Childbirth experience questionnaire (CEQ): development and evaluation of a multidimensional instrument

The Childbirth Experience Questionnaire (CEQ) was developed to study women's perceptions of first labour and birth. Pertinent domains were identified through literature studies, discussions in a group of four experienced midwives and one senior obstetrician and a group interview one month postpartum with twelve mothers, most of them primiparous. Identified areas of childbirth experience included intrapartum sense of security, experience of labour pain, partner's support, midwifery care and support, memories from the childbirth and experience of own performance. The items regarding memories came from the women's narratives one month postpartum. Items were drafted to cover the identified areas and formulated as positive and negative statements. The response format was a 4-point Likert scale ranging from 1 (Totally agree), 2 (Mostly agree), 3 (Mostly disagree) to 4 (Totally disagree). Memory of labour pain, sense of security and control were assessed with visual analogue scales (VAS). A draft version of the CEQ was pilot tested for face validity (i.e. comprehension and relevance) in 25 primiparous women. Based on the women's comments, the questionnaire was revised and comprised totally 28 items. A one-month follow-up frame was chosen. At that point the mothers were expected to have gone through the first relief phase but still have relatively fresh memories of the childbirth experience. Moreover, the women had little or no contact with childbirth caregivers after one month and were therefore less likely to give socially desirable responses.

The sample included 1177 women who participated in a study on labour progress and oxytocin augmentation between October 1998 and December 2003 at Sahlgrenska University Hospital in Gothenburg and Ryhov County Hospital in Jönköping, Sweden. The study is described in detail elsewhere [21]. In brief, the antenatal clinics provided written information about the study as well as information about the follow-up questionnaire to healthy nulliparous women in their third trimester. Inclusion criteria were nulliparity, a singleton fetus in cephalic presentation, uncomplicated pregnancy, spontaneous onset of active labor with regular contractions and an effaced cervix dilated not less than four centimeters and gestational age of between 37+0 and 41+6 weeks at admission to the delivery ward. Informed consent was obtained from all respondents who were mailed the questionnaire one month postpartum.

Descriptive statistics were computed to characterize item score distributions. Item response completeness and frequency distributions were examined to identify items with high missing data rates and/or high ceiling/floor effects. The VAS-scales scores were transformed to categorical values, 0-40 = 1, 41-60 = 2, 61-80 = 3 and 81-100 = 4. Ratings of positively worded statements and the pain items were reversed so that higher scores reflect more positive scoring.

Exploratory factor analyses using principal components analysis were performed to examine the construct validity of the CEQ. Both orthogonal (varimax) and oblique (promax) rotations were conducted. Oblique rotation was selected due to the fact that the dimensions were correlated (r = 0.23-0.63). Items with extreme ceiling effects were not entered in the analysis. The Kaiser rule (eigenvalue > 1.0) was applied for determining the number of dimensions to extract. Factor extraction was also guided by the clinical interpretability of the factors. Items with maximum loadings less than 0.40 were dropped.

Multitrait scaling analysis [22] was performed in order to confirm the derived factor structure and to test scaling assumptions for aggregating item ratings of the dimensions. Four assumptions were tested: item internal consistency (item-hypothesized scale correlations ≥ 0.40 and Cronbach's alpha ≥ 0.70), item discriminant validity (item-hypothesized scale correlation > item-other scale), equal item-hypothesized scale correlation (item-scale correlations roughly the same for all items in scale), item equal variance (variances of items in hypothesized scale roughly equal).

Item ratings were aggregated to scale scores for each respondent using the half scale method, i.e., mean values were computed when the respondent had answered at least half of the items in the scale [23].

The method of known-groups validation [24] was used to assess the ability of the CEQ to distinguish between subgroups known to differ on key sociodemographic or clinical variables. Based on previous research results [8‐13], it was hypothesized that the women who had a labour lasting more than twelve hours, women who had the labour augmented with oxytocin and women who had experienced an operative delivery would score lower on all scales. Scale scores were compared between groups with Mann-Whitney U-test. All significance tests were two-tailed and conducted at the 5 percent significance level. The Holm-Bonferroni correction method was used when multiple tests were performed [25]. Accordingly, the p-values were ranked from lowest to highest and the smallest p-value was compared to 0.05 divided by 4 (the total number of tests), the second smallest with 0.05 divided by 3 (the total number of tests minus one) and so forth. Effect sizes, as defined by Cohen [26], were computed as the difference between group mean scores divided by the pooled standard deviation of the two groups. In line with criteria suggested by Cohen, effect sizes of 0.2-0.5 was regarded as "small", 0.5-0.8 as "moderate" and above 0.8 as "large" [23].

All statistical analyses were performed using SPSS 15.0 for Windows (SPSS Inc., Chicago, IL, USA) with the exception of multitrait scaling analysis for which the MAP-R program was used [22].

In total, 25 items were initially entered in the exploratory factor analysis. Two items with maximum loadings less than 0.40 on any domain were excluded: "I have memory lapses from the labour process" and "Experienced pain in second stage". The item "The care given by my midwife made me feel secure" lowered Cronbach's alpha for its factor and was also excluded. After eliminating these 3 items, four dimensions met extraction criteria and were retained: Own capacity (8 items regarding sense of control, personal feelings during childbirth and labour pain), Professional support (5 items about information and midwifery care), Perceived safety (6 items regarding sense of security and memories from the childbirth), and Participation (3 items regarding own possibilities to influence the birthing situation). The four factors accounted for 28, 15, 6, and 5% of the variance, respectively, for a total of 54% (Table 2).

The multitrait scaling analysis showed that scaling assumptions were adequately met for all dimensions (Table 3). Item-scale correlations exceeded 0.40 for nearly all items and those less than 0.40 were still higher with their own scale than with competing scales (item discriminant validity). Cronbach's alpha coefficients were acceptable for group analyses (> 0.70) in all but the participation scale. Item-scale correlations, means and variances showed that items contributed roughly equally to its hypothesized scale.

Sensitivity of the model was examined by comparing mean scores on each subscale between women with labour duration more versus less than 12 hours; women with versus without oxytocin augmentation during labour; and women who had a spontaneous vaginal delivery versus those who had an operative delivery (emergency caesarean or instrumental vaginal birth). Women with prolonged labour scored lower in all scales, see Table 4. Effect sizes were, however, small (Own capacity, 0.47; Perceived safety, 0.42 and Professional support, 0.27) to trivial (Participation, 0.19). Women with oxytocin treatment scored significantly lower on all subscales. Effect sizes were moderate (Own capacity, 0.67; Perceived safety, 0.57), small (Participation, 0.29) or trivial (Professional support, 0.16). Women with operative births scored significantly lower on all four subscales, see Table 4. Effect sizes were moderate (Own capacity, 0.72; Perceived safety, 0.63; Participation, 0.54) or small (Professional support, 0.34), see Figure 1.