a	Labor induction (-)	Labor induction ( +)	aIR (95%CI) b
Cesarean delivery	9.2%	17.1%	1.75 (1.74–1.76)	8.7%	15.7%	1.77 (1.75–1.78)
Operative deliveryc	4.8%	6.0%	1.25 (1.24–1.27)	4.9%	6.2%	1.24 (1.22–1.25)
Chorioamnionitis	1.9%	3.2%	1.67 (1.64–1.69)	1.9%	3.1%	1.62 (1.59–1.65)
Umbilical cord prolapse	0.1%	0.2%	1.67 (1.58–1.77)	0.1%	0.2%	1.75 (1.63–1.87)
Uterine rupture in-labor	7.8 / 100,000	21.8 / 100,000	2.66 (2.17–3.25)	6.8 / 100,000	13.4 / 100,000	2.02 (1.52–2.68)
Third degree laceration	1.8%	2.2%	1.24 (1.22–1.27)	1.9%	2.4%	1.21 (1.19–1.24)
Forth degree lacerationd	0.3%	0.4%	1.20 (1.15–1.26)	0.3%	0.4%	1.12 (1.07–1.18)
Postpartum hemorrhagee	3.9%	5.7%	1.42 (1.40–1.44)	3.8%	5.5%	1.43 (1.41–1.45)
Length of stay ≥ 7 days	0.1%	0.2%	3.50 (3.26–3.75)	< 0.1%	0.1%	3.12 (2.85–3.42)
Severe maternal morbidityf	0.2%	0.4%	1.61 (1.54–1.68)	0.2%	0.3%	1.53 (1.45–1.62)
Maternal mortality g	0.7 / 100,000	4.8 / 100,000	7.34 (4.37–12.33)	0.4 / 100,000	4.6 / 100,000	12.78 (6.47–25.26)

^aGeneralized linear model was used to assess the association of labor induction and the targeted outcome measures (cesarean delivery, chorioamnionitis, umbilical cord prolapse, and uterine in labor), adjusted for presumed pre-induction clinical characteristics (maternal age, year, race / ethnicity, census-level median household income, primary payer, U.S. census division, hospital relative bed capacity, obesity, and gestational age). Cases without labor induction was set as the reference group.

^bModel followed the base cohort except for excluding obesity in the adjusting factor.

^cVacuum or forceps delivery.

^dIncluding anal sphincter laceration.

^eIncluding blood transfusion.

^fComposite assessment of 19 indicators per the Centers for Disease Control and Prevention definition (blood transfusion / eclampsia not included). The incidence rate of any hospital deliveries during the study period was 0.9%. Results for individual morbidity indicator are shown in Table S7.

^gEvent during the index hospitalization for delivery. The maternal mortality rate of any hospital deliveries from 2018 to 2022 was 7.9 / 100,000.

Abbreviations: aIR, adjusted-incidence rate ratio; CI, confidence interval.

Independent characteristics associated with cesarean delivery among 618,785 labor inductions are shown in Table 4. Older maternal age, Black / Asian individuals, obesity, 40-weeks’ gestation, induction failure including labor obstruction, abnormal fetal heart rate, chorioamnionitis, umbilical cord prolapse, and in-labor uterine rupture were associated with cesarean delivery. Specific to the cervical preparation approach, cesarean delivery rate was modestly higher for the ripening and mechanical dilation group (24.0% vs 15.7%, aIR 1.20, 95%CI 1.18–1.23) and the mechanical dilation alone group (18.4% vs 15.7%, aIR 1.05, 95%CI 1.04–1.07) compared to the ripening alone group.

Table 4.

Multivariable analysis for cesarean delivery.

Characteristic	CD rate	aRR (95%CI)
Age (y)
< 25	17.2%	1.07 (1.05–1.09)
25–29	16.3%	1.00 (reference)
30–34	16.9%	1.01 (1.00–1.03)
35–39	17.5%	1.06 (1.04–1.08)
≥ 40	21.4%	1.24 (1.21–1.28)
Year*
Before QT2/2020	17.4%	1.00 (reference)
QT2/2020 or later	16.8%	0.94 (0.93–0.95)
Race / ethnicity
White	16.0%	1.00 (reference)
Black	21.3%	1.11 (1.09–1.13)
Hispanic	16.4%	1.01 (0.99–1.03)
Asian	22.3%	1.18 (1.15–1.21)
Native American	14.9%	0.97 (0.89–1.06)
Other	18.8%	1.09 (1.06–1.13)
Unknown	16.2%	1.00 (0.96–1.03)
Census-level median household income
QT1 (lowest)	17.1%	1.02 (1.01–1.04)
QT2	16.4%	1.00 (reference)
QT3	17.0%	0.99 (0.97–1.01)
QT4 (highest)	17.8%	0.98 (0.96–1.00)
Unknown	17.4%	1.11 (1.03–1.20)
Primary payer
Medicaid	15.7%	1.05 (1.01–1.10)
Private including HMO	18.1%	1.24 (1.18–1.30)
Self-pay	14.9%	1.00 (reference)
Other	15.7%	1.13 (1.07–1.20)
Unknown	17.6%	1.19 (0.99–1.44)
Census division
Division 1 (New England)	19.6%	0.99 (0.95–1.02)
Division 2 (Mid-Atlantic)	19.3%	1.00 (0.98–1.03)
Division 3 (East North Central)	16.3%	1.00 (reference)
Division 4 (West North Central)	15.5%	1.02 (0.99–1.05)
Division 5 (South Atlantic)	18.1%	1.11 (1.09–1.14)
Division 6 (East South Central)	19.3%	1.42 (1.38–1.45)
Division 7 (West South Central)	16.9%	1.18 (1.15–1.21)
Division 8 (Mountain)	13.8%	0.89 (0.87–0.92)
Division 9 (Pacific)	15.7%	0.84 (0.82–0.86)
Hospital relative bed capacity
Small	16.9%	1.03 (1.01–1.05)
Mid	16.8%	1.00 (reference)
Large	17.3%	0.99 (0.97–1.01)
Obesity
No	16.1%	1.00 (reference)
Yes	23.9%	1.34 (1.32–1.36)
Gestational age at delivery
39 weeks	14.1%	1.00 (reference)
40 weeks	20.9%	1.25 (1.24–1.27)
Cervical preparation
Cervical ripening only	15.7%	1.00 (reference)
Cervical dilation only	18.4%	1.05 (1.04–1.07)
Both ripening and dilation	24.0%	1.20 (1.18–1.23)
Adjunctive intervention
AROM / oxytocin	7.2%	1.00 (reference)
AROM alone	6.2%	1.00 (0.98–1.02)
Oxytocin alone	11.9%	1.37 (1.34–1.39)
Neither two	12.5%	1.44 (1.42–1.46)
Induction failure / labor obstruction
No	12.8%	1.00 (reference)
Yes	56.1%	3.61 (3.56–3.65)
Abnormal fetal heart rate
No	9.0%	1.00 (reference)
Yes	32.0%	3.03 (2.99–3.07)
Chorioamnionitis
No	16.3%	1.00 (reference)
Yes	41.7%	1.53 (1.50–1.57)
Umbilical cord prolapse
No	16.9%	1.00 (reference)
Yes	92.7%	4.70 (4.45–4.96)
Uterine rupture in-labor
No	17.1%	1.00 (reference)
Yes	74.1%	2.12 (1.74–2.58)

Generalized linear model for multivariable analysis. All the listed covariates were entered in the model. *Study period was divided to reflect the inflection point of labor induction. Abbreviations: CD, cesarean delivery; aRR, adjusted-rate ratio; CI, confidence interval; QT, quartile.

During the study period, cesarean delivery rates among the labor induction group were overall similar, ranging from 16.2 to 18.6% (Fig. 5). Induction failure rate gradually decreased by 13.0% from 10.8% in the first year-quarter of 2018 to 9.4% in the last year-quarter of 2022 (P-trend < 0.001; Fig. 5).

Fig. 5.

Temporal trends of obstetric outcome measures. Among 618,785 hospital deliveries following labor induction from 2018 to 2022, temporal trends of cesarean delivery, induction failure including obstructed labor, and chorioamnionitis were evaluated. (↑) indicates statistically significant increase. (↓) indicates statistically significant increase. Line indicates modeled value. Dots indicate observed values, and bars indicate standard error. The vertical axis is truncated to 0–20% for visibility. Abbreviation: QT, year-quarter.

A classification-tree was constructed to visualize the patterns of pre-induction characteristics to assess cesarean delivery (Table S3). There were 45 patterns identified for cesarean delivery, and 4 patterns had the cesarean delivery rate of higher than 20%. Of those, patients aged ≥ 40 years who had obesity and delivered in the East South Central U.S. division (Kentucky, Tennessee, Mississippi, and Alabama; Table S2) had the highest rate of cesarean delivery (33.3%).

Secondary outcomes

The measured secondary outcomes were compared between 618,785 deliveries via labor induction requiring cervical preparation and 4,384,564 deliveries without (Table 3). Labor induction was associated with higher rates of operative vaginal delivery (6.0% vs 4.8%, aIR 1.25, 1.24–1.27), chorioamnionitis (3.2% vs 1.9%, aIR 1.67, 95%CI 1.64–1.69), umbilical cord prolapse (0.2% vs 0.1%, aIR 1.67, 95%CI 1.58–1.77), in-labor uterine rupture (21.8 vs 7.8 per 100,000, aIR 2.66, 95%CI 2.17–3.25), severe perineal laceration including fourth degree / anal sphincter tear (0.4% vs 0.3%, aIR 1.20, 95%CI 1.15–1.26), postpartum hemorrhage (5.7% vs 3.9%, aIR 1.42, 95%CI 1.40–1.44), and length of hospitalization of 7 days or longer (0.2% vs 0.1%, aIR 3.50, 95%CI 3.26–3.75) compared to non-induction. From 2018 to 2022, incidences of chorioamnionitis increased by 65.2% in the labor induction group (2.3% to 3.8%, P-trend < 0.001; Fig. 5).

The incidence rates of severe maternal morbidity (composite: 0.4% vs 0.2%, aIR 1.61, 95%CI 1.54–1.68), and maternal mortality during the index hospitalization (4.8 vs 0.7 per 100,000, aIR 7.34, 95%CI 4.37–12.33) were higher in the labor induction group compared to the non-induction group (Table 3). Similar results were seen when restricted to age < 35 years and non-obesity (Table 3). These rates were lower compared to that of any hospital deliveries from 2018 to 2022 (0.9% for severe maternal morbidity; and 7.9 per 100,000 for in-hospital maternal mortality, respectively). In an exploratory analysis of individual morbidity indicator (Table 5), labor induction had the largest association for amniotic fluid embolism with cardiac arrest, coagulopathy, and/or shock (unadjusted-IR 3.54, 95%CI 2.19–5.73).

Table 5.

Assessment of individual morbidity indicator (exploratory).

Morbidity indicator	Labor induction (-)	Labor induction ( +)	IR (95%CI) a
Amniotic fluid embolism (multi-morbidities) b	1.1	4.0	3.54 (2.19–5.73)
Cardiac arrest / ventricular fibrillation	1.9	6.5	3.34 (2.29–4.86)
Conversion of cardiac rhythm	2.3	7.3	3.19 (2.24–4.53)
Acute renal failure	23.0	63.8	2.77 (2.47–3.11)
Sickle cell disease with crisis	0.9	2.4	2.66 (1.47–4.81)
Hysterectomy	16.8	42.8	2.56 (2.22–2.94)
Aneurysm	1.6	4.0	2.53 (1.60–4.00)
Amniotic fluid embolism (any)	3.0	6.5	2.18 (1.53–3.11)
Pulmonary edema / acute heart failure	8.1	15.4	1.90 (1.51–2.38)
Ventilation	7.8	14.5	1.88 (1.49–2.37)
Acute respiratory distress syndrome	19.8	33.9	1.71 (1.47–1.99)
Disseminated intravascular coagulation	97.7	165.7	1.70 (1.58–1.82)
Air and thrombotic embolism	13.7	22.6	1.65 (1.38–1.99)
Shock	23.3	38.0	1.63 (1.42–1.88)
Sepsis	48.6	63.8	1.31 (1.18–1.46)
Severe anesthesia complications	4.7	4.9	1.04 (0.71–1.52)
Puerperal cerebrovascular disorders	14.0	11.3	0.81 (0.63–1.03)
Acute myocardial infarction	0.8	0	n/a

Incidence rate per 100,000 hospital deliveries is shown. Morbidity indicators with small number event are.

not shown per HCUP guidelines. ^a Unadjusted analysis for exploratory fashion. ^b Amniotic fluid embolism with any of the following additional indicators (cardiac arrest, coagulopathy, and shock). Indicators are shown in descending order. Abbreviations: IR, incidence rate ratio; CI, confidence interval.

Census division assessment

There was a heterogeneity in labor induction use, approach, and outcomes across the U.S. regions. Labor induction rate exceeded 15% in East North Central division (Wisconsin, Michigan, Illinois, Indiana, Ohio) (15.9%) whereas the rate was lower than 10% in Pacific division (Alaska, Washington, Oregon, California, Hawaii) (9.5%) (Figure S1). Labor induction rate in East North Central division was 58% higher compared to Pacific division (aRR 1.58, 95%CI 1.57–1.60).

Among the nine census divisions, labor induction rate continued to increase during the COVID-19 pandemic in East North Central division (Figure S2). Cervical preparation with mechanical dilation ranged from 14.1% in West South Central division (Oklahoma, Texas, Arkansas, Louisiana) to 47.5% in New England division (Maine, New Hampshire, Vermont, Massachusetts, Rhode Island, Connecticut) (P < 0.001; Figure S3).

Higher rates of cesarean delivery in the labor induction group compared to the non-induction group were consistent across the nine divisions (Table S4), but the effect size varied from 43% in West South Central division (aIR 1.43, 95%CI 1.40–1.46) to 104% in East North Central division (aIR 2.04, 95%CI 1.97–2.11). Census division-level labor induction rate was not associated with census division-level cesarean delivery rate (P = 0.394; Figure S4).

Discussion

Principle findings

Key results in this study were the following. First, the utilization of labor induction requiring cervical preparation for full-term, low-risk pregnancy increased until the second year-quarter of 2020 when the COVID-19 global pandemic hit the United States. Second, labor induction approach evolved during the study period. Third, the labor induction group had higher rates of cesarean delivery and other measured adverse outcomes compared to the non-induction group. Last, there was a clinically meaningful heterogeneity in labor induction use, approach, and outcomes across nine census divisions.

Results in the context of what is known

Temporal trends

The rise in inductions in the United States has been explored by prior investigators. The focus, however, had been time periods pre- and post-publication of the ARRIVE trial. In 2023, Nethery et al. concluded that while there was a 4-week period immediately after the trial’s publication with a 42% increase in induction, there was overall no change in the background trend of increasing elective induction of labor.³ This is in contrast to the conclusions of other authors who concluded the post-ARRIVE induction rate exceeded what was expected based off the pre-existing trend.^8,29 However, only one of the three aforementioned studies included data during the COVID-19 pandemic. These studies all focus on induction of labor for nulliparous women, whereas, the current study does not include parity data and includes nulliparous and multiparous patients which limits some of the comparisons to prior literature.

The present study corroborates the previously described rise in induction among low-risk patients, but also provides insight into the trend following the onset of the pandemic. There are a variety of reasons that labor induction among low-risk patients may have declined following the pandemic. First, patients may have preferred to go into spontaneous labor due to restrictions on number of support persons on labor and delivery and/or fears of acquiring COVID in the hospital. Additionally, hospital policies may have prevented induction of labor among low-risk patients. Notably, however, “Labor and Delivery Guidance for COVID-19” published in May of 2020 in American Journal of Obstetrics and Gynecology Maternal Fetal-Medicine specifically noted that “inductions for reaching 39 weeks” should not be postponed or re-scheduled, after patient counseling.³⁵

The trend in induction methods is of interest and requires further investigation. Here, this investigation found an increase in the use of combined cervical preparation methods over time. This pattern aligns with the growing body of evidence that suggests a combination of cervical preparation methods may decrease the time to delivery and decrease the rate of cesarean Sect. ³⁶ This study also found a rise in method alone for cervical ripening. It should also be noted that an on-going area of interest is the role of outpatient cervical preparation through mechanical dilation, which cannot be captured by the dataset utilized in this study. With respect to adjunctive methods, the increase in the combination of artificial rupture of membrane and oxytocin aligns with the 2024 ACOG guidance recommending amniotomy for patients undergoing induction.³⁷

Characteristics

Characteristics associated with induction identified by this study included obesity, advanced maternal age, and private insurance. It is unsurprising that maternal age ≥ 40 years old had the largest effect size. Advanced maternal age ≥ 40 years is a risk factor for stillbirth with rates of stillbirth at 39 weeks similar to the rates of stillbirth of younger women at 41 weeks.³⁸ For this reason, as of 2022, ACOG recommends induction of labor for at 39 weeks 0 day to 39 weeks 6 days for maternal age ≥ 40 years of age.³⁹

Outcomes

The present study found the cesarean delivery rate was 75% higher in the labor induction requiring cervical preparation group compared to the non-induction group. This is in contrast to findings in other studies that report a lower risk of cesarean delivery.^3,6,10,11 One possible explanation is that, in an effort to ensure patients being augmented rather than induced were excluded, all patients included had cervical preparation. This results in exclusion of patients who were induced that did not require cervical preparation, presumably presenting with a favorable cervix by Bishop score.¹² However, some recent data has contested the notion that an unfavorable cervix increases the risk of cesarean delivery.^6,40 Rather, risk of cesarean may instead be altered by whether or not a favorable cervix is achieved with cervical preparation during the induction process.

One further consideration is that both the ARRIVE trial and many post-ARRIVE studies focus solely on nulliparous deliveries whereas the current study includes both nulliparous and multiparous pregnant people given the lack of parity-specific data. However, a 2019 investigation assessed maternal outcomes for low-risk multiparous pregnancy undergoing elective induction of labor and found a decreased frequency of cesarean delivery compared with expectant management.⁴¹ Further studies must continue to determine whether parity status may affect the rate of cesarean delivery as well as maternal morbidity and mortality in induction of labor.

A variety of characteristics were associated with cesarean including older maternal age, Black / Asian individuals, obesity, and 40-weeks’ gestation. Other studies also support that advanced maternal age is a risk factor for failed induction.⁴² However, as described above, due to the increased risk of stillbirth, there is a role for induction of labor among advanced maternal age patients, especially those ≥ 40 years of age. Furthermore, while advanced maternal age is a risk factor for failed induction compared to younger patients, a randomized control trial of only patients ≥ 35 years old found lower rates of cesarean among patients randomized to labor induction between 39 weeks 0 days and 39 weeks 6 days of gestation compared to expectantly managed patients.⁴³

Likewise, previous studies have demonstrated that obesity is a risk factor for failed induction, a finding corroborated here.^44–47 However, obese patients are also more likely to have longer gestations resulting in more late-term and post-term pregnancy requiring induction.^48–50 A recent systematic review and meta-analysis of patients with obesity reported that full-term induction of labor may reduce the risk of cesarean compared to expectant management.⁵¹ Thus, the role for induction of labor among otherwise low-risk patients with obesity and older maternal age requires further investigation.

Clinical and research implications

Overall, the findings here using population-level data challenge the broad applicability of other studies that reported a reduced risk of cesarean. Given the conflicting data, further investigation regarding the risk of cesarean is warranted. Though further research is needed to corroborate the findings here, the data suggest that induction of labor among low-risk pregnancies is associated with cesarean section and maternal morbidity and mortality. However, although maternal mortality rate in the labor induction group was statistically higher compared to the non-induction group (4.8 vs 0.7 per 100,000), the mortality rate was numerically lower than any hospital delivery (7.9 per 100,000) during the same study period, unselected cesarean delivery (15.6 per 100,000),⁵² and all-cause mortality rate for women aged 15–44 years (42–136 per 100,000).^53,54

However, while coding artifacts, unmeasured confounding variables (e.g., medical indication for labor induction), lack of temporal event data, and possible sparse bias due to extremely low event rates may be affecting the analysis for mortality outcomes, the observed association cannot be downplayed and external validation studies are necessary to evaluate maternal mortality for patients with a low-risk pregnancy who are undergoing labor induction.

Moreover, as described above, additional research is needed to understand the risks and benefits of induction for obese patients and those of advanced maternal age. Future research also need to explore the racial and regional disparities detected by the present study. Finally, given the limitations of the database, the findings regarding cervical preparation and induction method should be interpreted with caution and require further investigation.

Strength and limitations

Strengths of this study include the recent time period studied, use of national-level data, and rigorous inclusion and exclusion criteria. Prior nationwide investigation to assess the temporal trends of labor induction did not include the COVID-19 era (ended in QT1/2020).²⁹.

There were, however, several limitations in this study. First, patients that did not require cervical preparation were not captured by the study design. There are limitations that are inherent to the use of national clinical databases, studies that rely on clinical coding, and the specific exposure definitions used within this study. These are extensively detailed in the Methods section under 2.3.2 Exposure Limitations.

In addition, indication for cesarean delivery, quality metrics such as patient satisfaction, neonatal information, and cause of death were not available, limiting the interpretation of the study findings. Although identification of labor induction followed the organization’s administrative coding schema,^26,27 actual medical record review to assess the accuracy of these codes were not assessable in this study, risking possible misclassification. Lastly, generalizability of observed results in other populations was not assessed in the current study. Ultimately, due to the retrospective nature of this study, this investigation cannot establish causation and can only draw associations and correlations.

Conclusions

This nation-level analysis suggested that the COVID-19 pandemic may have impacted the utilization of full-term labor induction for low-risk pregnancy in the United States. Higher risks of cesarean and adverse outcomes related to labor induction observed in the national data warrant further investigation in prospective evaluation.

Author contributions

CRediT author contributions: Conceptualization: Carolyn N. Rocha, Koji Matsuo, Kimeshia C. Thomas; Data curation: Koji Matsuo; Formal analysis: Koji Matsuo, Carolyn N. Rocha; Funding acquisition: Koji Matsuo, Joseph G. Ouzounian; Investigation: all authors; Methodology: Koji Matsuo, Carolyn N. Rocha; Project administration: Koji Matsuo; Resources: all authors; Software: Koji Matsuo; Supervision: K.M., Elizabeth B. Sasso, Joseph G. Ouzounian; Validation: Koji Matsuo; Visualization: Koji Matsuo; Writing—original draft: Carolyn N. Rocha, Ariane .C. Youssefzadeh, Koji Matsuo; Writing—review & editing: all authors.

Funding

Ensign Endowment for Gynecologic Cancer Research (Koji Matsuo). The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Data availability

The data on which this study is based are publicly available upon request at Healthcare Cost and Utilization Project, Agency for Healthcare Research and Quality. https://www.hcup-us.ahrq.gov/nisoverview.jsp

Declarations

Competing interests

The authors declare no competing interests.

Ethical approval

University of Southern California Institutional Review Board (HS-16–00481). Patient informed consent: not required.

Transparency

The manuscript’s corresponding author (Koji Matsuo) affirms that the manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned (and, if relevant, registered) have been explained. The National Inpatient Sample is developed for the Healthcare Cost and Utilization Project that is sponsored by the Agency for Healthcare Research and Quality, and the program is the source of the de-identified data used; race and ethnicity was grouped by the program; and the program has not verified and is not responsible for the statistical validity of the data analysis or the conclusions derived by the study team.

Tweetable statement

In this nationwide cross-sectional study, increasing trends of full-term labor induction for low-risk pregnancy may have stopped during the COVID-19 pandemic in the U.S.; labor induction was associated with cesarean delivery and other measured adverse outcomes.