Limitations of Vaginal Birth After Cesarean Success Prediction

Epistemic Silences and Experiential Knowledge in Decisions After a First Cesarean: The case of a vaginal birth after cesarean calculator

Evidence-based obstetrics can employ statistical models to justify greater use of cesareans, sometimes excluding experiential elements from informed decision making. Over the past decade, prenatal providers adopted a vaginal birth after cesarean (VBAC) calculator designed to support patients in making informed decisions about their births by estimating their probability for a VBAC. Among other factors, the calculator used race and ethnicity to make its estimate, assigning lower probabilities for a successful VBAC to Black and Hispanic patients. I analyze how a diverse group of women and their providers engaged with the VBAC calculator. Some providers used low calculator scores to remove a shared decision-making model by prescriptively counseling Black and Hispanic women who desired a VBAC into undergoing repeat cesareans. Consequently, women racialized by the calculator as Black or Hispanic used experiential knowledge to challenge the calculator's assessment of their supposed lesser ability to give birth vaginally.

Validated Calculators Predicting Cesarean Delivery After Induction: Accuracy in an External Population

OBJECTIVE

To evaluate the performance of two previously published calculators in predicting cesarean delivery after induction of labor in an external population.

METHODS

This was a cohort study including all nulliparous pregnant patients with singleton, term, vertex fetuses; intact membranes; and unfavorable cervices who underwent induction of labor between 2015 and 2017 at an academic tertiary care institution. Individual predicted cesarean risk scores were calculated with two previously published calculators. For each calculator, patients were stratified into three risk groups (lower, middle, and upper thirds) of approximately equivalent size. Predicted and observed incidences of cesarean delivery were compared with two-tailed binomial tests of probability in the overall population and in each risk group.

RESULTS

A total of 846 patients met inclusion criteria, and 262 (31.0%) had cesarean deliveries, which was significantly lower than overall predicted rates of 40.0% and 36.2% with the two calculators (both P<.01). Both calculators significantly overestimated risk of cesarean delivery in higher risk tertiles (all P<.05). The areas under the receiver operating characteristic for both calculators were 0.57 or less in the overall population and in each risk group, suggesting poor predictive value. Higher predicted risk tertile in both calculators was not associated with any maternal or neonatal outcomes except wound infection.

CONCLUSION

Both previously published calculators had poor performance in this population, with neither calculator accurately predicting the incidence of cesarean delivery. Patients and health care professionals might be discouraged regarding trial of labor induction by falsely high predicted risk-of-cesarean scores. We caution against widespread implementation of these calculators without further population-specific refinement and adjustment.

VBAC calculator 2.0: Recent evidence

Following criticism for the use of race as a biological predictor of vaginal birth after cesarean (VBAC), an updated version of the Society for Maternal-Fetal Medicine (SMFM) VBAC calculator has been published. The variable "African American" or "Hispanic" (yes/no), which produced systematically lower chances of VBAC for nonwhites has been replaced with "chronic hypertension requiring treatment" (yes/no). Although there are no published external validation studies to date, developers report accuracy (area under the curve and calibration) nearly identical to the original calculator and it is published online for immediate use. This review examines the history of the calculator, measures of its validity, and recent studies measuring its performance among Hispanics, Blacks, Asians, and others with lower range scores. Underprediction of successful VBAC is evident in the original calculator, especially as predicted VBAC decreases. These studies raise a concern about the use of calculator scores in clinical management, that is, discouraging or restricting access to labor after cesarean (LAC) for parents with lower calculator scores. This raises special concern for minority populations who experience increased cesarean-related morbidity, face obstacles accessing LAC care, and who may benefit disproportionately from increased LAC uptake. Although calculator developers have discouraged using calculator scores to restrict access to LAC, such uses are documented. It is not clear what effect the removal of race will have on calculator performance, and further study is required before calculator scores are used in counseling. This includes studies that include large numbers of low scoring and minority patients.

Predicting Race And Ethnicity To Ensure Equitable Algorithms For Health Care Decision Making

Algorithms are currently used to assist in a wide array of health care decisions. Despite the general utility of these health care algorithms, there is growing recognition that they may lead to unintended racially discriminatory practices, raising concerns about the potential for algorithmic bias. An intuitive precaution against such bias is to remove race and ethnicity information as an input to health care algorithms, mimicking the idea of "race-blind" decisions. However, we argue that this approach is misguided. Knowledge, not ignorance, of race and ethnicity is necessary to combat algorithmic bias. When race and ethnicity are observed, many methodological approaches can be used to enforce equitable algorithmic performance. When race and ethnicity information is unavailable, which is often the case, imputing them can expand opportunities to not only identify and assess algorithmic bias but also combat it in both clinical and nonclinical settings. A valid imputation method, such as Bayesian Improved Surname Geocoding, can be applied to standard data collected by public and private payers and provider entities. We describe two applications in which imputation of race and ethnicity can help mitigate potential algorithmic biases: equitable disease screening algorithms using machine learning and equitable pay-for-performance incentives.

Racial/Ethnic Differences in Socioeconomic Status and Medical Correlates of Trial of Labor After Cesarean and Vaginal Birth After Cesarean

Objectives: Black and Latinx women have higher rates of trial of labor after cesarean (TOLAC) compared with White women, but lower rates of vaginal birth after cesarean (VBAC). This study examined potential racial/ethnic differences in correlates of TOLAC and VBAC. Materials and Methods: The analytic sample includes term, singleton hospital births to women with one prior cesarean in birth certificate data for 2016. We estimated associations between medical factors (diabetes, hypertension, and prepregnancy obesity) and socioeconomic status (education level and insurance type) and TOLAC and VBAC using logistic regression, stratifying by race/ethnicity and testing whether coefficients differed across models. Results: Hypertension and obesity were more strongly related to reduced chances of TOLAC among White women than among women of color. For example, having a body mass index (BMI) between 30 and 39 (vs. normal BMI) was associated with a 6.3 percentage-point (pp) lower probability of TOLAC for White women, a 5.9 pp lower probability for Black women, and 2.9 pp lower probability for Latinx women. Paying out-of-pocket for birth was associated with a 5.5 pp increase in the probability of TOLAC among White women, versus a 3.2 pp decrease among Black women. Overweight and obesity were associated with lower probability of VBAC, but the magnitude of this association was smaller for Black and Latinx women than for White women. Conclusions: More research is needed to elucidate the underlying decision-making processes that lead to these associations. Future work should focus on ensuring equity in access to VBAC-supportive providers and hospitals and fostering informed decision-making after a prior cesarean.

Predicting vaginal birth after previous cesarean: Using machine-learning models and a population-based cohort in Sweden

Introduction

Predicting a woman’s probability of vaginal birth after cesarean could facilitate the antenatal decision‐making process. Having a previous vaginal birth strongly predicts vaginal birth after cesarean. Delivery outcome in women with only a cesarean delivery is more unpredictable. Therefore, to better predict vaginal birth in women with only one prior cesarean delivery and no vaginal deliveries would greatly benefit clinical practice and fill a key evidence gap in research. Our aim was to predict vaginal birth in women with one prior cesarean and no vaginal deliveries using machine‐learning methods, and compare with a US prediction model and its further developed model for a Swedish setting.

Material and methods

A population‐based cohort study with a cohort of 3116 women with only one prior birth, a cesarean, and a subsequent trial of labor during 2008‐2014 in the Stockholm‐Gotland region, Sweden. Three machine‐learning methods (conditional inference tree, conditional random forest and lasso binary regression) were used to predict vaginal birth after cesarean among women with one previous birth. Performance of the new models was compared with two existing models developed by Grobman et al (USA) and Fagerberg et al (Sweden). Our main outcome measures were area under the receiver‐operating curve (AUROC), overall accuracy, sensitivity and specificity of prediction of vaginal birth after previous cesarean delivery.

Results

The AUROC ranged from 0.61 to 0.69 for all models, sensitivity was above 91% and specificity below 22%. The majority of women with an unplanned repeat cesarean had a predicted probability of vaginal birth after cesarean >60%.

Conclusions

Both classical regression models and machine‐learning models had a high sensitivity in predicting vaginal birth after cesarean in women without a previous vaginal delivery. The majority of women with an unplanned repeat cesarean delivery were predicted to succeed with a vaginal birth (ie specificity was low). Additional covariates combined with machine‐learning techniques did not outperform classical regression models in this study.

Calculators Estimating the Likelihood of Vaginal Birth After Cesarean: Uses and Perceptions

INTRODUCTION

A calculator estimating likelihood of vaginal birth after cesarean (VBAC) has been promoted by the Society for Maternal-Fetal Medicine, but little is known about how it is used and perceived in practice. Cutoffs for prohibiting labor after cesarean are discouraged by the calculator's developers, but such uses may be widespread. The purpose of this study was to determine how calculators predicting VBAC are used and perceived in midwifery practices.

METHODS

Certified nurse-midwives and certified midwives currently providing care for labor after cesarean were surveyed between January 17, 2019, and February 7, 2019. Quantitative and text data were collected regarding the uses and perceptions of calculators among midwives and their colleagues. We compared these findings with midwives' perceptions of their ability to accommodate patient wishes for labor after cesarean. We used descriptive content analysis to evaluate themes occurring in text responses.

RESULTS

There were 1305 valid responses. A requirement to use calculator scores for patient counseling was reported by 527 (40.4%) of responding midwives. Over 1 in 5 midwives reported that scores were used to discourage or prohibit labor after cesarean. Almost half reported some or strong disagreement with physician colleagues regarding calculator use. Interprofessional agreement and disagreement centered on how scores are used to direct clinical care or restrict patient options. Calculator scores were used in more than twice as many midwives' practices to discourage rather than encourage labor after cesarean. Descriptive analysis of text revealed 4 themes regarding calculators: inconsistent use, information counseling, informed consent, and influence patient management or options.

DISCUSSION

Calculators predicting likelihood of VBAC success are widely used in midwifery settings and are more often used to discourage than to encourage labor after cesarean. Midwives reported both directive and nondirective counseling based on calculator scores.

Factors obstetricians, family physicians and midwives consider when counselling women about a trial of labour after caesarean and planned repeat caesarean: a qualitative descriptive study

BACKGROUND

Repeat caesarean sections (CSs) are major contributors to the high rate of CS in Canada and globally. Women's decisions to have a planned repeat CS (PRCS) or a trial of labour after CS (TOLAC) are influenced by their maternity care providers. This study explored factors maternity care providers consider when counselling pregnant women with a previous CS, eligible for a TOLAC, about delivery method.

METHODS

A qualitative descriptive design was implemented. Semi-structured, one-to-one in-depth telephone interviews were conducted with 39 maternity care providers in Ontario, Canada. Participants were recruited at 2 maternity care conferences and with the use of snowball sampling. Interviews were audio recorded and transcribed verbatim. Data were uploaded into the data management software, NVIVO 10.0 and analyzed using qualitative content analysis.

RESULTS

Participants consisted of 12 obstetricians, 13 family physicians and 14 midwives. Emergent themes, reflecting the factors maternity care providers considered when counselling on mode of delivery, were organized under the categories clinical/patient factors, health system factors and provider preferences. Maternity care providers considered clinical/patient factors, including women's choice … with conditions, their assessment of women's chances of a successful TOLAC, their perception of women's risk tolerance, women's preferred delivery method, and their perception of women's beliefs and attitudes about childbirth. Additionally, providers considered health system factors which included colleague support for TOLAC and time needed to mount an emergency CS. Finally, provider factors emerged as considerations when counselling. They included provider preference for PRCS or TOLAC, provider scope of practice, financial incentives and convenience related to PRCS, past experiences with TOLAC and PRCS and providers' perspectives on risk of TOLAC.

CONCLUSION

The findings highlight the multiplicity of factors maternity care providers consider when counselling women. Effectively addressing clinical, health care system and personal factors that influence counselling may help decrease non-medically indicated PRCS.

Considering Labor After Cesarean in a Woman with Class 3 Obesity

Because of the high cesarean rate in the United States, perinatal care providers are increasingly called upon to counsel women considering labor after cesarean. This counseling can be more complicated for women with class 3 obesity, defined as a body mass index at or exceeding 40. Although labor after cesarean may be less likely to result in successful vaginal birth after cesarean (VBAC) for this population, the risks of repeat cesarean are also higher for these women. This case report describes the benefits and risks of labor after cesarean, the chance of successful VBAC, and clinical recommendations to aid women in achieving VBAC when class 3 obesity is present.

Hospital variation in utilization and success of trial of labor after a prior cesarean

BACKGROUND

Trial of labor after cesarean delivery is an effective and safe option for women without contraindications.

OBJECTIVES

The objective of the study was to examine hospital variation in utilization and success of trial of labor after cesarean delivery and identify associated institutional characteristics and patient outcomes.

STUDY DESIGN

Using linked maternal and newborn hospital discharge records and birth certificate data in 2010-2012 from the state of California, we identified 146,185 term singleton mothers with 1 prior cesarean delivery and no congenital anomalies or clear contraindications for trial of labor at 249 hospitals. Risk-standardized utilization and success rates of trial of labor after cesarean delivery were estimated for each hospital after accounting for differences in patient case mix. Risk for severe maternal and newborn morbidities, as well as maternal and newborn length of stay, were compared between hospitals with high utilization and high success rates of trial of labor after cesarean delivery and other hospitals. Bivariate analysis was also conducted to examine the association of various institutional characteristics with hospitals' utilization and success rates of trial of labor after cesarean delivery.

RESULTS

In the overall sample, 12.5% of women delivered vaginally. After adjusting for patient clinical risk factors, utilization and success rates of trial of labor after cesarean delivery varied considerably across hospitals, with a median of 35.2% (10th to 90th percentile range: 10.2-67.1%) and 40.5% (10th to 90th percentile range: 8.5-81.1%), respectively. Risk-standardized utilization and success rates of trial of labor after cesarean delivery demonstrated an inverted U-shaped relationship such that low or excessively high use of trial of labor after cesarean delivery was associated with lower success rate. Compared with other births, those delivered at hospitals with above-the-median utilization and success rates of trial of labor after cesarean delivery had a higher risk for uterine rupture (adjusted risk ratio, 2.74, P < .001), severe newborn respiratory complications (adjusted risk ratio, 1.46, P < .001), and severe newborn neurological complications/trauma (adjusted risk ratio, 2.48, P < .001), but they had a lower risk for severe newborn infection (adjusted risk ratio, 0.80, P = .003) and overall severe unexpected newborn complications (adjusted risk ratio, 0.86, P < .001) as well as shorter length of stays (adjusted mean ratio, 0.948 for mothers and 0.924 for newborns, P < .001 for both). Teaching status, system affiliation, larger volume, higher neonatal care capacity, anesthesia availability, higher proportion of midwife-attended births, and lower proportion of Medicaid or uninsured patients were positively associated with both utilization and success of trial of labor after cesarean delivery. However, rural location and higher local malpractice insurance premium were negatively associated with the utilization of trial of labor after cesarean delivery, whereas for-profit ownership was associated with lower success rate.

CONCLUSION

Utilization and success rates of trial of labor after cesarean delivery varied considerably across hospitals. Strategies to promote vaginal birth should be tailored to hospital needs and characteristics (eg, increase availability of trial of labor after cesarean delivery at hospitals with low utilization rates while being more selective at hospitals with high utilization rates, and targeted support for lower capacity hospitals).