Introduction

Apathy, an important component of Mild Behavioral Impairment, is a psychological syndrome characterized by lack of motivation, interest, goal-directed activity, and diminished emotion [8-12].  In contrast, the dysphoric component of depression is characterized by sadness, feelings of guilt, negative thoughts and feelings, helplessness, despair, pessimism, selfcriticism, anxiety, and suicidal thoughts [13]. In older adults and people with neurologic illnesses, apathy has been linked to both functional and cognitive outcomes including incident frailty and dementia [14-20]. Current literature highlights apathy as a separate entity from depression, which is uniquely correlated (independent of depressive symptoms) to cognitive and functional decline in neurologic diseases like stroke, Parkinson’s disease (PD) and dementia [21-25]. Apathy predicts worse functional outcomes [26,27] and cognitive decline post stroke and in PD [19,27,28]. Further, recent studies indicate that apathy mediates the association between depression and cognitive decline in PD [ 29] and functional decline in persons with stroke and traumatic brain injury [30]. Funes and colleagues found that apathy mediates the relationship between depression and cognitive impairment [31].

Apathy is also a common behavioral phenomenon with a prevalence of 20-30% noted in community dwelling older adults [14,16,18]. Both van Dalen et al and Bock et al have identified an association between apathy in community dwelling older adults and incident dementia [15,32]. Furthermore, this association appears early in the dementia pathway as we found that apathy was associated with incident predementia syndromes [16]. Predementia syndromes are a constellation of cognitive deficits in varied domains that do not meet the criteria for dementia but are highly predictive of progression to dementia and may represent a prodromal phase [33-35]. Given an increasing body of literature that links apathy with incident dementia in community dwelling older adults, [15,20] assessment of apathy will be important in preventing or slowing progression to dementia, as it will allow for the identification of individuals who may benefit from treatment targeted to early neurodegenerative processes.

The majority of studies in community dwelling cohorts have employed varied measures of apathy, most of which are based on subscales of existing depression tools [14,16,18,20,36]. While an apathy specific tool like the Apathy Evaluation Scale is useful in clinical trials where apathy is the primary outcome, participants may find it burdensome to complete both depression and apathy assessment tools due to overlapping symptoms [37]. Moreover, assessing apathy based on existing depressive tools allows investigators to explore apathy in relation to other domains like dysphoria and cognitive concerns in well-established cohorts without increased questionnaire burden.

The Geriatric Depression Scale is a widely used tool, which was developed to assess depression, as a multi-domain construct, in older adults [38,39]. Data reduction methodologies, such as principal component analysis, have been frequently applied to identify distinct factors in the 30-item GDS Long Form [40-47]. These studies consistently identified dysphoria and withdrawal/ apathy factors, even though the total number of factors varied across cohorts. A few studies have conducted confirmatory factor analysis to validate the factor structure of the GDS across cohorts [41,42,44]. However, the primary objective of these studies was not to parse out apathy, an important predictor of cognitive and functional decline.

The goal of this study was to conduct a confirmatory factor analysis of the GDS, pooling 3 unique cohorts of older adults (community dwelling, international south Indian, and cognitive risk in primary care) in order to 1) develop a novel measurement model that distinguishes apathy from other domains of depression including dysphoria and cognitive concern and 2) evaluate if the measurement model distinguishes older adult populations with varied risk for cognitive decline. The resultant measurement model should identify higher levels of apathy, dysphoria, and cognitive concern in higher risk populations.

Methods

Sample Population

We pooled the baseline waves of three Albert Einstein COM older adult cohorts. All study protocols were approved by the Albert Einstein College of Medicine Institutional Review Board and participants completed written consent.

Community Cohort

A sample of 538 community dwelling older adults (≥65 years old) without dementia enrolled in the Central Control of Mobility and Aging Study from 2011-2017 were included in this study. The primary aims of this prospective cohort study are to determine the cognitive and neural predictors of mobility in late life. As previously described, participants were contacted by mail and telephone from population lists in Westchester County, NY [48]. A structured telephone interview was administered to potential participants to obtain verbal assent, assess medical history and rule out dementia [49]. Participants who passed the telephone interview received comprehensive neuropsychological, psychological, and mobility assessments as well as a structured neurological examination. Community participants were followed yearly.

Kerala Study

We included a sample of 238 participants enrolled in the Kerala-Einstein Aging Study (KES), a community-based cohort study that was originally established in 2008 and is based in the Kozhikode district in Kerala, India. In 2023, KES received new funding to focus on novel risk factors (apathy and traumatic brain injury) and the brain pathology for motoric cognitive risk syndrome (MCR), a predementia syndrome characterized by the presence of slow gait and subjective cognitive concerns in individuals without dementia or disability [50]. Participants similarly completed comprehensive neuropsychological, psychological, and mobility assessments as well as magnetic resonance imaging of the brain.

Cognitive Risk Primary Care

Our study also included 600 participants enrolled in the 5-Cog Battery to improve detection of cognitive impairment and dementia study, a randomized controlled trial in predominantly Black and Hispanic older adults from Bronx County, NY who are presenting to primary care with cognitive concerns [51]. The study will validate a novel 5-minute cognitive assessment coupled with an electronic medical record-embedded decision tree to overcome the barriers of current cognitive assessment paradigms in primary care and facilitate improved dementia care.

Geriatric Depression Scale

Participants in all three cohorts completed the 30-item Geriatric Depression Scale (GDS) Long Form.³⁹ The strength of the GDS in older adult populations is that it was developed for use in older adults including those with somatic and cognitive complaints. It utilizes a yes/no format and a score of 0–9 is considered normal; 10–19 indicates mild depression and ≥20 severe depression [52]. The GDS has been shown to be a reliable screening tool in people with mild cognitive impairment [53]. We utilized the GDS completed in the baseline assessment wave for all cohorts.

Statistical Analysis

Data from the three cohorts were pooled. After data harmonization, bivariate analyses by cohort were conducted using SPSS version 28 (SPSS Inc. Chicago, IL). Continuous variables were evaluated using ANOVA tests and categorical variables were evaluated using chi-square tests. Fisher’s exact test was used when the expected count for less than 5 in greater than 20% of the cells.

With the aim of partitioning apathy from other constructs that compose the GDS and with a PCA suggesting 3-component solution, we then conducted a confirmatory factor analysis (CFA) [54] using two R-based statistical analyses packages (i.e., less R and lavaan; RStudio, PBC, Boston, MA). Lavaan and lessR apply Maximum Likelihood Estimation (MLE), a nonlinear optimization algorithm commonly used in latent variable analyses tasks, allowing selection of parameter values that make the observed data most probable, given a sample space (i.e., a listwise inter-item correlation matrix). Informed by prior literature on the psychometric properties of the GDS, we used the CFA to parse apathy from two other, substantive components or factors reported to comprise the GDS: dysphoria and cognitive concern. We dropped items with low item-factor correlations and/or if items loaded strongly on both factors. Decisions to retain or drop items were assessed iteratively. Dysphoria, apathy, and cognitive concern scale scores were calculated as the mean of the items for each case. Given that individual items were dichotomous, computed scale scores were continuous, ranging from 0 to 1. Mean dysphoria, apathy, and cognitive concern scale scores for the community and memory clinic cohorts were then compared using independent t-tests.

Results

Descriptive Analyses

Table 1 highlights the distinct characteristics of the three cohorts included in the analyses. Most notably the cohorts significantly differed by: mean age, with the Kerala cohort being the youngest (68.9 ± 5.39 years); gender, with the Community cohort having more women (72.2%); ethnicity, with the Cognitive Risk Primary Cohort being most diverse; and years of education, with the Kerala cohort having the fewest years of education (9.9 ± 3.95 years). In addition, both the Kerala and Cognitive Risk Primary Care cohorts endorsed higher levels of depression (7.5 ± 6.15 and 7.5 ± 6.21, respectively) than the Community cohort. Indicators of locomotor and cognitive and motor function also significantly varied by cohort with, the Cognitive Risk Primary Care cohort having the slowest mean gait velocity (64.4 cm/s ± 19.14) and the highest proportion of participants with MCR (17.3%).

Table 1: Baseline Characteristics by Cohort.

Confirmatory Factor Analysis

Table 2 illustrates the item to factor correlations for the 3 factors: dysphoria, apathy, and cognitive concern. The factors were named based on the clinical relevance of the items that loaded highest on to these specific factors. All the dysphoria, apathy, and cognitive concern factors showed acceptable unidimensionality [55] with α=.76, .59, and .54, respectively.

Table 2: Confirmatory Factor Analysis item- Factor Correlation Matrix.

Factor Scores

Factor scores were calculated for the participants using the mean of the factor item responses, such the scores are continuous values between 0 and 1. To assess face validity, mean factors scores were compared by cohort. Overall, the Cognitive Risk Primary Care cohort had significantly higher mean dysphoria, apathy and cognitive concern scales (Figure 1). Also, the apathy scale score was elevated across cohorts, highlighting the prevalence of this risk factor. Table 3 shows the post hoc analysis. Of note, the Kerala and Cognitive Risk Primary Care cohorts did not differ significantly dysphoria or apathy score, suggesting that the Kerala cohort may have a similar risk profile.

Figure 1: Comparison of Baseline Scale Scores by Cohort.

Table 3: Bonferroni Multiple Comparisons of Mean Scale Scores by Cohort.