Depression and Anxiety in Older Adults with Dementia During the COVID-19 Pandemic

2.1Design and setting

We used a retrospective longitudinal study design. Westat collected data from the National Health and Trends Study (NHATS), led by the Johns Hopkins University Bloomberg School of Public Health and the University of Michigan’s Institute for Social Research. Moreover, the National Institute on Aging provides support for the NHATS. The Johns Hopkins Medicine Institutional Review Board approved the study.

2.2Participants

We drew participants from the NHATS—an ongoing nationally representative population-based study of Medicare beneficiaries aged≥65 years in the United States [15].

NHATS enrolled 8,245 participants at baseline in 2011 and followed up with participants or proxy respondents using annual in-person surveys. In 2015, a new sample was introduced to restore the sample to the original size by age and race groups. Details of the design and protocol of the NHATS are available elsewhere [16].

We included older adults who participated in the 2018 and 2019 rounds of the NHATS surveys and excluded those identified as decedents in the 2018 or 2019 rounds of the NHATS surveys. Therefore, our sample included 4,548 participants who attended two or more rounds of the NHATS surveys between 2018 and 2021. We combined data from two consecutive rounds of surveys (2018 and 2019, 2019 and 2020, and 2020 and 2021) to establish panel data in which each case had (a) baseline dementia status, demographics, functional impairment, and mental health outcomes, and (b) 1-year follow-up activity participation and mental health outcomes.

2.3Measurements

We identified dementia status in the NHATS survey based on the validated algorithm for surveys [17]: reports of physician diagnosis, proxy responses to the AD8 dementia screening interview [18], and cognitive testing. NHATS participants are classified into three groups based on the algorithm: probable dementia, possible dementia, and no dementia. A report by either the participant or a proxy respondent that a doctor told the sample person that they had dementia or Alzheimer’s disease was used to classify persons as probable dementia. Proxy respondents not reporting a diagnosis who gave answers to the AD8 that met the criteria for likely dementia (a score of 2 or higher) also were classified as probable dementia. Score cutpoints applied to cognitive tests were used for all others with cognitive test information. In this study, having dementia was defined as having probable dementia.

We used the year of assessment at follow-up (2019, 2020, or 2021) as a period indicator in the analysis. Additionally, we did not use a dichotomous classification (before and during the pandemic) because the exact date on which the NHATS interviews were conducted was unavailable from 2020 onwards. Moreover, we defined 1-year change in mental health as that before (2018–2019), during (2019–2020), and after the onset of the COVID-19 pandemic (2020–2021).

Mental health outcomes included depressive symptoms and anxiety. We assessed depressive symptoms using the Patient Health Questionnaire-2 [19] with scores ranging from 0 –6; high scores indicated more severe depressive symptoms than low scores. Scores≥3 indicated the presence of depressive symptoms [20]. Furthermore, we assessed anxiety using the Generalized Anxiety Disorder-2 scale [21] with scores ranging from 0–6; high scores indicated high anxiety levels, and scores≥3 defined the presence of anxiety [22].

Moreover, we measured activity participation for two physical and five social activities in the past month [15]. We coded all activity items using dichotomous variables (1 = yes; 0 = no). Additionally, we calculated sum scores for the physical and social activity items, with high scores indicating participation in more activities within the respective domain.

We defined covariates based on the literature [6, 8, 23] and included participants’ age, sex, race or ethnicity, education, living alone, functional impairment, proxy response, and place of residence. Functional impairment comprised the activity of daily living (ADL) and instrumental activity of daily living (IADL) impairments. We assessed ADL impairment based on difficulty or help required for bathing, eating, dressing, toileting, getting around inside the home, or leaving home. Moreover, we assessed IADL impairment based on difficulty or help required in cooking, managing finances, managing medications, shopping, and doing laundry [24]. We categorized ADL and IADL impairments as “no,” “moderate,” or “severe” based on the reported impairments (0, 1–2, and≥3, respectively). Lastly, we assessed living alone, functional impairment, proxy response, and place of residence in each NHATS round and treated them as time-varying variables in the analyses.

2.4Statistical analysis

We calculated the proportion and 95% confidence intervals (CIs) of mental health outcomes across the baseline dementia status. Moreover, we calculated the mean and 95% CIs of activity participation based on the baseline dementia status.

Furthermore, we performed multivariable binomial logistic regression analysis to test for differences in depression and anxiety across dementia statuses at baseline and year of follow-up assessment. The model used each outcome measure at follow-up as the dependent variable and dementia status at baseline as the independent variable. Additionally, the model included baseline demographics and functional impairment as independent variables. Next, we added activity participation at the follow-up to the model, as an independent variable. In these analyses, each case had a time variable (follow-up in 2019, 2020, or 2021) and variables at baseline and follow-up. These models accounted for the clustering of outcome measures among older adults. Since the 1-year mental health changes were explicit between 2019 and 2020, we examined the interaction between dementia status and year of follow-up using dummy variables as independent variables: (a) dementia×2020 and (b) dementia×2021, in addition to the main effects of dementia and year of follow-up.

The regression analyses used full information maximum likelihood to handle missing data [25]. Since our concerns were controlling for within-cluster correlation rather than the clustering level, we used a sandwich estimator instead of modeling random effects [26]. Lastly, we performed all analyses using Mplus for Windows, version 8.8 (Muthén & Muthén, Los Angeles, California, USA). p < 0.05 indicated statistical significance.

4.1Main findings of this study

Older adults with dementia had an increased risk of depressive symptoms and anxiety before and during the COVID-19 pandemic. The overall level of depressive symptoms increased between 2019 and 2020 because of the onset of the pandemic and lifestyle changes caused by several restrictions. However, the association between an increased risk of depressive symptoms and period was insignificant after adjusting for activity participation. Poor activity participation was associated with a high risk of depressive symptoms and anxiety. Furthermore, the association between dementia and worse mental health outcomes remained significant even after controlling for activity participation. However, there was no interaction effect between dementia and year of follow-up for depressive symptoms and anxiety.

Worse mental health outcomes among people with dementia were consistent with suggestions from the qualitative data collected during the COVID-19 pandemic [10–12]. Our findings further present the quantitative evidence, hence providing a significant contribution to the existing literature. In particular, our observation was based on data from a large population-based cohort, with most participants living in private homes. Thus, the results illuminate the fact that widespread mental health challenges exist in the community-dwelling population. Even before the pandemic, depressive symptoms and anxiety were common (a prevalence of 11% and 9% in 2018, respectively) among older adults with dementia [27, 28]. Depressive symptoms are associated with excess mortality in older adults [29], especially in combination with dementia [30]. Therefore, the dementia-specific increased risk of depressive symptoms and anxiety during the onset of the COVID-19 pandemic could have further increased mortality in this population. We partially attribute this elevated risk to the reduced level of activity participation during this period; however, the dementia-specific risk remained even after adjusting for activity participation. One possible explanation is that older adults with dementia may experience chronic psychological distress because of their independence to care in daily life stimulated by rapid lifestyle changes and reduced social contact in 2020. People with dementia may be more vulnerable to reduced contact than those without dementia since loneliness and social isolation are linked to an increased risk of dementia [31]. Although several states in the United States have eased restrictions (for example, relaxing physical distancing requirements and reopening businesses) since 2021 [32], reduced policy support for dementia-friendly initiatives may have caused limited funding and resources needed to sustain the activity participation of persons with dementia [33]. Nevertheless, dementia care and support should address emotional and social needs using accessible means under long-term restrictions. For example, person-centered technology, such as service delivery using videoconferencing platforms, would provide feelings of connectedness and improve the resilience and well-being of people with dementia [34].

4.2Strengths and limitations

The strength of this study lies in the use of nationwide representative cohort data from the United States. The data enabled us to test the longitudinal association between baseline dementia status and 1-year mental health changes. Furthermore, the NHATS provided data from multiple periods, including before and after the onset of the COVID-19 pandemic. The results highlight that mental health deterioration was more prominent during the pandemic onset than in other 1-year periods. However, this study has some limitations. First, since the response date was unavailable in the NHATS data, we could not distinguish between participants before and after the onset of the pandemic in 2020. Additionally, the level of social and physical activities in early 2020 may have been higher than that in April 2020 and afterward, when the government imposed stay-at-home quarantine and other restrictions. Moreover, Medicare claim data was not consistent with the data from NHATS 2018 survey and afterward. Therefore, we could not confirm the diagnosis or healthcare use for dementia or depression in our sample. Lastly, data regarding the prescription of antidepressants, which is common among patients with dementia, was unavailable [35].

4.3Conclusions

Despite these limitations, this is the first study to indicate that having dementia and a reduced level of activity participation independently affected the risk of depressive symptoms and anxiety during the COVID-19 pandemic. These results highlight the emotional and social needs of people with dementia during public health crises. Therefore, dementia care and support should be developed to enable people with dementia to engage in meaningful activities and experience feelings of connectedness.