Depressive Symptoms Partially Mediate the Association of Frailty Phenotype Symptoms and Cognition for Females but Not Males

Frailty, which is present in approximately 10% of older adults (Furtado et al., 2019; Wang et al., 2019), is a health state associated with a significant risk of future impairment and adverse health outcomes, such as hospitalizations, mortality, and loss of activities of daily living (Margioti et al., 2019). However, the prevalence for of frailty differs for males and females, with females experiencing more frailty symptoms than their male counterparts (Kane & Howlett, 2021). While there is a growing body of research regarding frailty, its causes, and its consequences, there remains a limited understanding of the mechanisms by which frailty leads to poor health outcomes. Such information is crucial because it can help to identify mechanisms and risk factors for the development of related health conditions.

Frailty is characterized as a state of greater vulnerability to stressors due to an age-related decline in multiple physiological processes (Borges et al., 2019; Wang et al., 2019). In research practice, frailty can be defined as the presence of at least three of five indicators: symptoms of fatigue, little to no physical activity, weight loss, decreased grip strength, and a slow walking pace (Sargent et al., 2018; Wang et al., 2019). Frail older adults have a greater risk of cognitive decline and accelerated declines (Chen et al., 2018; Margioti et al., 2019). For instance, Chen et al. found that frail older adults had a significantly greater decline in global cognitive scores compared to non-frail older adults (Chen et al., 2018). Although, sex differences for cognitive trajectories of frail older adults should be considered, as one study found that the association between frailty and cognitive decline differed by males and females (Thibeau et al., 2019).

While it may be unclear what processes may lead frail older adults to experience greater cognitive decline, researchers have speculated that chronic inflammation, mitochondrial dysfunction, metabolomic markers, and mental health may play important roles in this association (Ma & Chan, 2020). For example, evidence suggests that frailty and cognition may be linked by the same physiological pathways and markers such as inflammation (Ma & Chan, 2020; Sargent et al., 2018). Canon and Crimmins found that C-reactive protein—an inflammatory marker—mediated the relationship between muscle quality and cognitive decline in women (Canon & Crimmins, 2011). In addition to these biological and physiological mechanisms, research suggests that mental health disorders, particularly depression, may partially explain the association between frailty and cognitive decline. For example, previous research has shown that frail older adults are at increased risk of developing depression and being depressed, which may be potentially explained by decreased social activity, physical activity, or an unknown shared vulnerability to physiologic and mental stressors (M. Lohman et al., 2016; M. C. Lohman et al., 2017; Mezuk et al., 2012). Further, there is a well-established link between depression and poorer cognitive performance over time (Donovan et al., 2017; Gallagher et al., 2018; Robertson et al., 2013). However, research may suggest sex difference for depression and cognitive decline. A previous study showed that females, but not males, experienced cognitive decline due to depression (Anstey et al., 2021). While evidence suggests that depression may play a role in the connection between frailty and cognitive decline, the extent to which depression or depressive symptoms may mediate this association is unclear, especially how these pathways impact males and females differently. Such information may be important for monitoring the progression of or preventing adverse health outcomes among frail older adults.

Since there remain few effective means to slow or prevent cognitive decline, understanding potential mechanisms that connect frailty and cognition is an important step to help identify individuals at greater risk of cognitive decline. A better understanding of the potential mechanisms by which these risk factors influence cognitive decline, and how these differ by sex, will further strengthen preventive efforts by providing targets for treatment or screening among frail older adults. This study aims to evaluate whether depressive symptoms mediate the relationship between frailty phenotype and cognitive function. A stratified sub-analysis was completed to understand how the pathways of frailty symptoms, depressive symptoms, and cognition change by sex.

Data from the Health and Retirement Study (HRS) was used to conduct this secondary data analysis that used longitudinal data from wave 2012 to 2016. The HRS is an ongoing, nationally representative prospective sample of individuals aged 50 years and older interviewed biannually since 1992 (Heeringa & Connor, 1995). Half of the sample from wave 2012 was administered the enhanced face-to-face interview which included measures of physical characteristics such as height, weight, gait speed, strength, and other indicators of physical functioning (Crimmins et al., 2013). A total of 8970 were selected for enhanced face-to-face interviews, of which 8512 of these individuals consented and were eligible to participate in the physical measures. Of those that consented, 3708 completed the walking test. In addition, three individuals with a walking test were excluded for not completing the grip strength test, giving us a final analytic sample of 3,705, of which, 2136 were female and 1569 were male. Of these 3705 at baseline, 3336 (90.0% of the baseline sample) were at wave 2014, and 2873 (77.5% of the baseline sample), were at wave 2016.

Table 1 displays descriptive statistics for covariates, depressive symptoms (2014) and cognitive scores (2012 and 2016) for the whole population, while comparing males and females. The analytic sample included 3705 participants, with 1569 being males and 2136 females. The mean age of participants was 75.4 years (standard deviation (SD) = 6.7 years), with males and females having similar ages (75.5 vs. 75.3; p value = 0.33). For both males and females, the vast majority of responded reported being white (85.0% vs. 83.7%). There were significant differences between males and females in report educational attainment (p value <0.001). For males, the highest reported educational attainment category was college and above (28.4%), while females reported high school (35.6%) as the most frequent educational attainment.

At baseline, on average, females reported more frailty symptoms compared to males (1.4 vs. 1.1; p value <0.001). Similarly, at wave 2014, females reported more depressive symptoms compared to males (1.5 vs. 1.1; p value <0.001). Males at both wave 2012 (21.0 vs. 21.7; p value <0.0001) and 2016 (20.6 vs. 21.0; p value = 0.02) had lower cognitive scores.

Table 2 shows the results of the mediation analysis comparing an additional reported frailty symptom. When adjusting for confounders, there was a statistically significant natural direct effect from frailty symptoms to cognition; respondents with a one-unit increase in frailty symptoms in 2012 had significantly lower cognitive scores in 2016 (β= −0.32; 95% CI: −0.51, −0.13). Individuals who exhibited all five symptoms experienced a 1.60 lower cognitive score compared to those with no symptoms. There was also a significant natural indirect effect from frailty symptoms to cognition through depressive symptoms, suggesting that among an increase in frailty symptoms, a one-unit increase in depressive symptoms was associated with a lower cognitive score by 0.04 (β= −0.04; 95% CI: −0.07, −0.01). The total effect of frailty symptoms on cognitive score was −0.36 (β= −0.36; 95% CI: −0.55, −0.17); meaning individuals with all five frailty symptoms would have a 1.80 lower cognitive score, on average, compared to individuals with no symptoms. The proportion of the association of frailty symptoms and cognitive function scores mediated by depressive symptoms (natural indirect effect/total effect) was 11.1%.

The stratified sub-analysis demonstrated different results for males and females (Table 3). When adjusting for confounders, there was a statistically significant natural direct effect from frailty symptoms to cognition; respondents with a one-unit increase in frailty symptoms in 2012 had significantly lower cognitive scores in 2016 for both males (β= −0.39; 95% CI: −0.63, −0.16) and females (β= −0.24; 95% CI: −0.44, −0.05); however, males had a 0.15 greater decline per frailty symptoms compared to females. There was also a significant natural indirect effect from frailty symptoms to cognition through depressive symptoms for females but not males. This suggested that among an increase in frailty symptoms, a one-unit increase in depressive symptoms was associated with a lower cognitive score by 0.03 (β= −0.03; 95% CI: −0.06, −0.00) for females. The total effect of frailty symptoms on cognitive score was significant, but different, for males and females. Males had larger TE with a −0.43 (β= −0.43; 95% CI: −0.66, −0.02) lower cognitive score for each increase in frailty symptom; meaning male individuals with all five frailty symptoms would have a 2.15 lower cognitive score, on average, compared to individuals with no symptoms. Females had a TE effect of −0.28 (β= −0.28; 95% CI: −0.47, −0.08) lower cognitive score for each increase in frailty symptom; meaning female individuals with all five frailty symptoms would have a 1.40 lower cognitive score, on average, compared to individuals with no symptoms. The proportion of the association of frailty symptoms and cognitive function scores mediated by depressive symptoms for females was (natural indirect effect/total effect) was 11.7%.

This study demonstrated that a greater number of frailty symptoms were significantly associated with lower cognitive scores in older US adults 4 years later. However, these results differed by males and females, with males having lower cognitive scores with every increase in frailty symptom compared to females. Further, we found that the influence of increased frailty symptoms on cognition was at least partially mediated by an increase in depressive symptoms, with depressive symptoms accounting for approximately 11.1% of the total effect of frailty symptoms on cognition. The mediating effective was also significant and similar for females, with 11.7% of the total effect of the total effect being mediated by depressive symptoms, but was not significant for males. These results help to provide evidence for the role of frailty symptoms for both males and females as a predictor of poorer cognitive health outcomes and contribute to our understanding of depressive symptoms as a potential mechanism in the pathway of frailty and cognitive decline for females.

Our findings that increased frailty symptoms were associated with worse cognitive scores are consistent with a previously published systematic review that examined longitudinal and cross-sectional studies and showed that frailty and pre-frailty were associated with either cognitive decline or cognitive impairment (Robertson et al., 2013). With our study examining increases in reported frailty symptoms (from 0 to 5 symptoms), it is important to understand how these results compare to the categorization schema for pre-frailty and frailty by Fried and colleagues (pre-frailty = 1–2 symptoms; frailty = 3–5 symptoms) (Fried et al., 2001). One study found that after 2 years of follow-up, being frail at baseline was associated with a decline in their global cognitive score (β = - 1.48; 95% CI: −2.37, −0.59), compared to robust individuals (Chen et al., 2018). Additionally, these authors found that pre-frail individuals at baseline experienced a decline in their global cognitive score (β = - 0.24; 95% CI: −0.65, 0.18), compared to robust individuals; although, this was not statistically significant. We additional found that males experience worse cognitive scores compared to females for each additional frailty symptoms reported. The findings that males experience greater cognitive decline has been consistently reported previously (Bloomberg et al., 2021; McCarrey et al., 2016). The similarity of previous findings with the current study results suggests that increasing frailty symptoms lead to worsening cognitive scores, with males experience worse cognitive outcomes. One potential reason why each additional frailty symptom had a steeper cognitive decline for males compared to females is due to frail males having worse health outcomes overall (Kane & Howlett, 2021). Gordon and Hubbard (2020) proposed that females are more likely to be physiological resiliency than males; (Gordon & Hubbard, 2020) suggesting that even though females have higher rates of frailty, they may have more resilience to adverse effects of frailty.

The current study extends previous results on the association between frailty and cognition by elucidating one key mechanism explaining this association, increasing depressive symptoms and how it differs by sex. These findings are among the first to examine the longitudinal relationship of frailty and depression with cognitive performance in relation to frailty symptoms and to compare the results among males and females separately. Among those with an additional frailty symptom, we found that for each additional depressive symptom, there was a 0.04 decrease in cognitive scores. These findings are consistent with a previous meta-analysis that found depressive symptoms are associated poorer performance in several cognitive domains (Lee et al., 2012). However, when stratified by sex, we found that only females had a significant association with each additional depressive symptom which was associated with a decrease in cognitive scores. A recently published article found that depression among females, but not males, was associated with worsening cognitive score which supports these findings (Anstey et al., 2021). Overall, these results suggest that one mechanism that connected the underlying etiology between frailty and cognition may be due to psychological factors such as depression among females but not males. Further, we found that depressive symptoms partially mediated the relationship between increased frailty symptoms and cognitive function, with depressive symptoms accounting for 11.7% of the cognitive decline associated with an increase in frailty symptoms among females only. Robertson and colleagues hypothesized but did not test that depression may mediate the relationship between frailty and cognitive decline, which our findings confirm (Robertson et al., 2013).

Our findings help to expand the understanding of a potential mechanism for the associations between frailty symptoms, depressive symptoms, and cognition in older adults among females but perhaps not for males. One explanation could be that frailty decreases social activities and increases loneliness (Jarach et al., 2021), which then increases depression (Donovan et al., 2017). Then, depression could lead to increased vascular problems and poorer cognition (Donovan et al., 2017; Wu et al., 2021). The findings that frailty phenotype symptoms lead to depressive symptoms and worse cognitive scores help support this idea. Although we found that depressive symptoms may explain a proportion of the association between frailty symptoms and cognitive scores, most of the association with cognitive scores (88.3%) was due to direct effects, attributable to either frailty symptoms or other mediating pathways not considered in the present analysis. One alternative pathway could be through inflammation, which is related to the accumulations of chronic conditions and an age-related decline of the immune system (Perry & Teeling, 2013). Increased inflammation is associated with frailty and pre-frailty (Velissaris et al., 2017), depression (Miller & Raison, 2016), and cognitive decline (Sartori et al., 2012). Increased inflammation can also be acquired through an individual’s diet; dietary inflammation is also linked to frailty and pre-frailty (M.C. M. C. Lohman et al., 2019; Resciniti et al., 2019), depression (Wang et al., 2018), and cognitive decline (Frith et al., 2018), Further, the aging process increases an inflammatory marker, interleukin-6 and C-reactive protein, which helps to decreases lean body mass and muscle, leading to frailty (Hubbard et al., 2009), In addition, peripheral pro-inflammatory cytokines increase levels of central pro-inflammatory cytokines that generate neurotoxicity, leading to depression and cognitive impairment (Miller & Raison, 2016; Sartori et al., 2012). Additionally, previous studies have found that females tend to report more depressive symptoms compared to males (Shi et al., 2021), which may influence our findings that there was a significant effect of frailty to cognitive decline through depressive symptoms for females and not males. Further research is needed to understand additional mechanisms that contribute to the association between frailty and cognition and to confirm the sex differences between males and females.

Our study has some strengths worth noting. First, our study included a large sample of community-dwelling US adults, which followed individuals over time. This allowed us to establish temporality with frailty status, depression, and cognitive decline. Additionally, we were able to use advanced mediation methods. This study also had limitations worth noting. The dependence on observational data imposes limits on causal inference. However, the longitudinal design of the HRS data allows for the exposure to precede both the mediator and outcome. Second, we used a continuous variable of frailty symptoms and not the cutoff of frail versus robust, limiting the interpretation to increasing frailty symptoms instead of comparing frail to robust individuals, a commonly used classification schema in studies of frailty. Lastly, we used a modified version of the frailty phenotype model due to the overlap with the question asked on the CES-D with frailty measures (exhaustion). However, previous researchers have used and validated this categorization of frailty symptoms within the HRS (M. Lohman et al., 2014).

In conclusion, our study directly tested the mediatory role of depressive symptoms on the association of frailty symptoms on cognitive function. Our findings establish depressive symptoms as a potential mechanism underlying cognitive decline in presenting with frailty symptoms for females but not for males. Furthermore, we found that males had worse cognitive outcomes compared to females. These results highlight the importance of identifying frail and pre-frail individuals to monitor individuals at greater risk of cognitive decline. Females experiencing both frail and depressive symptoms should be under specific reflection, as depression by link frailty and cognitive decline for this population. Among those experiencing frailty symptoms, healthcare providers should test and monitor for depression, cognitive decline, and other related health-related indicators, such as physical and social activity.