Does fear of childbirth during pregnancy predict emergency caesarean section?

Authors

  • Rebecca Johnson,

    1. Child, Adolescent and Family Therapy Service, Edmund Street Clinic, Chesterfield, North Derbyshire, UK
    2. Clinical Psychology Unit, Department of Psychology, Sheffield University, Sheffield, UK
    Search for more papers by this author
  • Pauline Slade

    Corresponding author
    1. Clinical Psychology Unit, Department of Psychology, Sheffield University, Sheffield, UK
      * Dr P. Slade, Clinical Psychology Unit, Department of Psychology, Sheffield University, Sheffield, UK.
    Search for more papers by this author

* Dr P. Slade, Clinical Psychology Unit, Department of Psychology, Sheffield University, Sheffield, UK.

Abstract

Objective Caesarean section rates are rising dramatically in the UK. It has been estimated that they have increased from 10% to 22% of all births over 15 years. A Swedish study has suggested that fear of childbirth during pregnancy may increase the risk of emergency caesarean section. The aim of this study is to identify whether fear of childbirth can predict the occurrence of emergency caesarean section in a UK sample.

Design A prospective design using between-group comparisons.

Setting Sheffield, S. Yorkshire, UK.

Sample Four hundred and forty-three pregnant women, recruited at 32 weeks of gestation, over 16 years of age.

Methods Participants completed self-assessment, postal questionnaires assessing fear of labour and anxiety using the Wijma Delivery Expectancy Scale (W-DEQ) and the Speilberger State Trait Anxiety Scale (STAI), together with their expectations about their mode of delivery. Delivery information was gathered via birth summary sheets.

Main outcome measure Mode of delivery.

Results Emergency caesarean section was associated with previous caesarean section, parity, age and a score reflecting medical risk, but not fear of childbirth or anxiety measures. There were no differences in fear between women experiencing spontaneous-vertex, forceps/ventouse, emergency or elective caesarean deliveries. The W-DEQ was factor analysed and was found to measure four distinct domains: fear, lack of positive anticipation and the degree to which women anticipate isolation and riskiness in childbirth. However, these individual factors also failed to contribute to the prediction of mode of delivery. Primiparous women in the UK sample showed highly elevated fear scores when compared with a Swedish sample. Such discrepancies were not found for the multiparous sample.

Conclusions Fear of childbirth during the third trimester is not associated with mode of delivery in a UK sample. Possible cross-cultural differences are discussed.

Introduction

It has been hypothesised that women's fear of childbirth during the third trimester of pregnancy, if significantly high, may result in obstetric complications, negative delivery experience and/or an increased risk of emergency caesarean section1–3. This hypothesis was developed from the suggestion that maternal anxiety during pregnancy is related to complications during labour, about which there have been a number of inconsistent research findings4–7.

Evidence for a relationship between anxiety and obstetric problems has been demonstrated by studies using composite measures of complications8, and measures of psychological risk combined with medical risk factors9. Studies considering various obstetric outcomes separately have also demonstrated associations between anxiety during pregnancy and, for example, length of labour10,11; pre-eclampsia, prolonged labour and forceps delivery12; use of anaesthesia during delivery13; and preterm delivery14. Supporting evidence is also provided by research into the relationship between anxiety and levels of adrenaline and noradrenaline, which are known to be associated with abnormal uterine contractions and subsequent obstetric complications10,11,15,16.

However, other researchers have found that anxiety was unrelated to obstetric complications17–21, and was only weakly related to use of analgesia in the second stage of labour22. The inconsistency between studies can be accounted for by several factors. Firstly, the research tends not to take account of the possibility that the nature of the relationship between anxiety and obstetric complications may be characterised by specificity (i.e. specific dimensions of anxiety may interact with specific obstetric outcomes). Therefore, studies considering broad definitions of either variable may produce inconsistent results. Secondly, efforts to evaluate the hypothesis have been hampered by the use of flawed methodological designs. Researchers have measured obstetric complications inconsistently, for example, by collapsing a wide range of outcome variables into one score, or by failing to utilise sufficient sample sizes for sufficient power. Results are also limited by a failure to control for confounding variables, such as medical risk or women's knowledge of the possibility of complications. Finally, the overly broad conceptualisation of anxiety, and variation in the timing of its measurement, may contribute to the inconsistency across findings.

Some attempts at greater specificity have been made. Lederman et al.11, for example, sought to determine the specific dimensions of anxiety (i.e. anxiety about coping, safety or pain), which may be associated with prolonged labour or fetal heart rate deceleration. Other researchers have considered the impact of a specific dimension of anxiety associated with childbirth, fear of childbirth, and have designed an instrument for its measurement; the Wijma Delivery Expectancy/Experience Questionnaire (W-DEQ)23.

It has been hypothesised that fear of childbirth implies an increased risk of intrapartum complications, such as prolonged labour or fetal asphyxia, which may often be concluded by an emergency caesarean section3. In a Swedish sample, the scores of 97 women who were delivered by emergency caesarean section were compared with 194 controls selected from participants with a range of deliveries (excluding elective caesarean section deliveries) who were matched for age and parity. Three questionnaires (W-DEQ, State Trait Anxiety Inventory [STAI] and Stress Coping Inventory), were completed by women during their pregnancies, at 32 weeks of gestation3. A significant difference was found between the groups on all three measures, but particularly the W-DEQ, leading the authors to conclude that it is possible that fear of childbirth may increase the risk of subsequent emergency caesarean section.

It is not known whether the finding that high fear of labour may increase the risk of emergency caesarean section is culturally specific or can be generalised to a UK sample. The rates of emergency caesarean section in the local UK service, for example, are higher than in Linköping, Sweden, where previous studies have been carried out1–3. In addition, it is not known whether there are specific areas of focus of women's fear which will predict an increased risk of emergency caesarean section.

There are a number of reasons why further research to investigate the hypothesis that high fear of labour is related to obstetric complications, and in particular, an increased risk of emergency caesarean section, is important. Firstly, in addition to increasing the risk of having a negative delivery experience, antenatal fear of childbirth has been found to be associated with severe emotional imbalance postnatally2 and with potential for negative implications for the relationship between mother and child. It has also been argued that emergency caesarean section should be regarded as a pointer with respect to possible postnatal post-traumatic stress3,24. Secondly, there seems little contention that delivery by emergency caesarean section is potentially stressful with greater morbidity compared with spontaneous vaginal delivery. It is also financially costly to services.

Caesarean section rates are rising dramatically in the UK, being described by one author as ‘epidemic’25. In the early 1980s, for example, the rate of caesarean section in Britain was 10%, increasing to 16% in 199625, and 22% in 2000–200126. In many UK units, the rate has reached 25%. The suggestion that psychological factors might be associated with pregnancy outcome is of interest because it underpins the prospect that clinicians may be able to identify risk for labour complications via antenatal psychological screening, thereby potentially reducing the incidence of such complications through psychological intervention.

The aims of the present study were to consider whether women who subsequently undergo emergency caesarean section show higher levels of fear of childbirth during the third trimester than those who do not. Whether the specific measure of fear, the W-DEQ scores, can improve the identification of women at risk of emergency caesarean section, which is currently based on variables such as parity, previous caesarean section, medical risk and age, was also considered. In order to address methodological limitations evident in previous research, the study addressed whether fear accounts for the variance once women's knowledge of their risk, a potential confounding factor, is taken into account.

Hypothesis 1: Women who have emergency caesarean sections will have higher fear of labour (W-DEQ scores) at 32 weeks of gestation, than those who have spontaneous-vertex deliveries.

Hypothesis 2: W-DEQ scores will act as a significant predictor for emergency caesarean section when combined with known predictors, such as parity, age, diabetes and previous caesarean section.

Hypothesis 3: There will be a low positive correlation between scores on the W-DEQ, which has been designed and validated for a pregnant population, and a general anxiety measuring instrument (STAI)27.

Ethical approval to conduct the study was obtained from the North Sheffield Research Ethics committee, and reciprocal approval was given by the South Sheffield Research Ethics committee.

Method

Questionnaire booklets were sent to all pregnant women over 16 years of age, registered at one of two hospital sites in Sheffield, over a four-month period. All were contacted at 32 weeks of gestation, based on expected date of delivery from booking records. This time point was selected because evidence suggests that the focus of women's anxiety tends to shift away from the pregnancy and fetal wellbeing towards the experience of labour during this period24. Care records were checked to ensure the pregnancy was ongoing. No other inclusion or exclusion criteria were applied. A sample of 1200 women was sent an invitation from the clinical and nursing directors to take part in the study.

Although the rates of emergency caesarean section are lower in multiparous women, both nulliparous and multiparous women were included in the study to account for the possibility that fear of childbirth may be related to previous experience of labour.

A total of 443 women returned completed questionnaires. Delivery information was unavailable for 19 participants due to delivery either elsewhere or outside the data collection period. A usable sample of 424 remained, which gave a response rate of 35%.

The proportion of the sample expecting their first child (nulliparous) was 43.4%, while 56.6% had previous children (multiparous). This ratio is consistent with the 2793 women delivering in the two hospitals during the previous year (1999), of whom 43.0% were nulliparous. The average age of the total sample was 29.52 years, ranging between 16 and 43 years. Of the 424 participants, 11.3% (n= 48) deliveries were emergency caesarean sections. The frequency of emergency caesarean sections within the service during the previous year was 13.9%.

To obtain 80% power with 5% significance, using a two-tailed test, 48 participants were required for two group comparisons28, assuming a clinically significant difference is set at 10 points on the W-DEQ. This difference is based upon findings in the Swedish paper by Ryding et al.3. Five to 10 cases per predictor variable are required for logistic regression29, therefore the sample size in the present study was also adequate for a valid logistic regression analysis.

Participants were sent an invitation from the clinical and nursing directors to take part in the study, information and consent forms, and a questionnaire booklet covering demographic details, medical risk factors (see below), the woman's current perception regarding a caesarean and measures of fear (W-DEQ) and state/trait anxiety. The researchers were not aware of any names and addresses until the respondents themselves gave such information on returning the questionnaires. As part of the ethical approval access to information contained in the questionnaires, or whether women had agreed to participate, was not available to those clinically responsible for the women during pregnancy and labour. As a result, it was not possible to send reminders to non-respondents. Each participant gave consent for the researcher to receive a copy of her birth summary sheet. This forms part of normal hospital records about the delivery and includes information about the infant birthweight, augmentation of labour, type of delivery and indication, length of labour, pain relief and Apgar scores.

The W-DEQ is a 33-item questionnaire designed to measure fear of childbirth by means of women's cognitive appraisals regarding the delivery during pregnancy23. Internal reliability in the present study was high (Cronbach's α= 0.91), and compares favourably with reliability estimates calculated for the original Swedish version of the W-DEQ23 (Cronbach's α= 0.93).

The STAI27 is one of the most widely used measures of subjective anxiety30. It is a 40-item questionnaire separated into scales of state anxiety, considered to be transient and situation specific, and trait anxiety which is a more stable, personality-based construct. The minimum and maximum scores for each scale are 20 and 80, respectively.

The study employs a between-subjects design with one outcome variable: mode of delivery, which has four levels: emergency caesarean section; spontaneous-vertex delivery; assisted vaginal delivery (i.e. forceps/ventouse); and elective caesarean section. The study has two explanatory variables: fear of childbirth and state anxiety.

Information was gathered from demographic information sheets in the questionnaire booklets on three variables that are known risk factors for emergency caesarean section (age, parity and previous caesarean section). A fourth variable represents medical risk, and is a composite score with participants scoring one point for each of the following: multiple birth; breech or malposition; diabetes; and induction of labour for reasons other than postdates. For the purposes of the analysis, however, responses were divided into dichotomous categories as only three participants scored more than one point, and none more than two. The ‘yes’ category therefore included all responses of either one or two points. This information was gathered from the birth summary sheets, with the exception of diabetes, which was gathered via the questionnaires.

Participants were also asked whether they had any reasons for expecting a caesarean section at the time of completing the questionnaires. Again responses were divided into dichotomous categories, the ‘yes’ category included responses indicating any degree of possibility given by participants, and this formed the fifth variable (i.e. women's own perception of risk of caesarean section measured at 32 weeks of pregnancy).

The data were subjected to independent t tests to explore differences in fear and anxiety scores between groups defined by parity and awareness of complications. A one-way between-subjects analysis of variance (ANOVA) was used to look at differences between groups, defined by mode of delivery, on continuous variables such as W-DEQ and anxiety scores, and χ2 tests on categorical variables, such as parity, medical risk, previous caesarean section and expectation of a caesarean. A matched samples t test was used to control for age and parity. Principal components analysis (PCA) was conducted to explore whether particular subcategories of fear are related to emergency caesarean section. Factors identified in the PCA were then subjected to further analysis of variance. Finally, sequential logistic regression analysis was used to assess the additional contribution of W-DEQ scores to the prediction of emergency caesarean section, after known predictors.

Results

Comparison of demographic and obstetric characteristics of participants and the total population of women who delivered within the service in 1999 is shown in Table 1, showing reasonable equivalence of the sample group for these factors.

Table 1.  Demographic and obstetric information about the study sample and the hospital population during the previous year. Values are given as n (%).
 Full sample (n= 424)Hospital population during 1999 (n= 2793)
Age
<2020 (4.8)217 (7.7)
20–2454 (12.8)610 (21.8)
25–29122 (28.8)832 (29.8)
30–34152 (35.8)766 (27.4)
35–3967 (15.9)295 (10.6)
>407 (1.7)73 (2.6)
Parity
Nulliparous184 (43.4)1201 (43.0)
Mode of delivery
Spontaneous-vertex257 (60.6)1832 (65.6)
Forceps28 (6.6)138 (4.9)
Ventouse41 (9.7)245 (8.8)
Emergency caesarean section48 (11.3)389 (13.9)
Elective caesarean section47 (11.0)157 (5.6)
Breech3 (0.7)32 (1.1)

For analyses involving W-DEQ scores, excluding those using abbreviated W-DEQ scores following the PCA, 25 cases with missing values were deleted, leaving 48 cases in the emergency caesarean section group, 243 in the spontaneous-vertex group, 58 in the forceps/ventouse group and 47 in the elective caesarean section group, giving a total of 396 cases available for analysis (breech deliveries are not included in analyses). For all other analyses, the numbers in each group are as shown in Table 1, giving a total of 421.

Indications for emergency caesarean section are shown in Table 2. Suspected fetal distress and poor or no progress in labour accounted for 64.6% of the indications recorded.

Table 2.  Indications for emergency caesarean section within present study (n= 48), as recorded on participants' birth summary sheets.
Indicationn (%)
Suspected fetal distress21 (43.8)
Poor/no progress10 (20.8)
Malposition4 (8.3)
Breech presentation2 (4.2)
Abruption2 (4.2)
Severe pre-eclampsia2 (4.2)
Failed trial of forceps/ventouse1 (2.1)
Antepartum haemorrhage1 (2.1)
Booked elective caesarean section but started labour1 (2.1)
Fetal tachycardia1 (2.1)
Two or more of the above1 (2.1)
Not recorded2 (4.2)
Total48 (100)

The W-DEQ scores were normally distributed with the mean scores for nulliparous women higher in the present study, 65.41 (17.49), than in the Swedish study3, 54.1 (21.1) (P < 0.0001). Mean scores for multiparous women, however, were not different, 58.01 (20.09) (UK sample), and 57.7 (22.1) (Swedish sample). In the UK sample, nulliparous women showed significantly higher levels of fear than multiparous women (P < 0.0001).

An independent samples t test showed no difference between fear levels of women who were aware of the possibility of complications that may lead to a caesarean section and those who were not (P > 0.6).

There was a positive correlation between the W-DEQ and STAI scores (r= 0.5, P < 0.01). Only 25% of the variance was predicted by the STAI, indicating that the W-DEQ measures within the domain of anxiety, with sufficient variance left for the measurement of another dimension.

Groups defined by mode of delivery were compared with respect to W-DEQ and anxiety scores.

Three one-way ANOVAs, with mode of delivery as the independent variable, were conducted on the W-DEQ, State anxiety, and Trait anxiety scores, (see Table 3). After deletion of 25 cases with missing values, in the spontaneous-vertex and forceps/ventouse groups, 396 cases were available for analysis. The hypothesis that women who subsequently delivered by emergency caesarean section would have greater levels of fear of childbirth than controls was not supported. There were also no differences in state or trait anxiety between women who delivered by spontaneous-vertex delivery, forceps/ventouse or emergency or elective caesarean section. Indeed, the similarity in means of these measures is striking.

Table 3.  Group sum scores in mean (SD) on W-DEQ, State Anxiety (STAI Y-1), and Trait Anxiety (STAI Y-2) at 32 weeks of gestation of women subsequently delivered by emergency caesarean section, spontaneous-vertex delivery, forceps/ventouse delivery and elective caesarean section.
 Emergency caesarean section (n= 48)Spontaneous-vertex (n= 243)Forceps/ventouse (n= 58)Elective caesarean section (n= 47)FdfP
W-DEQ60.89 (20.11)60.19 (19.51)64.08 (18.53)62.81 (19.56)0.743>0.5
State anxiety40.64 (10.35)40.94 (10.96)41.88 (10.59)41.87 (12.66)0.233>0.9
Trait anxiety39.58 (9.66)39.12 (10.06)38.94 (9.82)37.59 (9.38)0.373>0.8

As the main purpose of the study was to explore differences specifically between the emergency caesarean section group and a control group who delivered spontaneously (spontaneous-vertex), and also in order to facilitate clarity of interpretation, subsequent analyses (excluding PCA) were conducted on these two groups of interest.

Table 3 has shown that there were no differences between the groups on fear or anxiety measures. There were, however, significant differences in parity, medical risk, previous caesarean section and expectation of a caesarean between women who delivered by emergency caesarean section and those with spontaneous-vertex deliveries (see Table 4).

Table 4.  Differences between women who delivered by emergency caesarean section and a control group of women with spontaneous-vertex deliveries (unmatched), on categorical variables. Values are given as n (%).
VariableEmergency caesarean section (n= 48)Spontaneous-vertex (n= 257)Odds ratio95% CI
  1. *P < 0.01

  2. **P < 0.0001

Reason to expect a caesarean section11 (22.9)19 (7.4)3.721.64–8.45
Medical risk14 (29.2)31 (12.1)3.01.45–6.21
Previous caesarean section10 (20.8)13 (5.1)4.92.02–12.06
Nulliparous30 (62.5)84 (32.7)3.41.81–6.51

As parity was significantly different within the emergency caesarean section group and the spontaneous-vertex group, it was decided to carry out a matched samples t test to control for age and parity. Women from the emergency caesarean section group were matched with the first available matching participant from the spontaneous-vertex group, in order to ensure random selection. Identity number, age and parity were the only information available to the researcher during the matching. A related samples t test also showed that the hypothesis that W-DEQ scores would be higher in the case group than controls was not supported at the 5% level of significance (P > 0.05).

The analyses so far clearly show that fear of childbirth, measured by the W-DEQ as a whole, is not associated with mode of delivery. However, in order to explore further whether particular aspects of fear might be related to caesarean section, it was necessary to consider whether subcategories of fear exist within the W-DEQ. PCA was therefore conducted on the 33 items from the W-DEQ.

PCA with orthogonal rotation (varimax method) yielded eight factors with eigenvalues greater than 1.0 (Kaiser's criterion), which accounted for 63.6% of the variance. In view of the conceptual clarity of the resulting components, and following examination of the scree plot, four factors were extracted, accounting for 49.4% of the variance. Loadings of variables on factors, communalities and percentage of variance are shown in Table 5. Variables are ordered and grouped by size of loading to facilitate interpretation, and interpretative labels are suggested for each factor. These are fear, lack of positive anticipation, isolation and riskiness.

Table 5.  Factor loadings, communalities (h2) and percent of variance for principal factors extraction with orthogonal varimax rotation on W-DEQ items.
Item F1*F2*F3*F4*h2
  1. *Factor labels: F1= Fear; F2= Lack of positive anticipation; F3= Isolation; F4= Riskiness.

(6)Afraid0.77<0.0020.140.200.66
(19)Panic0.76<−0.010.200.160.64
(17)Relaxed0.720.30<−0.0040.150.64
(12)Tense0.70<−0.0020.150.190.55
(5)Confident0.690.30<−0.030.250.62
(2)Frightful0.670.110.110.200.52
(16)Composed0.630.28<−0.06<−0.010.48
(8)Weak0.610.140.29<−0.050.48
(22)Self-confident0.600.47<0.06<0.090.60
(24)Pain0.58<−0.01<−0.07−0.240.40
(20)Hopelessness0.570.110.350.200.49
(25)Behave badly0.53<−0.0030.30−0.150.40
(4)Strong0.520.34<0.080.250.45
(27)Lose control0.46<−0.010.30−0.120.32
(10)Independent0.450.34−0.110.210.38
(14)Proud<0.010.770.22<−0.040.59
(17)Happy0.130.740.17<−0.020.59
(13)Glad<0.060.730.22<−0.020.59
(21)Longing for child<−0.090.580.16<−0.060.38
(1)Fantastic0.280.57<−0.070.210.45
(29)Natural0.140.56<0.060.190.37
(30)Obvious<0.090.470.11<0.04(0.25)
(22)Trust0.180.400.300.290.37
(15)Abandoned<0.030.230.83<0.080.75
(7)Deserted0.110.220.82<0.090.74
(3)Lonely0.210.200.76<0.050.67
(11)Desolate0.280.150.670.190.58
(33)Child injured0.11<0.01<0.020.830.69
(32)Child die<0.04<−0.04<0.020.810.66
(31)Dangerous0.130.160.180.490.31
(8)Safe0.220.380.150.430.39
(28)Funny<0.020.30−0.18<−0.005(0.12)
(26)Let happen<0.090.29<0.06<0.07(0.11)
Percent of variance19.0312.999.857.48 

Cronbach's α reliability estimates for the four factors were all acceptable: F1 (fear) α= 0.9; F2 (lack of positive anticipation) α= 0.81; F3 (isolation) α= 0.68; F4 (riskiness) α= 0.85. Three items had communalities less than 0.3, shown in parentheses in Table 5, and either loaded on none of the factors (items 26: ‘let happen’, and 28: ‘funny’) or had a low loading on one factor (item 30: ‘obvious’). It was therefore decided to exclude these items in subsequent analyses, thereby creating an abbreviated version of the W-DEQ, for which Cronbach's α= 0.91. Twenty-five cases previously deleted for missing values were now available for inclusion in subsequent analyses, indicating that the missing values were for these three items.

Having identified four factors within the W-DEQ, the data were analysed using one-way ANOVAs for differences between the delivery groups on factor scores (see Table 6). There were no differences between the four modes of delivery on either the abbreviated W-DEQ or the four factors.

Table 6.  Mean W-DEQ and factor scores (standard deviations) for participants in four delivery groups.
 Emergency caesarean section (n= 48)Spontaneous-vertex (n= 257)Forceps/ventouse (n= 69)Elective caesarean section (n= 47)P
Abbreviated W-DEQ54.60 (19.76)52.86 (19.35)57.57 (17.77)55.32 (17.93)>0.1
Factor 1 (Fear)36.69 (12.51)35.58 (12.34)29.90 (11.63)36.77 (11.53)>0.05
Factor 2 (Lack of positive anticipation)9.69 (5.90)6.39 (5.46)9.32 (5.39)10.13 (5.30)>0.5
Factor 3 (Isolation)4.77 (2.98)4.75 (3.65)4.80 (3.54)5.60 (3.83)>0.5
Factor 4 (Riskiness)3.46 (3.25)3.18 (3.54)3.55 (3.93)2.83 (3.09)>0.5

While it has been shown that levels of fear of childbirth do not vary significantly between women who subsequently experience different types of delivery, within the present sample, one of the main questions we wished to address was whether fear of childbirth, when considered in combination with other predictor variables, can improve the prediction of mode of delivery. A sequential logistic regression analysis was therefore performed to assess prediction of mode of delivery (spontaneous-vertex or emergency caesarean section), first on the basis of five predictors: parity; previous caesarean section; medical risk; reason to expect a caesarean section; and age, and then after the addition of a sixth variable, fear of childbirth (abbreviated W-DEQ scores). After deletion of four cases with missing values, data from 301 women were available for analysis: 253 women delivered spontaneously, 48 by emergency caesarean section. Table 7 shows the contribution of the individual predictors to the model.

Table 7.  Logistic regression analysis of mode of delivery as a function of parity, medical risk, age, previous caesarean section and fear of childbirth (W-DEQ scores).
VariablesOdds ratio (95% CI)
Medical risk2.48 (1.12–5.52)
Nulliparity9.11 (3.78–21.96)
Previous caesarean section9.94 (2.83–34.93)
Reason to expect a caesarean section1.95 (0.84–4.52)
Age1.09 (1.02–1.17)
Fear of childbirth (W-DEQ scores)1.00 (0.98–1.01)

The odds of having an emergency caesarean section are 9.11 times higher for nulliparous women, 9.94 times higher for women who have had a previous caesarean section and 2.48 times higher for women identified as having a medical risk (even taking other factors into account), while age and fear of childbirth scores do not show significant impact (the odds of having an emergency caesarean section are almost equal—1.09—for different ages and are exactly equal—1.0—for fear scores).

There was a good model fit on the basis of the five predictors alone, P < 0.0001, indicating that the first five predictors, as a set (as above), reliably distinguished between women who delivered by spontaneous-vertex delivery and those who required an emergency caesarean section. After addition of the fear of childbirth variable, significance was unchanged. Comparison of log-likelihood ratios for models with and without the additional variable showed no significant improvement with the addition of W-DEQ scores as a predictor, P > 0.69.

Discussion

In this UK sample of women, fear of childbirth in the third trimester was not related to the type of delivery they subsequently experienced. Known predictors (i.e. previous caesarean section, age, parity and medical risk), however, were shown to be associated with emergency caesarean section. These findings are at odds with previous research with a Swedish sample, in which fear of childbirth, also measured by the W-DEQ at 32 weeks of gestation, was found to be associated with an increased risk of an emergency caesarean section3.

The sample in the present study is similar to the population of pregnant women in the same geographic region, during the previous year, in terms of age range, parity and distribution of types of delivery. Scores on the W-DEQ were normally distributed and the reliability of the measure was high, comparing well with previous reliability estimates. Therefore, it is assumed that the present sample is representative of the population of pregnant English-speaking women in the Sheffield area, in terms of these variables.

In addition, the present study has addressed some of the methodological flaws evident in previous research in this area. For example, it aimed to investigate the specific relationship between fear of childbirth, a form of anxiety, and the precise outcome variable, type of delivery, rather than using a composite measure of obstetric complications. Confounding variables were accounted for through the use of multivariate analysis to investigate the additional contribution of fear of childbirth to the prediction of mode of delivery, after known predictors, such as parity, medical risk, age and previous emergency caesarean section, had been taken into consideration. A further potential variable worthy of consideration, women's awareness of possible complications, was also investigated in the present sample and was shown to be unrelated to fear levels. Finally, a prospective design was employed to increase the validity of self-report measures of fear of childbirth during the third trimester.

There is clearly a need to explain the discrepancy between the findings in the present study and those of Ryding et al.3. A series of possible explanations are considered. Firstly, a significant weakness of the present study is the 35% response rate, compared with 84% in the Swedish study. In the local service, if the pregnancy is progressing normally, women have no routine hospital-based contact during the third trimester, as they do in Sweden, therefore necessitating postal recruitment. As a result, a response bias may have occurred, as participants were women who chose to respond to a postal questionnaire. The higher levels of fear in this sample might suggest that women with lower levels of fear or anxiety did not assign as much priority to the task of returning the questionnaire booklet as those who felt that anxiety was an issue for them. However, evidence from studies of other populations would suggest that the presence of anxiety is likely to mitigate against participating in related research as a result of the need to avoid discomfort31. Face-to-face, clinic-based, recruitment may have reduced the likelihood of such a response bias.

However, STAI scores were comparable both with the Swedish sample3 and expected scores based on normative populations32. This would suggest that, rather than being due to a response bias, the higher mean W-DEQ score in the present study reflects an actual difference in the levels of fear of childbirth reported by the British and Swedish samples. It is interesting to note that the rates of caesarean section are much higher in the UK (22% for 2000–2001 and 18% for 1997–1998) than in Sweden (9.1%)3, although the differences in findings suggest the linkage between fear and emergency caesarean section is not the same. One explanation could be that emergency caesarean sections are carried out for different reasons within either country, however, on examination the indications presented by Ryding et al.3 are not different from those in the present study. Criteria for selection used in the two studies were similar.

The discrepancy in findings may be explained by the fact that the W-DEQ was originally designed and tested in the Swedish language. It is possible that translation of the measure into English has caused some distortion. Two items in particular were frequently queried or omitted by participants. It may be that these items, which ask women to rate how funny (item 28) and self-evident (item 30) they will feel the moment they have delivered, have a less ambiguous or more subjectively relevant quality in Swedish.

A further explanation of the discrepancy between findings may be that the lower levels of fear of childbirth reported by Swedish women reflect an actual difference in the way childbirth is conceptualised within either the culture or the medical systems of each country. This argument is supported by differences in other birth-related outcomes. For example, the incidence of low birthweight, defined as <2700 g, in Sweden is 4.4%33, compared with 8.9% in Sheffield. It is interesting to note that the mean W-DEQ score for all the participants in the present study is similar to the mean score for the group of women in the Swedish study who were considered to be extremely fearful of childbirth (i.e. the highest 10% of the Swedish W-DEQ scores), suggesting that women in the UK are generally more frightened of childbirth than Swedish women. Further investigation of these potential differences may provide valuable clues on how to reduce women's fear in this country.

The findings of non-significance in the study could also be due to insufficient sample size. However, the difference in mean fear levels between cases and controls was less than one point, suggesting that even with increased power the groups were unlikely to show differences that were clinically meaningful. It is however worthy of note that while the differences in scores appear large, the effect size in the Swedish study was relatively small at 0.23.

The W-DEQ, which had not previously been factor analysed, appears to measure four clear dimensions that are conceptually distinct. In addition to fear of childbirth, the scale reflects the degree to which women focus on the positive aspects of bearing a child, the risks inherent in labour and expected feelings of isolation during the procedure. This analysis has demonstrated the sophistication of the W-DEQ, however, as yet no association between any of the factors and the process of delivery has been established. It would be useful to investigate in future research whether scores on each factor have more predictive power then the W-DEQ overall.

Finally, although not predictive of an emergency caesarean section, there may be other negative sequelae to fear of childbirth. Women with serious fear of childbirth are more likely to experience dissatisfaction with their delivery1 and may be at risk of post-traumatic stress symptoms following the birth34,35. In terms of facilitating psychologic wellbeing, further research to investigate the nature of fear and the efficacy of interventions is therefore warranted.

Conclusion

The level of fear of childbirth, experienced by women at 32 weeks of gestation, in a UK sample, does not influence mode of delivery. Emergency caesarean section was associated with age, parity, previous caesarean section and medical risk, but not with W-DEQ scores. The W-DEQ has been shown to discriminate between different foci of fear, having four subscales: fear, lack of positive anticipation, isolation and riskiness, and may therefore be a useful tool in clinical practice. The mean fear score for the entire UK sample was similar to the mean score of the highest 10% of scores in the sample in a Swedish study, in which such high scores were taken to indicate serious fear of childbirth. Further research investigating cross-cultural differences in the field of obstetrics and gynaecology is recommended.

Acknowledgements

The authors would like to thank all women who participated in the study. Particular thanks also to Mr Peter Stewart, Consultant Obstetrician for obstetric advice and Ms Helen Spiby, Midwifery Research Sister, Ms Pattie Preston, Unit Manager of Obstetrics Gynaecology and Neonatology, Northern General Hospital and Ms Angela Culley, Unit Manager, Jessop Hospital for Women for midwifery input. Further thanks and appreciation to Jennie Lil, Karen Reader, Sue Gregory and Carol Kirkwood, for their practical contributions to the project.

Ancillary