Interpretable machine learning predicts postpartum hemorrhage with severe maternal morbidity in a lower-risk laboring obstetric population.

BACKGROUND: Early identification of patients at increased risk for postpartum hemorrhage (PPH) associated with severe maternal morbidity (SMM) is critical for preparation and preventative intervention. However, prediction is challenging in patients without obvious risk factors for postpartum hemorrhage with severe maternal morbidity. Current tools for hemorrhage risk assessment use lists of risk factors rather than predictive models. OBJECTIVE: To develop, validate (internally and externally), and compare a machine learning model for predicting PPH associated with SMM against a standard hemorrhage risk assessment tool in a lower risk laboring obstetric population. STUDY DESIGN: This retrospective cross-sectional study included clinical data from singleton, term births (>=37 weeks' gestation) at 19 US hospitals (2016-2021) using data from 58,023 births at 11 hospitals to train a generalized additive model (GAM) and 27,743 births at 8 held-out hospitals to externally validate the model. The outcome of interest was PPH with severe maternal morbidity (blood transfusion, hysterectomy, vascular embolization, intrauterine balloon tamponade, uterine artery ligation suture, uterine compression suture, or admission to intensive care). Cesarean birth without a trial of vaginal birth and patients with a history of cesarean were excluded. We compared the model performance to that of the California Maternal Quality Care Collaborative (CMQCC) Obstetric Hemorrhage Risk Factor Assessment Screen. RESULTS: The GAM predicted PPH with an area under the receiver-operating characteristic curve (AUROC) of 0.67 (95% CI 0.64-0.68) on external validation, significantly outperforming the CMQCC risk screen AUROC of 0.52 (95% CI 0.50-0.53). Additionally, the GAM had better sensitivity of 36.9% (95% CI 33.01-41.02) than the CMQCC screen sensitivity of 20.30% (95% CI 17.40-22.52) at the CMQCC screen positive rate of 16.8%. The GAM identified in-vitro fertilization as a risk factor (adjusted OR 1.5; 95% CI 1.2-1.8) and nulliparous births as the highest PPH risk factor (adjusted OR 1.5; 95% CI 1.4-1.6). CONCLUSION: Our model identified almost twice as many cases of PPH as the CMQCC rules-based approach for the same screen positive rate and identified in-vitro fertilization and first-time births as risk factors for PPH. Adopting predictive models over traditional screens can enhance PPH prediction.

Effects of the ARRIVE (A Randomized Trial of Induction Versus Expectant Management) Trial on Elective Induction and Obstetric Outcomes in Term Nulliparous Patients.

OBJECTIVE: To evaluate the effect of publication of the ARRIVE (A Randomized Trial of Induction Versus Expectant Management) trial on perinatal outcomes in singleton, term, nulliparous patients. METHODS: An interrupted time series analysis was performed using clinical data for nulliparous singleton births at 39 weeks of gestation or later at 13 hospitals in the Northwest region of the United States (January 2016-December 2020). A modified Poisson regression was used to model time trends and changes after the ARRIVE trial (August 9, 2018). Outcomes of interest were elective induction, unplanned cesarean births, hypertensive disorders of pregnancy, a composite of perinatal adverse outcomes, and neonatal intensive care unit admissions. RESULTS: The analysis included 28,256 births (15,208 pre-ARRIVE and 13,048 post-ARRIVE). The rate of elective labor induction was 3.6% during the pre-ARRIVE period (January 2016-July 2018) and 10.8% post-ARRIVE (August 2018-December 2020). In the interrupted time series analysis, elective induction increased by 42% (relative risk [RR] 1.42; 95% CI 1.18-1.71) immediately after the ARRIVE trial publication. Thereafter, the trend was unchanged compared with the pre-ARRIVE period. There was no statistically significant change in cesarean birth (RR 0.96; 95% CI 0.89-1.04) or hypertensive disorders of pregnancy (RR 0.91; 95% CI 0.79-1.06) immediately after the trial, and no change in trend. After the ARRIVE trial, there was no immediate change in adverse perinatal outcomes, but a statistically significant increase in trend of adverse perinatal events (1.03; 95% CI 1.01-1.05) when compared with a declining trend observed in the pre-ARRIVE period. CONCLUSION: Publication of the ARRIVE trial was associated with an increase in elective induction, and no change in cesarean birth or hypertensive disorders of pregnancy in singleton nulliparous patients giving birth at 39 weeks or later. There was a flattening of the pre-ARRIVE decreasing trend in perinatal adverse events.

Exclusive Breastfeeding in the Northwest: Disparities Related to Race/Ethnicity and Substance Use.

BACKGROUND AND OBJECTIVES: Mothers who are Black, Indigenous, and people of color (BIPOC) are disproportionately impacted by substance use in pregnancy and less likely to breastfeed. Our objectives were to assess relationships between substance use in pregnancy and exclusive breastfeeding at discharge (EBF) and race/ethnicity and EBF, and determine the extent to which substance use influences the relationship between race/ethnicity and EBF. METHODS: This is a retrospective cohort study of term mother-infant dyads using 2016 to 2019 data from a Northwest quality improvement collaborative, Obstetrical Care Outcomes Assessment Program. Stepwise and stratified multivariable logistic regression analyses were conducted to determine associations between independent variables consisting of characteristics, including maternal race/ethnicity and substance use, and the dependent variable, EBF. RESULTS: Our sample consisted of 84,742 dyads, 69.5% of whom had EBF. The adjusted odds of EBF for non-Hispanic Black and Hispanic mothers were half, and for American Indian/Alaska Native mothers two-thirds, that of White mothers (aOR [95% CI]: 0.52 [0.48, 0.57], 0.51 [0.48, 0.54], 0.64 [0.55, 0.76], respectively). Substance use did not mediate the association between race/ethnicity and EBF, but it modified the association. Among those reporting nicotine or marijuana use, Hispanic mothers were half as likely as White mothers were to exclusively breastfeed. Other factors associated with a lower likelihood of EBF included public or no insurance, rural setting, C-section, NICU admission, and LBW. CONCLUSIONS: Disparities in EBF related to race/ethnicity and substance use were pronounced in this study, particularly among Hispanic mothers with nicotine or marijuana use.

Customized GROW vs INTERGROWTH-21st birthweight standards to identify small for gestational age associated perinatal outcomes at term.

BACKGROUND: Fetal growth restriction is associated with stillbirth and other adverse pregnancy outcomes, and the use of the correct weight standard is an essential proxy indicator of growth status and perinatal risk. OBJECTIVE: This study aimed to assess the performance of two international birthweight standards for their ability to identify perinatal morbidity and mortality indicators associated with small for gestational age infants at term. STUDY DESIGN: This retrospective cohort study used data from a multicenter perinatal quality initiative, including a multiethnic dataset of 125,826 births from 2012 to 2017. Of the singleton term births, 92,622 had complete outcome data including stillbirth, neonatal death, 5-minute Apgar score <7, neonatal glucose instability and need for newborn transfer to a higher level of care or neonatal intensive care unit admission. The customized GROW and INTERGROWTH-21st birthweight standards were applied to determine small for gestational age (<10th percentile) according to their respective methods and formulae. The associations with adverse outcomes were expressed as relative risks with 95% confidence intervals and population attributable fractions. RESULTS: GROW and INTERGROWTH-21st classified 9578 (10.3%) and 4079 (4.4%) pregnancies as small for gestational age, respectively. For all of the outcomes assessed, GROW identified more small for gestational age infants with adverse outcomes than INTERGROWTH-21st, including more stillbirths, perinatal deaths, low Apgar scores, glucose instability, newborn seizure, and transfers to a higher level of care. Moreover, 13 of 27 stillbirths (48%) that were small for gestational age by either method were identified as small for gestational age by GROW but not by INTERGROWTH-21st. Similarly, additional cases of all other adverse outcome indicators were identified by GROW as small for gestational age, whereas INTERGROWTH-21st identified in only 1 category (glucose instability) 9 of 295 cases (3.1%), which were not identified as small for gestational age by GROW. CONCLUSION: Customized assessment using GROW resulted in increased identification of small for gestational age term infants that were at significantly increased risk of an array of adverse pregnancy outcomes.

A Simple Approach to Adjust for Case-Mix When Comparing Institutional Cesarean Birth Rates.

OBJECTIVE: This study aimed to develop a validated model to predict intrapartum cesarean in nulliparous women and to use it to adjust for case-mix when comparing institutional laboring cesarean birth (CB) rates. STUDY DESIGN: This multicenter retrospective study used chart-abstracted data on nulliparous, singleton, term births over a 7-year period. Prelabor cesareans were excluded. Logistic regression was used to predict the probability of CB for individual pregnancies. Thirty-five potential predictive variables were evaluated including maternal demographics, prepregnancy health, pregnancy characteristics, and newborn weight and gender. Models were trained on 21,017 births during 2011 to 2015 (training cohort), and accuracy assessed by prediction on 15,045 births during 2016 to 2017 (test cohort). RESULTS: Six variables delivered predictive success equivalent to the full set of 35 variables: maternal weight, height, and age, gestation at birth, medically-indicated induction, and birth weight. Internal validation within the training cohort gave a receiver operator curve with area under the curve (ROC-AUC) of 0.722. External validation using the test cohort gave ROC-AUC of 0.722 (0.713-0.731 confidence interval). When comparing observed and predicted CB rates at 16 institutions in the test cohort, five had significantly lower than predicted rates and three had significantly higher than predicted rates. CONCLUSION: Six routine clinical variables used to adjust for case-mix can identify outliers when comparing institutional CB rates.

Comparison of Midwifery and Obstetric Care in Low-Risk Hospital Births.

OBJECTIVE: To compare midwife and obstetrician labor practices and birth outcomes in women with low-risk pregnancies delivered in the hospital. METHODS: We conducted a retrospective cohort study of singleton births of 37 0/7-42 6/7 weeks of gestation at 11 hospitals between January 1, 2014, and December 31, 2018. Exclusions included intrapartum transfer from home-birth center, antepartum stillbirth, previous cesarean delivery, practitioner other than midwife or obstetrician, prelabor cesarean, prepregnancy maternal disease, and pregnancy complications or risk factors. Interventions (induction, artificial rupture of membranes, epidural, oxytocin, and episiotomy), mode of delivery, maternal outcomes (third- or fourth-degree laceration, postpartum hemorrhage, blood transfusion, and severe maternal morbidity), and newborn outcomes (shoulder dystocia, 5-minute Apgar score less than 7, resuscitation at delivery, birth trauma, and neonatal intensive care unit admission) were examined by practitioner type. We used modified Poisson regression models adjusted for individual confounders to assess risk ratios, stratified by parity, for health care provider type and perinatal outcomes. RESULTS: The study cohort comprised 23,100 births (3,816 midwife and 19,284 obstetrician). Compared with obstetricians, midwifery patients had significantly lower intervention rates, an approximately 30% lower risk of cesarean delivery in nulliparous patients (adjusted relative risk [aRR] 0.68; 95th% CI 0.57-0.82), and an approximately 40% lower risk of cesarean in multiparous patients (aRR 0.57; 95th% CI 0.36-0.89). Operative vaginal birth was also less common in nulliparous patients (aRR 0.73; 95th% CI 0.57-0.93) and multiparous patients (aRR 0.30; 95th% CI 0.14-0.63). Shoulder dystocia was more common in multiparous patients receiving midwifery care (aRR 1.42; 95th% CI 1.04-1.92). CONCLUSIONS: In low-risk pregnancies, midwifery care in labor was associated with decreased intervention, decreased cesarean and operative vaginal births, and, in multiparous women, an increased risk for shoulder dystocia. Greater integration of midwifery care into maternity services in the United States may reduce intervention in labor and potentially even cesarean delivery, in low-risk pregnancies. Larger research studies are needed to evaluate uncommon but important maternal and newborn outcomes.

Maternal and newborn outcomes with elective induction of labor at term.

BACKGROUND: A growing body of evidence supports improved or not worsened birth outcomes with nonmedically indicated induction of labor at 39 weeks gestation compared with expectant management. This evidence includes 2 recent randomized control trials. However, concern has been raised as to whether these studies are applicable to a broader US pregnant population. OBJECTIVE: Our goal was to compare outcomes for electively induced births at ≥39 weeks gestation with those that were not electively induced. STUDY DESIGN: We conducted a retrospective cohort study using chart-abstracted data on births from January 1, 2012, to December 31, 2017, at 21 hospitals in the Northwest United States. The study was restricted to singleton cephalic hospital births at 39+0-42+6 weeks gestation. Exclusions included previous cesarean birth, missing data for delivery type or gestational week at birth, antepartum stillbirth, cesarean birth without any attempt at vaginal birth, fetal anomaly, gestational diabetes mellitus, prepregnancy diabetes mellitus, and prepregnancy hypertension. The rate of cesarean birth for elective inductions at both 39 and 40 weeks gestation was compared with the rate in all other on-going pregnancies in the same gestational week. Maternal outcomes (operative vaginal birth, shoulder dystocia, 3rd- or 4th-degree perineal laceration, pregnancy-related hypertension, and postpartum hemorrhage) and newborn infant outcomes (macrosomia, 5-minute Apgar <7, resuscitation at delivery, intubation, respiratory complications, and neonatal intensive care unit admission) were also compared between elective inductions and on-going pregnancies at 39 and 40 weeks gestation. Logistic regression modeling was used to produce odds ratios for outcomes with adjustment for maternal age and body mass index. Results were stratified by parity and gestational week at birth. Duration of hospital stay (admission to delivery, delivery to discharge, and total stay) were compared between elective inductions and on-going pregnancies. RESULTS: A total of 55,694 births were included in the study cohort: 4002 elective inductions at ≥39+0 weeks gestation and 51,692 births at 39+0-42+6 weeks gestation that were not electively induced. In nulliparous women, elective induction at 39 weeks gestation was associated with a decreased likelihood of cesarean birth (14.7% vs 23.2%; adjusted odds ratio, 0.61; 95% confidence interval, 0.41-0.89) and an increased rate of operative vaginal birth (18.5% vs 10.8%; adjusted odds ratio, 1.8; 95% confidence interval, 1.28-2.54) compared with on-going pregnancies. In multiparous women, cesarean birth rates were similar in the elective inductions and on-going pregnancies. Elective induction at 39 weeks gestation was associated with a decreased likelihood of pregnancy-related hypertension in nulliparous (2.2% vs 7.3%; adjusted odds ratio, 0.28; 95% confidence interval, 0.11-0.68) and multiparous women (0.9% vs 3.5%; adjusted odds ratio, 0.24; 95% confidence interval, 0.15-0.38). Term elective induction was not associated with any statistically significant increase in adverse newborn infant outcomes. Elective induction of labor at 39 weeks gestation was associated with increased time from admission to delivery for both nulliparous (1.3 hours; 95% confidence interval, 0.2-2.3) and multiparous women (3.4 hours; 95% confidence interval, 3.2-3.6). CONCLUSION: Elective induction of labor at 39 weeks gestation is associated with a decrease in cesarean birth in nulliparous women, decreased pregnancy-related hypertension in multiparous and nulliparous women, and increased time in labor and delivery. How to use this information remains the challenge.

Cervical Dilation on Admission in Term Spontaneous Labor and Maternal and Newborn Outcomes.

OBJECTIVE: To examine associations between cervical dilation on admission and maternal and newborn outcomes in term spontaneous labor. METHODS: This is a retrospective cohort study of 11,368 singleton, term (37-43 6/7 weeks of gestation) spontaneously laboring women delivering in 14 hospitals in Washington State between 2012 and 2014 using chart abstracted data from the Obstetrics Clinical Outcomes Assessment Program. Women with prior cesarean delivery or ruptured membranes on admission were excluded. Pregnancy history, cervical dilation on admission, and outcomes were analyzed. Associations between early (less than 4 cm cervical dilation) and late (4 cm or greater cervical dilation) admission and outcomes were examined using general linear models with a log-link stratifying by parity. Results were reported as adjusted relative risks (RRs) with 95% confidence intervals (CIs). RESULTS: Early admission compared with late admission was associated with increased epidural use of 84.8% compared with 71.8% in nulliparous women and 66.3% compared with 53.1% in multiparous women (nulliparous RR 1.18, 95% CI 1.13-1.24; multiparous RR 1.25, 95% CI 1.18-1.32); oxytocin augmentation in 58.5% compared with 36.6% in nulliparous women and 45.9% compared with 20.7% in multiparous women (nulliparous RR 1.56, 95% CI 1.50-1.63; multiparous RR 2.14, 95% CI 1.87-2.44); and cesarean delivery of 21.8% compared with 14.5% in nulliparous women and 3.7% compared with 1.9% in multiparous women (nulliparous RR 1.50, 95% CI 1.32-1.70; multiparous women RR 1.95, 95% CI 1.47-2.57). Early admission was associated with increased neonatal intensive care unit admission for newborns of nulliparous women only (RR 1.38, 95% CI 1.01-1.89). Between 2012 and 2014, late admission increased 14.6% for nulliparous patients and 10.1% for multiparous patients, and the cesarean delivery rate decreased from 10.5% to 7.9% (P<.001) for all. CONCLUSION: Early admission (less than 4 cm cervical dilation) is a risk factor for increased medical intervention and cesarean delivery.