The “Greater Expectations?” study was funded by The Nuffield Trust, London, and the National Health Service Executive, Northern & Yorkshire Regional R&D, Durham, United Kingdom.
Josephine M. Green, BA(Hons), PhD, AFBPsS, CPsychol, Mother & Infant Research Unit, Department of Health Sciences, Area 4, Seebohm Rowntree Building, University of York, Heslington, York YO10 5DD, UK.
ABSTRACT: Background: Concern has increased about rising rates of cesarean section and other obstetric interventions, and it has been suggested that a change in women’s attitudes may be partly responsible. Our objectives were, first, to examine changes in women’s antenatal willingness to accept obstetric interventions between 1987 and 2000 and, second, to look at the relationship between willingness to accept obstetric interventions and mode of birth. Methods: Data on willingness to accept obstetric interventions were collected at 35–36 weeks of pregnancy using postal questionnaires, and follow-up of women was conducted 6 weeks postnatally. Data are presented for 977 women drawn from 8 maternity units in England who were due to give birth in April to May 2000. To address the first objective, data were compared with the parent study carried out in 1987. Results: The sample had significantly more positive antenatal attitudes toward obstetric interventions than the comparable sample in 1987 (F= 42.25, df= 1, p < 0.001). Willingness to accept obstetric interventions was related to mode of birth. Binary logistic regression controlling for age, education, and parity showed that women with high “willingness to accept intervention” scores had a nearly twofold increase in the odds of an operative or instrumental birth (OR 1.94, 95% CI 1.28–2.95) compared with women who had low scores. These attitudes also predicted epidural analgesia use, and differences in mode of birth were no longer significant when epidural use was included in the regression model. Conclusions: A shift toward greater willingness to accept obstetric interventions appears to have occurred since 1987, and this shift does appear to relate to mode of birth in the 2000 cohort but not in 1987. The findings suggest that epidural analgesia use mediates the link. (BIRTH 34:1 March 2007)
Considerable debate has been generated by rising rates of cesarean section and other obstetric interventions (1,2). Part of the debate is to try and explain why these changes are occurring. One of the aims of the National Sentinel Caesarean Section Audit Report, carried out in the United Kingdom in 2001 (2), was to explore the determinants of birth by cesarean section. In addition to the collection of clinical data, the views of women and obstetricians were surveyed. The audit concluded that the main primary indications for cesarean section as reported by clinicians were presumed fetal compromise (22%), failure to progress in labor (20%), and previous cesarean section (14%). Maternal request, which is increasingly being cited as a reason for the increase in the cesarean section rate (3–5), accounted for 7 percent of births by cesarean section according to the clinicians surveyed (2, p. 17). In the maternal survey, just 5.3 percent of women said that they would prefer a birth by cesarean section, and these were mainly women who had already had a baby by this method (2, p. 95).
Evidently, maternal request, as such, is unlikely to be having a major effect on overall cesarean section rates. However, we can conceptualize “maternal request” as being one end of an attitude spectrum. Women may not necessarily be initiating a request for cesarean sections or other obstetric interventions, but possibly they are more accepting of the suggestion if the question arises. As Anderson suggested, “How women view the care they want to receive in labour and delivery may have changed, moving from the notion of demedicalization that was common in the early 1980s to the increased demand for the use of medical technology found in today’s world” (1, p. 697).
To the best of our knowledge, no such shift in women’s attitudes has been demonstrated, nor data presented on how, if at all, such attitudes relate to mode of birth. The purpose of this article is to present such evidence. In 2000, we carried out a prospective study of women’s expectations and experiences of childbirth (6). The study was a replication of a similar one conducted in 1987 (7,8), and its purpose was to investigate changes in women’s expectations and experiences of maternity care. These data allow us to address the following 2 hypotheses raised by Anderson’s editorial (1): first, women’s willingness to accept interventions has increased since the 1980s and, second, willingness to accept interventions relates to mode of birth.
This prospective study used postal questionnaires that were based on those used in 1987. Two questionnaires were sent antenatally and 1 postnatally. Details of the 1987 methods and sample have been published previously (7,8).
Participating maternity units
Eight maternity units participated, 4 in the south of England that had taken part in the 1987 study and 4 serving similar populations in the north of England. The units ranged in size from 1,600 to 3,300 births per annum. All served semirural areas centered on towns where the major employment was agriculture related, light industry, service industries, or armed services. The Northern & Yorkshire Multi-Centre Research Ethics Committee and all 8 local Research Ethics Committees gave their approval for the study.
Recruitment and data collection
To protect women’s confidentiality, each maternity unit forwarded the introductory letters and questionnaires on our behalf to reach women at 28–29 weeks of pregnancy. Each unit was sent 300 envelopes containing a questionnaire and explanatory letter to forward to a consecutive series of women with an expected date of delivery on or after April 1, 2000. Questionnaires were to be sent to all women irrespective of whether they were planning to give birth at home or in a hospital and irrespective of parity. Women who wished to take part returned the completed questionnaire to the researchers in a prepaid envelope, and thereafter could be sent the second and third questionnaires directly at approximately 35 weeks’ gestation and 6 weeks after the expected date of delivery. The third questionnaires were sent to all women who had returned a valid first questionnaire, irrespective of whether or not a valid second questionnaire had been received.
The first antenatal questionnaire collected primarily demographic information, and the second focused on wants and expectations for the birth. The third questionnaire covered the events of the birth, and subsequent feelings.
The second antenatal questionnaire asked several questions which assessed willingness to accept birth interventions. Seven such questions had been included in both the 1987 and 2000 studies (Fig. 1). They concerned drugs for pain relief, continuous electronic fetal monitoring, acceleration of labor, episiotomy, postterm induction, forceps/ventouse, and elective cesarean section. The generic question about pain relief was used in preference to questions about specific forms of pain relief, although these were also asked, since we were trying to assess an overall attitude. These questions appeared at different points in the questionnaire, had not originally been conceived of as contributing to a scale, and did not necessarily have common response options. Each was therefore recoded onto a scale from 1 to 6, where 1 was the least positive attitude toward accepting the intervention and 6 the most positive. “Neutral” responses such as “I don’t mind” were coded as 4. Scores for the 7 questions were summed to create a “willingness to accept intervention” score with a possible range from 7 to 42. Given that all neutral responses were scored as 4, the “indifference point” on the scale is 7 × 4 = 28. Scores less than this amount represented a desire to avoid intervention and scores above, a positive desire to receive intervention.
The 1987 data set
The 1987 sample had been selected as described above, but a general practitioner unit and a rural satellite unit, both of which catered to low-risk women, had also been included. Women from these 2 units were omitted from the analyses presented here to ensure greater comparability of the samples.
Data analysis and presentation
Data were analyzed using Statistical Package for the Social Sciences, version 14 (9). For univariate and bivariate analyses, chi-squared and analysis of variance (General Linear Models) were used as appropriate. Binary logistic regression was used to investigate the relationships between mode of birth and multiple other variables, with the 2 levels being “instrumental or operative birth” (=1) and “unassisted vaginal birth” (=0). Where appropriate, we have cited values for Nagelkerke pseudo R2, which is the binary logistic regression equivalent of the R2 that is produced in standard multiple regression and gives an estimate of the proportion of variation accounted for by the model.
Comparisons between 1987 and 2000 data were made for the whole sample, and also checked by limiting analysis to the southern units (because the northern units were not sampled in 1987). Unless otherwise stated, all data presented are from the 2000 cohort. All logistic regression models were rerun in Stata (Release 9) (10), estimating robust standard errors to account for the clustering associated with women being sampled from 8 distinct maternity units. Results were not materially different and no inferences were changed.
The responses for the questionnaires were as follows: 1,432 first questionnaires were returned (60%), 1,272 second questionnaires (91%), and 1,286 postnatal questionnaires (92%). A total of 1,278 women reported their mode of birth, of whom 9.1 percent (n = 116) had a planned cesarean section, 11.6 percent (n = 148) had an unplanned cesarean section, 12.2 percent (n = 156) had an instrumental birth, and 67.1 percent (n = 858) had an unassisted vaginal birth.
This article is limited to women who were booked to give birth at 1 of the 8 consultant units and actually did so, and who returned all the questionnaires. In addition, it excludes any woman who had a planned cesarean section or who antenatally thought that she “probably” or “definitely” would have a cesarean section since the antenatal attitudes and expectations of such women are likely to be different. This adjustment left a sample of 977 women to be compared with 512 women meeting the same criteria from the 1987 sample, although the number for specific variables may be slightly less due to missing data.
Sample characteristics are shown in Table 1. In keeping with national statistics (11), women were older and had more years in full-time education than was the case in 1987. The samples did not differ in the proportion of first-time mothers (nullipara) who were included (44.3% compared with 46.2%). Education in the 2000 data set was coded in 4 categories in terms of highest qualification (“none”; “GCSE” [General Certificate of Secondary Education—exams usually taken in England at 16 yr] or equivalent; “A levels” [exams usually taken in England at 18 yr] or equivalent; and “degree”). These data were not available for the 1987 sample.
Table 1. Sample Characteristics
Year 2000(n= 977) No. (%)
Year 1987(n= 512) No. (%)
p < 0.001.
GCSEs are the exams usually taken in England at age 16 yr; A levels are the exams usually taken in England at age 18 yr.
Willingness to accept intervention scores could be calculated for 942 women. The possible range was 7–42, the actual range was 7–40, and the mean was 22.80 (SD 5.88). The latter was significantly higher than the mean score for the equivalent women in the 1987 sample (n = 494): 20.66 (SD 5.95) (F = 42.25, df = 1, p < 0.001) (Table 2).
Table 2. Willingness to Accept Intervention Scores in 2000 and 1987
95% CI for mean
Controlling for age and parity did not affect the result (data not shown). To check that the finding was not an artefact of differences in the sampling frames, the analysis was rerun limited to the 4 consultant units in the south of England that participated in both studies. This calculation yielded a similar result, although somewhat less significant: mean 22.35 (SD 6.05, n = 426) compared with the 1987 sample: mean 20.66 (SD 5.95, n = 494) (F = 17.98, df = 1, p < 0.01).
Finally, the analysis was rerun excluding any woman who had been told antenatally that her birth might not be straightforward. This calculation also made only marginal differences to the mean scores (data not shown).
Willingness to accept intervention and demographic variables
Willingness to accept intervention was strongly related to level of education (F = 9.58, df = 3, p < 0.001). The mean score for women with no educational qualifications was 24.33 (SD 5.34) compared with 21.44 (SD 5.98) for those with degrees, with the other groups falling in between. A negative correlation with age (r =−0.115, p < 0.01) was also observed; that is, younger women were more willing to accept interventions. Differences between nulliparas and women who had given birth before, although consistent with the age relationship, were not significant (data not shown). Since these demographic variables are interrelated, they are all included as covariates in subsequent multivariate analyses, with first birth coded yes/no, age in years, and highest educational qualification at 4 levels as described above.
Willingness to accept intervention and mode of birth
Analysis of variance indicated significant differences in the mean willingness to accept intervention scores of women with different modes of birth (F = 5.91, df = 2, p < 0.01). Women who had unassisted vaginal births had the lowest scores (mean 22.45, SD 5.77) and women who had instrumental deliveries had the highest scores (mean 24.28, SD 6.02), with women who had an unplanned cesarean section falling in between the 2 scores (mean 23.33, SD 6.15).
For ease of presentation, the sample was divided into 3 groups on the basis of their willingness to accept intervention scores: low (scoring <20, n = 291); medium (scoring 20–25, n = 339); and high (scoring >25, n = 310). Scores were approximately normally distributed and cutoffs were chosen to give groups of approximately equal sizes. Figure 2 shows the percentage of women in each of these 3 groups who had an operative or instrumental birth, and demonstrates that 30.6 percent of those with high willingness to accept intervention scores had an operative or instrumental birth compared with only 20.6 percent in the group who were least willing to accept intervention.
Binary logistic regression controlling for age, education, and parity confirmed that those with high scores had an almost twofold increase in the odds of an operative or instrumental birth compared with those who had low scores (OR = 1.94, 95% CI 1.28–2.95) (Table 3). Table 3 shows that parity and age were also both significantly related to the odds of an operative or instrumental birth. The Nagelkerke pseudo R2 for the model was 0.238; that is, 23.8 of the variance in mode of birth could be accounted for by these prelabor variables. Testing the same logistic regression model on the 1987 data showed no relationship between mode of birth and attitude to intervention scores and a much weaker relationship with age (data not shown).
Table 3. Odds Ratio for Operative or Instrumental Birth by Willingness to Accept Intervention Score in 3 Groups, with First Birth, Age, and Education
95% CI for OR
Willingness to accept intervention score
Exploring the link between antenatal attitudes and mode of birth
Our findings have demonstrated that willingness to accept intervention scores were significantly related to mode of birth. To explore possible mechanisms for this link, the logistic regression model was expanded to include intrapartum variables known to be related to mode of birth: induction of labor, acceleration of labor, and use of epidural analgesia. As Table 4 shows, when these variables are included, the willingness to accept intervention score no longer makes a significant contribution to the model. Instead, the major significant variable, apart from parity, is epidural analgesia use. Compared with women who did not have an epidural, women who did had 5.93 times greater odds of an operative or instrumental birth (95% CI 3.88–9.05), controlling for parity, age, education, induction, acceleration of labor, and antenatal willingness to accept intervention.
Table 4. Odds Ratio for Operative or Instrumental Birth by Willingness to Accept Intervention Score in 3 Groups, with First Birth, Age, Education, Induction, Acceleration of Labor, and Epidural Analgesia Use
95% CI for OR
Willingness to accept intervention score
The fact that antenatal attitudes are no longer significant when intrapartum variables are included suggests that the link with mode of birth is by means of these other variables. This hypothesis was investigated by entering age, education, parity, and willingness to accept intervention in separate logistic regression models to predict induction of labor, acceleration of labor, and use of epidural analgesia, respectively.
Prediction of induction of labor
Willingness to accept intervention was not related to the likelihood of induction of labor, controlling for parity, age, and education. Only parity was a significant predictor, with nulliparous women having 1.8 times greater odds of labor being induced compared with parous women: OR 1.82, (95% CI 1.34–2.47).
Prediction of acceleration of labor
Willingness to accept intervention was not related to the likelihood of acceleration of labor, controlling for parity, age, and education. As with induction, the only significant predictor of acceleration was parity, with nulliparous women having a fourfold increase in the odds of labor being accelerated compared with parous women: OR 4.09 (95% CI 2.95–5.67).
Prediction of epidural analgesia use
Willingness to accept intervention score was a significant predictor of epidural analgesia use, with high scorers having odds of epidural use 2.5 times higher than low scorers: OR 2.51 (95% CI 1.73–3.64), controlling for parity, age, and education. Parity was also significant with nulliparous women having a more than fivefold increase in the odds of using epidural analgesia compared with parous women: OR 5.13 (95% CI 3. 57–7.01). The Nagelkerke pseudo R2 for the model was 0.224; that is, 22.4 percent of the variance in epidural use could be accounted for by these prelabor variables.
Using data collected under comparable circumstances in 1987 and 2000, this investigation has been able to support the hypothesis that childbearing women’s willingness to accept birth technology has increased since the 1980s. The mean age of childbearing women and their number of years of full-time education have both increased since 1987. Since it was found that well-educated and older women were less willing to accept intervention, one might have expected any change over time to be cancelled out, but this was not the case.
The second hypothesis, that willingness to accept intervention was related to mode of birth, also received support. Women who had high willingness to accept intervention scores had a nearly twofold increase in odds of an operative or instrumental birth compared with women with low scores, controlling for parity, age, and education. However, antenatal attitudes did not remain significant predictors once known intrapartum correlates of operative and instrumental birth were included in the model. It was postulated that this finding was because these interventions were the link between attitudes and mode of birth, but no relationship was found between the willingness to accept intervention score and experiencing either induction or acceleration of labor. However, willingness to accept intervention score was a significant predictor of epidural analgesia use, and epidural use was strongly related to mode of birth. Compared with women who did not have an epidural, those who did have one had 5.93 times greater odds of an operative or instrumental birth (95% CI 3.88–9.05), controlling for parity, age, education, induction, acceleration of labor, and antenatal willingness to accept intervention. This finding is in keeping with numerous other studies that have shown use of epidural analgesia to be associated with higher rates of instrumental birth, although debate continues about the relationship with unplanned cesarean section (12–16). We must, of course, be cautious in attributing causality because some epidurals will be a response to a labor that is not progressing normally, rather than a cause. However, the strength of our data lies in its prospective design. Women’s attitudes were assessed in late pregnancy so these attitudes clearly predate the events of labor. We have shown that these attitudes were a significant predictor of epidural use. Thus, although some epidural use is undoubtedly a response to intrapartum events, it cannot be a complete explanation.
We have shown that women who said antenatally that they were willing to accept interventions were indeed more likely to do so, at least in the case of epidural analgesia and instrumental or operative birth. This finding may seem predictable, but it is intriguing that no such relationship was found within the 1987 data. One possible interpretation is that the ethos of women’s choice that is now espoused in the United Kingdom after publication of Changing Childbirth (17) has resulted in staff being more willing to intervene for a woman who wants interventions, even if they do not consider it clinically necessary. Given these differences between 1987 and 2000, we must, of course, be cautious in generalizing the findings to other settings. It would indeed be of interest to explore these kinds of relationships in other contexts where the caregivers’ ethos may be very different.
Another interesting aspect of the data was the persistent significance of parity and age. These variables are interlinked, in fact, not only because parous women are necessarily older on average but also because we have subsequently found that the increasing odds of an operative or instrumental birth with increasing age are limited to nulliparous women. This relationship was not present in the 1987 data. These issues will be explored in a subsequent article (in preparation).
Surprisingly little research has been conducted on women’s attitudes to obstetric intervention and even less that has related antenatal attitudes to birth outcomes. A booking clinic survey in Dublin (18) suggested a general desire for quick and easy labors, although a high level of agreement with the statements “I want to avoid a caesarean section in labour” and “I want to avoid a forceps or vacuum delivery” was also present. No data on intrapartum events were presented. Goldberg et al considered nulliparas’ antenatal preference for epidurals and demonstrated that, as in the present study, those who wished to use them were more likely to do so than those who did not (19). Mode of birth was not reported. Two further small studies have looked, postnatally, at the characteristics of women who used epidurals, both reporting differences in locus of control compared with women who did not use epidurals (20,21). We had anticipated that willingness to use an epidural might indeed be a marker for different attitudes toward the process of birth, which was why we initially looked at birth outcomes in terms of antenatal attitudes rather than actual use, and also why we devised the composite score. However, as we saw, once actual epidural use was included in the model, antenatal attitude to epidural use made relatively little contribution to the odds of an operative or instrumental birth.
This investigation lends support to Anderson’s suggestions (1) that women’s attitudes to interventions have become more favorable and that willingness to accept interventions is contributing to the decline in the rate of unassisted vaginal births. Epidural analgesia use would seem to be a mediating factor, especially since it is much more a result of individual choice than other intrapartum interventions. On the basis of these data, it would appear that having a negative antenatal attitude toward birth interventions has a highly protective effect. If health practitioners wish to stem the decline in rates of unassisted vaginal births, it may well be that encouraging a nontechnological ethos, and associated coping strategies, would be effective (22). Women who favor epidurals may be unaware of the disadvantages associated with epidural analgesia use, including doubling their odds of an assisted or operative birth. Greater awareness of these associations may temper women’s enthusiasm, especially if they have the support and encouragement to pursue other strategies.
We give our thanks to Sue Easton, who oversaw initial data coding and entry. The idea of a “willingness to accept intervention” scale was first suggested by colleagues from Mother & Child Health Research at La Trobe University, Melbourne, Australia, when one of the authors (J.M.G.) was visiting in 2002. Kate Pickett advised on multivariate analysis, and she and Mary Renfrew both offered critical comments on earlier drafts of this article.