Measuring anxiety in late life: A psychometric examination of the Geriatric Anxiety Inventory and Geriatric Anxiety Scale

Highlights

• Convergent validity was strong for Geriatric Anxiety Inventory (GAI) and Geriatric Anxiety Scale (GAS) scores.

• Good internal consistency and adequate reliability were found for GAI and GAS scores.

• GAI scores had better discriminant validity with a health rating compared with GAS scores.

• Psychometric properties were decreased, in participants with average versus those with superior delayed recall.

Abstract

We examined the psychometric properties, internal scale reliability and validity, of two geriatric anxiety measures: the Geriatric Anxiety Inventory (GAI) and Geriatric Anxiety Scale (GAS). We also determined the extent to which memory ability influenced the psychometric properties of these measures. Older adult participants (N = 110; M age = 75 years) completed self-report, clinician-rated and diagnostic psychiatric measures and a neuropsychiatric battery. GAI and GAS scores had good internal consistency, adequate reliability, and strong convergent validity. GAI scores had better discriminant validity than GAS scores relative to a health rating. Both measures had strong associations with depression scores. Psychometric properties were decreased in participants with average delayed memory recall compared with those with superior recall. Both measures had good psychometric support, particularly in those with strong memory abilities. Psychometric performance characteristics indicate that the GAI and GAS may be good alternatives to anxiety measures not designed specifically for older adults.

Introduction

Anxiety disorders represent the most common late-life mental health illnesses (Gum, King-Kallimanis, & Kohn, 2009). Anxiety symptoms are also pervasive, afflicting 15–56% of older adults in clinical settings (Bryant, Jackson, & Ames, 2008). Assessing anxiety with psychometrically sound geriatric anxiety measures is critical given the harmful health consequences of anxiety including increased risk of mortality (Van Hout et al., 2004), high comorbidity with depression (King-Kallimanis, Gum, & Kohn, 2009), and greater disability (Porensky et al., 2009). Moreover, identifying subthreshold anxiety symptoms is crucial given they are related to lower cognitive performance (Beaudreau and O’Hara, 2008, Beaudreau and O’Hara, 2009), disability and diminished quality of life (de Beurs et al., 1999), presence of more medical illnesses, and poor sleep (Mehta et al., 2003).

Many anxiety measures validated on younger samples have limitations for use with older adults. For instance, the Beck Anxiety Inventory (BAI; Beck & Steer, 1993) has good internal consistency in older samples (e.g., Wetherell & Gatz, 2005), but relies heavily on self-reported somatic symptoms, making the BAI a poor choice for use with medically ill older adults receiving home-care (Diefenbach, Tolin, Meunier, & Gilliam, 2009) or those seen in primary care settings (Gould, Beaudreau, & Huh, 2013). Reverse-scored items also pose a challenge because they can be confusing to some older adults. In the case of the Penn State Worry Questionnaire (Meyer, Miller, Metzger, & Borkovec, 1990), reverse-scored items load on a separate factor (Hopko et al., 2003), resulting in diminished reliability and validity.

Two recently developed, elder-specific measures, the Geriatric Anxiety Inventory (Pachana et al., 2007) and the Geriatric Anxiety Scale (Segal, June, Payne, Coolidge, & Yochim, 2010), assess a broad array of anxiety symptoms and show promise for use in varied clinical and research settings. Modeled after the Geriatric Depression Scale, the 20-item Geriatric Anxiety Inventory (GAI) uses a simple response scale of ‘agree’ or ‘disagree’. The 30-item Geriatric Anxiety Scale (GAS) measures somatic, cognitive, and affective anxiety symptoms rated on a dimensional, four-point scale of anxiety severity. Psychometric studies support excellent internal consistency and strong convergent validity with measures of anxiety and worry for the GAI (Byrne et al., 2010, Diefenbach et al., 2009, Pachana et al., 2007) and for the GAS (Segal et al., 2010, Yochim et al., 2010). Discriminant validity has been demonstrated through small correlations (r > .40) with demographic characteristics and cognition for GAI scores (Byrne et al., 2010, Pachana et al., 2007, Yochim et al., 2010) and GAS scores (Segal et al., 2010, Yochim et al., 2010); however, more evidence of discriminant validity from depression and health-related variables is needed for both measures. An area in which the development of the GAI exceeds that of the GAS is the use of the GAI to identify the presence of anxiety disorders. Studies have identified a clinical cut score of >8 to detect the presence of any anxiety disorder in geropsychiatric patients (Pachana et al., 2007) and homebound older adults (Diefenbach et al., 2009).

With regards to cognition, the GAI has been used to detect anxiety in older patients with a range of memory abilities including those from a memory clinic (Byrne, Pachana, Arnold, Chalk, & Appadurai, 2008) and those in long-term care settings (Gerolimatos, Gregg, & Edelstein, 2013). Yochim and colleagues (2010) established that GAS scores in community-residing older adults were not strongly associated with reading ability or processing speed, but the extent to which cognition affects the reliability and validity of the GAS is unknown. The factor structure of the GAI was examined in individuals with dementia, cognitive impairment no dementia, and no impairment (Diefenbach, Bragdon, & Blank, 2013), but no studies have determined whether cognitive abilities affect the GAI's reliability or validity in a largely non-impaired sample.

The GAI and GAS show promise as geriatric anxiety measures. Previous validation studies have primarily examined validity of the GAI and GAS with self-report measures of anxiety, worry, or depression; and none with measures of posttraumatic stress symptoms. Previous studies also have not determined whether cognitive abilities are related to psychometric properties of the GAI or GAS or whether psychometric properties differ for the GAI and GAS depending on participants’ memory ability. The first and primary aim of the present study was to examine the psychometric properties, specifically internal consistency, item-total reliability, convergent validity, and discriminant validity of the GAI and GAS in a sample of community-dwelling older adults. The present study builds on Yochim et al. (2010) findings of their direct comparison of GAI and GAS scores with self-report measures of anxiety, depression, and health and a memory assessment. We replicated and extended the evaluation of GAI and GAS scores to include comparisons with a clinician-rated measure of anxiety severity, self-report measures of post-traumatic stress and worry, and a structured diagnostic interview. The second aim was to examine the extent to which memory functioning influenced the psychometric properties of the GAI and the GAS. The third aim was to identify clinical cut scores for the GAI and GAS compared with a structured diagnostic interview.