key: cord-0755119-gy09p2e9
authors: Crandall, Carolyn J; Larson, Joseph; Cene, Crystal Wiley; Bellettiere, John; Laddu, Deepika; Jackson, Rebecca D; Schumacher, Benjamin T; Stefanick, Marcia L
title: Relationship of Social Connectedness with Decreasing Physical Activity during the COVID-19 Pandemic among Older Women Participating in the Women’s Health Initiative Study
date: 2022-05-21
journal: J Gerontol A Biol Sci Med Sci
DOI: 10.1093/gerona/glac108
sha: c16920b876ad2beefd6ed024a2c59dea0a3ce4dc
doc_id: 755119
cord_uid: gy09p2e9

BACKGROUND: Aging is generally accompanied by decreasing physical activity, which is associated with a decline in many health parameters, leading to recommendations for older adults to increase or at least maintain physical activity (PA). METHODS: We determined relationships between social connectedness and decreasing or increasing PA levels during the COVID-19 pandemic among 41,443 participants of the Women’s Health Initiative Extension Study. Outcomes of logistic regression models were decreasing PA activity (reference: maintaining or increasing) and increasing PA activity (reference: maintaining or decreasing). The main predictor was social connectedness as a combined variable: not living alone (reference: living alone) and communicating with others outside the home more than once/week (reference: once/week or less). We adjusted for age, race, ethnicity, body mass index, physical function level, and education. RESULTS: Compared with participants who were not socially connected, socially connected participants had lower odds of decreasing PA (adjusted odds ratio 0.91, 95% confidence interval 0.87-0.95). Odds of increasing PA (vs. decreasing or maintaining PA) were not significantly different among socially connected and not socially connected participants. Associations between social connectedness and decreasing PA did not significantly differ by age (<85 vs. ≥85 years), race/ethnicity (non-Hispanic White vs. other races/ethnicity), education (college vs. <college), use of technology to stay in touch with others, or RAND physical function score (≤75 vs. >75). CONCLUSIONS: Social connectedness was associated with lower odds of decreasing PA among older women during the pandemic. These findings could inform the development of future interventions to help older women avoid decreasing PA..

Well-documented benefits of physical activity include reduced risk of heart disease, hypertension, stroke, several cancers, including breast and colorectal cancer, diabetes, falls, and premature death, as well as improved sleep, balance, and joint mobility.(1-4) Given these benefits, and recognizing the aging of the U.S. population (5, 6) , it is particularly important to ensure adequate physical activity among older persons. Concerns about a decrease in PA levels during the coronavirus disease 2019 (COVID-19) pandemic (7, 8) are of major public health importance, considering that the majority of U.S. adults were not obtaining recommended levels of physical activity (PA) even before the pandemic, with fewer than 20% of women aged 65 years and older engaging in sufficient physical activity(1, 2).

The pandemic also posed challenges to maintaining social connectedness (e.g. social isolation, network size, social integration, social support), which could also adversely influence health outcomes; social isolation is associated with increased risk of death. (9) Also, among women aged between 65 and 99, social isolation and loneliness have each been independently associated with higher risk of cardiovascular disease.(10) Findings from an online survey conducted in April 2020 of over 1000 North American adults aged 50 and over suggested that even light physical activity during the COVID-19 pandemic may have alleviated some of the negative mental health impacts that older adults were experiencing while isolated and adhering to social distancing guidelines.(11) A Scottish survey (mean participant age 32 years) similarly found that less PA was associated with greater negative mood.(12) How social connectedness influences PA levels public health crises that mandate social distancing, such as the COVID-19 pandemic, is important to understand. While lack of social support during the COVID-19 was reported as a reason for not doing PA among adults in Denmark. (13) , and staying in touch with family and friends and staying active were each rated as top priorities in the midst of the COVID-19 pandemic in a subset of older A c c e p t e d M a n u s c r i p t 6 women assigned to a physical activity intervention in a randomized controlled trial (14, 15) , one U.S. study found no association of social support with PA during the pandemic. (16) To our knowledge, no published studies have ascertained whether social connectedness predicts greater likelihood of maintaining or increasing PA during the COVID-19 pandemic among older women in the U.S..

To examine this question, we adapted a conceptual model (Figure 1 ) from the National Academies of Sciences, Engineering, and Medicine 2020 report on Social Isolation and Loneliness in Older Adults.(17) Both quantity and quality of social connections influence physical activity. For this analysis, we were interested in the impact of social connections on physical activity. Based on previous studies, age, race/ethnicity (as a proxy for structure inequities) and socioeconomic status may influence the quality and quantity of an individual's social connections. These factors have also been associated with physical activity, therefore we considered these factors as confounders and adjusted for them in our analyses.. We considered the potential moderating effects of age, race/ethnicity, and physical function. The goal of this study was to determine the relationships between social connectedness (expressed as number of persons living in the household and the frequency of communication with others living outside the home) and decreasing PA levels or increasing PA among older women during the COVID-19 pandemic. The well-characterized cohort of older women residing across the U.S. (aged 71-104 years) participating in the Women's Health Initiative Extension study provided the opportunity to explore the hypothesis that social connectedness during the pandemic (defined as not living alone and having more than one communications per week with others living outside the household) would be associated with lower odds of decreasing PA level but would not be significantly associated with increasing PA level during the COVID-19 pandemic.

. A c c e p t e d M a n u s c r i p t 7

Between 1993 and 1998, 161,808 postmenopausal women aged 50-to 79 years were enrolled in the Women's Health Initiative (WHI) at 40 clinical sites. The WHI consisted of the randomized clinical trials, which tested three interventions (menopausal hormone therapy, calcium plus vitamin D supplementation, and/or low-fat dietary pattern intervention), and the WHI Observational Study, which was designed to determine important causes of morbidity and mortality among postmenopausal women.(18) Women with less than three years of predicted survival or who planned to move within three years were excluded from participating, with additional exclusions for each clinical trial. After the clinical trial and observational study phase was completed (1993) (1994) (1995) (1996) (1997) (1998) (1999) (2000) (2001) (2002) (2003) (2004) (2005) , all active study participants were invited to continue participating for five years at their respective clinical centers (2005) (2006) (2007) (2008) (2009) (2010) , after which those who were still actively participating were invited to continue ongoing follow-up in the WHI Extension Study (2010-present) . Each participant provided written informed consent to participate. The WHI Extension Study includes a substudy, the Women's Health Initiative Strong and Healthy (WHISH) trial, which is testing whether increasing physical activity will reduce heart disease and stroke in older women. Institutional review board approval was obtained by the institutions affiliated with four WHI regional centers and/or the WHI Coordinating Center at the Fred Hutchinson Cancer Research Center, which assumed the role of IRB of record for each participant.

In 2020, all participants who were alive and had consented to be contacted by WHI (n = 64,350) were invited to complete the WHI COVID-19 survey which included items about self-rated well-being, medical history (high blood pressure, diabetes, cancer, autoimmune disease), change in living arrangement, number of people living in the household, access to A c c e p t e d M a n u s c r i p t 8 visitors, restriction of exit and entry to home, exposure to persons suspected of being COVID-19 infected, death of family or close friends due to COVID-19, COVID-related symptoms, COVID-19 testing (frequency, nasal swab, throat swab, saliva test, blood test, test results), hospital stays or treatments for COVID-19, access to medication and health care utilization during the pandemic, degree of concern regarding the pandemic, type and frequency of communication with others outside the home, the use of technology to stay in touch with others, alcohol use, smoking, and physical activity. The COVID-19 survey included the items from the Perceived Stress Scale and the Patient-Reported Outcomes Measurement Information System Emotional Distress-Anxiety Short Form 4 (PROMIS) scale (https://www.healthmeasures.net/explore-measurement-systems/promis/intro-to-promis/listof-adult-measures). Also, the questionnaire include an item asking "Has anyone in your family or a close friend died from COVID-19?" Response choices were "no" and "yes".

Of the 64,350 participants eligible to receive the COVID-19 survey, 49,695 (77.2%) participants completed the survey (Figure 2) . We excluded data from participants who reported ever having had a positive COVID test (n = 311), those who did not provided data regarding change in physical activity during vs. before the pandemic (main outcome) or social connectedness (main predictor)(n = 6,651), and women for whom information regarding covariates were missing (1,290), resulting in an analytic sample of 41,443 participants.

On the COVID-19 questionnaire, participants were asked, "Over the past month, how would you describe your level of physical activity or exercise compared to your average physical activity level before the COVID-19 pandemic began?" Response choices included:

A c c e p t e d M a n u s c r i p t 9 "much less", "somewhat less", "about the same", "somewhat more", and "much more". We defined maintaining PA as "about the same". For the statistical analyses, we created two binary (yes/no) outcome variables. The first variable was decreased physical activity ("much less" or "somewhat less" vs. "about the same", "somewhat more", or "much more"). The second variable was increased physical activity ("somewhat more" or "much more" vs.

"much less", somewhat less", or "about the same").

Regarding living alone, participants were asked to report the number of participants living in the household: "Including yourself, how many people living in the same household with you?" Response choices were: "1", "2", ""3, "4", "5 or more", and not applicable. For statistical analyses, we collapsed the categories to a binary variable of not living alone (1, 2, 3, 4, or 5 or more living in the same household with you) or living alone.

Regarding communication with others outside the home, participants were asked, "How often do you communicate with others who live outside your home?" The responses choices were: "every day","several X/week", "1-2 X/week", "once/week", or "rarely/never".

For statistical analyses, we collapsed the categories into a binary (yes/no) variable: more than once per week ("once/week", "1-2X/week", "several X/week", or "every day") vs. once per week or less ("rarely/never", "once/week"). In addition, participants were asked "compared to the months before the outbreak began, would you say that this is…: "more often than before", "about the same as before", of "less often than before".

Based on two binary variables described above (living alone and communication with others outside the home), we defined social connectedness as a combined variable: not living alone and communicating with others more than once per week.

A c c e p t e d M a n u s c r i p t

At baseline, we used a self-assessment questionnaire for information regarding age, race, ethnicity, smoking, and highest education attained. On annual questionnaires, as well as on the COVID-19 survey, participants updated their medical history, including information regarding cardiovascular disease, peripheral artery disease, cancer, diabetes, and hypertension.

From annual questionnaires, we accessed information regarding physical function and body mass index (BMI). For each participant, we used data from the most recently-collected survey data (within the last two years) prior to the COVID-19 survey. Physical function was assessed using the RAND 36-item health survey physical functioning construct (range 0-100, higher score indicates more favorable health state). 

We used logistic regression to determine associations between social connectedness (primary predictor) and decreasing physical activity (reference not decreasing physical activity during vs. before the COVID-19 pandemic (primary outcome). The secondary pre-A c c e p t e d M a n u s c r i p t 11 specified outcome, examined in a separate logistic regression model, was increasing physical activity (reference: not increasing) during vs. before the COVID-19 pandemic.

We adjusted the logistic regression models for potential moderators (age and education) as well as potential confounders selected a priori: BMI and physical function level prior to COVID-19 pandemic, race, and ethnicity. All models were adjusted for Women's Health Initiative Study component (clinical trial, observational study) and

Women's Health Initiative Strong and Health Trial intervention assignment (intervention, control, not randomized). In the subset of participants who provided information regarding hours per week of physical activity prior to the pandemic (n = 26,847), we performed a sensitivity analysis in which we repeated the main analysis described above before and after adjustment for tertile of hours per week of physical activity prior to the pandemic. In another sensitivity analysis, we repeated the logistic regression models using maintained physical activity as the reference group. Specifically, we compared decreased (vs. maintained physical activity), and increased (vs. maintained) physical activity.

To test the hypothesis that associations between social connectedness and PA during (versus before) the COVID-19 pandemic differed based on characteristics selected a priori, we used statistical interaction terms (social connectedness * characteristic). In this way, we tested for effect modification by age (<85 vs. ≥85 years), education (no ≤high school vs. ≥ college), race and ethnicity (Non-Hispanic White vs. all other race and ethnicities), using technology to stay in touch with others (yes vs. no), physical function score (RAND 36-item health survey physical functioning score ≤75 vs. >75), and change in the frequency of communication with others outside the household during (compared with before) the pandemic.

A c c e p t e d M a n u s c r i p t

Sociodemographic and clinical characteristics of the study participants are presented in Table 1 . Mean (SD) participant age was 83.2 (5.4) years; mean (SD) BMI was 26.1 (5.2) .

Of the total group of 41,443 women, 15,679 women (38%) were aged 85 years and older.

The majority of participants reported being White (n = 37,469, 90%) and non-Hispanic (40,215, 97%) with 465 participants identifying with more than one racial group. Forty-seven percent of participants reported living alone and 7% of participants reported communicating with others outside the home once per week or less.

Of the 41,443 participants, 20,092 participants (48%) were socially connected (reported communicating with others outside the household more than once per week and were not living alone); 21,351 (52%) were not socially connected.

Compared with women who were not socially connected, characteristics of those who were socially connected were similar. Socially connected women were more likely to have a high RAND physical function score (29% vs. 24% had score ≥90) and rate their well-being as excellent (53% vs. 48%); they were less likely to stay in touch with others by speaking in person (37% vs. 42%). 22,547 (54%) reported that their PA during the COVID-19 was less than before the pandemic; 15,332 (37%) reported that their PA during the pandemic was the same as before the pandemic, and 3,564 (9%) reported PA during the A c c e p t e d M a n u s c r i p t 13 pandemic being more than before the pandemic. Women who reported decreasing PA were less likely to have high physical function scores, less likely to report excellent/very good well-being, and more likely to report being very concerned about the COVID-19 pandemic.

We present results of two sets of models for the two pre-specified outcomes:

decreased PA (primary outcome) and increased PA (secondary outcome) ( Table 2) .

Compared with women who were not socially connected, in unadjusted models, women who were socially connected had significantly lower odds of decreasing their PA levels during the There was no statistically significant difference in the odds of increasing PA between women who were not socially connected and women who were socially connected.

In a sensitivity analysis among participants who filled the CHAMPS physical activity questionnaire prior to the pandemic, results were very similar before compared to after adjustment for tertile of pre-pandemic physical activity (Supplemental Table 3 ).

In a second sensitivity analysis, which examined increased (vs. maintained) physical activity and decreased (vs. maintained) physical activity), results were very similar to those of the primary models (Supplemental Table 4 ).

A c c e p t e d M a n u s c r i p t 14

Associations between social connectedness and decreased PA during vs. before the COVID-19 pandemic did not significantly differ by age (85 years-old or older versus younger than 85 years-old), race and ethnicity (non-Hispanic White versus all other race and ethnicities), education (college graduate yes vs. no), physical function level prior to the pandemic (RAND physical function score ≤75 vs. >75), use of technology to stay in touch with others (women who did not use technology vs. women who did use technology, and change in frequency of communication with others outside the household during (compared with before) the COVID-19 pandemic (all interaction p values >0.05) ( Table 3 ).

In this large cohort of older women from across the U.S., women who were socially connected were significantly less likely to decrease their PA levels during the COVID-19 pandemic than women who not socially connected. These associations between social connectedness and decreasing PA were similar among women aged 85 years and over vs. <85 years-old, women who had lower vs. higher pre-pandemic physical function score, women who reported being college graduates vs. not being college graduates, and using vs. not using technology to stay in touch with others. We had hypothesized a priori that social connectedness would be associated with increased or maintained PA during the pandemic.

Our findings were consistent with the pre-specified hypothesis.

We cannot directly compare our results with those of previous similar studies. To our knowledge, studies have not directly assessed whether the indicators of social connectedness that we examined are associated with the odds of decreasing PA in older women during the A c c e p t e d M a n u s c r i p t 15 COVID-19 pandemic in the U.S.. However, our results are generally consistent with published studies of PA during the pandemic, such as: living with smaller number of persons in the household was associated with "unhealthy lifestyle" (composite of physical activity, nutritional components, and other features) among adults in Spain, (23) , that anxiety negatively influenced the intention to do physical activity among adults in Italy,(24) that living alone was significantly related to declines in PA among adults in the U.S. (25), that more social support was associated with higher moderate-vigorous PA among adolescents in the U.S. (26), that greater social media use (Facebook, Facebook Messenger, Instagram, WhatsApp, TikTok) is related to higher step count (assessed via smartphone application) among adult psychiatric outpatients in Spain,(27) greater social support was associated with PA during the pandemic among student athletes,(28) and that higher social support is associated with better self-reported health behavior among hospitalized patients in Thailand.(29) Our findings are also consistent with studies reporting that no longer meeting PA guidelines during the COVID-19 pandemic was associated with loneliness among U.S. adults (30).

However, results of one study suggested that greater social support was not significantly associated with PA during the pandemic among U.S. adults (not specific to older persons) (16) and another study performed in Denmark found that the largest decline in minutes per week of physical activity during the COVID-19 pandemic occurred among adults who lived with their parents (40% decline), and the smallest decline in physical activity occurred among persons living alone (8% decline).(13) However, neither of the latter two studies specifically focused on older women.

While the majority of results of previously-published studies are consistent with the results and hypothesis of the current study, they are not necessarily representative of the general population of older women in the U.S., because they examined participants residing A c c e p t e d M a n u s c r i p t 16 outside of the U.S. ((13, 23, 24, 27-29) , focused on special populations such as adolescents or teenagers (26, 28), psychiatric outpatients (27), hospitalized patients (29), and/or did not provide results specific to older women (23-25), (13, 16, 27, 29) Finally, results of one study suggested that greater social support was associated with PA during the pandemic among student athletes (not specific to older persons). (28) In the broader context of interventions that have been tested to increase PA among older women, The 2018 Physical Activity Guidelines Advisory Committee Scientific Report for the Secretary of U.S. Health and Human Services found that, among older adults, the effectiveness of interventions to increase physical activity were consistently positive, but that the magnitude of the effect was not easy to determine.(https://health.gov/our-work/physicalactivity/current-guidelines/scientific-report) Indeed, the limited number of studies without an active control group and the diverse range of physical activity outcomes precludes the use of meta-analysis to provide a summary of intervention effectiveness. One systematic review reported that interventions had a small effect on physical activity (Cohen's d = 0.14, 95% Scientific Report to conclude that "strong evidence demonstrates that physical activity interventions that target older adults have a small but positive effect on physical activity when compared with minimal or no-treatment controls."

Our results have public health relevance. Decreasing PA during the COVID-19 pandemic has been associated with higher depression, anxiety, and stress symptoms (Australian survey (31)), and people who exercised more during the pandemic had higher psychological resilience (cross-sectional U.S. survey) (32). Among identical twins in the U.S., a decrease in PA during the pandemic was associated with higher anxiety levels. (33) If acute changes to behavior, such as decreased PA, are sustained for longer periods of time, A c c e p t e d M a n u s c r i p t 17 they could be associated with higher risk for conditions linked with low PA, such as cardiovascular disease and diabetes.

It is difficult to ascertain cause and effect in these associations. Anxiety may have caused a decrease in PA, but the decrease in PA may have caused anxiety. For example, a study of persons in Norway aged 18-81 years found that reduced physical activity during the COVID-19 lockdown was associated with higher risk of anxiety and depression.(34) The safety of PA outdoors might be influenced by where someone lives. Also, the pandemic may increase sedentary behavior (which may have its own relationships with social connectedness) along with decreases in PA. Finally, the potential dependence of using devices to communicate with others outside the household may be a surrogate for greater wealth (allowing the purchase of electronic devices).

Our study has several potential limitations. We only had information regarding hours per week of pre-pandemic PA levels in a subset of participants (26,847 of 41,443 participants). Second, we did not have information regarding the potential influence of COVID-19 mitigation strategies on the use of certain venues for PA among our study participants, which could in part explain our results. Third, we did not have information regarding social network size, social integration, loneliness, or relationship quality. Finally, we lacked recent information regarding social support and marital status, although we were able to focus on the number of people living in the household. Results of this study cannot be generalized to social connectedness and PA during non-pandemic times, as the current study was designed specifically to evaluate PA during the COVID-19 pandemic. However, our results may apply to circumstances that result from other types of public health emergencies.

Strengths of our study include the large number of participants, the detailed information regarding covariates including physical function and technology use, and the A c c e p t e d M a n u s c r i p t 18 racial/ethnic diversity of the study cohort. In addition, we use a temporally-anchored selfreport of physical activity for analyses regarding the change of activity prior to and during the pandemic in relation to social connectedness. Our results are consistent regardless of whether the comparator was defined as women who maintained PA or defined as women who did not increase (or decrease) PA, i.e. when the reference was maintained or increased, or maintained or decreased, PA. This consistency further supports our hypothesis that older women who had greater social connectedness would be less likely to decrease their PA during the COVID-19 pandemic.

In conclusion, these findings provide insights into the potential importance of maintaining communication with others outside the household and not living alone in efforts to avoiding decreases in PA among older women during future local and national crisis periods, and could inform the development of future interventions designed to help older women avoid decreasing PA more generally. 15 Video calls, email, or social media 16 Includes participants with unknown race and/or ethnicity A c c e p t e d M a n u s c r i p t M a n u s c r i p t 39 Figure 2 

Charles Kooperberg Investigators and Academic Centers: (Brigham and Women's Hospital

Department of Health and Human Services. Physical Activity Guidelines for Americans 2nd Edition

Department of Health and Human Services

Department of Health and Human Services

United States Census. Projections for the United States

Department of Health and Human Services AfCL. 2020 Profile of Older Americans

A tale of two pandemics: How will COVID-19 and global trends in physical inactivity and sedentary behavior affect one another?

Physical Activity Among Predominantly White Middle-Aged and Older US Adults During

Health Initiative Strong and Healthy (WHISH) intervention survey

Women's Health Initiative Strong and Healthy Pragmatic Physical Activity Intervention Trial for Cardiovascular Disease Prevention: Design and Baseline Characteristics

Health Behaviors at the Onset of the COVID-19 Pandemic

Board on Health Sciences Policy

Division of Behavioral and Social Sciences and Education. Social isolation and loneliness in older adults : opportunitiies for the health care system

Design of the Women's Health Initiative clinical trial and observational study. The Women's Health Initiative Study Group

Test-retest performance of a mailed version of the Medical Outcomes Study 36-Item Short-Form Health Survey among older adults

The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item selection

The RAND 36-Item Health Survey 1.0. Health Econ

CHAMPS physical activity questionnaire for older adults: outcomes for interventions

The assessment of lifestyle changes during the COVID-19 pandemic using a multidimensional scale

Outbreak and Physical Activity in the Italian Population: A Cross-Sectional Analysis of the Underlying Psychosocial Mechanisms

The authors express their thanks to the WHI study participants, WHI investigators, and WHI study staff for their tremendous dedication and commitment.

Program Office: (National Heart, Lung, and Blood Institute, Bethesda, Maryland) Jacques Rossouw, Shari Ludlam, Joan McGowan, Leslie Ford, and Nancy Geller

A c c e p t e d M a n u s c r i p t 21 A c c e p t e d M a n u s c r i p t 28