Interpersonal Processes of Care and Cesarean Delivery in Two Health Care Settings

Abstract

Objectives. We examined whether interpersonal processes of care (IPC) were associated with cesarean delivery.

Methods. We performed a cross-sectional study of 1308 postpartum women at Kaiser Permanente Medical Center in Walnut Creek, CA (KP-WC), and San Francisco General Hospital (SFGH) from 2004 to 2006. Using interview and medical record data, logistic regression analyses estimated the odds of cesarean delivery as a function of IPC domains.

Results. After adjustment for demographic and reproductive factors, women at KP-WC who reported higher scores for their provider’s “elicitation of patient concerns and responsiveness” were less likely to have delivered by cesarean, whereas women who reported higher scores for “empowerment and self-care” were more likely. At KP-WC, women who reported low English proficiency were less likely to have delivered by cesarean than women who reported high proficiency. At SFGH, none of the IPC measures were significant; however, younger age was associated with a lower risk of cesarean delivery, whereas higher educational attainment was associated with an increased risk.

Conclusions. To reduce record-high rates of cesarean delivery, more emphasis should be placed on addressing the nonmedical factors associated with operative delivery.

In the last 13 years, the rate of cesarean delivery in the United States increased by 59%, from 20.7% in 1996 to an all-time high of 32.9% in 2009.^1,2 This increase was observed among all groups of women, regardless of age, race/ethnicity, or state of residence. Compared with vaginal delivery, cesarean delivery is associated with increased morbidity and mortality as well as higher costs, and there are little data to suggest that perinatal outcomes have improved as cesarean rates have increased.³ Although there are clinical indications for performing a cesarean delivery, several nonmedical factors have been associated with its use, such as patient preferences, health insurance status, physician and institutional practice patterns, practice guidelines, and liability concerns.^4–6

In the early 1990s, the Department of Health and Human Services identified the need to reduce the cesarean rate. Despite national goals^7,8 and recommendations from the American College of Obstetricians and Gynecologists,⁹ the national cesarean delivery rate has steadily increased since 1996; in 2009, one third of all women who gave birth had a cesarean delivery.² To achieve the overall goal of reducing unwarranted cesarean deliveries, we need a more comprehensive understanding of the influence of medical and nonmedical factors on operative delivery.

We focused on the role of interpersonal processes of care (IPCs) and the likelihood of cesarean delivery. IPCs describe various aspects of the provider–patient relationship, including communication, patient-centered decision-making, and interpersonal style. IPCs could influence the decision to deliver by cesarean, yet very few studies have analyzed these issues. For example, better provider–patient communication could help educate patients about the risks vs benefits of operative delivery and reduce the number of women who request an elective or repeat cesarean delivery.

METHODS

In this cross-sectional study conducted from March 2004 through August 2006, bilingual (English and Spanish) and bicultural female interviewers recruited women from the postpartum wards at 2 health care facilities in northern California: the Kaiser Permanente Medical Center in Walnut Creek (KP-WC), which is part of a group-model integrated health care delivery system, and San Francisco General Hospital (SFGH), the only public hospital in San Francisco County.

On the postpartum floor, a brief screening interview was conducted to assess patients’ eligibility for the study. Eligible participants were women who were hospitalized for the delivery of a live-born, singleton infant, who self-reported their race/ethnicity, and were comfortable responding in English or Spanish. The 45-minute structured interview assessed demographic and health service characteristics, reproductive factors, and several dimensions of interpersonal processes of care. Information about use of prenatal care and delivery characteristics was abstracted from the medical record. The research assistants who conducted the interviews and performed the medical record review were trained at the same time, used the same procedure manual, and were supervised by the study investigators.

A total of 185 women refused to participate in the study at KP-WC and 13 refused at SFGH. The most common reasons for refusal were timing (not a convenient time for the mother to be interviewed) and the presence of visitors. Because of differences in the patient characteristics between the 2 health care settings, we stratified all analyses by recruitment site and categorized as KP-WC or SFGH.

The primary dependent variable was cesarean delivery for the index pregnancy. We deleted from the analyses women who were missing data on type of delivery (11 women from KP-WC, none from SFGH) or who reported that they did not receive any prenatal care (none from KP-WC, 7 women from SFGH).

We measured our primary independent variables, the IPCs, using 7 multi-item scales that we adapted from 2 existing instruments. The first instrument was the original version of the Interpersonal Processes of Care Survey,¹⁰ which was developed for adults from diverse racial/ethnic groups. The second instrument was the Prenatal Interpersonal Processes of Care (PIPC) Survey,¹¹ in which the original IPCs were adapted and validated for ethnically diverse, pregnant women. We analyzed measures of provider–patient communication (elicitation of patient problems and responsiveness, explanations, and empowerment and self-care), patient-centered decision-making, and the prenatal provider’s interpersonal style (respectfulness, perceived discrimination, and emotional support and reassurance). Whereas the original instruments referred to “doctors” (IPC instrument) or “providers” (PIPC instrument), we used the term “doctors and other health professionals” since many women receive their prenatal care from midwives. The questions that assessed IPC specifically referred to the index pregnancy. The scores for each scale ranged from 1 to 5; a higher score indicated that the woman reported more favorable experiences, such as more explanations or less discrimination.

Maternal race/ethnicity (hereafter called ethnicity) was an additional primary independent variable. Ethnicity was self-reported and categorized as African American, Asian/Pacific Islander (hereafter called Asian), Latina (referent), White, or multiethnic or other. Only 58 of women reported their ethnicity as multiethnic or other, and they were excluded from the analyses.

We analyzed maternal and infant characteristics, health service, and reproductive factors as potential confounders of the relationship between IPC and cesarean delivery. We categorized maternal age as younger than 22, 22 to 35 (referent), or older than 35 years. We defined younger age as younger than 22 years rather than 20 because very few women at KP-WC were younger than 20 years. We categorized educational attainment as fewer than 12 years (< high school), 12 years (high school; referent), or more than 12 years (> high school). We dichotomized marital status as married (married or living with partner; referent) or not married (single, divorced, widowed, or other). We dichotomized maternal birthplace as United States (referent) or not United States. We created a variable that combined parity and previous history of cesarean delivery, which we categorized as multiparous with no prior cesarean delivery (referent), multiparous with prior cesarean delivery, or primiparous.

We used self-reported maternal height and prepregnancy weight to compute body mass index (BMI; defined as weight in kilograms divided by height in meters squared), which we categorized as healthy (BMI < 25 kg/m²), overweight (BMI = 25–29.9 kg/m²), or obese (BMI ≥ 30 kg/m²). We categorized English proficiency as good or fluent (referent), average, or poor or none. We dichotomized self-reported insurance status as uninsured or insured, aggregating all types of insurance into a single group. All women at KP-WC were insured, so we included the insurance variable only in the analyses of the SFGH participants.

We created dichotomous variables for maternal diabetes (a combination of pregestational and gestational diabetes) and hypertension. We also created an indicator variable for delivery complications if any of the following maternal or infant conditions were noted in the medical record: chorioamnionitis, disseminated intravascular coagulation, pregnancy-induced hypertension, pre-eclampsia, placenta previa, maternal bleeding or hemorrhage, placental abruption, breech position, shoulder dystocia, other malpresentation, or umbilical cord prolapse.

We analyzed 2 infant characteristics that we obtained from the medical record. Gestational age was categorized as fewer than 33, 33 to 37, or 38 to 42 weeks (referent). No infant had a gestational age that exceeded 42 weeks. We categorized infant birth weight as less than 2500, 2500 to 4000 (referent), or more than 4000 grams.

We performed paired t tests and compared mean IPC scores by health care setting. We conducted contingency table analyses to compare the distribution of participant characteristics according to health care setting. To assess site-based variation in risk of cesarean delivery, we performed separate regression analyses for each site (KP-WC and SFGH). We performed unadjusted and adjusted logistic regression analyses to compute odds ratios and 95% confidence intervals for the risk of cesarean delivery. The IPC scales were scored so that a higher score represented a more favorable response; in the logistic regression models, a higher score, such as more elicitation of concerns or greater responsiveness, would be expected to be associated with lower rates of cesarean delivery (odds ratio [OR]< 1). To create the final multivariable models, we used the Hosmer and Lemeshow test to determine the variables that were the best fit for cesarean delivery. We used an iterative process of elimination to determine the variables retained in the final model; we retained all IPC variables, all variables that were significant at the P < .1 level, and all variables that improved the model fit. Because women who have had a prior cesarean delivery are at greater risk for repeat cesarean, in separate analyses we excluded women who had already delivered by cesarean and reran the adjusted regression models to further assess the association between the IPC scores and risk of cesarean delivery. We conducted the data analyses with SAS/STAT software version 9.2 (SAS Institute, Cary, NC).¹²

RESULTS

A total of 1308 women (445 from KP-WC and 863 from SFGH) met the inclusion criteria and were included in the analysis. At SFGH in 2005, the rate of cesarean delivery was 24.5% overall and 22.6% among study participants (P = .3); at KP-WC, the overall rate in 2005 was 26.4% and the rate among study participants was 30.1% (P = .1).

The distribution of maternal and infant characteristics varied significantly by site (Table 1 ). Over half of all women at the KP-WC site were White and one quarter were Latina, whereas three quarters of SFGH participants were Latina and only 4% were White. Delivery complications were more common among the KP-WC participants (27.4%) than among participants at SFGH (7.0%); the higher rate at KP-WC is probably because this facility is a subregional referral center with a level III neonatal intensive care unit (i.e., a nursery that routinely provides intensive care) and therefore receives high-risk patients from the neighboring Kaiser Permanente facilities. Women at KP-WC were also more likely to have delivered by cesarean than women at SFGH (30% [n = 134] vs 23% [n = 195]; P = .003).

TABLE 1—

Characteristics of Participants in Study of Interpersonal Processes of Care and Cesarean Delivery, by Health Care Setting: San Francisco Bay Area, 2004–2006

Characteristic	KP-WC, No. (%)	SFGH, No. (%)	P a
Ethnicity			< .001
African American	23 (5.2)	103 (11.9)
Asian	65 (14.6)	76 (8.8)
Latina	121 (27.2)	652 (75.6)
White	236 (53.0)	32 (3.7)
Age, y			< .001
< 22	39 (8.8)	201 (23.3)
22–35	343 (77.1)	592 (68.6)
> 35	63 (14.2)	70 (8.1)
Education, y			< .001
< 12 (< high school)	35 (8.0)	447 (52.6)
12 (high school)	109 (25.0)	238 (28.0)
> 12 (> high school)	292 (67.0)	165 (19.4)
Missing	9	13
Married	419 (94.2)	660 (76.5)	< .001
Foreign-born	137 (30.8)	674 (78.1)	< .001
Parity–cesarean profile			.1
Primiparous	200 (44.9)	383 (44.4)
Multiparous, no prior cesarean delivery	173 (38.9)	374 (43.3)
Multiparous, prior cesarean delivery	72 (16.2)	106 (12.3)
Body mass index, kg/m2 (prepregnancy)			.03b
Normal (< 25.0)	258 (59.0)	289 (50.7)
Overweight (25.0–29.9)	104 (23.8)	172 (30.2)
Obese (≥ 30.0)	75 (17.2)	109 (19.1)
Missing	8	293
English proficiency			< .001
Good or fluent	379 (85.6)	307 (35.6)
Average	38 (8.6)	130 (15.1)
Poor or none	26 (5.9)	426 (49.4)
Missing	2	0
Insured during pregnancy	445 (100.0)	846 (98.0)	.003
Diabetes (pregestational or gestational)	38 (8.5)	40 (4.6)	.005
Hypertension	47 (10.6)	58 (6.7)	.02
Delivery complications	122 (27.4)	60 (7.0)	< .001
Infant gestational age, wk			.04
< 33	10 (2.3)	18 (2.1)
33–37	49 (11.1)	140 (16.2)
38–42	383 (86.7)	705 (81.7)
Missing	3	0
Infant birth weight, g			.21
< 2500	30 (6.8)	51 (5.9)
2500–4000	347 (78.2)	709 (82.2)
> 4000	67 (15.1)	103 (11.9)
Missing	1	0
Delivery type			.003
Vaginal	311 (70.0)	668 (77.4)
Cesarean	134 (30.1)	195 (22.6)

Note. KP-WC = Kaiser Permanente Medical Center in Walnut Creek, CA; SFGH = San Francisco General Hospital.

^aChi-square test.

^bTwenty-three percent of the data were missing.

Women at SFGH reported significantly higher mean scores for the 3 IPC communication measures than women at KP-WC, which indicates greater satisfaction with “responsiveness,” “empowerment and self-care,” and “explanations” (Table 2). At SFGH, women reported higher mean scores on the domain “we decided together,” which represents collaborative decision-making, than did women at KP-WC. Scores for 2 of the 3 interpersonal style measures (“emotional support” and “respectfulness”) were also significantly higher at SFGH than KP-WC, but there were no differences in the dimension of perceived discrimination. In posthoc statistical power calculations for a t test difference in mean IPC scores (α = 0.05), the observed power ranged from 99% to 100%.

TABLE 2—

Interpersonal Processes of Care as Reported by Postpartum Women, by Health Care Setting: San Francisco Bay Area, 2004–2006

Interpersonal Processes of Care	KP-WC, Mean Scorea (SD)	SFGH, Mean Scorea (SD)	P b
Communication
Elicitation of patient problems and responsiveness	4.62 (0.57)	4.78 (0.58)	< .001
Empowerment and self-care	4.34 (0.70)	4.66 (0.64)	< .001
Explanations	4.55 (0.68)	4.77 (0.63)	< .001
Patient-centered decision-making	2.86 (1.34)	4.33 (0.93)	< .001
Interpersonal style
Emotional support and reassurance	4.16 (0.84)	4.63 (0.74)	< .001
No perceived discrimination	4.94 (0.17)	4.95 (0.24)	.29
Respectfulness	4.70 (0.56)	4.81 (0.58)	< .001

Note. KP-WC = Kaiser Permanente Medical Center in Walnut Creek, CA; SFGH = San Francisco General Hospital.

^aThe scores for each scale ranged from 1 to 5; a higher score indicated that the woman reported more favorable experiences, such as more explanations or less discrimination.

^bPaired t test.

In unadjusted logistic regression analyses of participants at KP-WC, we found that 2 IPC subdomains were associated with cesarean delivery. Women who reported higher mean scores on elicitation of patient problems and responsiveness measures were less likely to have a cesarean delivery, whereas women who reported higher scores on the provider’s respectfulness were more likely to have a cesarean delivery (Table 3 ). In unadjusted logistic regression analyses of participants at SFGH, the only IPC measure that was associated with cesarean delivery was respectfulness; higher mean scores on respectfulness were associated with a lower risk of cesarean delivery (in contrast to the direction of the association at KP-WC).

TABLE 3—

Unadjusted Odds Ratios for Cesarean Delivery, by Health Care Setting and Interpersonal Processes of Care: San Francisco Bay Area, 2004–2006

	Cesarean Deliveries at KP-WC		Cesarean Deliveries at SFGH
Interpersonal Processes of Care	OR (95% CI)	P a	OR (95% CI)	P a
Communication
Elicitation of patient problems and responsiveness	0.39 (0.19, 0.82)	.01	1.27 (0.73, 2.23)	.4
Empowerment and self-care	1.14 (0.74, 1.75)	.57	0.83 (0.52, 1.30)	.41
Explanations	1.04 (0.62, 1.74)	.89	0.82 (0.50, 1.37)	.46
Patient-centered decision-making	0.98 (0.82, 1.17)	.82	1.15 (0.92, 1.45)	.22
Interpersonal style
Emotional support and reassurance	1.33 (0.89, 1.99)	.16	1.15 (0.80, 1.66)	.46
No perceived discrimination	0.42 (0.11, 1.57)	.2	0.85 (0.41, 1.77)	.67
Respectfulness	2.09 (1.10, 3.96)	.02	0.58 (0.36, 0.94)	.03

Note. CI = confidence interval; KP-WC = Kaiser Permanente Medical Center in Walnut Creek, CA; OR = odds ratio; SFGH = San Francisco General Hospital. At KP-WC, the total sample was 445 and the number of cesarean deliveries was 134; at SFGH, the total sample was 857 and the number of cesarean deliveries was 193.

^aChi-square test.

When we used multivariable logistic regression to adjust for potential confounders (Table 4 ), the association of provider’s elicitation and responsiveness with cesarean delivery at KP-WC remained significant; women who reported higher scores on elicitation and responsiveness were 84% less likely to deliver by cesarean. In addition, at KP-WC the association of empowerment and self-care with cesarean delivery became statistically significant; women who reported higher mean scores on empowerment and self-care were twice as likely to have delivered by cesarean.

TABLE 4—

Adjusted Odds Ratios for Cesarean Delivery, by Health Care Setting, Interpersonal Processes of Care, and Participants’ Characteristics: San Francisco Bay Area, 2004–2006

	Cesarean Deliveries at KP-WC		Cesarean Deliveries at SFGH
Variable	OR (95% CI)	P a	OR (95% CI)	P a
Interpersonal processes of care
Communication
Elicitation of patient problems and responsiveness	0.16 (0.06, 0.44)	< .01	1.32 (0.66, 2.64)	.43
Empowerment and self-care	1.90 (1.02, 3.54)	.04	0.94 (0.54, 1.64)	.84
Explanations	1.56 (0.80, 3.05)	.2	0.80 (0.42, 1.52)	.49
Patient-centered decision-making	0.94 (0.75, 1.18)	.61	1.13 (0.86, 1.48)	.39
Interpersonal style
Emotional support and reassurance	1.42 (0.81, 2.50)	.23	1.09 (0.70, 1.70)	.71
No perceived discrimination	1.05 (0.15, 7.19)	.96	0.66 (0.27, 1.65)	.38
Respectfulness	1.92 (0.86, 4.27)	.11	0.76 (0.41, 1.38)	.36
Participants’ characteristics
Race/ethnicity
African American	…	…	1.24 (0.44, 3.51)	.69
Asian	…	…	0.87 (0.30, 2.55)	.8
Latina	…	…	0.46 (0.16, 1.26)	.13
White (Ref)	…	…	1.00
Age, y
< 22	0.34 (0.09, 1.27)	.11	0.48 (0.28, 0.82)	< .01
22–35	1.00		1.00
> 35	1.31 (0.68, 2.54)	.42	1.24 (0.68, 2.24)	.48
Education, y			y
< 12 (< high school)	2.39 (0.57, 10.1)	.23	1.14 (0.70, 1.86)	.59
12 (high school)	1.00		1.00
> 12 (> high school)	1.26 (0.62, 2.59)	.52	1.94 (1.11, 3.38)	.02
Foreign-born
Yes	0.87 (0.43, 1.77)	.69	…	…
No	1.00		…	…
English proficiency
Good or fluent (Ref)	1.00		1.00
Average	0.57 (0.15, 2.27)	.43	0.71 (0.34, 1.47)	.36
Poor or none	0.04 (0.005, 0.33)	< .01	1.61 (0.86, 3.05)	.14
Parity-cesarean profile
Primiparous	1.90 (1.01, 3.55)	.05	3.50 (2.16, 5.66)	< .01
Multiparous, no prior cesarean delivery	1.00		1.00
Multiparous, prior cesarean delivery	72.9 (27.8, 191)	< .01	19.6 (11.0, 34.8)	< .01
Diabetes, pregestational or gestational
Yes	1.65 (0.66, 4.11)	.29	1.47 (0.61, 3.56)	.4
No (Ref)	1.00		1.00
Hypertension
Yes	…	…	1.39 (0.66, 2.94)	.38
No	…	…	1.00
Delivery complications
Yes	5.97 (3.31, 10.8)	< .01	5.27 (2.82, 9.85)	< .01
No (Ref)	1.00		1.00
Infant birth weight, g
< 2500	…	…	0.96 (0.42, 2.16)	.92
2500–4000 (Ref)	…	…	1.00
> 4000	…	…	2.61 (1.52, 4.47)	< .01

Note. CI = confidence interval; KP-WC = Kaiser Permanente Medical Center in Walnut Creek, CA; OR = odds ratio; SFGH = San Francisco General Hospital. At KP-WC, the total sample was 434 and the number of cesarean deliveries was 132; at SFGH, the total sample was 844 and the number of cesarean deliveries was 191. For KP-WC model, adjusted R² = 0.51, Hosmer and Lemeshow goodness-of-fit P = .83; for SFGH model, adjusted R² = 0.34, Hosmer and Lemeshow Goodness of fit test P value goodness-of-fit P = .73. Ellipses indicate that variables were not retained in the final model as their P values were ≥ .1.

^aChi-square test.

As hypothesized, other maternal factors were independently associated with cesarean delivery (Table 4). At KP-WC, women with poor or no English proficiency were 96% less likely to deliver by cesarean than were women with higher English proficiency. On the other hand, an increased risk of cesarean at KP-WC was associated with primiparity (OR = 1.9), multiparity with prior cesarean delivery (OR = 72.9), and delivery complications (OR = 5.97).

At SFGH, women who were younger than 22 years were half as likely to have delivered by cesarean as women who were 22 to 35 years old (OR = 0.48; Table 4). In addition, women who had more than 12 years of education (OR = 1.9), were primiparous (OR = 3.5), or were multiparous with a history of cesarean delivery (OR = 20) were more likely to deliver by cesarean than were women from the reference categories. Women who experienced delivery complications (OR = 5.3) and women who delivered infants who weighed more than 4000 grams (OR = 2.6) were 2.5 to 5 times more likely to have a cesarean delivery than were women without delivery complications or women who delivered infants of normal birth weight.

In posthoc statistical power calculations for the multiple regression model for KP-WC (with α = 0.05; 18 predictors; R² = 0.35; and sample size = 434), the observed power was 100%. Similarly, for the SFGH regression model (with α = 0.05; 23 predictors; R² = 0.22; and sample size = 844), the observed power was 100%.

In additional adjusted logistic regression analyses that were restricted to women who did not have a prior history of cesarean delivery, at KP-WC the significant IPC scores were as follows: elicitation of patient problems (OR = 0.09; 95% confidence interval [CI] = 0.03, 0.31), emotional support and reassurance (OR = 2.26; 95% CI = 1.06, 4.82), and respectfulness (OR = 2.68; 95% CI = 1.04, 6.94). The Hosmer and Lemeshow goodness-of-fit P value was .996. In the same group of women at SFGH, none of the IPC scores were significant (all P > .3) and the goodness-of-fit P value was .69.

DISCUSSION

In 2 distinct cohorts of childbearing women in northern California, we found that several dimensions of interpersonal processes of care, which reflect the patient–provider experience during the prenatal period, were independently associated with risk of cesarean delivery. Among women at KP-WC, a group-model integrated delivery system, higher mean scores on the provider’s “elicitation of patient problems and responsiveness” conferred a lower risk of cesarean delivery, whereas higher scores on “empowerment/self-care” were associated with a higher risk of cesarean delivery. In subgroup analyses of women without a history of cesarean delivery, we found that higher mean scores on the provider’s “elicitation of patient problems and responsiveness” remained inversely associated with cesarean delivery. This indicates that among the KP-WC women who had a primary cesarean delivery, the IPC measures are fairly robust and still explain some of the risk associated with this delivery method. If women at the KP-WC site tend to prefer cesarean deliveries, then the relationship between the provider’s empowerment, emotional support, and respectfulness and cesarean delivery may be a result of patient preferences.

On the other hand, we did not find an independent association between IPC measures and cesarean delivery at the SFGH site, a public hospital. The variation in the relationship between IPC measures and cesarean delivery could be attributable to differences in expectations among the patients who are served at these 2 facilities, or to differences in practitioner or institutional practice patterns. Because the IPC scales assess women’s experiences during prenatal care, our results suggest that targeted efforts for prenatal providers could have a measurable impact on outcomes. Our results do not prove that efforts to increase providers’ elicitation of patient concerns will reduce the rate of cesarean delivery or that improving the patient’s sense of empowerment, emotional support, or respectfulness will increase the rate of cesarean delivery; instead, the results indicate that further study is needed to assess the ways in which prenatal experiences affect delivery outcomes. This includes evaluating the role of maternal request on both primary and repeat cesarean delivery.

Also noteworthy were the institutional differences in mean IPC scores for patient-centered decision-making, suggesting that women at KP-WC felt they were consulted less about the medical recommendations they received than women at SFGH. These score differences in patient-centered decision-making were significant across each of the 4 ethnic groups (P < .001) and warrant further investigation in this patient population.

As expected,¹³ the strongest independent predictor of cesarean delivery was prior cesarean delivery. Although the American College of Obstetricians and Gynecologists recommends that most pregnant women with a history of 1 low, transverse cesarean delivery should be counseled to consider vaginal birth after cesarean delivery (VBAC),⁹ we found that VBAC was very uncommon. At KP-WC, there were 72 deliveries among multiparous women with prior cesareans; only 12 women had a successful VBAC (data not shown). At SFGH, there were 106 deliveries among women with previous history of cesarean, and 9 women had successful VBAC. However, providers’ concerns about the risk of ruptured uterus may override concerns about patient choice for vaginal delivery, thereby reducing the VBAC rate. For example, beginning in the early 2000s and to improve patient safety, efforts were made at KP-WC to speed up cesarean deliveries if there was concern about possible ruptured uterus.

Similar to what was reported with other study populations,^13–15 primiparous women were more likely to have a cesarean delivery than multiparous women who delivered vaginally in prior births. Although our data did not indicate whether cesarean deliveries were elective, the higher risk of cesarean delivery among primiparous women is disturbing because women with 1 cesarean delivery are much more likely to have subsequent cesarean deliveries, thus perpetuating the cycle of high cesarean delivery rates.

As others have reported,^14,16 we found that younger women had a decreased risk of cesarean delivery. In addition, educational attainment and English proficiency were independently associated with cesarean delivery in our study. As previously reported,¹⁶ women at SFGH who had higher educational attainment were more likely to have a cesarean delivery, which is consistent with national data on minority births since SFGH is a public hospital that provides services to an urban, underserved population. At the Kaiser Permanente site, we found that women who reported greater English proficiency were more likely to deliver via cesarean, which is consistent with another study¹⁷ and indicates a need to address linguistic issues that influence the provision of care, especially at health care centers that serve women with limited English proficiency.

Our findings are subject to several limitations. First, our study sites were an urban public hospital (SFGH) and a suburban health maintenance organization, or HMO (KP-WC), and our results may not be applicable to other settings. In addition, we could not determine whether a cesarean delivery was an elective procedure or clinically indicated. Although high BMI is known to increase the risk of cesarean delivery, we did not include this variable in the SFGH regression analyses because of the large proportion of missing data. Lastly, women were recruited and interviewed after delivery, and women who delivered by cesarean may have consented to participate or recalled events differently than women who had vaginal deliveries. However, because we interviewed women after delivery, as other studies have done,¹⁸ participants were able to account for their entire pregnancy, rather than only prenatal experiences at an early point in their pregnancy and at a particular gestational age.

Notwithstanding these limitations, our study has several strengths, including the recruitment of a multiethnic group of women during the immediate postpartum period in 2 distinct health care settings in the same geographic area. Our final multivariable models were robust and fit the data well (for KP-WC model, adjusted R² = 0.51, Hosmer and Lemeshow goodness-of-fit P = .83; for SFGH model, adjusted R² = 0.34, Hosmer and Lemeshow goodness-of-fit P = .73).

In our cross-sectional study of women at 2 distinct health care settings in the San Francisco Bay Area, we found that interpersonal and nonmedical factors were associated with cesarean delivery. Clinicians should place more emphasis on improving provider–patient communication during prenatal care and reducing nonmedical barriers to vaginal delivery.