Materials and Methods

The Mode of Delivery Preferences Among Diverse Populations of Women study was conducted at the University of California, San Francisco (UCSF), between 2008 and 2014. This report describes the primary outcomes of this study. Methods have been described elsewhere. In brief, women receiving prenatal care at UCSF were sent letters describing the study. Women who opted in or who did not return the postage-paid response card were contacted to assess eligibility and interest.

Inclusion criteria included being 24-40 weeks’ pregnant with a singleton or twin gestation and having the ability to complete an English-language interview. Exclusion criteria included the inability to complete a face-to-face interview, having triplets or higher-order multiple gestation, or an inability to speak English. Women were not excluded on the basis of medical comorbidities or the number of prior cesarean deliveries. Participation consisted of 1 face-to-face interview that included a sociodemographic and attitudinal questionnaire and a series of preference elicitation exercises specific to the participant’s clinical context. All participants received a $40 gift card.

This study was approved by the UCSF Committee on Human Research. All participants provided written consent.

We assessed 3 types of mode-of-delivery preferences using ELICIT, a computerized preference elicitation tool previously developed by our group. The first type was a stated preference, for which we simply asked the participant, “If you could choose, which type of delivery would you want to have?” with response options of definitely a vaginal birth, probably a vaginal birth, probably a cesarean delivery, and definitely a cesarean delivery.

The second type focused on the strength of the stated preference (focusing on those preferring vaginal delivery), which we assessed using the standard gamble metric (see explanation in the following text).

Finally, because planned vaginal and planned cesarean delivery can involve various interventions and outcomes, such as induction of labor, use of antibiotics for chorioamnionitis, operative vaginal delivery, and surgical complications, the third type of preference we measured focused on the value women attach to experiencing these interventions and outcomes. These utilities were measured using the time tradeoff metric (see explanation in the following text).

To measure the strength of preference for vaginal delivery among women who preferred this delivery mode, we opted to use a standard gamble exercise that reflected a simplified choice a woman might realistically face in thinking through whether to opt for a planned vaginal vs a planned cesarean delivery. To gain an understanding of how all the participants felt about many of other the potential outcomes of these 2 delivery approaches, regardless of their preferred delivery mode, we used the time tradeoff. This is a metric that was developed specifically for use in health care evaluations, and we and others have found to be more easily comprehended by study participants, particularly when a large number of scenarios, without a clear worst case, must be assessed.

Participants were assigned to 1 of 4 preference elicitation protocols, depending on their clinical context. After identifying their preferred delivery mode, women preferring vaginal delivery were presented with a choice between certainty of an uncomplicated planned cesarean delivery (the intermediately ranked outcome) or a gamble between a specified probability of an uneventful, spontaneous vaginal birth with no adverse outcomes (the ideal outcome) and the complementary probability of having the attempted vaginal birth end in a cesarean delivery (the undesired outcome for women preferring vaginal delivery). For this assessment, the ideal, intermediate, and undesired outcomes were assigned and standardized for all women identifying vaginal delivery as their preferred delivery mode. The probability of the gamble was varied until the woman was indifferent between the 2 choices (ie, until the participant found the gamble between the ideal and undesired outcomes to be equivalent to certainty of the intermediate outcome). The strength-of-preference score was calculated at this indifference point.

In theory, a woman strongly preferring vaginal delivery would choose the gamble, even with a high chance of ending in a cesarean delivery, whereas a woman with a weaker preference would choose the gamble only if the likelihood of a cesarean delivery was relatively low. For example, if a woman had a stated preference for vaginal delivery but indicated she would opt for a planned cesarean if the chance labor would end in a cesarean was only 10%, her strength-of-preference score would be 0.10, indicating a weak preference for vaginal delivery. But if she would only opt for a planned cesarean delivery if her chance of cesarean delivery was as high as 90%, her strength-of-preference score for vaginal delivery would be 0.90, indicating a strong preference for vaginal birth.

All participants then completed a series of time tradeoff exercises to generate utilities for potential labor interventions and potential outcomes of planned delivery approaches specific to their clinical context. A similar scale was used, regardless of preferred delivery mode. The narratives for each clinical context were developed after a qualitative analysis of 6 focus groups; these narratives then underwent an iterative process of feedback and revision based on participant responses and expert opinion to develop the complete set of scenarios.

For each clinical scenario, the participant was presented with a choice between living her full life expectancy experiencing each outcome vs living a shorter life expectancy with her preferred delivery mode without interventions or complications. The life expectancy was then varied until the participant reached her indifference point. In this case a score of 0 resulted if the respondent indicated she would give up her entire lifetime to avoid the scenario being assessed, whereas a score of 1 resulted if she indicated she would not give up any time (meaning it is nearly equivalent to her ideal outcome; outcome descriptions by protocol presented in the Appendix ; Supplementary Tables 1-5 ). Lower scores indicate a greater number of years women would be willing to give up from her projected life span to achieve their desired outcome; for example, for a woman with a life span of 80 years, a score of 0.925 (equals [80-6]/80) indicates she would give up 6 years of life to avoid the undesirable outcome.

Participants in the prior cesarean and breech presentation contexts additionally evaluated a range of maternal and neonatal morbidity outcomes related to uterine rupture and head entrapment, respectively, because these are central components of counseling for women considering these options.

The primary analyses consisted of determining the percentage of women with stated preferences for 1 delivery mode over the other in each clinical context, calculating the mean strength-of-preference score for spontaneous, uncomplicated vaginal delivery among those who had a stated preference for vaginal delivery in each clinical context, and calculating the mean time tradeoff utilities for the potential interventions and outcomes of planned vaginal vs planned cesarean for the entire sample of women in each of the 4 protocol groups, using SAS version 9.3 (SAS Institute, Cary, NC).

Results

In this population of 240 women, the mean age was 33.0 years (SD ± 6.0 years), and approximately half (45%) were nulliparous ( Table 1 ). Approximately 23% had a prior cesarean delivery (38 with 1 prior, 10 with 2 prior, and 2 with 3 prior), and 36% had a prior vaginal delivery. The population was racially/ethnically diverse (56% white, 21% African American, 14% Asian/Asian American, and 9% Latina) and well educated (65% college graduates). Overall, 90.8% of the participants had a stated preference for vaginal delivery, which varied from 83% of women with a prior cesarean delivery to 93% of the women without a medical indication for a cesarean delivery ( Table 2 ).

On average, these women strongly preferred vaginal delivery, as indicated by the strength-of-preference scores ranging from a mean of 0.59 for women with a prior cesarean (suggesting they would accept a 59% chance of their planned vaginal birth after cesarean ending in cesarean delivery before opting for a planned repeat cesarean delivery) to 0.75 for the women without an indication for cesarean delivery (suggesting they would accept a 75% chance of their attempted vaginal delivery ending in a cesarean delivery before choosing a planned cesarean delivery).

We also found that, regardless of clinical context, participants assigned high preference scores (utilities) to having a planned vaginal delivery without interventions that concluded with a vaginal birth without complications, ranging from a mean of 0.993 for women carrying twins to 0.998 for women with breech presentations or a prior cesarean delivery ( Table 3 ). We observed more variation in the scores assigned to a planned cesarean delivery ending with a cesarean birth and no complications, ranging from 0.933 among participants carrying twins to 0.974 among participants in the prior cesarean group.

Planned vaginal delivery ending in cesarean had higher mean utilities than operative vaginal delivery in all contexts except for breech presentation, ranging from 0.934 to 0.973 for planned vaginal delivery ending in cesarean delivery, and from 0.906 to 0.959 for operative vaginal delivery. Among women carrying twins, experiencing a combined delivery (vaginal delivery for the presenting twin followed by a cesarean delivery for the second twin) received a lower mean utility (0.918) than scenarios with uncomplicated cesarean delivery for both twins, whereas experiencing a vertex vaginal delivery of the presenting twin followed by breech extraction of the second twin received a utility (0.936) similar to planned or unplanned uncomplicated cesarean delivery of both twins.

Our participants reported decrements in utility associated with labor interventions and short-term complications. For example, women with a prior cesarean or no indication for cesarean delivery assigned scores of 0.986 and 0.975, respectively, to induction at 39 weeks ending in vaginal delivery and slightly lower scores of 0.979 and 0.958 to augmentation of spontaneous labor with oxytocin resulting in a vaginal birth.

Planned vaginal delivery requiring antibiotic administration for chorioamnionitis and ending with vaginal delivery was valued similarly to planned vaginal delivery with induction at 41 weeks that ended in a cesarean delivery. And vaginal deliveries that resulted in major perineal lacerations received lower utilities for both groups (0.944 for prior cesarean; 0.952 for absent medical indication) than either planned or unplanned uncomplicated cesarean delivery.

As expected, scenarios including serious maternal and/or neonatal complications were assigned substantially lower scores ( Table 4 ). In the prior cesarean group, uterine rupture resulting in severe infant complications or infant death received the lowest utility scores (from 0.506 to 0.598), whereas deliveries ending in maternal complications but no infant complications received higher utilities. Among women with breech presentations, a vaginal breech delivery with head entrapment without long-term sequelae received a mean utility of 0.952, whereas head entrapment scenarios requiring further interventions and/or resulting in long-term sequelae had substantially lower scores. Planned vaginal breech delivery with head entrapment resulting in neonatal death received the lowest mean score (0.432), indicating participants would give up more than half of their life expectancy to avoid this devastating outcome.

Results

In this population of 240 women, the mean age was 33.0 years (SD ± 6.0 years), and approximately half (45%) were nulliparous ( Table 1 ). Approximately 23% had a prior cesarean delivery (38 with 1 prior, 10 with 2 prior, and 2 with 3 prior), and 36% had a prior vaginal delivery. The population was racially/ethnically diverse (56% white, 21% African American, 14% Asian/Asian American, and 9% Latina) and well educated (65% college graduates). Overall, 90.8% of the participants had a stated preference for vaginal delivery, which varied from 83% of women with a prior cesarean delivery to 93% of the women without a medical indication for a cesarean delivery ( Table 2 ).

On average, these women strongly preferred vaginal delivery, as indicated by the strength-of-preference scores ranging from a mean of 0.59 for women with a prior cesarean (suggesting they would accept a 59% chance of their planned vaginal birth after cesarean ending in cesarean delivery before opting for a planned repeat cesarean delivery) to 0.75 for the women without an indication for cesarean delivery (suggesting they would accept a 75% chance of their attempted vaginal delivery ending in a cesarean delivery before choosing a planned cesarean delivery).

We also found that, regardless of clinical context, participants assigned high preference scores (utilities) to having a planned vaginal delivery without interventions that concluded with a vaginal birth without complications, ranging from a mean of 0.993 for women carrying twins to 0.998 for women with breech presentations or a prior cesarean delivery ( Table 3 ). We observed more variation in the scores assigned to a planned cesarean delivery ending with a cesarean birth and no complications, ranging from 0.933 among participants carrying twins to 0.974 among participants in the prior cesarean group.

Planned vaginal delivery ending in cesarean had higher mean utilities than operative vaginal delivery in all contexts except for breech presentation, ranging from 0.934 to 0.973 for planned vaginal delivery ending in cesarean delivery, and from 0.906 to 0.959 for operative vaginal delivery. Among women carrying twins, experiencing a combined delivery (vaginal delivery for the presenting twin followed by a cesarean delivery for the second twin) received a lower mean utility (0.918) than scenarios with uncomplicated cesarean delivery for both twins, whereas experiencing a vertex vaginal delivery of the presenting twin followed by breech extraction of the second twin received a utility (0.936) similar to planned or unplanned uncomplicated cesarean delivery of both twins.

Our participants reported decrements in utility associated with labor interventions and short-term complications. For example, women with a prior cesarean or no indication for cesarean delivery assigned scores of 0.986 and 0.975, respectively, to induction at 39 weeks ending in vaginal delivery and slightly lower scores of 0.979 and 0.958 to augmentation of spontaneous labor with oxytocin resulting in a vaginal birth.

Planned vaginal delivery requiring antibiotic administration for chorioamnionitis and ending with vaginal delivery was valued similarly to planned vaginal delivery with induction at 41 weeks that ended in a cesarean delivery. And vaginal deliveries that resulted in major perineal lacerations received lower utilities for both groups (0.944 for prior cesarean; 0.952 for absent medical indication) than either planned or unplanned uncomplicated cesarean delivery.

As expected, scenarios including serious maternal and/or neonatal complications were assigned substantially lower scores ( Table 4 ). In the prior cesarean group, uterine rupture resulting in severe infant complications or infant death received the lowest utility scores (from 0.506 to 0.598), whereas deliveries ending in maternal complications but no infant complications received higher utilities. Among women with breech presentations, a vaginal breech delivery with head entrapment without long-term sequelae received a mean utility of 0.952, whereas head entrapment scenarios requiring further interventions and/or resulting in long-term sequelae had substantially lower scores. Planned vaginal breech delivery with head entrapment resulting in neonatal death received the lowest mean score (0.432), indicating participants would give up more than half of their life expectancy to avoid this devastating outcome.

Comment

We measured preferences for planned vaginal vs planned cesarean delivery in the context of prior cesarean, twins, breech presentation, and absent medical indication for cesarean, finding that in all of them, the majority of women highly value vaginal delivery. We also identified substantial variation in preferences for interventions that are often used during labor to help achieve a vaginal delivery. Women in our study assigned relatively low utilities to common labor interventions as well as less frequent minor and major morbidities of planned vaginal delivery. Furthermore, the variation in and decrements for utilities for common interventions and morbidities were similar across clinical contexts, demonstrating similar preferences, regardless of clinical history.

These findings highlight the importance of improved communication between patients and providers regarding processes that may occur during labor and delivery to help create realistic expectations and allow the opportunity for elicitation of informed patient preferences. The results of our 3-level approach to preference assessment emphasize the need to ensure women have meaningful opportunities not only to attempt vaginal delivery in various contexts but also to receive patient-centered information and participate in shared decision making about labor and delivery.

Our finding that women would accept a 59-75% chance of their attempted vaginal birth ending in a cesarean delivery before opting for a planned cesarean delivery underscores the intensity of the preference for vaginal delivery in this population. However, the utilities for labor processes and outcomes reveal this preference to be nuanced.

Although in the context of natural labor the mean strength-of-preference scores for spontaneous, uncomplicated vaginal delivery suggest a high risk tolerance for unplanned cesarean, the full range of utilities assessed suggests women are less accepting of labor requiring interventions, even when those interventions are consistent with the goal of vaginal birth. Moreover, these utilities suggested participants viewed some routine labor interventions as equivalent to uncomplicated cesarean deliveries, an unexpected finding, given the strength of preference for vaginal delivery. For example, the finding that an uncomplicated operative vaginal delivery received similar utilities to vaginal delivery with major perineal laceration and lower utilities than cesarean delivery suggests participants may have a different understanding than clinicians would anticipate of the risks and benefits of an uncomplicated operative vaginal delivery as compared with other outcomes. Alternatively, it may mean that although they would like to have a vaginal delivery, the prospect of a major laceration is more worrisome to them than the prospect of a cesarean delivery. These data emphasize the importance of patient education and counseling to improve concordance between patient and provider expectations regarding indications for and potential benefits of labor interventions as well as the potential morbidity of cesarean delivery.

Appropriate patient education that enables women to place labor interventions and potential maternal and neonatal outcomes in context is critical to ensuring informed patient choices. In counseling women about labor, providers must explain that many interventions facilitate the shared goal of achieving vaginal delivery.

Our work suggests education about goals for birth should extend beyond simply asking what type of birth a woman hopes to achieve but also should include a discussion of how common labor interventions may be utilized to achieve that goal. Such framing also allows providers the opportunity to discuss the rarity of major maternal or neonatal complications, which women in our study strongly desired to avoid.

Although the women in our study reported that they would give up nearly half of their remaining life span to avoid outcomes such as neonatal death, such events are exceedingly rare. The absolute risk of term neonatal hypoxic ischemic encephalopathy is 0.46 per 1000 women undergoing trial of labor after cesarean ; for women with breech presentations, long-term data show planned cesarean is not associated with a reduction in the risk of death or neurodevelopmental delay compared with planned vaginal delivery. Similarly, a recent landmark trial of women with a cephalic presenting twin demonstrated no differences in composite perinatal morbidity and mortality between women randomized to planned vaginal vs planned cesarean delivery.

Women in these clinical scenarios are commonly counseled about rare adverse outcomes, and, although women and providers are strongly motivated to avoid these unlikely but devastating events, the appropriate presentation of the likelihood of these outcomes is crucial to allow women to place them in context with other tradeoffs inherent to delivery decisions. Our findings underscore the need for research investigating optimal counseling techniques that promote fully informed decisions while avoiding overweighting low probability events.

In addition to improving the quality of obstetric care, enhanced patient education and greater focus on patient preferences could potentially contribute to reducing the cesarean rate. Despite expressed preferences for vaginal birth, the vast majority of US women with prior cesarean(s), twins, and breech do not undergo planned vaginal delivery. Provider and system factors contribute to delivery decisions and are potential points of intervention to reduce the cesarean rate. Metz et al, for example, have suggested provider factors are a potentially significant reason for disparities in rates of trial of labor after cesarean among women who are good candidates. Similarly, Bernstein et al found when women with a prior cesarean perceived their provider to have a preference, they were more likely to choose their perceived provider-preferred delivery mode.

Patient preferences along with safety and medical capacity should be considered, and providers have a role in offering balanced counseling incorporating realistic framing of risks and benefits. This process may first and foremost help patients make medically appropriate decisions in alignment with their goals, which may ultimately reduce the cesarean delivery rate.

This study is not without limitations. First, this population was limited to English-speaking women within the San Francisco Bay Area receiving care at an academic medical center with multiple resources available. Trial of labor after cesarean, singleton vaginal breech deliveries, and vaginal delivery of a nonvertex second twin were offered with careful attention to patient selection. Women may have chosen this site for delivery based on these resources, and, as a result, their preferences may not be generalizable.

In addition, although the population was ethnically and racially diverse, they reported high educational attainment and income. However, a significant strength of this study is that it is one of the first to study the complexity of mode-of-delivery preferences in several clinical contexts. The use of both standard gamble and time tradeoff methods of preference elicitation allowed nuanced exploration of women’s values for birth processes and outcomes, and the measurement of these preferences in 4 differing clinical contexts enabled us to assess how women’s preferences differ (or remain the same) in these contexts.

In this racially/ethnically diverse population of high-income, highly educated women, our work suggests that most women highly value uncomplicated vaginal birth but that preferences for obstetric interventions facilitating vaginal delivery vary substantially. These data highlight the importance of preparing patients for birth by providing clear information, appropriately contextualized risk education, and ongoing opportunities for communication. Providers must offer balanced patient counseling that appropriately frames rare adverse events and manages patient expectations about labor processes. Shared decision making encouraging fully informed patient preferences is critical to achieving patient-centered care in obstetrics.