Effectiveness of sorting tests for detecting cognitive decline in older adults with dementia and other common neurodegenerative disorders: A meta-analysis

Highlights

• Sorting tests detect cognitive impairment in common neurodegenerative disorders.

• Sorting tests seem to be particularly suitable for dementia.

• Sorting tests appear suited to screening for cognitive impairment in older adults.

Abstract

The demand for simple, accurate and time-efficient screens to detect cognitive decline at point-of-care is increasing. Sorting tests are often used to detect the ‘executive’ deficits that are commonly associated with behavioural-variant frontotemporal dementia (bvFTD), but their potential for use as a cognitive screen with older adults is unclear. A comprehensive search of four databases identified 142 studies that compared the sorting test performance (e.g. WCST, DKEFS-ST) of adults with a common neurodegenerative disorder (e.g. Alzheimer’s disease, vascular dementia, bvFTD, Parkinson’s disease) and cognitively-healthy controls. Hedges’ g effect sizes were used to compare the groups on five common test scores (Category, Total, Perseveration, Error, Description). The neurodegenerative disorders (combined) showed large deficits on all scores (g -1.0 to -1.3), with dementia (combined subtypes) performing more poorly (g -1.2 to -2.1), although bvFTD was not disproportionately worse than the other dementias. Overall, sorting tests detected the cognitive impairments caused by common neurodegenerative disorders, especially dementia, highlighting their potential suitability as a cognitive screen for older adults.

Introduction

The number of older adults with a neurodegenerative disorder is increasing as the population ages (Ecomonics, 2009), with 131.5 million people predicted to be living with dementia by 2050 worldwide (Prince et al., 2015). Cognitive decline is a defining feature of dementia, but is also common in many other neurodegenerative disorders, such as Parkinson’s disease and motor neuron disease (Cui et al., 2015; Mihaescu et al., 2019; Muslimovi et al., 2007). Cognitive decline often occurs many years prior to receiving a formal diagnosis (e.g., in Alzheimer’s dementia; AD), highlighting the importance of early detection (Bäckman et al., 2005). Indeed, there are multiple benefits to detecting cognitive impairments early in their course, including (i) reduced hospital admissions, readmissions, and outpatient costs (McCarten et al., 2010; Torisson et al., 2013); (ii) lower rates of delirium, morbidity and mortality (Lee. et al., 2008); and (iii) improved psychological and behavioral symptoms, and carer outcomes (McCarten et al., 2010).

Accurate cognitive screens conducted at or near the point-of-care are recommended to assist with the early detection of cognitive decline (Robinson et al., 2015). These screens are designed to be quick and easy to administer, and provide immediate information about a patient’s risk of cognitive impairment. Effective screening can, in turn, inform interventions and other investigations, while also facilitating timely diagnoses (Borson et al., 2013; Robinson et al., 2015).

The most common point-of-care cognitive screen is the Mini Mental Status Examination (MMSE), which is quick to administer (8−15 min) and interpret (Folstein et al., 1975; Mitchell, 2013; O’Bryant et al., 2008; Sheehan, 2012). The MMSE detects dementia with 80% sensitivity and 81% specificity in memory clinic settings, but is less accurate when used to assess the cognitive decline associated with Parkinson’s disease and mild cognitive impairment (Athey et al., 2005; Hu et al., 2014; Mitchell, 2009). As with other commonly-used screens, such as the Montreal Cognitive Assessment (MoCA) and Addenbrooke’s Cognitive Exam (ACE), the MMSE assesses multiple cognitive domains (Mathuranath et al., 2000; Nasreddine et al., 2005). Although this feature is particularly pertinent to dementia because the diagnostic criteria require cognitive decline in two or more domains (e.g. AD; McKhann et al., 2011), it is less relevant when screening older adults for cognitive impairment. Instead of domain-specific scores, summary scores are used for screening and those derived from multidomain screens are not necessarily more sensitive than those from single-domain tests (Brodaty and Moore, 1997; Kingery et al., 2011; Larner, 2016; Summers et al., 2019).

Sorting tasks are one example of a single-domain cognitive test. Although often used as part of a comprehensive neuropsychological assessment (Strauss, 2006), they are not included in common cognitive screens, nor are they routinely used when screening older adults for cognitive impairment. Theoretically, sorting tests are thought to assess inductive reasoning but, clinically, they are often described as assessing ‘executive’ functioning, although the latter construct conflates a number of cognitive abilities (Floyd et al., 2010; Schneider and McGrew, 2018; Strauss, 2006).

Sorting tests assess a person’s ability to sort stimuli (e.g., cards) according to specific categories (usually colour, shape, number) and then switch between these categories (Feldman and Drasgow, 1959; Grant and Berg, 1948; Heaton et al., 1993). They are relatively quick to administer (range: 5–25 min.) and score (Strauss, 2006), making them well-suited to point-of-care cognitive assessments. A number of sorting tests have been developed over the years, starting with the Weigl Color-Form Sorting Test, followed by the Berg-Wisconsin Card Sorting Test, the Verbal Visual Test, and the California Card Sort Test, which has since been incorporated into the Delis Kaplan Executive Functioning System (Delis et al., 2001, 1992; Feldman and Drasgow, 1959; Goldstein and Scheerer, 1941;Grant and Berg, 1948). Most of these tests generate composite scores, labelled Category scores (number of categories correctly sorted) or Total scores (also termed ‘global’ or ‘total correct’ scores, tallying the number of successful sorts). Many sorting tests also generate Perseveration scores (number of repeated responses after failing to shift category) and Error scores (when sorts do not fit a single category) (Delis et al., 2001; Heaton et al., 1993). Some also produce a Description score, which assesses a person’s ability to articulate the category or rule underpinning their sort (e.g., colour, shape, number, Delis et al., 2001).

Sorting tasks are amongst the most sensitive to brain damage (Delis et al., 1992; Reitan, 1993) and have been used to detect the cognitive decline caused by a variety of neurodegenerative disorders, such as dementia (Byrne et al., 1998), Parkinson’s disease (Hobson et al., 2007; Paolo et al., 1996) and amyotrophic lateral sclerosis (Barbagallo et al., 2014; Evans et al., 2015). In particular, sorting tests are often used in the assessment of frontotemporal dementia (FTD) because deficits in reasoning and ‘executive’ functioning are thought to be a distinguishing feature of FTD (Strauss, 2006). Perseverative speech has additionally been reported in behavioural-variant FTD (bvFTD), with the Perseveration score provided by some sorting tests potentially measuring this characteristic (Strauss, 2006). Sorting tests have therefore been used by clinicians to differentiate between AD and FTD, despite limited research support for this practice (Hutchinson and Mathias, 2007; Roca et al., 2013).

Research comparing the sorting test performance of older adults who have a neurodegenerative disorder to that of cognitively-healthy peers is now quite extensive. However, the collective findings have yet to be evaluated. Consequently, our understanding of whether the most common neurodegenerative disorders perform differently on these tests, and whether a specific test best detects the cognitive decline associated with these disorders, is limited. The current meta-analysis therefore examined whether sorting scores differentiate between older adults with a neurodegenerative disorder and their cognitively-healthy peers in order to assess the potential usefulness of sorting tests as a cognitive screen in point-of-care settings. Only the most common older-age neurodegenerative disorders were considered, namely, Parkinsonian disorders (PD), motor neuron disease (MND), mild cognitive impairment (MCI), and ‘other’ disorders (human immunodeficiency virus [HIV], normal pressure hydrocephalus [NPH], multiple sclerosis [MS], Huntington’s disease [HD]), as well as the most common dementia subtypes (AD, bvFTD, vascular [VaD], lewy body [LBD], semantic [SD], primary progressive aphasia [PPA], and not otherwise specified [dementia NOS]). Dementia was of particular interest because cognitive screens are commonly used to assist with its early detection. Given that deficits in ‘executive’ functioning are thought to characterise bvFTD, and sorting tests are commonly used to assess these deficits, this dementia subtype was a specific focus.