The pregnancy-related anxiety scale: A validity examination using Rasch analysis

Highlights

• The psychometric properties of the Pregnancy-related Anxiety Scale were examined using Rasch Analysis.

• The scale demonstrated good fit to the Rasch model and the ability to discriminate between individuals.

• Scale properties can be enhanced up to interval-level of measurement using developed conversion algorithms.

• Confirmatory Factor Analysis supported the scale structure that demonstrated the best Rasch model fit and excellent reliability.

• The Pregnancy-related Anxiety Scale shows good psychometric properties and is useful for both clinicians and researchers.

• The subscales permit accurate estimation of specific pregnancy-related anxiety aspects enabling more targeted interventions.

Abstract

Background

Pregnancy-related anxiety is increasingly recognised as a common condition that is associated with many deleterious outcomes for both the mother and infant (e.g., preterm birth, postnatal depression). Limitations in the psychometric properties and/or breadth of existing scales for pregnancy-related anxiety highlight the need for a psychometrically sound measure to facilitate effective screening and possible early interventions. The recently developed Pregnancy-related Anxiety Scale (PrAS) was evaluated using Rasch analysis to explore how the scale's psychometric properties could be fine-tuned.

Method

A sample of 497 pregnant women completed the PrAS. Data were subjected to Rasch analysis, and the resulting scale structure examined using Confirmatory Factor Analysis.

Results

After minor modifications, the Rasch model with 33-items and 8-factors demonstrated good fit, unidimensionality and excellent targeting and internal consistency. Confirmatory Factor Analysis confirmed the final structure, and Cronbach's alpha demonstrated excellent reliability.

Limitations

The use of the same sample for all analyses was a potential limitation due to the possibility of sample-specific influences.

Conclusions

The Rasch analysis further supports the internal construct validity of the PrAS. Ordinal to interval score conversions provide added precision to the analysis of the PrAS scores. The Rasch results, together with previous validation evidence, point to the PrAS as a comprehensive and psychometrically sound screening scale for pregnancy-related anxiety. The PrAS offers clinicians the ability to screen for pregnancy-related anxiety. The subscales provide additional insights into a woman's pregnancy-related anxiety and her specific areas of concern, enabling more targeted interventions.

Introduction

Pregnancy was once regarded as a time of protection from mental health disorders, however, research evidence increasingly points to the common experience of emotional disturbances in pregnancy (Austin, 2003). Symptoms of general anxiety and diagnosed anxiety disorders (Diagnostic and Statistical Manual), for instance, are estimated to affect approximately 35% and 8.5% of pregnant women, respectively (Lee et al., 2007, Sutter-Dallay et al., 2004). These estimations demonstrate the greater prevalence of mood disorders in pregnancy when compared to women of similar age in the general population which has been reported to be 19.8% and 3%, respectively (American Psychiatric Association 2000, Hinz and Brähler, 2011). Moreover, consideration needs to be given to a specific anxiety experienced antenatally, called pregnancy-related anxiety¹ (Huizink et al., 2004). Evidence supports the distinction of pregnancy-related anxiety from other anxiety types. This evidence includes (but is not limited to) consistent associations with negative obstetric and paediatric outcome (i.e. preterm birth, postnatal depression, Robertson et al., 2004, Wadhwa et al., 1993). These consistent associations, however, are not regularly reported for other anxiety types (e.g. general anxiety). Pregnancy-related anxiety, which is regarded to affect up to 14.4% of expectant mothers (Poikkeus et al., 2006), is the focus of this paper.

Pregnancy-related anxiety is a contextually tied type of state anxiety characterised by a pregnancy specific fears and worries. These include, for example, childbirth, the wellbeing of the mother and baby and impending motherhood. As stated above, there are various unfavourable outcomes associated with this anxiety type. For the mother, these may include increased nausea, poor psychological adjustment and negative views of motherhood (Hart and McMahon, 2006, Swallow et al., 2004). Furthermore, pregnancy-related anxiety has been shown to predict birth weight and gestational age independent of general anxiety and anxiety disorders (Blackmore et al., 2016; Wadhwa et al., 1993). Pregnancy-related anxiety has also been identified as independently predicting postnatal depression (Austin et al., 2007, Heron et al., 2004, Milgrom et al., 2008, Sutter-Dallay et al., 2004). Postnatal depression, with an estimated prevalence of 13%, has well-documented consequences (Milgrom et al., 2011 provides a review), such as reducing a woman's capacity to care equally for herself and her baby (Beestin et al., 2014, Leigh and Milgrom, 2008). The deleterious effects of postnatal depression are such that Australian perinatal care guidelines recommend routine postnatal depression screening (Australian Health Ministers Advisory Council, 2012). In consideration of all the adverse associations, the importance of prenatal screening for pregnancy-related anxiety cannot be underestimated. Recent calls for pregnancy-related anxiety measures to be included in clinical interventions (Blackmore et al., 2016) highlight that antenatal screening may afford valuable intervention opportunities.

While the need for early screening for pregnancy-related anxiety has been identified in the research literature, (Australian Health Ministers Advisory Council 2012, NSW Department of Health 2009), this need is not reflected in current clinical practice. There are no formal antenatal screening procedures for pregnancy-related anxiety in Australia, the United States of America or the United Kingdom (NICE 2008, Office of Women's Health 2014 ). In Australia, prenatal screening for pregnancy-related anxiety does not tend to take place due to the limitations of current anxiety measures. For instance, some general anxiety scales (e.g., State Trait Anxiety Inventory) have been utilised for pregnancy-related anxiety yet these scales potentially overlook facets of this specific type of anxiety by failing to acknowledge its characteristics, such as fear of childbirth or appearance-related concerns (to name just two). Therefore, any existing psychometric properties of general anxiety scales (such as the State Trait Anxiety Inventory) cannot be assumed to comprehensively cover all the relevant aspects of pregnancy-related anxiety. Moreover, the overlap of many anxiety symptoms and common pregnancy experiences (e.g. poor sleep) further suggest substantial limitations of general measures (Meades & Ayers, 2011).

Other more specific instruments (e.g. Pregnancy Related Anxiety Questionnaire-Revised, PRAQ-R, Pregnancy Related Thoughts (Huizink, 2000, Rini et al., 1999), may specifically target a pregnant woman's concerns and have demonstrated good, and in some cases excellent reliability; but evidence of content and construct validity is lacking. The PRAQ-R has shown evidence of predictive validity (Huizink et al., 2003), yet in the majority of these studies, pregnancy-related anxiety was assessed as a component of prenatal stress (which is considered a different construct to anxiety). Given these specific limitations, the predictive validity of the PRAQ-R may be limited. Furthermore, many existing pregnancy specific scales lack the broadness of scope for this multidimensional construct such as the Pregnancy Specific Anxiety Scale (PSAS, Mancuso et al., 2004). This scale consists of a single item rated in terms of the intensity of panic, concern, anxiety and fear experienced. The lower reliability (a = 0.65–0.72) of the PSAS has been attributed to the exploratory nature of the scale. Therefore, the use of instruments that have limitations related to both validity and reliability to screen for this anxiety may result in some women with pregnancy-related anxiety being misdiagnosed, and/or overlooked (a full and detailed review of existing scales which demonstrates the need for the development of this scale is provided by Brunton, Dryer, Saliba and Kohlhoff, 2015).

The Pregnancy-related Anxiety Scale (PrAS) was developed to fill the abovementioned need (Brunton, Dryer, Saliba & Kohlhoff, 2017a). The PrAS contains 33-items with responses ranging from 1 (not at all) to 4 (very often). Higher scores are indicative of increased pregnancy-related anxiety. The PrAS is multi-dimensional scale with nine subscales. These are: Childbirth Concerns which contains six items consistent with concerns about birth; Body Image Concerns a subscale with five questions about appearance-related concerns; Attitudes Towards Childbirth, three items relating to a women's attitude towards birth; Worry About Motherhood with three items assessing motherhood worry; Acceptance of Pregnancy contains three items about a woman's acceptance of pregnancy; Anxiety Indicators, four anxiety symptoms items; Attitudes Towards Medical Staff, three items concerning attitudes towards the doctors/midwives; Avoidance with three items tapping into avoidance-type behaviour; and Baby Concerns, which asks three questions about the unborn baby's welfare (see, Brunton, Dryer, Saliba and Kohlhoff, 2017a, Brunton, Dryer, Saliba and Kohlhoff, 2017b; for details on the full development and initial validation).

The PrAS improves on existing scales by its development on sound theoretical and psychometric grounds. Theoretically, the scale was derived from a comprehensive literature review ensuring it contains the main fears and worries characteristic of this anxiety (e.g., childbirth, appearance, baby concerns). The PrAS also includes subscales identified as important characteristics/dimensions of pregnancy-related anxiety but not previously considered by existing measures. These include subscales such as Acceptance of Pregnancy, which assesses pregnancy wantedness. Likewise, the inclusion of the Anxiety Indicators subscale provides more depth to the evaluation of this anxiety. Assessment of anxiety symptomology is another facet rarely considered by existing scales, however, is an important consideration given that anxiety is always accompanied by physical symptoms (Healey, 2006). Therefore, the PrAS is a more comprehensive assessment of this unique anxiety than previously offered by other existing scales.

In an initial validation study of the PrAS (Brunton, et al., 2017b), the scale was shown to have good content validity, with the items/subscales found by an Expert Review Panel to be consistent with the key characteristics of pregnancy-related anxiety. The scale's construct validity was confirmed using a variety of statistical techniques (i.e., Principal Components Analysis, correlation, multiple regression). These statistical techniques provided support for the underlying principle of construct validity, i.e., the variable of interest was found to correlate more highly with similar measures than with measures of different traits (Campbell & Fiske, 1959). Finally, the PrAS's ability to predict group membership further supports its concurrent validity (Brunton, Dryer, Saliba and Kohlhoff, 2017a, Brunton, Dryer, Saliba and Kohlhoff, 2017b). The initial validity results and the good to excellent internal consistency reliability both overall and at subscale level (ranging from α 0.80 to 0.93, evaluated on several samples) point to the PrAS as having promising psychometric properties. Notwithstanding this initial validity evidence, to add to the accumulating validity evidence, further examination of the scale's psychometric properties will be done by more thorough statistical techniques. Specifically, Rasch analysis will provide more detailed psychometric data, as described below. This further examination, which is the aim of the current article, means the PrAS will only be one of a few pregnancy specific scales that have been evaluated using Rasch analysis.