Higher amyloid is associated with greater loneliness among cognitively normal older adults during the COVID-19 pandemic

Participants

Community-dwelling older adults were recruited for this cross-sectional study from an ongoing longitudinal cohort study at the Knight Alzheimer Disease Research Center (Knight ADRC) at Washington University in St. Louis (Bollinger et al., 2021). The inclusion criteria for the cohort study were: (1) ≥65 years old, (2) CN as determined by a Clinical Dementia Rating^® (CDR; Morris, 1993) score of 0, and (3) had CSF biomarkers and/or neuroimaging (PET and/or MRI) within two years of study enrollment. Individuals were excluded if they had a medical or psychiatric diagnosis that could interfere with longitudinal follow-up or neuroimaging or adversely affect cognition. Participants included in this secondary analysis had amyloid PET data processed within two years of study enrollment. The Knight ADRC recruitment procedures have been published previously (Morris et al., 2019).

Recruitment

At the beginning of the COVID-19 pandemic, individuals were approached by phone or e-mail and invited to participate. Written informed consent was obtained, and participants completed surveys between 27 April 2020 and 8 June 2020 regarding their mental health with staff via phone or secure e-mail using a Research Electronic Data Capture (REDCap; Harris et al., 2009) survey link. Upon completion, participants could elect to enter a drawing for a $50 gift card. This study was approved by the Institutional Review Board at Washington University in St. Louis (reference number: 201807135).

Measures

The global CDR score (Morris, 1993) from the participant’s most recent annual clinical visit at the Knight ADRC was used to assess the level of impairment from normal to severe symptomatic AD. The CDR is scored as follows: 0 (CN), 0.5 (very mild dementia), 1 (mild dementia), 2 (moderate dementia), and 3 (severe dementia). A CDR score of 0 was required for enrollment in this study.

PET amyloid imaging was used to obtain AD biomarker data through previously described methods (Su et al., 2013, 2015). Participants were injected with either [11C] Pittsburgh compound B (PiB) or [18F]-Florbetapir (AV45), and dynamic scans were acquired. The post-injection time window for quantification was 30–60 minutes for PiB and 50–70 minutes for AV45. A PET Unified Pipeline was implemented to process the data. In short, a region of interest (ROI) segmentation approach was used with FreeSurfer 5.3 (Martinos Center for Biomedical Imaging, Charlestown, MA, USA). Based on the ROI segmentation, a tissue mask was created, and partial volume correction was used (Su et al., 2013). For each ROI, a standard uptake ratio (SUVR) was acquired with the cerebellar gray matter serving as a reference region. Cortical regions affected by AD were used to obtain a summary value (mean cortical uptake ratio) of the partial-volume corrected SUVR (Mintun et al., 2006). Centiloids were used to synchronize measures from both PiB and AV45 tracers, with global amyloid deposition scaled from 0–100. Centiloid cutoffs for PiB and AV45 mean cortical SUVR were 16.4 and 20.6, respectively (Klunk et al., 2015). A value higher than these cutoffs is considered preclinical AD.

Participants answered questions related to the COVID-19 pandemic to capture temporally relevant personal and social characteristics. Demographic characteristics and six COVID-19-related questions were collected via self-report. Annual income was recoded to reflect the federal poverty level (Office of the Assistant Secretary for Planning and Evaluation, 2020) and median household income for adults over 65 in the U.S. (Semega et al., 2020): <$20,000 (low income), $20,000–$50,000 (below U.S. median), and over $50,000 (above U.S. median). COVID-19-related questions included whether the participant had been diagnosed with COVID-19, had practiced social distancing, and four questions to assess financial strain related to the pandemic: “Since the start of the COVID-19 pandemic in the U.S., have you (1) involuntarily lost a job; (2) taken a cut in wage, salary, self-employed income, or similar; or (3) newly applied for public assistance or unemployment?” Participants also rated their financial stress on a scale from 0 (no financial stress) to 10 (severe financial stress).

Loneliness was defined as negative thoughts and feelings of being isolated and disconnected from others (Russell, Peplau, & Ferguson, 1978). Loneliness was measured using version 1 of the 20-item UCLA Loneliness Scale (Russell, 1996). Participants reported how often they experienced subjective feelings of isolation, feeling left out, and lacking companionship on a scale from 1–5. Total scores on this measure range from 20–80, with lower scores indicating less loneliness. This measure does not have established cutoff scores to indicate levels of loneliness, and it is recommended to examine the full range of scores when using this measure. This instrument has high internal consistency (0.96) and a test-retest correlation of 0.73. In our sample, internal consistency was high, with Cronbach’s α = 0.95.

Anxiety was defined as feeling nervous, worried, or frightened (Marks & Lader, 1973). Anxiety was measured using the Patient-Reported Outcomes Measurement Information System (PROMIS) Emotional Distress–Anxiety short form 4a (Pilkonis et al., 2011) and the Hospital Anxiety and Depression–Anxiety subscale (HADS-A; Zigmond & Snaith, 1983). The PROMIS Emotional Distress–Anxiety short form 4a measures feelings of being fearful, overwhelmed, and uneasy over the past seven days on a scale from 1 (never) to 5 (always). Total raw scores are converted to uncorrected t-scores and range from 40.3–81.6, with a mean score of 50 and standard deviation of 10. Higher scores indicate greater anxiety. This measure has demonstrated content validity across diverse populations (Pilkonis et al., 2011). The HADS-A measures how often participants have felt tense, frightened, and restless over the past week on a scale from 1 (not at all or only occasionally) to 5 (most of the time or very often; (Zigmond and Snaith, 1983). Total scores range from 0–21, with scores greater than 10 indicating clinically meaningful anxiety. This measure has excellent reliability (intraclass correlation coefficient = 0.92) and good internal consistency (Cronbach’s α = 0.87; (Djukanovic et al., 2017). In our sample, internal consistency measures for the PROMIS and HADS-A were Cronbach’s α = 0.79 and 0.76, respectively.

Depression was defined as feelings of sadness, hopelessness, or a lack of interest or pleasure in completing daily activities (Herrman et al., 2022). Feelings of depression were assessed using the Patient Health Questionnaire 9 (PHQ-9; (Kroenke et al., 2001). The PHQ-9 includes all nine Diagnostic and Statistical Manual Fourth Edition (American Psychiatric Association, 2000) diagnostic criteria for major depressive disorder and measures how often a participant has felt bothered by symptoms such as little interest in doing things, trouble concentrating, and feeling bad about oneself in the past two weeks on a scale from 0 (not at all) to 3 (nearly every day). Scores on the PHQ-9 range from 0–27, with higher scores indicating more severe depression (Kroenke et al., 2001). The PHQ-9 has internal reliability (Cronbach’s α = 0.86–0.89), excellent test-retest reliability, and established criterion and construct validity (Kroenke et al., 2001). Internal reliability in our sample was Cronbach’s α = 0.75.

Statistical analyses

Data were analyzed using SPSS 27.0 (IBM Corp., Armonk, NY; IBM Corp., 2017). Descriptive analyses were performed to examine demographics, COVID-19-related questions, and psychosocial characteristics (loneliness, anxiety, and depression). Multiple linear regression was used to estimate the associations between loneliness and amyloid accumulation. We controlled for relevant covariates identified through domain knowledge and clinical interest, including anxiety, depression, age, gender, level of financial stress, living situation (alone versus with others), and practice of social distancing, which were added to the model through forced entry. Because the cohort study included two measures of anxiety (PROMIS and HADS-A), we included only the PROMIS Anxiety measure for anxiety in this model given its higher precision in scoring compared to the HADS-A. Ninety-five percent confidence intervals (CIs) for regression coefficients were calculated for all independent variables.

Underlying data

The full dataset supporting the findings of this study could not be made publicly available as some of the data was collected by the Knight Alzheimer Disease Research Center (ADRC) who requires a separate request. This can be made through ADRC’s website: https://knightadrc.wustl.edu/data-request-form/. In their request, applicants will need to provide the following information: purpose, background and preliminary data, methods, inclusion and exclusion criteria, analytic plan, sample size justification and a list of data variables. A list of variables available for request from the Knight ADRC may be found on their website: https://knightadrc.wustl.edu/professionals-clinicians/request-center-resources/guidelines-data-available/.

Digital Commons@Becker: Higher amyloid correlates to greater loneliness during the COVID-19 pandemic. https://doi.org/10.48765/m113-0447 (Kehrer-Dunlap et al., 2022).

This project contains the dataset of psychosocial measures.

Extended data

Digital Commons@Becker: Higher amyloid correlates to greater loneliness during the COVID-19 pandemic. https://doi.org/10.48765/m113-0447 (Kehrer-Dunlap et al., 2022).

This project contains the following extended data:

Data are available under the terms of the Creative Commons Attribution 4.0 International license (CC-BY 4.0).