全文

Objective
To develop an instrument to the assess quality of care during pregnancy and delivery as perceived by women who recently gave birth.

Study design
Prospective design from focus group interviews to validation of the questionnaire. The focus groups consisted of seven care providers, ten pregnant women and six women who recently gave birth. With the results of the focus group interviews, a draft questionnaire of 52 items was composed and its psychometric properties were tested in a first cohort of 300 women who recently gave birth (sample I) by means of exploratory factor analysis (EFA) and reliability analysis. The final version was further explored by confirmatory factor analyses (CFA) in another sample of 289 women (sample II) with similar characteristics as sample I.

Results
EFA in sample I suggested an 18-item scale with two components concerning the quality of care during pregnancy: ‘personal treatment’ (11 items, Cronbach's alpha (α) = 0.87) and ‘educational information’ (7 items, α = 0.90); the ‘delivery’ scale showed a single domain (7 items, α = 0.88). CFA in sample II confirmed both factor structures with an adequate model fit. Overall, satisfaction with care was highest among women who only received midwife-led care, while women who were referred to an obstetrician during pregnancy reported less satisfaction.

Conclusions
The 25-item PCQ, primarily based on the experiences and perceptions of pregnant women and women who recently gave birth, showed adequate psychometric properties evaluating the quality of care during pregnancy and delivery. This user-friendly instrument might be a valuable instrument for future research to further evaluate the quality of care to pregnant women.

Keywords: Patient perspective, Satisfaction, Quality of care, Pregnancy, Childbirth.

1. Introduction
In the Netherlands (and to a lesser extent in other countries like the United Kingdom, New Zealand, and Canada), the obstetric care system gives low-risk pregnant women the opportunity to plan childbirth supervised by an independent community midwife, at home or in hospital [1]. As a consequence, there are three possibilities for obstetric care: (1) supervised by a community midwife only, (2) supervised by a community midwife followed by referral to an obstetrician and (3) obstetrician-led care only. Due to the complexity of the leveled obstetric care system [2], pregnant women often see different care providers, which may interfere with a personal treatment and continuity of care, and might negatively influence women's satisfaction with received care. Patient satisfaction as an indicator of quality of care is becoming more and more important [3] and [4]. This evaluation of care from a clients’ perspective has to be primarily based on the experiences of the target group, but the number of validated questionnaires based on patient interviews appears to be small. Sawyer et al. recently reviewed the literature and concluded that: “despite the interest in measures of satisfaction there are only a small number of validated measures of satisfaction with care during labor and birth” [5]. Most of the questionnaires are not based on patients’ input, or lack the validation process with exploratory and confirmatory factor analyses in adequate samples.

To our knowledge, no proper (useful) instrument has yet been developed to evaluate women's perception of obstetric care, following a rigorous methodological protocol (including focus groups, reliability analyses, and factor analyses in samples with sufficient statistical power) [6] and [7].

Therefore, the primary aim of this study was to develop an instrument with adequate psychometric properties that assesses quality of care received during pregnancy and delivery. As a proxy for construct validity (secondary aim), we compared the perceived quality of care of women who went through different obstetric routes. We hypothesized that women who received care from an independent midwife or obstetrician during the entire pregnancy and delivery had the highest level of satisfaction, while those who had to be referred from primary community care to secondary hospital care during labor were less satisfied.

 
2. Materials and methods
2.1. Procedure
Four different focus group interviews were conducted prior to constructing the Pregnancy and Childbirth Questionnaire (PCQ). The first group, consisting of five independent midwives and two clinical midwives, explored what care providers think women find important regarding obstetric care. The second group consisted of five nulliparous pregnant women, and the third group of five multiparous pregnant women. The fourth group consisted of six originally low-risk women who became high-risk during pregnancy or labor and recently gave birth in hospital. These patients were selected based on date of delivery, parity, route of care, and location of delivery, and were invited by their midwife or the obstetric department where they gave birth. There was no financial compensation for participants. The focus group interviews were conducted by a staff member from the medical psychology department of Tilburg University, assisted by a psychologist and a midwife. In all focus groups, an open interview strategy was used with a topic list as a back-up in case several items did not come up spontaneously. All focus groups were recorded with permission of the participants. Subsequently, the recorded texts were transcribed, sorted by subject, and evaluated by a panel of experts (ST, AP, VP). Based on the panel's consensus, 52 candidate items remained. Several care providers and some women who had participated in the focus group interviews reviewed the candidate items on clear formulation and user-friendliness. Questions were formed in positive and negative statements, rated on a five-point Likert scale from 1 (‘totally agree’) to 5 (‘totally disagree’). After recoding, higher scores indicated higher quality of care.

Three hospitals and five midwifery practices in The Netherlands invited women to complete an online questionnaire six weeks after childbirth. The first cohort was used to examine reliability and latent structure. These analyses were used to generate a more refined version of the questionnaire, which was then distributed to a second group of women with similar characteristics to test the model fit. Construct validity was examined by testing hypotheses according to satisfaction with care in low-risk and high-risk women.

2.2. Participants
A first cohort of 917 women (study I) was invited to fill in the online questionnaire. A few weeks after delivery, suitable mothers received an e-mail or letter of invitation from their midwife or the obstetric department of the hospital. Exclusion criteria were: fetal death, and not being able to understand Dutch sufficiently. Of all invited women of the first cohort, 300 women (33%) fully completed the online questionnaire. Subsequently, 289 of the 710 women (36%) of a second cohort in 2013 with the same characteristics fully completed the final version (study II). A flowchart is presented in Fig. 1. The procedure of questionnaire development with focus groups was approved by the ethical review board of Tilburg University. The Medical Ethics Committee of the Máxima Medical Centre Veldhoven decided that ethical approval was not required.

2.3. Measurements
The questionnaire consisted of the PCQ and several aspects regarding demographics, obstetrical features and the newborn's health. Furthermore, depending on the participants’ situation, there were items about the hospital and/or midwifery practice, and items about referral and collaboration between professionals in different levels of the care system.

2.4. Statistical analyses
Statistical analyses were performed using SPSS (version 20, IBM, Chicago, Illinois, USA). Confirmatory factor analyses were performed using AMOS (version 18, IBM, Chicago, Illinois, USA). To test for differences between the two samples, χ2-analyses were used for dichotomous data. Differences in scores between groups were analyzed using t-test (two-tailed) and one-way ANOVA with Tukey post hoc analysis. Furthermore, the assumption of normality was checked by reviewing the distribution of answers per item, with negatively posed items recoded so that higher scores corresponded with higher quality of care. Missing values were replaced using expectation maximization imputation, with values based on age, parity, level of education, referral, location of delivery, and mode of delivery.

In sample I, exploratory factor analyses (EFA) were performed on the full 52-item questionnaire, with 37 items about care during pregnancy and 15 about delivery. Principal component analyses with oblimin rotation were used to select factors for retention. Factor loadings >0.40 were considered important. Internal consistency analyses were conducted using Cronbach's alpha (α) for the total scale and its subscales, with α ≥ 0.70 considered the minimum acceptable criterion of internal reliability [6].

In sample II, confirmatory factor analyses (CFA) were performed to test stability of the factor structures as found in study I. The comparative fit index (CFI), normed fit index (NFI), and the root mean square error of approximation (RMSEA) are generally considered good parameters to evaluate model fit. Adequate model fit can be assumed with a CFI ≥ 0.80, NFI ≥ 0.80, and RMSEA ≤ 0.05 for good and ≤0.08 for adequate fit [8] and [9].

 
3. Results
As shown in Table 1, characteristics of the participants in the two samples were similar.

Most items were distributed normally, except for six pregnancy-items and two delivery-items (kurtosis > 3), which were therefore eliminated. Based on face validity and redundancy, 13 of the 37 pregnancy-items and 5 of the 15 delivery-items were eliminated beforehand.

The EFA on the pregnancy-items (Table 2a), suggested a two-component solution with Eigenvalues of 7.6 and 1.9. After oblimin rotation, a two-factor model explaining 53% of the variance was revealed: a factor ‘personal treatment’ with 11 items and a factor ‘educational information’ with 7 items. The resulting 18 items showed good overall internal consistency (α = 0.92) as well as for each subscale: α = 0.87 (‘personal treatment’, 11 items) and α = 0.90 (‘educational information’, 7 items).

The EFA on the delivery-items (Table 2b) showed a single-factor structure with an eigenvalue of 3.9. The 7 items explained 56% of the variance and showed good internal consistency (α = 0.86). As can be seen in Table 2a and Table 2b, the items refer to aspects of communication, autonomy, involvement, professionalism, educational information, teamwork, and partner involvement.

The model with 18 pregnancy-items and 7 delivery-items was tested in sample II (n = 289). First, EFA was repeated to verify the factor solutions found in sample I. Again the same dimensions were found with no overlapping items on the subscales, and a total explained variance of 56% and 54% in the pregnancy-scale and delivery-scale, respectively. Subsequently, CFAs were performed on the data of sample II. CFA confirmed the two scales with an adequate model fit on both the pregnancy-scale (CFI: 0.90, NFI: 0.86, and RMSEA: 0.09 (lower limit 0.08)), and delivery-scale (CFI: 0.98, NFI: 0.96, and RMSEA: 0.06 (lower limit 0.03)).

After merging the two samples, Cronbach's alpha of the total and subscales still showed good internal reliability (Table 3). Means, ranges and Cronbach's alpha's of the total sample (N = 589) are shown in Table 3.

Mean scores on the three subscales of the PCQ according to four different levels of care are shown in Fig. 2. Higher scores concerning ‘personal treatment’ and ‘educational information’ during both pregnancy and delivery were found among low-risk women who received exclusively midwife-led care. Satisfaction with care was less positive when women were referred during pregnancy. Post hoc Tukey analysis showed that with regard to satisfaction with personal treatment during pregnancy, the group with referral during pregnancy differed significantly from all other groups (F(1,3) = 23.93, p < 0.01). Scores on educational information were significantly higher among women who were low-risk during the entire pregnancy. With regard to care during delivery, the highest scores were found among low-risk women who gave birth supervised by their independent midwife, followed by those who received obstetrician-led care from the beginning of their pregnancy (Fig. 2). With regard to location of delivery, personal treatment during delivery was perceived more positively among women who gave birth at home compared to women who planned hospital delivery with their independent midwife (t = −2.29, p = 0.03).

4. Comment
This study aimed to develop an instrument that assesses quality of care during pregnancy and childbirth. The PCQ contains 25 items, divided into two scales: 18 items referring to pregnancy (11 ‘personal treatment’-items, 7 ‘educational information’-items) and 7 items referring to ‘personal treatment during delivery’. Both scales showed good psychometric properties.

Evaluation of quality of care during pregnancy showed that women who were referred during pregnancy reported less quality of care. Scores on educational information were significantly better among women who were low-risk during the entire pregnancy. With regard to personal treatment during delivery, women without referral scored significantly better than women who were referred to an obstetrician during pregnancy or delivery. In the low-risk group, perception of quality of care during delivery was higher after home delivery, compared to low-risk women with a planned hospital delivery. These findings, in combination with studies showing that home birth is as safe as hospital birth for low-risk women [10], [11], and [12], are in favor of offering low-risk women a home birth. Also, close collaboration between different care providers might improve the referral process and perceived quality of care [2].

Furthermore, results with regard to overall quality of care showed higher scores among women who only received midwife-led care. In addition, women who had been referred during pregnancy reported lower quality of care. These women received care in more than one institution from several care providers. It is hypothesized that with an increasing number of care providers, the individual involvement of each care provider decreases, resulting in diffusion of responsibility [13]. Referral during pregnancy may interfere with a systematic way of providing clients with information and might cause inconsistent advice, information, and protocols. Continuity of care appeared to be an important aspect of the quality of care, represented in both the pregnancy-scale (items 6–9) and delivery-scale (items 3–6), but these items did not come up as a separate dimension during factor analyses.

There is increasing attention to satisfaction with care and patient-centered healthcare. Especially when giving birth, women believe that their feelings and values should be understood by professionals, from whom they seek empathy and personal commitment, not just information [14]. Goberna-Tricas et al. [14] found that women make a clear distinction between technical and interpersonal skills of obstetric care providers. More ‘personal’ care is associated with higher levels of satisfaction [15]. Accordingly, this study found ‘personal treatment’ to be the main factor in satisfaction with care during both pregnancy and delivery. The extent to which ‘personal’ care is important to women also depends on demographic and personality characteristics; for low-risk women preferring a hospital birth, the assumed safety of the hospital appeared to be more important than the type of care provider [16].

The psychological concept of ‘feeling in control’ is important in childbirth satisfaction and refers to a woman's own perception of how she performed during childbirth [17]. Although it is apparent that the personal experience of childbirth will influence perception of obstetric care, items about ‘feeling in control’ are assigned to birthing experience rather than satisfaction with quality of care.

The current questionnaire is developed in a country with home and hospital births, but the perception of quality of care is general and not restricted to a specific care system. Therefore, the currently developed questionnaire can also be used in countries without home births.

In a recent review, Sawyer et al. [5] evaluated nine questionnaires about satisfaction with care during labor and birth. Three were designed for particular types of births only (e.g. operative births), and three of the remaining six questionnaires were neither based on interviews with patients nor validated with factor analyses. The other questionnaires had poor reliability, an unclear factor structure, or were difficult to assess on content and face validity as it was not reported how the items were selected [5].

Wiegers [18] and [19] developed a questionnaire to evaluate the quality of care as perceived by women who gave birth. However, several methodological requests were not met in the study: for example, there was absence of several focus group interviews with subgroups of pregnant and women who recently gave birth, the use of subscales with poor Cronbach's alpha (<0.70) or even single items, and no CFA being performed.

The current study has several limitations. Data of the Dutch national perinatal registry [1] showed 48% of women being primiparous, while in the current study primiparae (63%) were over-represented. Due to a recruitment bias, more high-risk women participated in this study (82% versus 72%). Accordingly, the number of home deliveries is lower in this study (5% versus 17%) [1]. There is no reason, however, to suggest that this will interfere with the psychometric characteristics of our instrument. The fact that only Dutch-speaking women were included questions whether the findings can be generalized for the whole population (including different ethnic minorities). Also, because of the anonymous responses, sending a reminder to the non-responders was not possible, resulting in a response rate of about one third. Furthermore, there were limited data on the course of pregnancy, the occurrence of major life events during pregnancy, and personality characteristics of the participants. Also, partners were not involved in the focus groups. In future research these aspects should be taken into account.

The main strength of the current study is related to the questionnaire based on several focus groups and therefore primarily based on the experiences of the target group. For example, an important aspect of quality of care that came up during the focus group interviews was ‘whether the care provider was able to put my mind at ease’. Furthermore, the interviews were valuable in discovering the importance of partner involvement, communication between professionals, the number of check-ups, and whether the educational information satisfied the women's needs.

Furthermore, this questionnaire is not limited to either prenatal care or care during childbirth, but contains items about the whole process of care, divided into pregnancy- and delivery-scales. Also, it includes the use of two separate but comparable and large samples for stepwise validation of the questionnaire.

What do the findings mean in practice? Postpartum women are vulnerable to emotional disturbances and a woman's perception of her childbirth has immediate and long-term effects on her well-being and relationship with her newborn [20] and [21]. This questionnaire might help clinicians to detect possible shortcomings of obstetric care with regard to maternal satisfaction. Subsequent training of all professionals involved in obstetric care will further improve the quality of care.

In conclusion, this study shows that the 25-item PCQ constitutes a valid and user-friendly instrument to evaluate satisfaction with care as perceived by women who recently gave birth. Future research should further confirm the construct validity and reliability of the PCQ.

 
Acknowledgment
Development of this questionnaire was part of the ZonMw-programme “Zwangerschap en Geboorte”, project “Vorming van een Regionaal Consortium Verloskunde Brabant”, dossier number: 50-50200-98-007.