Validation of the German version of the pregnancy-related anxiety scale (PrAS): psychometric properties across all trimesters of pregnancy

For this cross-sectional study, pregnant women in all trimesters were recruited using a multi-faceted approach. First, recruitment took place via Facebook and Instagram pages focusing on pregnancy in any way. Second, women were recruited using posters and flyers at local gynecological offices, enabling access to the online study by a QR-code. To encourage completion of the questionnaire, an incentive was offered (a chance to win one out of ten €15 gift cards). The survey was conducted online with the use of the platform EFS by QuestBack (www.​unipark.​de). After providing informed consent online, participants completed the survey on a smartphone or computer in an environment of their choice. Women were informed about their right to withdraw from the study at any given time. Additionally, they were advised to seek professional help if needed and numbers were provided to support services (i.e., German National Suicide and Crisis Line). The survey took approximately 15 min to complete. Nulliparous and multiparous women in all trimesters of pregnancy were included. Since we intended to collect data from a community sample, participants with mental disorders were also included. Exclusion criteria were younger age than 18 years and insufficient German language skills. The study protocol was approved by the Ethics Review Board of the University of Dresden (approval no. SR-EK-282,062,020) and the study was performed in accordance with the Declaration of Helsinki. All data were stored anonymously and in accordance with the German General Data Protection Regulation. Originally, 572 women took part in the study. However, after exclusion due to incomplete data, the sample consisted of 443 pregnant women. As previously suggested, not less than ten cases per indicator variable represent an acceptable sample size for CFA. With a total sample of N = 443, this prerequisite was fulfilled [26].

Reliability was estimated with Cronbach’s α using IBM SPSS statistics version 28 for windows. Additionally, internal consistency was also estimated with McDonald’s ω, which is a less biased estimate than Cronbach’s α to provide extensive comparability [42]. Confirmatory factor analysis (CFA) was performed using IBM SPSS Amos version 28. Fit indices included Root Mean Square Error of Approximation (RMSEA), Standardized Root Mean Square Residual (SRMR), Comparative Fit Index (CFI), and Tucker-Lewis Index (TLI). Good (and adequate, respectively) model fit is indicated by RMSEA ≤ 0.06 (0.06–0.08), SRMR ≤ 0.05 (0.05–0.08), and CFI, as well as TLI ≥ 0.95 (0.90–0.95) [43]. IBM SPSS statistics version 28 for windows was used to yield correlation coefficients for convergent and divergent validity and conduct secondary analyses with tests for group differences. To test for significant differences between women with planned versus unplanned pregnancies, different trimesters, and parity, Welch´s t-test and Welch´s ANOVA were used, since some of the subscales of the PrAS showed no homogeneity of variance. Post-hoc comparisons were computed with the Games-Howell-Test [44, 45].

The final sample included N = 443 women (M_age = 31.9, SD = 4.2). Over 90% of women were German nationals, with most being married or cohabitating. The women in this sample were well educated with roughly 81% having a general qualification for university entrance. More than 86% stated that the current pregnancy was planned and approximately half of women were expecting their first child. Table 1 provides full details.

Analysis of the internal consistency by Cronbach’s α resulted in excellent reliability of 0.90 and a Cronbach´s α ≥ 0.76 for all eight subscales. Further analyses reached a McDonald’s ω = 0.89 for the total scale. Thus, both measures indicate sound reliability [46].

To confirm the underlying factor structure of the translated scale, data were subjected to a theory-based CFA. For the original English version of the PrAS, two similar models showed adequate fit in previous validation studies and were tested in the current sample [24, 25]. The Kaiser-Meyer-Olkin measure of sampling adequacy test resulted in a coefficient of 0.85 and Bartlett’s Test of Sphericity was significant (χ²(496) = 8146.65, p < .001). Both results confirmed data suitability in accordance with previous guidelines [47, 48]. Initially, we tested a model with eight factors (Model 1). Those factors included: Childbirth Concerns, Body Image Concerns, Attitudes Towards Childbirth, Worry About Self, Baby Concerns, Acceptance of Pregnancy, Avoidance, Attitudes Towards Medical Staff. Model 1 (χ²(436) = 1262.57, χ²/df = 2.896) yielded a good fit to the data evidenced by the fit indices RMSEA = 0.065, SRMR = 0.069, CFI = 0.895, and TLI = 0.880. We also tested a model with nine factors (Model 2) in which items 15 and 16 were detached from the scale Worry About Self, building an additional scale Worry About Motherhood. Model 2 (χ²(428) = 1094.305, χ²/df = 2.557) was a slightly better fit to the data, with RMSEA = 0.059, SRMR = 0.062, CFI = 0.915, TLI = 0.902. Since both analyses resulted in similar results, we proceeded with the eight-factor model (Model 1) as proposed for the original version of the PrAS [24]. This way, the comparability of the English and German version of the PrAS can be ensured. Standardized factor loadings of Model 1 are shown in Fig. 1.

To test criterion-related validity of the PrAS with other measures, correlations between the PrAS and its subscales and convergent (i.e., PRAQ-R2 and its subscales, EPDS, and DASS-Anxiety subscale) and a divergent measure (BRS) were calculated. As expected, the PrAS and the PRAQ-R2 showed strong linear relationships between the scales overall and the related subscales. For example, the PrAS subscales Baby Concerns and Body Image Concerns were strongly correlated with the PRAQ-R2 subscales Worries of Bearing a Physically or Mentally Handicapped Child and Concerns about Own Appearance, respectively. There were moderate correlations between the PrAS subscales Childbirth Concerns and Attitudes about Childbirth and the PRAQ-R2 Fear of Giving Birth as well as for the PrAS with the EPDS and the DASS-Anxiety. The divergent correlation of the PrAS sum score and almost all subscales (except Avoidance) with the BRS resulted in a weak negative relationship. See Table 2 for further details.

The sum scores of the total scale as well as the subscales of the PrAS were tested for significant differences with regard to planned pregnancy, trimester, and parity.

This study has several strengths. To our knowledge, the translated version of the PrAS is the second German questionnaire for the assessment of PrA and offers expedient features. Since its subscales assess integral aspects of anxiety during pregnancy, a more comprehensive as well as differentiated depiction of PrA across all trimesters is possible. Even though the PrAS-G consists of 32 items and is comparably longer than the PRAQ-R2, its applicability in a clinical context is recommendable since the PrAS offers a wide range of information on PrA. The differentiation of its scales allows for the identification of individual profiles of PrA in pregnant women. Thus, the PrAS represents a useful diagnostic tool for the assessment PrA for women across all trimesters of pregnancy.

However, there are also some limitations that need to be addressed. The study design was cross-sectional and therefore we cannot provide data on intraindividual changes in PrA over the course of pregnancy. A recent study showed that sum scores of the PRAQ-R2 seem to be relatively stable during pregnancy, but scores of subscales change [60]. Since the PrAS consists of more subscales, studies with longitudinal designs should be conducted to examine trajectories of PrA and its diverse facets as reflected by the PrAS. Besides, our sample included participants with and without mental disorders. Further research on levels of PrA in women with preexisting mental disorders could be particularly relevant, since a higher fear of childbirth in women who suffered from anxiety and depression even before pregnancy has been shown [61]. Furthermore, participants filled in the survey online, which allows for low-threshold participation. In addition, most women were highly educated. Both aspects might have caused a selection bias and future studies should incorporate more diverse samples and use strategies to enhance representability of the sample [62].

Future studies should also include men to further decrease the relative neglect of studies on the peripartum mental health of (expectant) fathers [63, 64]. As several items of the PrAS are ineligible for men (e.g. ‘I worry that I will tear or need to be cut during the birth’, ‘I feel scared that I will never regain my figure), a version for men should be developed. This could provide a more holistic approach to parental peripartum mental health and could prevent (expectant) fathers from experiencing clinically relevant symptoms of anxiety in the long run. This would also prevent children and the entire family from suffering further negative consequences caused by spill-over effects [65, 66].

Furthermore, the study has been conducted during the ongoing COVID-19 pandemic, which might have led to higher scores in the PrAS than before the pandemic as has been shown in previous studies [67‐70]. However, the psychometric properties of the PrAS are most likely unaffected by this. Thus, with its factorial structure, convergent and divergent validity with other measures, and the ability to identify differences in symptom scores between subsamples of women, the PrAS is a magnificent tool for the assessment of PrA.

Taken together, our findings demonstrate that the PRAS-G has sound psychometric properties and is recommendable both for clinical practice and scientific purposes. In relevant analyses, high reliability, the eight-factorial structure, as well as convergent and divergent validity were confirmed. Furthermore, the PrAS-G showed the ability to discern between women who either planned or did not plan their pregnancy, were in varying trimesters, and differed in parity.

This knowledge will improve the possibilities to interpret unique compositions of PrA in individual women. Expectant mothers seem to differ in their experience of PrA and tailored interventions on an individual level as well as in public health campaigns are needed to tackle the most pressing aspects of PrA with regard to influencing factors, such as parity [71]. With this approach adverse effects of PrA can be prevented and will instead improve birth experience and well-being of expectant mothers and fathers [8]. Future research should also examine, if childhood development and mental health of children as well as couple and family relationships could be influenced in a positive manner by developing and applying adequate interventions for PrA [8, 16, 72]. Thus, the PrAS-G will be a useful tool for application in a clinical and research context.