Biopsychosocial Correlates of Subjective Cognitive Decline and Related Worry in the Canadian Longitudinal Study on Aging

In 2018, it was estimated that over 50 million people were living with dementia globally, and this number is predicted to increase to over 150 million by 2050 [1]. Since there is currently no cure, focus has shifted toward prevention and improving quality of life for persons living with dementia [2]. A review by Blackburn et al. [3] suggests that using a measure of subjective cognitive decline (SCD) may allow us to identify early stages of dementia years before clinical diagnosis. SCD is defined as a self-perceived decline in cognition among otherwise cognitively healthy people [4]. The idea that SCD may be linked to future dementia, as an intermediate between normal cognitive aging and mild cognitive impairment (MCI), was first proposed by Reisberg [5]. Since this initial acknowledgment of SCD, much research has been done to better understand the link between SCD and dementia [6]. A study by Reisberg et al. [7] found that the risk of developing MCI or dementia within 7 years is greater for individuals with SCD compared to individuals without. Additionally, a study by van Harten et al. [8] found that reporting consistent SCD increased the risk of MCI. Rabin et al. [9] working model outlining the course of cognitive decline highlights that SCD can occur in the late stage of preclinical Alzheimer’s disease (AD), before detectable impairment on standardized tests. Once cognition declines to a point where it is detectable on standardized tests, an individual shifts from preclinical AD to MCI [9].

There is increasing recognition of the potential importance of SCD-related worry as a risk factor for future objective cognitive decline. Research by Jessen et al. [10] has found that individuals reporting SCD-related worry have almost double the risk of developing dementia, compared to individuals reporting SCD without worry. Further research by Jessen et al. [11] found that subjective memory impairment (also known as SCD) is only associated with an increased risk of dementia when it is accompanied by a concern, while people reporting SCD without worry are at no greater risk than controls. This finding suggests that SCD without concerns may be associated with normal age-related decline and is not an indicator of future objective decline [12]. Based on these and similar findings, worry or concern related to SCD has been included as one of the SCD-plus criteria, that is, a feature thought to increase the likelihood of progression of SCD to AD [12, 13]. Furthermore, Wagner et al. [14] found that dementia risk increased with consistent subjective complaints and was even higher for people who also expressed worry about their complaints. In this study [14], the risk of developing AD in participants with consistent SCD but without worries was double that of participants without SCD, and the risk for participants with consistent SCD with worries was almost four times that of participants without SCD. Snitz et al. [15] also found that SCD with worry was associated with progression to MCI, whereas SCD without worry was not.

A biopsychosocial model of dementia suggests that cognition is not solely determined by biological and physical health factors, but that psychological and social factors also impact cognitive health and dementia risk [16]. Abundant evidence exists linking biological and health factors with higher risks of cognitive decline including age, genetic factors, sensory loss, obesity, smoking, physical inactivity, and hypertension [17]. Recently, in their report of the Lancet Commission on dementia prevention, Livingston et al. [18] identified potentially modifiable risk factors for dementia which include physical health factors (i.e., diabetes, hearing loss, hypertension, obesity) as well as psychosocial factors such as education, depression, and social isolation. While there is a robust body of research highlighting the role of biological and health risk factors of cognitive decline, less research has focused on psychosocial correlates. Potential psychosocial correlates of cognitive decline include, but are not limited to, depression, social support, caregiver status, and personality traits [16].

A considerable amount of evidence shows that depression is associated with increased risk of cognitive decline [19‒21]. It has been suggested that there is a dose-dependent relationship between depressive episodes and risk of MCI and dementia, with each depressive episode being associated with a 14% increased risk for dementia [22]. Similarly, a study by Wilson et al. [23] found that compared to those without symptoms of depression, the rate of cognitive decline among individuals with 4 symptoms of depression was 20% more rapid.

In addition, certain personality traits have also been associated with cognitive health and aging. The five-factor model of personality categorizes personality into five domains: extraversion, conscientiousness, agreeableness, openness, and neuroticism [24] (note that some established personality scales measure emotional stability, the inverse of neuroticism, in lieu of neuroticism). The three personality traits most linked to cognition are conscientiousness (i.e., being organized, dependable, and motivated), openness (i.e., being curious, creative, and emotional), and neuroticism (i.e., becoming angry easily, being anxious, or depressed) [25, 26]. Research has found that high neuroticism is associated with worse cognitive functioning, while high openness is associated with better cognitive functioning [27]. Longitudinal studies have also found that low conscientiousness increases the risk of future dementia [28, 29]. Ausén et al. [30] identified progressive patterns of personality change that coincided with cognitive decline, such that increased personality changes were seen during the transition from normal aging, to SCD, to MCI. These changes related to the level of cognitive decline include increased proneness to stress, anxiety, and agitation, and lower extraverted behavior.

Indicators of stress have also been linked to cognitive aging and risk for dementia. Forrester et al. [31] proposed a framework of minority stress which emphasizes that minority stressors (i.e., racism, homophobia, stigma, discrimination, etc.) result in increased stress, which negatively impacts cognitive aging outcomes. Further, minority stress has been acknowledged as a risk factor for cognitive decline for lesbian, gay, bisexual, and transgender individuals such that lesbian, gay, bisexual, and transgender individuals may experience accelerated cognitive decline [32]. Consistent with the framework for minority cognitive aging, Stinchcombe and Hammond [33] found that racial minority status, as well as lower perceived social standing, was associated with lower cognitive performance. In a separate study and using a sample of adults aged 65 and older, it was found that higher subjective social status is positively associated with episodic memory [34]. Similarly, evidence shows that increasing functional social support (i.e., one’s perception that they are loved, cared for, and valued and that their needs are being fulfilled) and emotional support (i.e., having someone to talk to) may have a protective effect against cognitive decline, whereas increasing instrumental social support (i.e., the quantity of support) did not [35, 36]. Other stressors, such as caregiving, have been associated with cognitive health, such that caregivers show lower cognitive test performance relative to noncaregivers [37].

Based on the existing literature describing risk and protective factors of dementia, the present study aimed to identify prospective correlates of SCD in a sample of midlife and older Canadians. It was hypothesized that, in addition to commonly studied biological and health determinants of cognitive aging (e.g., poor sensory function, self-rated health), psychosocial variables (e.g., depression, personality traits, social support) would also emerge as correlates of SCD in the sample. For example, given robust evidence linking mental health and cognition [20], we hypothesized that increased symptoms of depression would increase the risk of SCD. Based on evidence linking personality and cognition (e.g., [25, 26]), it was hypothesized that emotional stability, conscientiousness, and openness would be negatively associated with SCD. We expected that individuals who experience minority stress or report low social support would be more likely to report SCD. Finally, among those who report SCD, we hypothesized that a similar pattern of results would emerge when participants’ degree of SCD-related concern was the outcome.

Data reported here are from two time points (i.e., baseline and first follow-up) of the Canadian Longitudinal Study on Aging (CLSA), a 20-year national study following more than 50,000 Canadians aged 45–85 at baseline. Recruitment began in 2010 with baseline data collection taking place between 2011 and 2015. Data are collected every 3 years for up to 20 years or until participant death, with the first follow-up taking place between 2015 and 2018 [38]. Recruitment for the CLSA included using a subset of the Canadian Community Health Survey Healthy Aging, provincial health care registration databases, and random digit dialing of landlines. To be eligible, participants must speak either English or French and not have a cognitive impairment at baseline. Other exclusion criteria included individuals living in Canadian territories or certain remote areas, living on First Nations reserves or First Nations settlements, being a full-time member of the Canadian Armed Forces, and institutionalized populations. Informed consent was provided by all participants before participating in the study [38].

The CLSA consists of two cohorts, a comprehensive cohort (n = 30,097) and a tracking cohort (n = 21,241). The tracking cohort completed telephone interviews, whereas the comprehensive cohort underwent face-to-face in-home interviews, as well as in-depth data collection at data collection sites located in 11 cities across Canada. Because certain measures of interest to this study were only collected within the comprehensive cohort (e.g., personality traits), only data from the comprehensive cohort were analyzed in this study. Data collected in the comprehensive cohort at baseline included questions regarding health and demographics as well as physical, psychological, and cognitive testing. To maintain participant retention, 18 months after each data collection, all participants are contacted by phone to complete a 30-min Maintaining Contact Questionnaire (MCQ) [38]. All participants gave written informed consent. Ethical review of the CLSA protocol was conducted by the Ethical, Legal, and Social Issues Committee, falling under the jurisdiction of the Canadian Institutes of Health Research (CIHR), and research ethics board approval was then acquired from each research site. The University of Ottawa REB approved the analyses presented here.

At first follow-up, participants were asked whether they had experienced SCD with the question “Do you feel like your memory is becoming worse”? The possible response options were yes and no. If participants responded yes, they were then asked, “Does this worry you”? This follow-up question gave a measure of SCD-related worry and had 5 possible answers including strongly agree, agree, undecided, disagree, and strongly disagree. To improve model fit in the present study, the two ends of the response scale with small cell sizes (strongly agree and strongly disagree) were merged with their nearest neighbor (agree or disagree, respectively), resulting in three categories of responses: strongly agree/agree, undecided, and disagree/strongly disagree. Many studies have used a single question to measure of SCD [39, 40], as well as including a measure of worry within the responses [10, 14].

Age was recorded as a continuous variable. At baseline, participants were asked to identify as male or female; as such, we were unable to determine if participants reported their sex (i.e., biological attributes) or gender (i.e., socially constructed roles, behaviors, expressions, and identities). For purposes of this study, we refer to participants as men and women (gender). Participants were asked to report their total household income with possible responses falling into five categories: less than 20,000 CAD, 20,000 CAD to 50,000 CAD, 50,000 CAD to 100,000 CAD, 100,000 CAD to 150,000 CAD, or 150,000CAD or more. Participants reported their highest level of education. Responses were collapsed into the following four categories: < secondary school education, secondary school graduate, some postsecondary, and postsecondary graduate. Marital/partner status was also asked with possible options including single, married/common law, widowed, divorced, and separated. In this analysis, widowed, divorced, and separated were combined into one category. Race/ethnicity was self-reported and categorized as White, Black, and other non-White. Participants were asked if they self-identify their sexual orientation as heterosexual, homosexual, or bisexual.

Self-reported general health was recorded by asking participants if, in general, they would say their health was excellent, very good, good, fair, or poor. Hearing and vision were both self-reported using the same scale of excellent, very good, good, fair, or poor. Body mass index (BMI) (kg/m²) was calculated using participants’ weight and height. Participants were also asked to report health professional-diagnosed hypertension. Alcohol use in the past 12 months was measured using a categorical variable with responses combined into the categories: regular drinker (at least once a month), occasional drinker, and did not drink in the last 12 months. Participants were also asked if they had smoked at least 100 tobacco cigarettes in their life and if so, their current frequency of smoking (i.e., daily, occasionally, not at all). Based on the participants’ smoking habits, they were categorized as nonsmokers (≤100 cigarettes), former smokers (>100 cigarettes but does not currently smoke), or current smokers (>100 cigarettes and smokes occasionally or daily).

The Physical Activity Scale for the Elderly (PASE) [41] was used to measure participants’ physical activity. The PASE is a brief scale designed to assess physical activities such as walking, housework, yard work, and caring for others, done over the past 7 days. The questionnaire asks both frequency of activities and duration. A measure of physical activity is calculated based on frequency of activity (hours/day over the past 7 days), multiplied by the activities PASE weight (determined by its intensity level). All scores are then summed, with possible scores ranging from 0 to 739 and higher scores indicating greater physical activity. The PASE has high test-retest reliability (r = 0.75) and strong convergent validity [41].

The Rey Auditory Verbal Learning Test (RAVLT) (immediate and 5-min delayed recall) was used to measure baseline memory [42]. The original RAVLT has been found to have good test-retest reliability (r = 0.51–0.86) [43] and is widely used in neuropsychological testing [44]. An abbreviated version of the RAVLT was administered in the CLSA, consisting of only trial 1 of the 5 RAVLT trials and one delayed recall trail [45]. The RAVLT required participants to listen to a list of 15 words and immediately recall them (90-s recall period). After 5 minutes, during which participants completed the animal fluency task and the mental alternation test, they were asked to recall as many of the initial words as they could (60-s recall period). For each test, the participant received one point for each correct word that was recalled, creating two scores between 0 and 15. For this analysis, the immediate and delayed scores were combined to create a total score between 0 and 30. A description of participants’ performance on this task in the CLSA is reported elsewhere [45, 46].

Depressive symptoms were measured at baseline using the Center for Epidemiologic Studies Short Depression Scale (CESD-10) [47]. The CESD-10 contains 10 questions assessing depressive symptoms over the past week. The following four response options are given for each question: all of the time (5–7 days), occasionally (3–4 days), some of the time (1–2 days), or rarely or never (less than 1 day). Total scores ranged from 0 to 30, with a cutoff score of 10 indicating a negative screen for depressive symptoms and scores of 10 or more indicating a positive screen for depression [47]. The CESD-10 has demonstrated good reliability and validity, as well as adequate but acceptable sensitivity and specificity in identifying depression [48, 49]. Additionally, the CESD-10 has been found to measure depressive constructs equivalently across various populations regardless of age, level of education, or language used during administration (i.e., French or English) [50].

The Ten-Item Personality Inventory (TIPI) was used to measure personality traits at baseline [51]. This scale measures the big-five personality traits which include extraversion, agreeableness, conscientiousness, emotional stability (the inverse of neuroticism), and openness to experiences. Participants were asked questions regarding their personality with possible responses ranging from strongly disagree to strongly agree (1–7). For each personality trait, two items were averaged, with one being reversed scored and higher scores indicating more of that trait. The TIPI has been described as having adequate convergent validity (mean r = 0.77) and test-retest reliability (mean r = 0.72) [51].

Social support availability (SSA) was measured at baseline using the Medical Outcomes Study Social Support Survey (MOS-SSS) [36]. This validated scale includes 19 items with possible answers as follows: none of the time (1), a little of the time (2), some of the time (3), most of the time (4), and all of the time (5). The MOS-SSS produces an overall score (0–100) reflective of perceived social support, with higher scores indicating greater overall perceived support. The MOS-SSS has been found to have high convergent validity (range r = 0.72 to r = 0.90) and discriminant validity, as well as high internal-consistency reliability (range r = 0.91 to r = 0.97) [36].

During the first MCQ, participants were asked to report their perceived social standing in their community using the MacArthur Scale of Subjective Social Status [52]. Participants were asked to think of a ladder with 10 steps, with the bottom (step 1) representing people with the lowest social standing in their community and the top (step 10) representing people with the highest social standing in their community. They were then asked to place themselves on this ladder (minimum 1, maximum 10). The MacArthur Scale of Subjective Social Status has been found to have moderate test-retest reliability in the community (k = 0.58, ICC = 0.64) with reliability being better among older adults (≥55 years) [53].

Caregiving status was assessed by asking a series of questions regarding various caregiving roles, which resulted in participants being classified as a caregiver or a noncaregiver. Caregivers are defined as individuals who provided assistance (excluding financial assistance) in the past 12 months to another person (e.g., friends, family) due to a health condition or limitation.

All data analyses were conducted using Stata/SE 15.1, College Station, TX, USA: StataCorp LLC. Differences in baseline characteristics between participants that reported SCD at first follow-up and those who did not are reported in Table 1.

The primary analysis consisted of a Poisson regression model treating SCD (i.e., dichotomous variable) as the outcome and biological and psychosocial variables from baseline as explanatory variables. The vce (robust) option was applied to obtain Huber-White robust estimates of the standard errors. In cases where the outcome is common (i.e., >10%), logistic models may overstate the relative risk (RR) [54]. Thus, a robust Poisson model was used, as it is the preferred choice for estimating RRs for binary response variables [55]. The comprehensive cohort sample size at first follow-up included n = 27,765 before removal of missing data. The proportion of missing data in the outcome variable, SCD, was 0.75% (n = 209). For explanatory variables, the proportion of missing data ranged from 0.03% to 6.21%, with the highest missingness appearing in household income (6.21%). After removing missing data through listwise deletion, the final sample size was n = 21,920. For hypothesis testing, we examined the crude relationship between baseline memory and SCD. Next, variables were entered into the model simultaneously, producing a multivariable model. We report the RR and corresponding 95% confidence intervals (CIs).

In a secondary analysis of participants who reported SCD, the amount of worry related to their SCD was examined (n = 12,707). Using an ordinal logistic regression model, the relationship between potential biopsychosocial correlates and the level of SCD-related worry (outcome) were explored. Variables were entered into the model simultaneously. After the removal of participants who did not respond to the question regarding SCD-related worry (n = 13), the analytic sample for the secondary analysis was n = 12,694. Odds ratios (ORs) and corresponding CIs are reported. Alpha (α) was set to 0.05.

Participant characteristics are presented in Table 1. More than half of the sample (58%) reported SCD at follow-up. Statistically significant differences (p < 0.05) were found for all variables except for education, sexual orientation, hypertension, and caregiving status. Of those who reported SCD, 49.1% were men and 50.9% were women. Participants who reported SCD were also older (mean age 62.4 compared to mean age 61.3). Furthermore, participants who reported SCD tended to have lower self-rated general health and lower levels of physical activity and were more likely to have hearing and vision problems, as well as a positive screen for depression, in comparison with individuals who did not report SCD.

Results from the Poisson regression model treating SCD as the outcome variable are presented in Table 2. In the crude analysis, baseline memory was negatively associated with SCD at follow-up, with each additional recalled word being associated with a small reduction in the risk of SCD (RR = 0.99, CI: 0.99–1.00). No relationships were found between SCD and most health and lifestyle variables, including hypertension, vision problems, physical activity, alcohol intake, and current smoker status. Statistically significant results include a positive association between age and SCD, such that increasing age was associated with a greater risk of SCD after controlling for covariates, though the RR was 1.005 (CI: 1.00–1.01). In terms of marital status, participants currently in a relationship (married/common law) (RR = 1.10, CI: 1.05–1.15) or with a history of a married/common law relationship (widowed/divorced/separated) (RR = 1.08, CI: 1.02–1.13) were at greater risk of SCD when compared to their single counterparts. There was also a small significant negative association between BMI and SCD (RR = 0.99, CI: 0.99–0.99) such that with each increase in BMI score, the risk of SCD decreased.

In comparison with the lowest income bracket, individuals with higher income reported a greater risk of SCD. A similar relationship was found between education and SCD; postsecondary school graduates were at greater risk of SCD when compared to those with less than a secondary school education. Women were also at greater risk of SCD when compared to men. No statistically significant associations were found between SCD and sexual orientation, race, and SSA. A negative relationship between perceived social standing and SCD was identified such that as one’s perceived social standing increased, they had a reduced risk of SCD.

Self-reported hearing problems were associated with an increased risk of SCD (RR = 1.09, CI: 1.06–1.12). Additionally, self-rated general health had a negative association with SCD, with those who reported greater general health having a reduced risk of SCD. No associations were found between caregiving status and SCD. A positive screen for depression was significantly related to SCD (RR = 1.13, CI: 1.10–1.16). All personality traits (i.e., extraversion, agreeableness, conscientiousness, emotional stability, and openness) were negatively associated with SCD, with conscientiousness having the largest association (RR = 0.95, CI: 0.94–0.96). For example, more conscientious individuals were at lower risk of SCD.

Results from the ordinal logistical regression model treating SCD-related worry as the outcome are presented in Table 3. When compared to men, women had increased odds of reporting greater SCD-related worry (OR = 1.33, CI: 1.23–1.44). Interestingly, among individuals who reported experiencing SCD, those who were older had small decreased odds of reporting SCD-related worry (OR = 0.99, CI: 0.98–0.99). Higher baseline memory was associated with reduced odds of SCD-related worry (OR = 0.98, CI: 0.97–0.99), along with higher BMI (OR = 0.99, CI: 0.98–1.00). Participants with higher levels of education had an increased odds of reporting SCD-related worry, though the results from the secondary school graduate/some postsecondary school category were not statistically significant. Further, no relationship was found between SCD-related worry and being married/common law, though participants who were widowed/divorced/separated had increased odds of reporting SCD-related worry compared to single participants. No relationship was found between income and SCD-related worry.

Caregivers had increased odds of reporting SCD-related worry (OR = 1.11, CI: 1.04–1.19). A positive screen for depression was also associated with increased SCD-related worry (OR = 1.49, CI: 1.34–1.66). No relationship was found between agreeableness or openness and SCD-related worry. Extraversion, conscientiousness, and emotional stability were associated with reduced odds of SCD-related worry. Within the personality traits, emotional stability showed the strongest association, with individuals who scored higher on emotional stability having reduced odds of reporting SCD-related worry (OR = 0.86, CI: 0.84–0.89).

In terms of minority stress variables, bisexual participants had an increased likelihood of SCD-related worry (OR = 1.90, CI: 1.09–3.31) compared to heterosexual participants. No association was found between SCD-related worry and homosexuality within our sample. Similarly, no relationships were found between race, perceived social standing, or SSA and SCD-related worry.

A robust body of evidence exists highlighting biological and health risk factors for SCD, whereas less research to date has focused on psychosocial correlates of SCD and SCD-related worry. In this study, we sought to identify prospective biopsychosocial correlates of SCD and SCD-related worry in a sample of midlife and older adults. First, we observed that objectively assessed memory was prospectively negatively associated with SCD at first follow-up in the CLSA, suggesting that SCD may be indicative of memory deterioration across time, as measured by standardized neuropsychological tests. While there are inconsistent findings in the literature with respect to the relationship between SCD and cognitive functioning [2, 56], our preliminary findings demonstrate a link between lower memory scores and memory concerns 3 years later.

While there was a clear association between objective memory and SCD within our sample, many of the commonly studied risk factors for cognitive aging and dementia were not associated with SCD risk or ran contrary to the expected direction of the associations. For example, factors that have been found to have a negative association with cognitive decline, such as income and education [31], were found to be positively associated with SCD within our sample. Although contrary to most community-based studies, a few studies have reported similar findings regarding education and SCD. For example, in a longitudinal study Comijs et al. [57] found that individuals who reported SCD but did not show an objective cognitive decline tended to have higher levels of education. They suggested that individuals with higher education may express memory complaints due to overall decline in well-being and a greater sensitivity to cognitive changes [57]. Similarly, van Oijen et al. [40] suggested that individuals with high levels of education may be able to notice subtle declines in cognitive abilities and would thus be more likely to report SCD compared to people with lower educational levels. Furthermore, physical activity, a known protective factor against cognitive decline [58], and other health-related risk factors such as hypertension, alcohol intake, and smoking were not associated with SCD in our multivariable model.

These findings suggest that, rather than being related to common risk factors for dementia, SCD may instead be associated with psychosocial factors such as depression, perceived social status, and personality traits [56]. Our results show that psychosocial factors are more consistently associated with subsequent SCD. Specifically, a positive screen for depression was associated with an increased risk of SCD. This aligns with previous research findings that depression accelerates changes in the brain that lead to a decline in cognitive abilities and may be an early indicator of impairment [21]. On the other hand, since depression was associated with SCD in our sample, when holding baseline memory constant, it is possible that SCD may also be indicative of individuals’ mental health in addition to their cognitive health. Similar findings stating that SCD may be a reflection of depression rather than objective cognitive decline have been reported [59, 60]. Other research has suggested that negative effect and the associated lower level of self-esteem which results from mood disorders, such as depression, may increase the likelihood that individuals will negatively evaluate their cognitive abilities, or even result in poorer scores on tests of cognition [61]. Other studies with similar results have concluded that SCD without objective cognitive decline or impairment may reflect psycho-affective or health problems rather than an underlying cognitive impairment [57].

In our sample, as individuals’ perceptions of their social standing increased, their risk of SCD decreased. This aligns with previous research that has found associations between lower social standing and poorer cognition [33, 34]. This association may be explained by the impact of social stress on brain health, outlined in Forrester et al.’s [31] framework for minority stress, as well as an increase in stress-related biological factors which can impact cognition, and may have direct and indirect effects on cognitive health [62]. Surprisingly, we observed no associations between SCD and our minority stress variables of sexual orientation and race. In a recent paper, Flatt et al. [63] found that the prevalence of SCD in sexual and gender minorities in the USA was higher compared to non-sexual and gender minority adults, although this relationship was largely attenuated after accounting for depression. Similar findings have also been reported and may be explained by the minority stress model not applying to SCD, a subjective measure, after adjusting for other biopsychosocial variables [64]. The inclusiveness of our model (i.e., the inclusion of numerous biopsychosocial variables) may have led to attenuated relationships between minority stressors and SCD.

Interestingly, all five personality traits were associated with a reduced risk of SCD, with conscientiousness and emotional stability having the strongest negative associations. These results are in line with previous research that found that subjective memory complaints were associated with low conscientiousness and high neuroticism (i.e., low emotional stability) [26, 59, 65]. Conscientious individuals have been found to have better cognitive function and show less decline over time [26]. Luchetti et al. [26] even noted that the effect of conscientiousness on cognition was larger than other clinical and lifestyle risk factors, such as smoking and physical inactivity. Neuroticism has also been consistently associated with cognitive complaints in the literature [66]. Work by Graham et al. [67] found that neuroticism is negatively associated with cognitive resilience, stating that individuals with low neuroticism had better cognitive abilities than expected given the level of pathology. Further, progressive personality shifts during cognitive decline have been identified, such that an increase in negative emotions may be seen during the transition from normal aging to SCD and SCD to MCI. Therefore, an increase in personality traits associated with negative emotions (e.g., neuroticism) may occur earlier than an objective decline in cognition, meaning personality traits may be an early identifier of dementia risk [30].

Some research has questioned the role of SCD in predicting future progression to MCI and dementia [13] because the majority of individuals reporting SCD do not progress to cognitive decline [68, 69]. However, evidence shows that those who report worry associated with SCD are at a higher risk of future dementia [10, 70]. Despite these findings, there is a lack of research identifying individuals who are more likely to report SCD-related worry. Our second analysis sought to understand the characteristics of individuals who reported higher worry related to their SCD. Results showed that higher SCD-related worry was associated with being a woman, being younger, and having a positive screen for depression. Interestingly, no association was found between lesbian/gay identities and SCD-related worry; however, bisexual participants reported higher SCD-related worry in comparison with heterosexual participants. Other work has shown greater health disparities among bisexual people in comparison with heterosexuals, lesbians, and gays, attributable to bisexual-specific stigma, lack of visibility of bisexual identities, and lack of bisexual affirmative care and support [71]. Taken alongside our results related to sexual orientation and SCD, additional research is clearly needed to determine the relationship between minority stress and SCD/SCD-related worry. For example, it may be worthwhile to tease apart the role of minority stress and other forms of stressors as they contribute to SCD and SCD-related worry as minority stress may be unique and have a differential impact when compared to general life stress.

Furthermore, certain personality traits (i.e., extraversion, conscientiousness, emotional stability) were associated with reduced odds of reporting SCD-related worry. Emotional stability had the largest negative association with SCD-related worry. This may be due to the characteristics of this personality trait (i.e., less negative feelings, calm). These results are similar to previous studies which have found an association between high neuroticism (the inverse counterpart to emotional stability) and greater SCD and SCD-related worry [57]. Identifying and minimizing worry associated with SCD may be an important factor to reduce the risk of future cognitive decline.

The present study has multiple strengths, including the large population-based sample size from locations across Canada. Though, we wish to acknowledge that the large sample size may increase the likelihood of statistically significant results that are not clinically significant. Given that SCD is still an under researched area, even small effect sizes may be of interest as they may generate new hypotheses and advance our understanding of SCD. Additionally, the breadth and depth of the CLSA dataset allowed for a thorough examination of potential correlates of SCD, whereas existing studies are commonly limited to only a few explanatory variables. However, it is important to note that other confounding factors not included in this study may contribute to SCD (e.g., medication use, illness, and anxiety/stress). The inclusion of multiple psychosocial and minority stress variables provides new insights, given that these topics are understudied and yet very important within the concept of SCD. The prospective nature of our study is also a strength because we were able to examine associations between baseline characteristics and SCD 3 years later. Due to the longitudinal nature of the CLSA, future research may be able to identify potentially causal relationships between biopsychosocial variables, SCD, and objective cognitive decline.

Our study is not without its limitations. Most prominent is the brief measure of SCD and SCD-related worry, using a simple two-question approach. Although similar measures are commonly used [39, 40] and have the advantage of ease of use in large epidemiological samples such as the CLSA, many in-depth scales exist which may be able to better differentiate between individuals with and without SCD [69]. Further, due to cell sizes and statistical power, we did not explore the intersectional/multiplicative effects of multiple social locations on cognition (e.g., race, gender, and sexual orientation). With respect to minority stress, it may be useful for future research to examine participants’ self-reported minority stress experiences and not just their minoritized identities as they relate to cognition. Additionally, many measures within the study are self-reported (i.e., income, physical activity, vision, and hearing problems). The use of self-reported data introduces the possibility of discrepancies between participants’ reported and actual behaviors; for example, participants may over-report their physical activity level.

At baseline, the CLSA was comprised of a large portion of mid-aged healthy individuals, for whom SCD may not yet be predictive of cognitive impairment [69]. Instead, the relatively young age of the sample may explain the reported associations, or lack thereof, between psychosocial factors and SCD. Future longitudinal research should focus on the trajectory of individuals reporting SCD in terms of future cognitive decline and dementia risk, also considering the psychosocial correlates presented here.

This study builds on existing literature describing biological, health behavior, and psychosocial correlates of SCD and SCD-related worry. Many known risk and protective factors of cognitive decline were not associated with SCD within our sample. Rather, psychosocial variables (i.e., depression, perceived social status, and personality traits) showed a more consistent association with SCD in the sample. Greater SCD-related worry, which is believed to increase the risk of later dementia, was associated with specific personality traits, depression, age, gender, and sexuality. The results from this study support the importance of psychosocial factors in identifying individuals with SCD and, in addition to confirming previous work, highlights the contribution of personality traits and perceived social standing to cognitive complaints. Psychosocial variables, in addition to health variables, may warrant attention as part of standard cognitive screening.

All participants gave written informed consent before participation. Ethical review of the CLSA protocol was conducted by the Ethical, Legal, and Social Issues Committee, falling under the jurisdiction of the CIHR, and research ethics board approval was then acquired from each research site. The University of Ottawa REB approved the analyses presented here, approval number H-07-21-7271.

The authors have no conflicts of interest to declare.

This research was made possible using the data/biospecimens collected by the CLSA. Funding for the CLSA is provided by the Government of Canada through the CIHR under grant reference: LSA 94473 and the Canada Foundation for Innovation. This research has been conducted using the CLSA dataset Baseline Tracking Dataset version 3.2 and Baseline Comprehensive Dataset version 3.1, under Application Number 170321. The CLSA is led by Drs. Parminder Raina, Christina Wolfson, and Susan Kirkland. Ms. Shawna Hopper holds a master’s scholarship from the Social Sciences and Humanities Research Council of Canada. Ms. Nicole G. Hammond is funded by the Frederick Banting and Charles Best Canada Graduate Scholarship Doctoral Awards (CGS-D) program. This project was funded through a grant from the Alzheimer’s Society of Canada Research Program awarded to Dr. Arne Stinchcombe.

Shawna Hopper contributed to study design, conducted the statistical analyses, and drafted and revised the manuscript. Nicole Hammond contributed to the statistical analyses and revised the manuscript. Dr. Taler contributed to the study design and manuscript revisions. Dr. Stinchcombe secured funding, contributed to study design, secured research ethics board approval, contributed to the statistical analyses, and revised the manuscript.

Data are available from the CLSA (www.clsa-elcv.ca) for researchers who meet the criteria for access to de-identified CLSA data.

Disclaimer: The opinions expressed in this manuscript are the author’s own and do not reflect the views of the Canadian Longitudinal Study on Aging (CLSA).