Materials and Methods
The “Mode of Delivery Preferences among Diverse Populations of Women” study was conducted at the University of California, San Francisco, from 2008-2012. Details of this study have been described elsewhere. Briefly, women who received prenatal care were sent letters that described the study that included an “opt-in/opt-out” response card. Patients who returned the card with “opt in” checked off or who did not return the card were contacted by a research associate who further described the study and assessed the woman’s eligibility and interest in participation. Additionally, patients who contacted the research associate after seeing a flyer or hearing about the study by word of mouth were enrolled if they met eligibility criteria, which included being English-speaking and <36 weeks of gestation. Participants received $40 remuneration for each face-to-face interview. Institutional review board approval was obtained from the University of California, San Francisco, Committee on Human Research. Written informed consent was obtained from all participants.
Between 24-36 weeks of gestation, participants underwent a face-to-face interview, during which they completed a questionnaire that included items that were related to sociodemographic characteristics (age, race/ethnicity, education, employment, marriage status, and income), pregnancy history, their preferred delivery mode, and a 9-item depression measure (Patient Health Questionnaire [PHQ-9]). The PHQ-9, which is recommended by American College of Obstetricians and Gynecologists for perinatal depression assessment, has been validated in obstetrics and gynecology clinical settings and is used commonly in research and clinical practice to assess symptoms of depression, to make a preliminary diagnosis of depression, or to categorize depression severity. Scores range from 0–27; higher scores indicate more depressive symptoms (specifically, 0–4, 5–9, 10–14, and >15 represent minimal, mild, moderate, and severe levels of depression, respectively).
During the baseline interview, participants also completed a series of standard gamble exercises with the use of a computer tool our group developed for preference elicitation. The standard gamble exercise yields a preference score that ranges from 0-1, with 0 defined as the least desired outcome of a decision being considered and 1 defined as the most preferred outcome. Scores for intermediately ranked outcomes are generated by presenting the assessor with a hypothetical choice between certainty of experiencing the intermediate ranked outcome and a gamble between experiencing the preferred outcome vs experiencing the least desired outcome. The probability of experiencing the preferred vs the least desired outcome is varied until the assessor is indifferent between certainty of the intermediary outcome and the gamble.
As planned vaginal delivery is the most common delivery approach in the United States, we focused on the strength of preference for vaginal delivery. For this measurement, participants who had a stated preference for vaginal delivery were presented with a choice between certainty of having an uncomplicated planned cesarean delivery and a gamble between an uncomplicated vaginal delivery (their preferred delivery mode) vs undergoing labor and ending with an uncomplicated cesarean delivery (their less desired delivery mode in this exercise). The probability that their labor would end in a cesarean delivery was varied until the woman was indifferent between the 2 choices. Stronger preferences for vaginal delivery are reflected in a higher score, indicating women would accept a greater chance that labor would end in a cesarean delivery before opting for an uncomplicated planned cesarean delivery. The preference score for vaginal delivery was calculated as the probability of having the planned vaginal delivery end in a cesarean delivery at her indifference point. For example, if a woman who had a stated preference for vaginal delivery but indicated that she would opt for a planned cesarean delivery if the chance that her planned vaginal birth would end in cesarean delivery was 25%, her preference score for vaginal delivery would be assigned a value of 0.25. On the other hand, if a woman with a stated preference for vaginal delivery indicated that she would opt for a planned cesarean delivery only if the chance of labor ending in a cesarean delivery was 75%, she would have a preference score of 0.75 for vaginal delivery. As this analysis focused on the strength of preference for a vaginal delivery, participants who had a stated preference for cesarean delivery (ie, those who indicated they would “probably” or “definitely” choose to have a cesarean delivery) were assigned a preference score of 0 for vaginal delivery.
A telephone interview was conducted at 8-10 weeks after delivery during which participants again completed the PHQ-9, and the delivery mode undergone was assessed. At 6-8 months after delivery, participants had a face-to-face interview during which they completed the PHQ-9 a third time.
The primary outcome for this analysis was PHQ-9 score at 8-10 weeks after delivery; PHQ-9 score at 6-8 months after delivery was a secondary outcome. The primary predictors were the strength of preference for vaginal delivery and delivery mode undergone. Univariable and multivariable regression analyses were performed to identify predictors of PHQ-9 score at each of the postpartum time points. In the multivariable analyses, we included the primary predictors, an interaction term between delivery mode and strength of preference for vaginal delivery, and covariates that included the antepartum PHQ-9 score, parity, preterm delivery (defined as delivery at <37 weeks of gestation), age, race/ethnicity, education, income, employment status, and relationship status. We used a backward elimination procedure to remove hypothesized predictors with probability values that exceeded 0.20 for the final multivariable model. The data contained missing values so we fit the models to 20 multiply imputed datasets created with SAS PROC MI (version 9.2; SAS Institute Inc, Cary, NC). The imputation model included all predictor and outcome variables that were used in regression models. Imputed values for binary and categoric variables were rounded and truncated to the nearest category. Parameters and standard errors were estimated by the combination of the results across the 20 imputed datasets, according to Rubin’s rules and Meng and Rubin. A probability value of < .05 was considered statistically significant. All analyses were implemented with SAS software (version 9.2).
Results
One hundred sixty participants completed the baseline assessment. More than one-third of these women (37.5%) were ≥35 years old. The participants comprised an ethnically and racially diverse group with a range of previous birth experiences ( Table 1 ). Slightly less than one-half of the women identified as white (47.5%); more than one-quarter (26.3%) were African American; 8.8% were Latina, and 17.7% identified as being a member of another racial/ethnic group. Nulliparous women comprised 33.1% of the sample, and 30.6% had a history of a cesarean delivery.
Characteristic | Total |
---|---|
Age, y a | 31.9 ± 5.9 |
<25 y, n (%) | 22 (13.8) |
25-29 y, n (%) | 30 (18.8) |
30-34 y, n (%) | 48 (30.0) |
≥35 y, n (%) | 60 (37.5) |
Race/ethnicity, n (%) | |
African American | 42 (26.3) |
Latina or Hispanic | 14 (8.8) |
Other b | 28 (17.6) |
White | 76 (47.5) |
Educational attainment, n (%) | |
High school or less | 34 (21.2) |
Some college | 32 (20.0) |
College degree or higher | 94 (58.8) |
Employed, n (%) | 86 (54.1) |
Married or living with partner, n (%) | 131 (81.9) |
Annual household income, n (%) | |
<$25K | 39 (24.7) |
$25K-50K | 24 (15.2) |
$50K-100K | 34 (21.5) |
≥$100K | 61 (38.6) |
Primary prenatal care provider, n (%) | |
Obstetrician | 86 (54.8) |
Midwife | 69 (43.9) |
Other c | 2 (1.3) |
Nulliparous women, n (%) | 53 (33.1) |
Multiparous women, n (%) | |
Vaginal deliveries only | 58 (36.3) |
Cesarean deliveries only | 40 (25.0) |
Vaginal and cesarean deliveries | 9 (5.6) |
Baseline 9-item Patient Health Questionnaire score a d | 5.8 ± 4.1 |
Strength of preference for vaginal delivery a e | 0.658 ± 0.35 |
a Data are given as mean ± standard error
b Includes Asian/Pacific Islander (n = 26) and Native American (n = 2)
c Primary prenatal care provider was “other” and specified as homebirth midwife (n = 1) and several providers (n = 1)
d Scores range from 0–27; higher scores indicate more depressive symptoms
e Measured by the standard gamble metric; scores range from 0 (preference for cesarean delivery) to 1 (strongest preference for vaginal delivery).
Antepartum, most women (92.4%) had a stated preference for a vaginal delivery. When we included the women who preferred cesarean delivery, the mean and median vaginal delivery preference scores were 0.658 (SD ±0.352) and 0.750 (interquartile range, 0.500–0.994). These utility scores show that, on average, these women indicated that they would opt for a planned cesarean only if the probability that an attempted vaginal birth would end in a cesarean delivery reached 65.8%.
Approximately one-quarter of the participants (26%) delivered by cesarean. The incidence of postpartum depression (moderate and severe; defined as >9 on the PHQ-9) was 9.1% at 8-10 weeks and 12.8% at 6-8 months after delivery. Antepartum PHQ-9 score, income, and identification as Asian/Pacific Islander/Native American and delivery mode all emerged as significant independent predictors of PHQ-9 score at 8-10 weeks after delivery ( Table 2 ). However, at the 6- to 8-month assessment, the association with mode of delivery was no longer present, and only antepartum PHQ-9 score and income remained as predictors of PHQ-9 score at 6-8 months after delivery ( Table 3 ).
Characteristic | Univariable regression coefficient B (95% confidence interval) | P value | Multivariable regression coefficient a B (95% confidence interval) | P value |
---|---|---|---|---|
Age, y | .461 | |||
<25 | Reference | |||
25-29 | −0.41 (−2.29 to 1.47) | .670 | ||
30-34 | −1.27 (−2.97 to 0.43) | .143 | ||
≥35 | −0.73 (−2.39 to 0.93) | .387 | ||
Race/ethnicity | .094 | .102 | ||
African American | 1.39 (0.10–2.67) | .035 | 0.85 (−0.61 to 2.31) | .253 |
Latina or Hispanic | 1.17 (−0.98 to 3.33) | .286 | 1.15 (−0.76 to 3.06) | .241 |
Other b | 1.32 (−0.16 to 2.81) | .080 | 1.37 (0.18–2.57) | .025 |
White | Reference | Reference | ||
Educational attainment | .864 | |||
High school or less | 0.17 (−1.21 to 1.54) | .813 | ||
Some college | 0.32 (−1.08 to 1.72) | .653 | ||
College degree or higher | Reference | |||
Employed | −0.21 (−1.30 to 0.88) | .702 | ||
Married or living with partner | −0.40 (−1.81 to 1.01) | .578 | ||
Annual household income | .018 | .015 | ||
<$25K | −0.08 (−1.41 to 1.26) | .910 | −0.82 (−2.27 to 0.63) | .268 |
$25K-50K | 1.90 (0.29–3.52) | .021 | 0.70 (−0.80 to 2.19) | .363 |
$50K-100K | −0.98 (−2.35 to 0.40) | .164 | −1.39 (−2.52 to −0.26) | .016 |
≥$100K | Reference | Reference | ||
Multiparous women | 1.40 (0.30–2.50) | .012 | 0.66 (−0.29 to 1.62) | .173 |
Delivery at <37 weeks of gestation | −0.90 (−3.08 to 1.28) | .418 | −1.51 (−3.27 to 0.25) | .093 |
Antepartum 9-item Patient Health Questionnaire score c | 0.44 (0.31–0.57) | < .001 | 0.41 (0.29–0.53) | < .001 |
Cesarean delivery d | 1.67 (0.46–2.89) | .007 | 1.13 (0.01–2.26) | .048 |
Vaginal delivery preference score e | ||||
When cesarean delivery f | 0.24 (0.02–0.47) | .034 | 0.25 (0.03–0.47) | .027 |
When vaginal delivery f | −0.08(−0.23 to 0.07) | .318 | −0.02 (−0.19 to 0.14) | .761 |
Interaction term between mode of delivery and preference score | 0.28 (0.00–0.55) | .047 |
a Final multivariable regression model includes the predictors that were retained in the backward elimination procedure ( P < .20) and significant interaction term with actual delivery mode by vaginal preference score
b Includes Asian/Pacific Islander (n = 26) and Native American (n = 2)
c Unstandardized regression coefficient for every 1 point increase on the 9-item Patient Health Questionnaire scale; scores range from 0–27 (higher scores indicate more depressive symptoms)
d Measured at the mean preference score for vaginal delivery (0.658); measured by the standard gamble metric; scores range from 0 (preference for cesarean delivery) to 1 (strongest preference for vaginal delivery)
e Unstandardized regression coefficient for every 0.10 point increase in vaginal preference score
f Reported effect of vaginal preference score at each mode of delivery undergone.
Characteristic | Univariable regression coefficient B (95% confidence interval) | P value | Multivariable regression coefficient a B (95% confidence interval) | P value |
---|---|---|---|---|
Age, y | .287 | |||
<25 | Reference | |||
25-29 | −1.07 (−1.18 to 3.32) | .351 | ||
30-34 | −0.74 (−2.76 to 1.29) | .475 | ||
≥35 | −0.22 (−2.21 to 1.78) | .831 | ||
Race/ethnicity | .663 | |||
African American | 0.85 (−0.66 to 2.37) | .269 | ||
Latina or Hispanic | 0.93 (−1.49 to 3.35) | .454 | ||
Other b | 0.33 (−1.47 to 2.13) | .721 | ||
White | Reference | |||
Educational attainment | .631 | |||
High school or less | 0.57 (−1.06 to 2.20) | .493 | ||
Some college | 0.61 (−0.97 to 2.18) | .449 | ||
College degree or higher | Reference | |||
Employed | −0.82 (−2.09 to 0.45) | .205 | −0.87 (−2.05 to 0.31) | .148 |
Married or living with partner | −0.84 (−2.51 to 0.84) | .328 | ||
Annual household income | .034 | .043 | ||
<$25K | −0.26 (−1.79 to 1.28) | .742 | −1.38 (−2.83 to 0.08) | .064 |
$25K-50K | 2.59 (0.71–4.48) | .007 | 1.21 (−0.46 to 2.87) | .156 |
$50K-100K | −0.10 (−1.71 to 1.52) | .907 | −0.49 (−1.90 to 0.92) | .493 |
≥$100K | Reference | Reference | ||
Multiparous | 1.33 (−0.02 to 2.69) | .054 | ||
Delivery at <37 weeks of gestation | 0.02 (−2.52 to 2.55) | .990 | ||
Antepartum 9-item Patient Health Questionnaire score c | 0.52 (0.38–0.65) | < .001 | 0.50 (0.36–0.63) | < .001 |
Cesarean delivery | 1.35 (−0.15 to 2.86) | .079 | ||
Vaginal delivery preference score d | −0.02 (−0.21 to 0.16) | .819 |