key: cord-0809279-m5trb3mb
authors: Kobayashi, Lindsay C.; O'Shea, Brendan Q.; Joseph, Carly; Finlay, Jessica M.
title: Acute relationships between mental health and cognitive function during the COVID-19 pandemic: Longitudinal evidence from middle-aged and older US adults
date: 2022-04-08
journal: SSM Ment Health
DOI: 10.1016/j.ssmmh.2022.100097
sha: 9694af19bf517be095219d1e13c35f1e12e7981e
doc_id: 809279
cord_uid: m5trb3mb

BACKGROUND: The acute impacts of COVID-19-related mental health concerns on cognitive function among middle-aged and older adults are unknown. We investigated whether between-person (BP) differences and within-person (WP) changes in loneliness, anxiety, and worry about COVID-19 were related to cognitive function and abilities in a longitudinal cohort of middle-aged and older United States (US) adults over a nine-month period during the COVID-19 pandemic. METHODS: Data were from bimonthly questionnaires in the nationwide COVID-19 Coping Study from August/September 2020 through April/May 2021 (N = 2262 adults aged ≥55). Loneliness was assessed with the 3-item UCLA Loneliness Scale, anxiety with the 5-item Beck Anxiety Inventory, and COVID-19 worry on a 5-point Likert-type scale. Cognitive outcomes were assessed with the 6-item Patient Reported Outcomes Measurement Information System (PROMIS®) Cognitive Function and Abilities scales. Marginal structural models incorporating inverse probability of treatment and attrition weights as well as sampling weights estimated the BP and WP relationships between the mental health predictors and PROMIS® cognitive scores over time. RESULTS: In any given month, experiencing a loneliness, anxiety symptom, or COVID-19 worry score higher than the sample mean (BP difference) or higher than one's personal mean across the nine-month period (WP change) was negatively associated with cognitive function and abilities in that month. The observed magnitudes of associations were stronger for BP differences than for WP changes and were the strongest for anxiety symptom scale scores. CONCLUSIONS: Elevated loneliness, anxiety symptoms, and worry about COVID-19, both relative to other adults and to one's usual levels, were acutely associated with worse perceived cognitive function and abilities over a nine-month period during the COVID-19 pandemic in the United States. The long-term impacts of mental health symptoms experienced during the pandemic for population cognitive health should be explored.

The coronavirus disease 2019 (COVID-19) pandemic has had far-reaching impacts on the mental health of populations worldwide (1) (2) (3) (4) . Middle-aged and older adults are typically at higher risk of severe COVID-19 morbidity and mortality, making social distancing, vaccination, and other preventive measures important for this population. Middle-aged and older adults may also experience unique worries and mental health impacts of the pandemic (5) (6) (7) . From March through June 2020, the US National Poll on Healthy Aging identified that 41% of US adults aged 50-80 years felt a lack of companionship, and 56% felt isolated from others (8) . In June 2020, a national survey conducted by the US Centers for Disease Control and Prevention (CDC) identified that 20.3% of adults aged 45-64 and 8.1% of adults aged ≥65 had an anxiety or depressive disorder, and 29.5% of those aged 45-64 and 15.1% of those aged ≥65 had at least one mental or behavioural health symptom (9) . These prevalence estimates are thought to be higher than those observed in the year prior to the pandemic (10, 11) . Although adverse mental health effects among middle-aged and older adults during the pandemic are postulated to influence the future burden of dementia (12, 13) , it remains unclear whether acute changes in middle-aged and older adults' mental health during the pandemic are related to cognitive outcomes.

Among middle-aged and older adults, mental health symptoms are plausibly related to long-term cognitive outcomes, due to their direct effects on neurobiology and indirect effects on behavioural coping mechanisms that are dementia risk factors, such as physical inactivity, alcohol use, and withdrawal from social networks (14) (15) (16) (17) . Pre-pandemic evidence from meta-J o u r n a l P r e -p r o o f analyses of population-based, longitudinal cohort studies consistently indicates that elevated anxiety symptoms are associated with increased risks of subsequent Alzheimer's disease and related dementias (18, 19) . Similarly, loneliness and social isolation have consistently been associated with increased risks of Alzheimer's disease and related dementias in populationbased, longitudinal cohort studies (20) (21) (22) (23) . Depression is also a well-established risk factor for dementia (24) . However, the short-term relationship between depression and cognition is challenging to disentangle, as cognitive impairment can be a symptom of depression and the awareness of cognitive impairment may cause people to experience depressed mood (25) . This pre-pandemic body of evidence is based on studies with long-term follow-up periods, given the long pre-clinical period of dementia. Whether there are short-term, acute associations between mental health symptomatology (except for depression) and cognitive function is less well understood, both before and during the pandemic.

If there are short-term, acute relationships between mental health and cognitive function in the aging population during the COVID-19 pandemic, there may be longer term implications of the pandemic period for the future population burden of Alzheimer's disease and related dementias. We thus aimed to investigate whether between-person differences and withinperson changes in loneliness, anxiety, and worry about COVID-19 were related to self-reported cognitive function and abilities on a month-to-month basis in a longitudinal cohort of middleaged and older United States (US) adults. We hypothesized that experiencing higher than average loneliness, anxiety, and worry about COVID-19, both on between-persons and withinperson levels, would be associated with worse self-reported cognitive function and abilities.

J o u r n a l P r e -p r o o f Data were from the COVID-19 Coping Study, a national longitudinal cohort study of adults aged ≥55 years residing in all 50 US states, the District of Columbia, and Puerto Rico (26) . Details of the study design, recruitment, and methodology are available elsewhere (26) . In brief, participants were recruited through multi-frame non-probability sampling from 2 April through 31

May 2020 and completed monthly online questionnaires through the end of May 2021 (13 months of follow-up). The present analysis used longitudinal data on mental health exposures and cognitive outcomes assessed at the four-month (August/September 2020; analytic baseline), six-month (October/November 2020), eight-month (December 2020/January 2021), Participants were excluded from this analysis if they were missing any baseline covariate data (N=252/4,401, 6%), if they had more than two missing follow-up questionnaires (N=1,861/4,149, 45%), if they had more than two missing observations for any of the mental health exposure variables over the follow-up (N=62, 3% for loneliness; N=9, <1% for anxiety; N=1, <1% for worry about COVID-19), or if they were missing inverse probability of treatment weights (N=22, 1% for loneliness; N=26, 1% for anxiety; N=27, 2% for worry about COVID-19).

The final analytical sample sizes were N=2,204 for the loneliness model, N=2,253 for the anxiety model, and N=2,260 for the worry about COVID-19 model, with a total of 2,262 unique individuals contributing data to at least one model.

Loneliness in the past week was assessed with the 3-item UCLA Loneliness Scale (27) , which included the items: "How often did you feel you lacked companionship?", "How often did you feel left out?", and "How often did you feel isolated from others?", with response options of J o u r n a l P r e -p r o o f "hardly ever" (1 point), "some of the time" (2 points), and "often" (3 points). The scale had a range of 3 to 9, with higher scores indicating higher loneliness.

Anxiety symptoms in the past week were assessed with the 5-item Beck Anxiety Inventory, adapted from the US Health and Retirement Study (28) , which included the items: "I had fear of the worst happening", "I was nervous", "I felt my hands trembling", "I had a fear of dying", and "I felt faint", with response options of "never" (1 point), "hardly ever" (2 points), Zika virus infection (30) , dementia (31) , and general health (30) . This scale was newly developed for the current study, given the novelty of the COVID-19 pandemic in 2020. The scale is intended to capture the construct of worry, which is a functional psychological response to a perceived threat (e.g., COVID-19) that can keep the body in a prolonged state of physiological arousal, especially when reduction or avoidance of the threat is not possible (32) .

The scale was developed with survey methodologists and pilot-tested for comprehension and appropriateness with a group of adults aged ≥55 years who did not become part of the study sample, as described in detail elsewhere (26) . Response options were "not at all worried" (0 points), "slightly worried" (1 point), "somewhat worried" (2 points), "moderately worried" (3 points), and "extremely worried" (4 points), for a range of 0 to 4 with higher scores indicating greater worry (26) .

All three variables were analysed continuously, to capture symptom burden while avoiding loss of information through categorisation. To differentiate between between-person (BP) and within-person (WP) relationships, we derived two variables for each mental health exposure for each person. To create the BP exposure variables, we first calculated the sample J o u r n a l P r e -p r o o f grand means for the loneliness, anxiety, and worry scales across all time points. Then, for each scale, we centred each person's score at each time point to the sample grand mean. To create the WP exposure variables, we first calculated the person-specific grand means for the loneliness, anxiety, and worry scales across all time points. Then, for each scale, we centred each person's score at each time point to their personal grand mean. Both scales give a series of statements about cognitive function or abilities in the past 7 days, with statements such as "My thinking has been slow" and "I have had trouble concentrating" (Cognitive Function scale), and "My mind has been as sharp as usual" and "My memory has been as good as usual" (Cognitive Abilities scale). Response options were "Never", "Rarely (once)", "Sometimes (two or three times)", "Often (about once a day)", "Very often (several times a day)" for the Cognitive Function scale and "Not at all", "A little bit", "Somewhat", "Quite a bit", and "Very much" for the Cognitive Abilities scale. The Cognitive Function scale was reversecoded, so that higher scores indicate better cognitive outcomes.

Scores on the PROMIS scales have been validated against clinical diagnoses, they are considered relevant to assessing daily quality of life as affected by perceived cognitive function and abilities, and they have been recommended for use in COVID-19 research to harmonize measures and data across studies of older adults (33) . The PROMIS Cognitive Function and Abilities raw scores were converted to T-scores using the HealthMeasures Scoring Service, which is the most accurate method for scoring a PROMIS instrument (33) .

The T-score is a standardized score calibrated to the US general population with a mean of 50 J o u r n a l P r e -p r o o f and standard deviation of 10 (33). A higher T-score represents greater cognitive function or abilities. Within-person changes in PROMIS T-scores of 2 to 6 points are thought to represent the minimal degree of change that patients perceive as important or meaningful (35) . More information about the HealthMeasures calibration and interpretation of the T-scores can be found elsewhere (36).

Potential confounders of the relationships between the mental health exposures and cognitive outcomes were: age (continuous, in years), sex (male; female), race/ethnicity (non-Hispanic white; other racial/ethnic group), highest educational attainment (high school or less; some college or two-year associate's degree; four-year college or university degree; postgraduate or professional degree), pre-COVID-19 employment status (employed; unemployed; retired), relationship status (single, never married; single, divorced/separated; single, widowed; married or in a relationship), previous physician diagnosis of hypertension, diabetes, heart disease, asthma, chronic obstructive pulmonary disease, cancer, other physical or mental health condition (all yes; no), use of a mobility aid (yes; no), pre-COVID-19 social isolation (score out of five on the English Longitudinal Study of Ageing social isolation index, with one point assigned for each of: less than monthly contact with each of children, other family, or friends; living alone; and not belonging to any social organizations or clubs) (37) , and number of depressive symptoms according to the 8-item Center for Epidemiologic Studies Depression (CES-D) scale (38) . The covariates were measured at the initial cohort baseline (April/May 2020), except for depressive symptoms, which was assessed one month prior to each analytical time point and treated as a time-varying confounder.

As we aimed to estimate the time-varying BP and WP relationships between anxiety, loneliness, and worry about COVID-19 with perceived cognitive function and abilities, we considered depressive symptoms to be a potential time-varying confounder and mediator of the J o u r n a l P r e -p r o o f relationships under study. In this scenario, antecedent depressive symptoms would predict both the current mental health exposure and cognitive outcome under study (i.e., the confounding effect), while current depressive symptoms could mediate any effects of the prior mental health exposures on the cognitive outcomes (i.e., the mediating effect).

To address this potential time-varying confounding structure in our longitudinal data, we fit marginal structural models using inverse probability of treatment weights (IPTWs) (39) . The We conducted a sensitivity analysis to further assess whether study attrition between the cohort baseline (April/May 2020) and analytic baseline (August/September 2020) could have affected our results, which would happen if attrition was differential according to loneliness, anxiety, and COVID-19 worry as well as cognitive status. We compared loneliness, anxiety, COVID-19 worry, and self-rated memory scores at the cohort baseline according to attrition status between the cohort baseline and analytical baseline. Self-rated memory scores were examined in this sensitivity analysis, as the PROMIS Cognitive Function and Abilities scales

were not included at the cohort baseline. All analyses were conducted using Stata 17.0SE

(College Station, TX).

The University of Michigan Health Sciences and Behavioral Sciences Institutional Review Board approved the COVID-19 Coping Study protocol (HUM00179632) and all participants provided informed consent. Table 2 shows the MSM estimates for the BP and WP relationships between the three mental health exposures and cognitive function and abilities. BP differences and WP changes in all three of the mental health exposure variables were negatively associated with cognitive function and abilities ( Table 2 ). For all three mental health exposure variables, the BP associations were stronger in magnitude than the WP associations, and the strongest magnitudes of association were observed for anxiety symptoms and cognitive outcomes ( Table   2 ). When models were additionally conditioned on the time-varying histories of all three mental health exposures, the results were similar (Table 3 ). In this model set, having an anxiety score that was one unit greater than the sample grand mean at any time point was associated with a 4.54 (95% CI: -5.57, -3.51) lower cognitive function T-score and a 5.09 (95% CI: -6.31, -3.86) lower cognitive abilities T-score at that same time point (Table 3) . Similarly, having an anxiety score that was one unit greater than one's personal grand mean at any time point was associated with a 3.03 (95% CI: -4.20, -1.86) lower cognitive function T-score and a 3.54 (95% CI: -5.05, -2.03) lower cognitive abilities T-score at that same time point (Table 3) . -0.01 (-0.10, 0.09) Note: Time-varying inverse probability of treatment weights were applied to each model to account for confounding by age, sex, race/ethnicity, highest level of education, relationship status, pre-COVID-19 employment status, use of any mobility aid, pre-COVID-19 social isolation score, and presence of physician-diagnosed hypertension, diabetes, heart disease, asthma, chronic obstructive pulmonary disease, cancer, or other limiting long-standing health condition, and time-varying history of depressive symptoms. All covariates were measured at the cohort baseline (April/May 2020), except for depressive symptoms, which were measured monthly and treated as a time-varying confounder. BP and WP relationships for each mental health exposure were modelled separately. All estimates applied population weights based on 2018 American Community Survey data and are weighted for attrition. 0.00 (-0.09, 0.10) Note: Time-varying inverse probability of treatment weights were applied to each model to account for confounding by age, sex, race/ethnicity, highest level of education, relationship status, pre-COVID-19 employment status, use of any mobility aid, pre-COVID-19 social isolation score, and presence of physician-diagnosed hypertension, diabetes, heart disease, asthma, chronic obstructive pulmonary disease, cancer, or other limiting long-standing health condition, and time-varying history of depressive symptoms, loneliness, anxiety symptoms, and worry about COVID-19. All covariates were measured at the cohort baseline (April/May 2020), except for depressive symptoms, loneliness, anxiety symptoms, and worry about COVID-19, each of which were measured monthly and treated as time-varying confounders. BP and WP relationships for each mental health exposure were modelled separately. All estimates applied population weights based on 2018 American Community Survey data and are weighted for attrition.

In our sensitivity analysis, we found that loneliness, anxiety, COVID-19 worry, and selfrated memory at the cohort baseline (April/May 2020) were similar among those who remained in the study at the analytical baseline (August/September 2020) compared to those who dropped out, indicating that study attrition between the cohort baseline and analytical baseline does not bias our results (Supplemental Table 4 ).

In this longitudinal study of US adults aged ≥55 over a nine-month period during the COVID-19 pandemic, we found that elevated loneliness, anxiety symptoms, and worry about COVID-19, both relative to other adults and to one's usual level, were acutely associated with worse perceived cognitive function and abilities. The associations for anxiety symptoms were of magnitudes considered to be of importance to patients in evaluating their own cognitive changes. For example, a one-unit increase on the anxiety scale relative to one's usual level was associated with PROMIS T-score decrements of 3.03 to 3.54 points, approximately one-third of a population standard deviation. The associations for loneliness and worry about COVID-19

were weaker in magnitude. Given the prevalence of clinically meaningful anxiety symptoms during the COVID-19 pandemic, both the short-term and long-term impacts of anxiety J o u r n a l P r e -p r o o f experienced during the COVID-19 pandemic for population cognitive health should be further investigated.

Our results are consistent with pre-pandemic evidence from population-based, longitudinal cohort studies indicating that elevated anxiety symptoms and anxiety disorders are associated with increased risks of Alzheimer's disease and related dementias (16, 17) . Anxiety is thought to affect cognitive health directly through neurobiological effects, and indirectly through behavioural coping mechanisms that are dementia risk factors, such as smoking, physical inactivity, and withdrawal from engagement in social relationships (22) . We did not identify any studies that have investigated the time-varying, short-term monthly relationships between anxiety symptoms and cognitive outcomes in middle-aged and older adults during or before the COVID-19 pandemic.

Two studies conducted prior to the pandemic found that between-persons differences but not within-person changes in anxiety were associated with measures of cognitive function (40, 41) . These somewhat conflict with ours, as we observed associations between withinperson changes in anxiety and cognitive function and abilities scores. One possible explanation for these inconsistent results is that the anxiety symptoms observed in this study during the COVID-19 pandemic may have been stronger in this sample than in these pre-pandemic samples, such that its relationship with cognitive function was more apparent in the present study. However, comparisons between these previous studies and ours may be difficult, as the previous two studies used different anxiety scales than ours, they had smaller sample sizes and multi-year follow-up times, and one was restricted to multiple sclerosis patients (40, 41) . A small case-control study identified that older adults with generalized anxiety disorder were more likely to have prevalent cognitive impairment than anxiety-free controls (42) , but inference cannot be made about the timing of cognitive impairment onset relative to anxiety onset with that study design. A handful of studies among patients with specific chronic diseases, such as Parkinson's J o u r n a l P r e -p r o o f disease, have identified correlations between anxiety symptoms and cognitive function, but these studies are primarily cross-sectional in design (43) (44) (45) (46) . Additional large, population-based longitudinal studies are needed to further elucidate the short-and long-term relationships between anxiety symptomatology and cognitive aging outcomes.

Previous studies on the relationship between loneliness and cognitive outcomes are primarily either cross-sectional in design or longitudinal studies with long-term follow-ups (47).

We did not identify any studies investigating the short-term, monthly associations of BP differences or WP fluctuations in loneliness and cognitive function. Notwithstanding, our results are consistent with those from prior longitudinal cohort studies indicating that loneliness is associated with accelerated cognitive decline and increased dementia risk (18) (19) (20) 48, 49) .

Prolonged loneliness is thought to be a stressor that may affect physiological cardiovascular risk pathways that lead to vascular dementias (50), although evidence for these pathways is mixed (51) . Like anxiety, loneliness is thought to adversely influence health and social behaviours that are dementia risk factors (22, 34, 51) . Future research should further investigate the associations between short-term, within-person fluctuations in loneliness in relation to cognitive aging outcomes, especially in the context of changes to population loneliness during the COVID-19 pandemic.

The PROMIS cognitive measures are self-reported. Hence, they may be subject to recall or social desirability bias, and do not objectively capture neuropsychological deficits. However, they are valuable measures that capture the impact of cognition on quality of life, which is highly meaningful to patients. They have been validated against clinical outcomes and are widely used in research studies, thereby improving comparability of our findings to existing literature (31) .

This study was conducted during a major infectious disease pandemic, and as such does not include those who were too ill to take part, such as those with severe COVID-19. Our data were collected online, and results may not be generalizable to non-Internet users (24) . However, J o u r n a l P r e -p r o o f conducting this study online allowed us to recruit a large sample and rapidly collect relevant data during a period when in-person contact was restricted due to COVID-19 risk (52) . Although we applied general population weights developed using American Community Survey data to our models, our results may be subject to residual bias if there are unmeasured drivers of study participation that are correlated with the mental health exposures and cognitive outcomes, but uncorrelated with the sociodemographic variables included in weighting (53) . This sample had a higher proportion of non-Hispanic Whites and adults with higher educational attainment than the general US population aged ≥55 years. Population weighting may thus not fully account for any selection bias of this type that could be introduced by differences in race/ethnicity and education across exposure and outcome groups.

Strengths of this study include the timeliness of data collection during the COVID-19 pandemic, large sample size, national scope, commonly used mental health and cognitive function scales, and rich covariate data. We accounted for potential sampling bias and attrition bias through model weighting, and sensitivity analyses further indicated that attrition does not bias our results. We accounted for time-varying confounding by depressive symptoms. Prior to the COVID-19 pandemic, existing research linking mental health symptoms or disorders to cognitive aging outcomes had been largely based on measures of mental health symptoms or diagnoses assessed once at study baseline, despite their time-varying nature. We were able to investigate how within-person fluctuations in anxiety symptoms, loneliness, and COVID-19 worry were related to perceived cognitive function and abilities. However, despite our approach to account for time-varying confounding by depression and other mental health symptoms and our analysis of within-person changes in mental health, our results may still be subject to confounding by unmeasured factors such as personality. We did not identify any other studies that have examined these time-varying relationships on a short-term scale, highlighting the uniqueness of the present study and the contribution it makes to the literature.

In this large, nation-wide study of US adults aged ≥55 years over a nine-month period during the COVID-19 pandemic, we identified that elevated loneliness, anxiety symptoms, and worry about COVID-19 were associated with worse perceived cognitive function and abilities.

These associations were observed both in comparison to other adults of similar age and to one's own usual levels of loneliness, anxiety, and worry about COVID-19. These findings were independent of a range of sociodemographic, social, and health-related covariates, including depressive symptomatology. Adverse mental health effects are a mechanism through which the COVID-19 pandemic is thought to influence the future burden of dementia among the older population (54, 55) . However, it remains unclear whether acute changes in population mental health during the pandemic will have long-term implications. Future research in other populations and settings should evaluate how within-person fluctuations in mental health symptomatology in the wake of the pandemic may affect cognitive outcomes during aging. 

Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Writing -review & editing

COVID-19 mental health impact and responses in low-income and middle-income countries: reimagining global mental health

Impact of COVID-19 pandemic on mental health in the general population: A systematic review

Mental health and the COVID-19 pandemic

Mental health before and during the COVID-19 pandemic: a longitudinal probability sample survey of the UK population

COVID-19 and the consequences of isolating the elderly

Social Isolation and Loneliness Among San Francisco Bay Area Older Adults During the COVID-19 Shelter-in-Place Orders

The mental health consequences of COVID-19 and physical distancing: the need for prevention and early intervention

Loneliness Among Older Adults Before and During the COVID-19 Pandemic

Mental Health, Substance Use, and Suicidal Ideation During the COVID-19 Pandemic -United States

Prevalence of depression symptoms in US adults before and during the COVID-19 pandemic

Anxiety reported by US adults in 2019 and during the 2020 COVID-19 pandemic: Population-based evidence from two nationally representative samples

Mental health of older adults during the COVID-19 pandemic: lessons from history to guide our future

The Impact of COVID-19 Infection and Enforced Prolonged Social Isolation on Neuropsychiatric Symptoms in Older Adults With and Without Dementia: A

Social disconnectedness, perceived isolation, and symptoms of depression and anxiety among older Americans (NSHAP): a longitudinal mediation analysis

Articles Lancet Public Health

Neurobiology of loneliness: a systematic review

Is Physical Activity Associated with Less Depression and Anxiety During the COVID-19 Pandemic? A Rapid

Alcohol Research: Current Reviews -Co-Occurring Alcohol Use Disorder and Anxiety: Bridging Psychiatric, Psychological, and

Anxiety and risk of dementia: Systematic review and meta-analysis of prospective cohort studies

Updating the evidence for an association between anxiety and risk of

Alzheimer's disease: A meta-analysis of prospective cohort studies

Loneliness and risk of dementia

Loneliness increases the risk of allcause dementia and Alzheimer's disease

Does loneliness contribute to mild cognitive impairment and dementia? A systematic review and meta-analysis of longitudinal studies

The Association between Social Engagement, Loneliness, and Risk of Dementia: A Systematic Review and Meta-Analysis

Dementia prevention, intervention, and care: 2020 report of the Lancet Commission

Depression and cognition in older adults

Cohort profile: the COVID-19 Coping Study, a longitudinal mixed-methods study of middle-aged and older adults' mental health and well-being during the COVID-19 pandemic in the USA

UCLA Loneliness Scale (Version 3): reliability, validity, and factor structure

The HRS Psychosocial Working Group. Health and Retirement Study: Psychosocial and Lifestyle Questionnaire

Cancer-related health worries and psychological distress among older adult, long-term cancer survivors

Psychological Predictors of Health Anxiety in Response to the Zika Virus

Dementia worry: A psychological examination of an unexplored phenomenon

Worry about health in smoking behaviour change. Behaviour Research and Therapy

Cognitive function: a brief guide to the PROMIS® cognitive function instruments. Patient-Reported Outcomes Measurement Information System: Dynamic Tools to Measure Health Outcomes From the Patient Perspective. Chicago: Northwestern University

Minimal important change (MIC): a conceptual clarification and systematic review of MIC estimates of PROMIS measures. Quality of Life Research

Social isolation, loneliness, and health behaviors at older ages: longitudinal cohort study

The CES-D scale: A self-report depression scale for research in the general population

Constructing inverse probability weights for marginal structural models

Causal Inference: What If

Association of a negative wealth shock with all-cause mortality in middle-aged and older adults in the United States

Analysis of longitudinal complex sample survey data

Anxiety and depression affect performance on the symbol digit modalities test over time in MS and other immune disorders

Effects of between-person differences and within-person changes in symptoms of anxiety and depression on older age cognitive performance

Cognitive impairment in late-life generalized anxiety disorder

Association of depression, anxiety, and type D personality with cognitive function in patients with coronary artery disease

State anxiety predicts cognitive performance in patients with Parkinson's disease. Neuropsychol

Diabetes and anxiety adversely affect cognition in multiple sclerosis

Cognitive function in Parkinson's disease: Association with anxiety but not depression

Loneliness and cognitive function in the older adult: A systematic review

Social isolation and loneliness: relationships with cognitive function during 4 years of follow-up in the English Longitudinal Study of Ageing

Loneliness, depression and cognitive function in older U.S. adults

Loneliness matters: A theoretical and empirical review of consequences and mechanisms

Loneliness, social isolation, and behavioral and biological health indicators in older adults

Online health survey research during COVID-19

A structural approach to selection bias

J o u r n a l P r e -p r o o f Highlights:-We investigated mental health and perceived cognition over nine months during COVID-19 -Elevated loneliness, anxiety symptoms, and worry about COVID-19 relative to other adults of similar age and to one's usual levels were associated with worse cognition -Anxiety symptoms had the strongest negative associations with perceived cognition -Results were independent of depressive symptomatology and other confounders -Impacts of mental health symptomatology during the pandemic on long-term cognitive aging should be studied further J o u r n a l P r e -p r o o f

☐ The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.☒ The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Lindsay C. Kobayashi reports financial support was provided by National Institute on Aging. Jessica M. Finlay reports financial support was provided by National Institute on Aging. Jessica M. Finlay reports financial support was provided by National Center for Advancing Translational Sciences.