Comparison of Cognitive and Physical Decline as Predictors of Depression Among Older Adults

Abstract

Activities of daily living (ADL) limitations and cognitive impairment have been identified as key risk factors for depression among older adults. However, little has been done to compare the strength of these relationships. The current study describes the prevalence and compares the independent and joint associations of ADL and cognitive limitations with depression among older adults in the US. Analyses are based on a sample of 30,923 observations on 13,545 unique respondents from three waves (2012, 2014, and 2016) of the Health and Retirement Study. Linear and logistic multivariate regression models with random and individual fixed effects were estimated. Findings indicate that both cognitive and ADL limitations are associated with depression; however, across all models, ADL limitations have a much stronger association. Further, in our most rigorous models, having both limitations is not significantly different from having just ADL, and not cognitive, limitations.

Keywords

activities of daily living cognitive function depression

What this paper adds

Explores the independent and joint associations of ADL and cognitive limitations with depression among a nationally representative sample of older adults in the US.

Finds that both ADL and cognitive limitations are associated with depression, but ADL limitations are a stronger predictor of depression once the comorbidity of these limitations is taken into account.

Finds that cognitive limitations alone are not significantly associated with depression and that having both cognitive and ADL limitations does not increase depression risk over having just ADL limitations.

Applications of study findings

The study findings can inform complex diagnosis and treatment of depression among older adults with multi-morbidity.

The results highlight the need for advocacy surrounding policies and interventions to support older adults with ADL limitations.

The results can inform future research on the mechanisms that underlay the association between ADL limitations and depression, as well as cognitive limitations and depression.

Introduction

Major depressive disorder is a substantial public health concern for older adults (Blazer, 2003; Fiske et al., 2009). National data suggest the prevalence of depression in the past year among adults 65 and older in the United States is between 5 and 10%; however, this is likely an underestimate due to under-reporting of symptoms (Cheruvu & Chiyaka, 2019). The potential debilitating effects of depression among older adults include increased risk of mortality, self-neglect, suicide, and social dysfunction (Blazer, 2003). Depressive symptoms involve feelings of sadness, lack of interest in activities, decreased appetite and motivation, and sleep disruption, all of which may be related to physical and cognitive decline (Lin & Wu, 2011).

Physical decline in older adults, most commonly indicated by limitations in activities of daily living (ADL), is measured as difficulties with eating, bathing, dressing, using the toilet, and transferring (Katz et al., 1963). The CDC estimates approximately 14% of adults 65 and older in the United States report ADL limitations (CDC, 2016). Instrumental activities of daily living (IADLs), more complex activities such as managing finances and food preparation, are also used as a measure of physical decline; however, ADLs capture older adults’ functional status at the most fundamental level and are considered key indicators of quality of life (Edemekong et al., 2020). A large body of work across diverse populations, countries, and datasets has documented a strong association between physical decline, particularly ADL limitations, and depression among older adults (Bowen & Ruch, 2015; He et al., 2019). Physical limitations can prevent individuals from exercising, socializing with friends and family, and participating in daily activities and routines—all of which can lead to depression (Choi & Kim, 2007). There is also the possibility of a reciprocal relationship, in which individuals with depression have less motivation and interest in taking proactive steps to improve health (e.g., better diet and increased exercise)—which can lead to physical impairment (Bowen & Ruch, 2015). One study explicitly examined these reciprocal relationships and found a stronger effect from physical impairment to depression than in the reverse direction (Bowen & Ruch, 2015).

Cognitive decline presents another significant health problem for older adults. The estimated prevalence of cognitive impairment without dementia (CIND) among adults aged 70 and older in the US is 22% (Gure et al., 2013). In addition, approximately 8% of Americans aged 70 and older were living with dementia in 2019 (Freedman & Kasper, 2021). CIND and dementia can be assessed with objective assessment tools (e.g., neurocognitive assessment batteries) and CIND can also be assessed with subjective assessment tools (e.g., self-reported memory complaints). Studies have documented an association between both objective (Oi, 2017; Xu et al., 2019) and subjective (Mogle et al., 2020) cognitive impairment and depression among older adults. The association between cognitive impairment and depression may also be reciprocal, though this may be a bigger concern when subjective measures of cognitive impairment are used (Patnode et al., 2020).

Physical and cognitive limitations often co-occur among older adults (Di Carlo et al., 2000; Gure et al., 2013; Thomas, 2001; Xiang & An, 2015), though there are a lack of studies using longitudinal methods and population-based data to explore the comorbidity of physical and cognitive limitations among older adults in the US. Even fewer studies explore the associations of physical and cognitive limitations with depression in this population. For example, Di Carlo et al. (2000) and Thomas (2001) used cross-sectional methods to describe the prevalence of physical limitations and cognitive impairment among older adults, drawing on the Italian Longitudinal Study on Aging (Di Carlo et al., 2000) and the Canadian Study of Health and Aging (Thomas, 2001). In addition, one study using data from the Aging, Demographics, and Memory Study (ADAMS) (a supplement to the HRS) found that older adults with CIND and dementia had more ADL limitations and higher rates of depression compared to those with normal cognition (Gure et al., 2013). However, this study was cross-sectional and used logistic regression to examine the association between cognition (e.g., normal, CIND, and dementia) and having one or more ADL limitations (controlling for depression and other sociodemographic characteristics). Finally, another study using the HRS found that having cognitive impairment and depression was more strongly associated with ADL limitations compared to cognitive impairment or depression alone (Xiang & An, 2015). This study differs from the current study in that the authors used functional limitations (e.g., ADL and IADL) as their outcome variable and drew on earlier HRS survey data, from 1998 to 2010.

Therefore, while prior research demonstrates that cognitive and physical limitations are comorbid conditions that are both strongly related to depression among older adults, no study to our knowledge has compared the independent and joint associations of ADL and cognitive limitations with depression, nor accounted for the comorbidity of these two types of limitations. Another gap is that there are likely unobserved differences between those with physical and cognitive limitations and that these differences may also be associated with depression. For example, older adults with physical or cognitive limitations may have had early exposures to adverse events or trauma, which are generally unmeasured in most studies, and these may also be associated with depression. Alternatively, those with depression may report worse cognitive well-being, which may be problematic if cognitive impairment is assessed through subjective measures (Patnode et al., 2020).

The Current Study

Our study focuses on the following research question: what are the joint and independent associations of cognitive and ADL limitations with depressive symptoms among older adults in the US? The current study aims to address several gaps in prior research. First, we take a unique approach by considering the relative strengths, as well as the comorbidity, of physical and cognitive decline as predictors of depressive symptoms among older adults. Second, we utilize three recent waves of a large, longitudinal, nationally representative dataset that comprehensively examines the well-being of older adults in the US. Finally, we take several steps to address potential selection bias and reverse causality. We use objective measures of cognitive limitations which reduce the possibility of depressive symptoms and other unmeasured characteristics affecting reports of cognitive decline. In addition, we estimate models including a lagged measure of depressive symptoms from the prior wave and individual fixed effects to explicitly address unobserved differences between those with and without limitations that could also be associated with depressive symptoms.

We contribute to the literature in several ways. First, understanding the joint associations of ADL limitations and cognitive impairment with depressive symptoms is important to inform complex diagnosis and treatment among older adults with multi-morbidity given that “the cumulative effect of multiple conditions is likely to be more substantial than the effect of only one condition” (Xiang & An, 2015). Further, determining the comparative strength of the independent associations may inform future intervention research and clinical work that targets depressive symptoms among older adults with ADL limitations, cognitive impairment, or both. It may also help providers identify older adults at high risk for developing depressive symptoms.

Methods

Data

The sample is drawn from the HRS, a nationally representative panel study of community-dwelling older adults aged 51 and older in the US (Health and Retirement Study, 2020). Since 1992, the HRS has conducted surveys biennially with a national area probability sample of households containing oversamples of Black and Hispanic individuals as well as residents of Florida (Heeringa & Connor, 1995). Response rates varied from the first baseline interviews conducted in 1992 to the interviews completed in 2016 (Fisher & Ryan, 2018).

Study Population

Analyses are based on pooled cross-sections of primary respondents who participated in at least one wave of the 2012, 2014, or 2016 waves of the HRS, were 51 or older, and were not living in a nursing home at any wave. These sample exclusions produced 32,845 observations on 14,334 unique respondents. We further limited the sample to observations not missing on depression (96%), the dependent variable, and any other analysis variable. Our final analysis sample consists of 30,923 repeated observations on 13,545 unique respondents. We base our estimates on complete case analysis because including cases with imputation on the dependent variable is not recommended (von Hippel, 2007) and only 2% of observations are missing information on other variables.

Primary Measures

Our key variables of interest were entered as time-varying variables, measured at each wave that the respondent appeared in the data (2012, 2014, and 2016). Depressive symptoms (the outcome) were measured using the 8-item Centers for Epidemiological Studies Depression Scale (CES-D), a commonly used and well-validated instrument measuring depressive symptoms experienced in the previous week (Radloff, 1977). The eight items in the instrument were summed to create a continuous total depressive symptoms composite score (0–8), and a binary measure representing clinical depression using a cut-off score of greater than or equal to 3 depressive symptoms (Turvey et al., 1999).

ADL limitations were based on participants’ reports of having difficulty eating, bathing, dressing, getting in and out of bed, or using the toilet due to a health or memory problem (Katz et al., 1963). Affirmative responses to these items were summed to create a total continuous ADL limitations composite score (0–5), and a binary measure was constructed for experience of any ADL limitation. To measure cognitive functioning, we utilized the TICS-27 scale which includes an immediate and delayed 10-word recall test, a series 7 subtraction test, and a backward count from 20 test (Crimmins et al., 2011). The continuous score, ranging from 0 to 27, with higher scores indicating higher cognitive functioning, was used to construct a binary indicator for any cognitive impairment, based on a previously validated cut-off score of <12 (Crimmins et al., 2011). The primary independent variable in our main analyses combines these binary ADL and cognitive limitations measures into a 4-level categorical variable: (1) no limitations; (2) cognitive only; (3) ADL only; and (4) both types.

In supplementary analyses we expand our binary measure of cognitive impairment to a 3-level categorical variable indicating: no impairment (>=12); cognitive impairment—no dementia (CIND) (7–11); and dementia (<7). We then combine this variable with the binary ADL limitation variable to create a more nuanced measure of limitations: (1) no limitations; (2) CIND only; (3) dementia only; (4) ADL only; (5) ADL and CIND; and (6) ADL and dementia.

Covariates

The covariates in our study were selected based on prior literature identifying factors that may confound the associations between limitations and depression (Bowen & Ruch, 2015; Mogle et al., 2020) and were measured as non-time varying variables taken from the first wave (2012, 2014, or 2016) that the respondent appeared in the data. Sociodemographic variables included respondents’: sex (male or female); race/ethnicity (non-Hispanic white, non-Hispanic Black, Hispanic, other); age (51–64, 65–74, 75–84, 85–107); marital status (married/partnered, divorced/separated, widowed, never married); living arrangement (alone or with others); highest level of education (less than high school, high school graduate/GED, some college, college and above); household income (including: salary, social security, disability insurance, retirement pensions, and alimony) collapsed into quartiles; whether working for pay; whether receiving Medicaid (as an additional marker of socioeconomic status); and area of residence (urban, suburban, and ex-urban). Respondents’ health and health behaviors included: self-rated health (poor, fair, good, very good, and excellent); number (0–8) of diagnosed chronic diseases (high blood pressure, diabetes, cancer, lung disease, heart disease, stroke, psychiatric problems, and arthritis); whether respondent smokes; and whether respondent reports drinking alcohol three or more days per week.

Analytic Strategy

We first present descriptive statistics of all the variables in the analysis, for the full pooled sample and stratified by the four categories of the type of limitation: neither ADL nor cognitive limitation; cognitive, but not ADL, limitation; ADL, but not cognitive, limitation; and both types of limitations. Descriptive analyses are weighted to reflect the complex survey design of the HRS.

Next, we estimate multivariate regression models to examine the associations of type of limitation with the continuous measure of the number of depressive symptoms, using linear regression, and the binary measure for clinical depression (>=3 symptoms), using logistic regression. For each outcome, we estimate three models. Because our analyses are based on pooled cross-sections of the HRS data, with up to three waves of observations nested within respondents, our first two models (controlling for all previously described covariates) incorporate random effects (akin to mixed effects models) to take explicit account of this type of data structure (Laird & Ware, 1982). In the second model, we add a control for a lagged measure of depressive symptoms, which takes account of the unobserved factors that are associated with prior depressive symptoms and either ADL or cognitive limitations to address potential selection bias and reverse causality. In the third model, we include individual fixed effects, which estimate effects of changes in limitations on changes in depressive symptoms over time only within individuals, taking account of all unobserved static differences between individuals.

We do not apply HRS survey weights in regression models for several reasons. First, random and fixed effects models do not easily accommodate survey weights. Second, applying weights to complex models may produce biased estimates as the effects of a small number of observations in some cells can be overestimated (Gelman, 2007). Finally, we estimated basic models with and without weights and the general pattern of results was very similar (available upon request).

Results

Table 1 presents weighted descriptive statistics of the variables in our analyses for the full pooled sample of 30,923 repeated observations on 13,545 unique respondents, and separately by the types of limitations. The overwhelming majority (67%) of observations had neither cognitive nor ADL limitations; 16% had cognitive, but not ADL, limitations; 11% had ADL, but not cognitive, limitations; and 7% had both types of limitations. In the full sample, the average number of ADL limitations was .29 of a possible 5, with 15% of observations having at least one ADL limitation. Among, those who had an ADL, but not a cognitive, limitation, the mean was 1.84 ADL limitations; and 2.24 for those with both types. The average cognitive score was 15.67 of a possible 27, with 17% of the sample classified as having a cognitive limitation, based on the cut-off score of less than 12 (Crimmins et al., 2011). Average cognitive scores were 17.25 for those with neither limitation; 8.77 for those with no cognitive, but ADL limitation; 16.01 for those with ADL, but not cognitive limitation, and 8.13 for those with both types of limitations. These figures suggest that those with both types of limitations have more ADL limitations than those with just ADL limitations, but they do not have worse cognitive scores than those with just cognitive limitations.

Table 1.

Descriptive Statistics of All Analysis Variables.

	Full Sample	Type of limitation
	Full Sample	No limitations	Cognitive only	ADL only	Both limitations
N	30,923	20,653	4837	3371	2062
% of sample	100%	67%	16%	11%	7%
Limitations
Total ADL Limitations (0–5)**	.29 (.01)	0	0	1.84 (.38)	2.24 (.04)
One or more ADL Limitations**	.15	0	0	1.00	1.00
Total Cognitive Score (0–27)**	15.67 (.08)	17.25 (.04)	8.77 (.05)	16.01 (.09)	8.13 (.08)
Cognitively impaired (<12)**	.17	0	1.00	0	1.00
Supplementary limitations measure
No limitations	.67
CIND only	.13
Dementia only	.03
ADL Limitations only	.11
ADL + CIND	.05
ADL + dementia	.02
Depressive symptoms
Total depressive symptoms (0–8)**	1.46 (.03)	1.07 (.23)	1.64 (.05)	3.06 (.08)	3.47 (.09)
Depression (>=3 symptoms)**	.15	.09	.16	.39	.46
Covariates
Female**	.49	.48	.48	.54	.63
Race/Ethnicity**
Non-Hispanic White	.74	.79	.56	.69	.47
Non-Hispanic Black	.12	.09	.24	.13	.27
Hispanic	.10	.08	.16	.12	.21
Other, Non-Hispanic	.04	.04	.05	.05	.05
Age at baseline**
51–64°years	.62	.66	.46	.59	.43
65–74°years	.24	.24	.25	.22	.21
75–84°years	.11	.09	.20	.13	.20
95–103°years	.03	.01	.08	.05	.03
Marital status**
Married or partnered	.56	.61	.46	.46	.32
Divorced or separated	.21	.20	.22	.28	.27
Widowed	.11	.08	.18	.13	.26
Never married	.12	.11	.14	.13	.15
Lives alone**	.29	.26	.34	.35	.38
Education**
Less than high school	.12	.07	.32	.14	.46
GED or high school graduate	.31	.29	.37	.35	.29
Some college	.27	.28	.21	.32	.17
College and above	.30	.37	.10	.19	.09
Income**
First quartile	.20	.12	.39	.34	.56
Second quartile	.22	.19	.28	.27	.27
Third quartile	.26	.28	.20	.23	.11
Fourth quartile	.33	.40	.13	.16	.05
Employed**	.52	.61	.34	.25	.09
Medicaid**	.09	.04	.20	.19	.34
Urban/rural*
Urban	.52	.54	.47	.47	.48
Suburban	.23	.22	.26	.24	.27
Ex-Urban	.25	.24	.27	.28	.25
Self-rated health**
Poor	.06	.02	.07	.21	.28
Fair	.19	.13	.26	.39	.39
Good	.31	.32	.34	.27	.21
Very good or excellent	.44	.52	.33	.13	.11
Number of chronic diseases**	1.88 (.02)	1.62 (.02)	2.11 (.03)	2.96 (.05)	3.17 (.06)
Currently smokes**	.16	.14	.20	.24	.16
Three or more drinks per week**	.20	.23	.13	.13	.10
Wave
2012	.32	.32	.33	.31	.33
2014	.30	.30	.30	.30	.31
2016	.38	.39	.37	.39	.36

Significance tests between all analysis variables by categories of the limitation variable were estimated using chi-square and bivariate regression.

**p < .001, *p < .05.

Figures in table are means and (SD) or proportions.

Descriptive analyses are weighted to reflect the complex survey design of the HRS.

The average number of depressive symptoms was 1.46 of a possible 8 symptoms with 15% of observations having depression (>=3 symptoms). Depressive symptoms and rates of depression were much higher among those with both types of limitations (3.47 and 46%, respectively) and those with ADL, but not cognitive, limitations (3.06 and 39%), compared to those with neither limitation (1.07 and 9%). Depressive symptoms were also higher for those with cognitive, but not ADL, limitations (1.64 and 16%), but the difference was much smaller.

All covariates were measured at the first time that respondents appeared in the sample, in either the 2012, 2014, or 2016 wave of the HRS, which we refer to as baseline. The sample was approximately half female, 74% non-Hispanic white, 12% non-Hispanic Black, 10% Hispanic, and 4% other race/ethnicity. Most of the sample (62%) was in the youngest group (51–64), 24% was 65–74, 11% was 75–84, and 3% was in the oldest group (85+). Over half of the sample (56%) was currently married or partnered, 21% was divorced or separated, 11% was widowed, 12% was never-married, and 29% was living alone. Only 12% of the sample had not completed high school, with 31%, 27%, and 30%, having only a high school diploma, some college, and a college degree or more, respectively. Approximately half (52%) of the sample was working for pay and 9% reported Medicaid receipt. Approximately half of the sample lived in an urban area with one-quarter each in suburban or ex-urban areas. Almost half of the sample (44%) rated their health as very good or excellent with only 6% reporting poor health, with an average of 1.88 chronic diseases (of 8). Overall, 16% of the sample reported currently smoking, and 20% reported having an alcoholic drink three or more days per week.

All of the covariates differed at statistically significant levels by the presence and type of limitation, highlighting the importance of controlling for them in our multivariate models. In general, those with both types of limitations were the most socioeconomically disadvantaged and the least healthy across all indicators, followed by those with ADL, but not cognitive, limitations. One covariate that did not follow this pattern was having alcoholic drinks three or more days per week, with 23% of those with no limitation endorsing this pattern while only 10% of those with both types of limitations did so. This may be due to the possibility that those with health and cognitive limitations may be restricted from drinking due to medical reasons or that drinking may be a marker of greater opportunity for socializing as opposed to a marker of poor health behavior or worse mental health.

Table 2 presents results from three linear multivariate regression models of respondents’ depressive symptoms (0–8) on the type of limitation. Results from Model 1, including random effects and all previously discussed covariates, indicate that compared to having neither type of limitation, having a cognitive, but not ADL, limitation is associated with .15 more depressive symptoms; having an ADL, but not a cognitive, limitation is associated with .85 more depressive symptoms, and having both limitations is associated with 1.07 more depressive symptoms. These differences were all highly statistically significant. Results from Model 2, including a measure of depressive symptoms at the prior wave, were very similar to Model 1. Compared to having neither limitation, having cognitive only, ADL only, or both is associated with .15, .85, and 1.02 more depressive symptoms, respectively.

Table 2.

Associations of ADL and Cognitive Limitations with Depressive Symptoms (Continuous: 0-8).

	Model 1: Base model	Model 2: Add lagged DV	Model 3: Individual fixed effects
Type of limitation (ref: Neither limitation)
Cognitive limitations only	.15*** (5.08)	.15*** (4.02)	.05 (1.33)
ADL limitations only	.85*** (19.44)	.85*** (15.92)	.61*** (13.97)
Both limitations	1.07*** (18.34)	1.02*** (14.26)	.72*** (12.38)
Gender (ref: Male)
Female	.18*** (6.59)	.14*** (5.29)
Race (ref: Non-Hispanic White)
Non-Hispanic Black	−.10** (−2.85)	−.06 (−1.85)
Hispanic	.14** (2.97)	.08 (1.80)
Other, Non-Hispanic	.14* (2.09)	.10 (1.44)
Marital status (ref: Married or partnered)
Divorced or separated	.26*** (5.57)	.09* (2.08)
Widowed	.17** (3.15)	−.03 (−.60)
Never married	.24*** (4.20)	.10 (1.71)
Age (ref: 51–64)
65–74 years	−.55*** (−14.78)	−.23*** (−7.03)
75–84 years	−.59*** (−13.03)	−.23*** (−5.59)
95–103 years	−.85*** (−11.70)	−.39*** (−5.12)
Self-rated health (ref: Poor health)
Fair health	−.79*** (−10.83)	−.38*** (−5.13)
Good health	−1.42*** (−19.64)	−.66*** (−8.74)
Very good/Excellent health	−1.72*** (−23.51)	−.85*** (−11.11)
Education (ref: Less than high school)
GED/HS grad	−.05 (−.99)	−.04 (−.81)
Some college	−.07 (−1.46)	−.07 (−1.48)
College +	−.09 (−1.79)	−.05 (−1.03)
Income: (Ref: 1st income quartile)
2nd Income quartile	−.10* (−2.31)	−.06 (−1.56)
3rd Income quartile	−.25*** (−5.29)	−.12** (−2.79)
4th Income quartile	−.31*** (−5.93)	−.13** (−2.72)
Urban/rural (ref: Urban)
Suburban	.03 (.78)	.04 (1.24)
Ex-Urban	−.07* (−2.22)	.01 (.36)
Medicaid	.31*** (5.78)	.21*** (4.00)
# of chronic diseases	.17*** (15.33)	.11*** (10.47)
Employed	−.13*** (−3.93)	−.04 (−1.34)
Currently smokes	.37*** (8.98)	.24*** (6.15)
Drinks (3+ per week)	.08* (2.37)	.06 (1.88)
Lives alone	.05 (1.16)	.04 (.98)
Lagged depression		.36*** (34.01)
Wave (ref: 2012 wave)
2014 wave	−.02 (−1.27)
2016 wave	−.05** (−2.63)	−.03 (−1.35)
Constant	2.59*** (26.76)	1.26*** (12.92)	1.47*** (116.20)
Total observations	30,923	17,177	30,923
Unique observations	13,545	9,335	13,545

***p < .001, **p < .01, *p < .05.

Figures in table are unstandardized coefficients from random effects (and individual fixed effects – Model 3) linear regression models and robust z-statistics in parentheses.

Sample size is reduced in Model 2 because sample is limited to respondents with observations in at least two waves of data because of the inclusion of depression from the prior wave.

Results in Model 3, incorporating individual fixed effects, indicate that compared to having neither limitation, having an ADL, but not a cognitive, limitation increased depressive symptoms by .61, while having both types increased depressive symptoms by .72. In this model, having a cognitive limitation only, compared to neither type, did not significantly increase depressive symptoms. In post-hoc analyses (not shown), we tested whether having both types of limitations was statistically different from having ADL limitations only, as these associations across all the models were very similar. We found that in Model 3 (including individual fixed effects), these associations were not significantly different from one another, indicating that having both types of limitations did not lead to higher depressive symptoms than having only ADL limitations.

Table 3 presents results of three logistic regression models focusing on the binary outcome measure of clinical depression (>=3 symptoms). Results from Model 1 indicate that compared to having neither limitation, having a cognitive, but not ADL, limitation was associated with 46% higher odds (OR = 1.46) of having depression; having an ADL, but not cognitive, limitation was associated with nearly 4 times the odds (OR = 3.92) of depression; and having both types was associated with 5.60 times greater odds (OR = 5.60) of depression. In Model 2 (including a measure of depression from the previous wave), results showed that compared to having neither limitation, having a cognitive, but not ADL, limitation was associated with 27% higher odds of depression and having ADL only or having both types of limitation was associated with approximately 3 times higher odds of depression (OR = 2.72 and 3.13, respectively).

Table 3.

Associations of ADL and Cognitive Limitations with Depression (Binary: >=3 Symptoms).

	Model 1: Base model	Model 2: Add lagged DV	Model 3: Individual fixed effects
Type of limitation (ref: No limitations)
Cognitive limitations only	1.46*** (4.92)	1.27** (3.25)	1.21 (1.87)
ADL limitations only	3.92*** (17.01)	2.72*** (12.41)	2.56*** (9.02)
Both limitations	5.60*** (16.93)	3.13*** (11.08)	3.11*** (8.42)
Gender (ref: Male)
Female	1.58*** (6.66)	1.28*** (4.50)
Race (ref: Non-Hispanic White)
Non-Hispanic Black	.74*** (−3.56)	.88 (−1.94)
Hispanic	1.37** (3.23)	1.16 (1.88)
Other, Non-Hispanic	1.25 (1.45)	1.19 (1.25)
Marital status (ref: Married or partnered)
Divorced or separated	1.57*** (4.59)	1.10 (1.20)
Widowed	1.41** (2.92)	.86 (−1.64)
Never married	1.71*** (4.35)	1.14 (1.32)
Age (ref: 51–64)
65–74 years	.33*** (−11.58)	.69*** (−4.98)
75–84 years	.36*** (−9.64)	.71*** (−4.15)
95–103 years	.24*** (−8.08)	.73* (−2.23)
Self-rated health (ref: Poor health)
Fair health	.46*** (−7.10)	.86 (−1.71)
Good health	.19*** (−14.36)	.65*** (−4.33)
Very good/Excellent health	.10*** (−18.27)	.42*** (−7.68)
Education (ref: Less than high school)
GED/HS grad	.93 (−.73)	.98 (−.28)
Some college	.92 (−.86)	.93 (−.94)
College +	.91 (−.79)	.99 (−.14)
Income: (Ref: 1st income quartile)
2nd Income quartile	.91 (−1.14)	.94 (−.87)
3rd Income quartile	.64*** (−4.38)	.82* (−2.38)
4th Income quartile	.54*** (−5.01)	.78* (−2.42)
Urban/rural (ref: Urban)
Suburban	1.02 (.20)	1.02 (.26)
Ex-Urban	.78** (−3.09)	.97 (−.48)
Medicaid	1.34** (3.12)	1.20* (2.41)
# of chronic diseases	1.38*** (13.29)	1.18*** (7.87)
Employed	.83* (−2.32)	.95 (−.77)
Currently smokes	1.85*** (7.77)	1.36*** (4.73)
Drinks (3+ per week)	1.28** (2.83)	1.15 (1.87)
Lives alone	1.16 (1.71)	1.12 (1.71)
Lagged depression		6.83*** (28.42)
Wave (ref: 2012 wave)
2014 wave	.96 (−.77)
2016 wave	.94 (−1.16)	.98 (−.28)
Constant	.12*** (−11.81)	.09*** (−16.08)
Total observations	30,923	17,177	5546
Unique observations	13,545	9335	1938

***p < .001, **p < .01, *p < .05.

Figures in tables are odds ratios from random effects (and individual fixed effects – Model 3) logistic regression models and robust z-statistics in parentheses.

Sample size is reduced in Model 2 because sample is limited to respondents with observations in at least two waves of data because of the inclusion of depression from the prior wave.

Sample size is reduced in Model 3 because in logistic regression with individual fixed effects only observations with a change in value on the binary outcome variable are included.

Finally, in model 3, including individual fixed effects and similar to model 3 from Table 2, we find that compared to neither type of limitation, having a cognitive, but not an ADL, limitation was not significantly (p = .062) associated with having depression. Compared to having neither type of limitation, having only an ADL limitation and having both types was associated with approximately 3 times greater odds of having depression (OR = 2.56 and 3.11, respectively). In post-hoc analyses, the associations between having an ADL limitation only and having both types of limitation with depression were not significantly different from each other in both Models 2 and 3 of Table 3, indicating again that having both types of limitations did not increase the risk of depression to a greater degree than having only ADL limitations.

Supplementary analyses (Supplementary Tables 1 and 2 available online) considering a more nuanced measure of cognitive limitations, which separated out CIND from dementia, found very similar results. ADL limitations alone were much stronger predictors of depression than either CIND or dementia alone. In our more conservative individual fixed effects models, the associations of having both types of limitations (CIND and ADL or dementia and ADL) with depression were not statistically different than having just ADL limitations.

Discussion

Depression is a major debilitating condition in later life (Blazer, 2003). Previous studies have shown associations of physical and cognitive limitations with increased likelihood of having depression among older adults (Jiang et al., 2002; Oi, 2017). However, no prior study has compared the relative strengths of the associations between ADL and cognitive limitations with depressive symptoms nor considered the potential comorbidity of these conditions. In this study, we addressed this gap using nationally representative longitudinal data on older adults in the US. We chose to focus on ADL limitations, as opposed to IADL limitations, because ADL limitations are a more robust indicator of serious physical decline in functioning (Mlinac & Feng, 2016). This study provides an important contribution to the literature because determining the comparative effects of these types of limitations may inform future intervention research and clinical work that targets depressive symptoms among older adults living with ADL limitations, cognitive limitations, or both.

Our descriptive analysis reveals a profile of physical and cognitive limitations among people who were 51 years or older in the United States. Consistent with prior work, we find that this population is relatively healthy and independent as an overwhelming majority of the sample (67%) did not have any ADL or cognitive limitations (Jindai, 2016; Luck et al., 2010). Nonetheless, we also find that one-third of this population has at least one type of limitation. Similar to prior work documenting socioeconomic, gender, and racial disparities in ADL and cognitive limitations in later life (Choi et al., 2018; Fernández-Blázquez et al., 2021), we found that compared to those with none or with only one type of limitation, individuals with both types of limitations were more likely to have low levels of income and education and were more likely to be of Black or Hispanic racial/ethnic background. This finding underscores the importance of expanding psychosocial interventions that are accessible to historically marginalized populations with lower socioeconomic status.

We found that both cognitive and ADL limitations are associated with higher depressive symptoms and higher likelihood of clinical depression (>=3 symptoms) based on the CES-D scale (Turvey et al., 1999). This is consistent with previous studies that found a strong effect of ADL limitations and cognitive impairment on depression in later life (Bowen & Ruch, 2015). Building on this prior work by explicitly accounting for the possibility that the comorbidity of these limitations may be driving associations, we find that ADL limitations are independently predictive of depressive symptoms and may have stronger impacts on depression compared to cognitive limitations. Importantly, in more conservative models where we more rigorously account for unobserved differences between those with and without limitations, we find that cognitive limitations alone are not significantly associated with depression, and that having both types (cognitive and ADL) is not more strongly associated with depression than having just ADL limitations. These results indicate that ADL limitations by themselves, regardless of the presence of cognitive limitations, are a strong predictor of depression (approximately 3 times greater odds of having depression) and that having cognitive limitations in addition to ADL limitations does not significantly increase risk. Furthermore, our supplementary analyses confirmed that the strength of the association of having both dementia and ADL limitations with depression was not significantly different from the association of just having ADL limitations alone with depression. Taken together, our findings suggest that cognitive impairment (both CIND and dementia) may lead to depression when ADL limitations are present. In fact, previous studies have suggested that cognitive impairment tends to precede ADL limitations (Dodge et al., 2005; Puente et al., 2014). Therefore, the results of the current study highlight the importance of detecting cognitive impairment to intervene in the early phase of the potential decline in physical limitations, all of which may serve to mitigate depression in later life.

To that point, our results suggest that having cognitive impairment alone may not have as strong of an association with depressive symptoms compared to having ADL limitations alone. This is partially inconsistent with previous studies which showed the strong relationship between cognitive limitations and depressive symptoms (Oi, 2017; Xu et al., 2019). However, in these studies, physical and cognitive limitations were not independently investigated and cognitive limitations were measured as continuous scores on scales of cognitive function. In contrast, we accounted for comorbidity of these limitations and, utilized a validated cut-off point in order to examine the influence of having cognitive impairment, not just relatively lower cognitive functioning. Furthermore, we find a significant association of having dementia, on depression as well as having CIND depression. This finding is important given and results are inconsistent in the literature. For example, Richard et al. (2013) showed that incidence of mild cognitive impairment (MCI) was not associated with depression but incidence of dementia was (Richard et al., 2013). Future studies should investigate how the measurement of cognitive impairment (e.g., objective or subjective measures) contributes to the observed relationship between cognitive impairment and depression among older adults. It is also important to acknowledge that individuals living with cognitive impairment may under-report their depressive symptoms, which may explain some of the weaker associations with depression (Bruce et al., 2008). Most importantly, our results point to the importance of considering the comorbidity of cognitive limitations with ADL and other physical limitations in the onset and severity of depression among older adults.

Our study is not without limitations. First, although we utilized three waves of the HRS and took careful steps to reduce selection bias and reverse causality, there is still a possibility that there are time-varying factors that drive both limitations and depressive symptoms. For example, the observed association could be a result of reverse causality, such that depression could also be causing physical decline and cognitive limitations. Second, although validated (Crimmins et al., 2011), the measurement of depressive symptoms and the cut-off point for clinical depression are far from perfect. Clinical diagnoses are rarely available in population-based survey data that have rich contextual variables like the HRS. Finally, we did not apply HRS survey weights in our regression models which may limit the generalizability of findings. However, basic models estimated with and without weights showed a very similar pattern of results (available upon request).

Despite these limitations, our findings are valuable as this is one of the first attempts to investigate the independent associations and joint effects of physical and cognitive limitations with depression in later life. As depression in later life is a serious health concern, understanding the combination of determinants can have significant implications for public health. This study adds to the evidence that ADL limitations may have stronger associations with depression than cognitive impairment. Much intervention and clinical work has focused on supporting ADLs among older adults as these are key indicators of functional status and quality of life. The results of this study suggest that attention should also be paid to providing mental health-related psychosocial supports for older adults with ADL limitations as these may be key drivers of depression.

Supplemental Material

Supplemental Material - Comparison of Cognitive and Physical Decline as Predictors of Depression Among Older Adults

Supplemetary Material for Comparison of Cognitive and Physical Decline as Predictors of Depression Among Older Adults by Clara Scher, Lenna Nepomnyaschy, and Takashi Amano in Journal of Applied Gerontology

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Clara Scher

Supplemental Material

Supplemental material for this article is available online

References

Blazer

D. G.

(2003). Depression in late life: Review and commentary. The Journals of Gerontology: Series A, 58(3), M249–M265. https://doi.org/10.1093/gerona/58.3.M249

Bowen

M. E.

Ruch

(2015). Depressive symptoms and disability risk among older white and latino adults by nativity status. Journal of Aging and Health, 27(7), 1286–1305. https://doi.org/10.1177/0898264315580121

Bruce

J. M.

Bhalla

Westervelt

H. J.

Davis

Williams

Tremont

(2008). Neuropsychological correlates of self-reported depression and self-reported cognition among patients with mild cognitive impairment. Journal of Geriatric Psychiatry and Neurology, 21(1), 34–40. https://doi.org/10.1177/0891988707311032

CDC . (2016). QuickStats: Percentage of adults with activity limitations, by age group and type of limitation — national health interview survey, † United States, 2014. MMWR. Morbidity and Mortality Weekly Report, 65. https://doi.org/10.15585/mmwr.mm6501a6

Cheruvu

V. K.

Chiyaka

E. T.

(2019). Prevalence of depressive symptoms among older adults who reported medical cost as a barrier to seeking health care: Findings from a nationally representative sample. BMC Geriatrics, 192(1), Article 192, https://doi.org/10.1186/s12877-019-1203-2

Choi

N. G.

Kim

J. S.

(2007). Age group differences in depressive symptoms among older adults with functional impairments. Health & Social Work, 32(3), 177–188. https://doi.org/10.1093/hsw/32.3.177

Choi

Martin

L. G.

Langa

K. M.

(2018). Trends in the prevalence and disparity in cognitive limitations of Americans 55-69 Years old. The Journals of Gerontology. Series B, Psychological Sciences and Social Sciences, 73(suppl_1), S29–S37. https://doi.org/10.1093/geronb/gbx155

Crimmins

E. M.

Kim

J. K.

Langa

K. M.

Weir

D. R.

(2011). Assessment of cognition using surveys and neuropsychological assessment: The health and retirement study and the aging, Demographics, and memory study. The Journals of Gerontology. Series B, Psychological Sciences and Social Sciences, 66(Suppl 1), i162–i171. https://doi.org/10.1093/geronb/gbr048

Di Carlo

Baldereschi

Amaducci

Maggi

Grigoletto

Scarlato

Inzitari

(2000). Cognitive impairment without dementia in older people: Prevalence, vascular risk factors, impact on disability. The Italian Longitudinal Study on Aging. Journal of the American Geriatrics Society, 48(7), 775–782. https://doi.org/10.1111/j.1532-5415.2000.tb04752.x

10.

Dodge

H. H.

Kadowaki

Hayakawa

Yamakawa

Sekikawa

Ueshima

(2005). Cognitive impairment as a strong predictor of incident disability in specific ADL-IADL tasks among community-dwelling elders: The Azuchi Study. The Gerontologist, 45(2), 222–230. https://doi.org/10.1093/geront/45.2.222

11.

Edemekong

P. F.

Bomgaars

D. L.

Sukumaran

Levy

S. B.

(2020). Activities of daily living. In StatPearls. StatPearls Publishing. http://www.ncbi.nlm.nih.gov/books/NBK470404/

12.

Fernández-Blázquez

M. A.

Noriega-Ruiz

Ávila-Villanueva

Valentí-Soler

Frades-Payo

Del Ser

Gómez-Ramírez

(2021). Impact of individual and neighborhood dimensions of socioeconomic status on the prevalence of mild cognitive impairment over seven-year follow-up. Aging & Mental Health, 25(5), 814–823. https://doi.org/10.1080/13607863.2020.1725803

13.

Fisher

G. G.

Ryan

L. H.

(2018). Overview of the health and retirement study and introduction to the special issue. Work, Aging and Retirement, 4(1), 1–9. https://doi.org/10.1093/workar/wax032

14.

Fiske

Wetherell

J. L.

Gatz

(2009). Depression in older adults. Annual Review of Clinical Psychology, 5(1), 363–389. https://doi.org/10.1146/annurev.clinpsy.032408.153621

15.

Freedman

Kasper (2021). National health and aging trends study chart book: Key trends, measures and detailed tables. University of Michigan. https://micda.isr.umich.edu/wp-content/uploads/2022/03/NHATS-Companion-Chartbook-to-Trends-Dashboards-2020.pdf

16.

Gelman

(2007). Struggles with survey Weighting and regression modeling. Statistical Science, 22(2), 153–164. https://doi.org/10.1214/088342306000000691

17.

Gure

T. R.

Langa

K. M.

Fisher

G. G.

Piette

J. D.

Plassman

B. L.

(2013). Functional limitations in older adults who have cognitive impairment without dementia. Journal of Geriatric Psychiatry and Neurology, 26(2), 78–85. https://doi.org/10.1177/0891988713481264

18.

Ren

Zhou

Gong

Liu

Yang

Xiong

Wang

Liu

Zhang

(2019). Association between activities of daily living disability and depression symptoms of middle-aged and older Chinese adults and their spouses: A community based study. Journal of Affective Disorders, 242, 135–142. https://doi.org/10.1016/j.jad.2018.08.060

19.

Health and Retirement Study . (2020). 2016 RAND HRS Fat File public use dataset. Produced and distributed by the University of Michigan with funding from the National Institute on Aging (grant number NIA U01AG009740).

20.

Heeringa

Connor

(1995). Technical description of the health and retirement survey sample design. University of Michigan.

21.

Jiang

Tang

Futatsuka

(2002). The impact of ADL disability on depression symptoms in a community of Beijing elderly, China. Environmental Health and Preventive Medicine, 7(5), 199–204. https://doi.org/10.1007/BF02898005

22.

Jindai

Nielson

C. M.

Vorderstrasse

B. A.

Quinones

A. R.

(2016). Multimorbidity and functional limitations among adults 65 or older, NHANES 2005–2012. Preventing Chronic Disease, 13, Article 160174. https://doi.org/10.5888/pcd13.160174

23.

Katz

Ford

A. B.

Moskowitz

R. W.

Jackson

B. A.

Jaffe

M. W.

(1963). Studies of illness in the aged: The index of ADL: A standardized measure of Biological and psychosocial function. JAMA, 185(12), 914–919. https://doi.org/10.1001/jama.1963.03060120024016

24.

Laird

N. M.

Ware

J. H.

(1982). Random-effects models for longitudinal data. Biometrics, 38(4), 963–974. https://doi.org/10.2307/2529876

25.

Lin

I.-F.

H.-S.

(2011). Does informal care attenuate the cycle of ADL/IADL disability and depressive symptoms in late life? The Journals of Gerontology. Series B, Psychological Sciences and Social Sciences, 66(5), 585–594. https://doi.org/10.1093/geronb/gbr060

26.

Luck

Luppa

Briel

Riedel-Heller

S. G.

(2010). Incidence of mild cognitive impairment: A systematic review. Dementia and Geriatric Cognitive Disorders, 29(2), 164–175. https://doi.org/10.1159/000272424

27.

Mlinac

M. E.

Feng

M. C.

(2016). Assessment of activities of daily living, self-care, and independence. Archives of Clinical Neuropsychology, 31(6), 506–516. https://doi.org/10.1093/arclin/acw049

28.

Mogle

Hill

N. L.

Bhargava

Bell

T. R.

Bhang

(2020). Memory complaints and depressive symptoms over time: A construct-level replication analysis. BMC Geriatrics, 57(1), Article 57. https://doi.org/10.1186/s12877-020-1451-1

29.

(2017). Inter-connected trends in cognitive aging and depression: Evidence from the health and retirement study. Intelligence, 63, 56–65. https://doi.org/10.1016/j.intell.2017.05.004

30.

Patnode

C. D.

Perdue

L. A.

Rossom

R. C.

Rushkin

M. C.

Redmond

Thomas

R. G.

Lin

J. S.

(2020). Screening for cognitive impairment in older adults: Updated evidence report and systematic review for the US preventive services task force. JAMA, 323(8), 764–785. https://doi.org/10.1001/jama.2019.22258

31.

Puente

A. N.

Terry

D. P.

Faraco

C. C.

Brown

C. L.

Miller

L. S

. (2014). Functional impairment in mild cognitive impairment evidenced using performance-based measurement. Journal of Geriatric Psychiatry and Neurology, 27(4), 253–258. https://doi.org/10.1177/0891988714532016

32.

Radloff

L. S.

(1977). The CES-D Scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1(3), 385–401. https://doi.org/10.1177/014662167700100306

33.

Richard

Reitz

Honig

L. S.

Schupf

Tang

M. X.

Manly

J. J.

Mayeux

Devanand

Luchsinger

J. A.

(2013). Late life depression, mild cognitive impairment and dementia. JAMA Neurology, 70(3), 374–382. https://doi.org/10.1001/jamaneurol.2013.603

34.

Thomas

V. S.

(2001). Excess functional disability among demented subjects? Findings from the Canadian study of health and aging. Dementia and Geriatric Cognitive Disorders, 12(3), 206–210. https://doi.org/10.1159/000051259

35.

Turvey

C. L.

Wallace

R. B.

Herzog

(1999). A revised CES-D measure of depressive symptoms and a DSM-based measure of major depressive episodes in the elderly. International Psychogeriatrics, 11(2), 139–148. https://doi.org/10.1017/s1041610299005694

36.

von Hippel

P. T.

(2007). 4. Regression with missing Ys: An improved strategy for analyzing multiply imputed data. Sociological Methodology, 37(1), 83–117. https://doi.org/10.1111/j.1467-9531.2007.00180.x

37.

Xiang

(2015). The impact of cognitive impairment and comorbid depression on disability, health care utilization, and costs. Psychiatric Services, 66(11), 1245–1248. https://doi.org/10.1176/appi.ps.201400511

38.

Jiao

Zhu

Guo

Zhu

(2019). Prevalence and potential associated factors of depression among Chinese older inpatients. The Journal of Nutrition, Health & Aging, 23(10), 997–1003. https://doi.org/10.1007/s12603-019-1270-2

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB