RESULTS
A total of 2006 respondents to the survey provided complete information for all model variables. Of these respondents, 59.5% were aged 50–64 and 52.6% were female (Table 1). The sample was comprised of non-Hispanic Whites (71.1%), Hispanics (11.6%), non-Hispanic Blacks (10.5%), and Other racial/ethnic backgrounds (6.8%). Four in 10 individuals had high school or less education (40.1%) and 39.0% had household incomes of less than $60,000. Approximately one in 6 individuals (15.5%) reported being in fair or poor health, and 45.9% felt isolated (some of the time or often) from others.
Overall | Changes in activity levela | Changes in physical functiona | |||
---|---|---|---|---|---|
Less physical activity | Less time on feet | Worsened physical conditioning | Worsened mobility | ||
Total (n = 2006) | 100.0 | 36.9 | 35.1 | 26.9 | 20.9 |
Age | * | ||||
50–64 (n = 993) | 59.5 | 35.1 | 33.2 | 26.4 | 20.8 |
65–80 (n = 1013) | 40.5 | 39.5 | 38.0 | 27.7 | 21.1 |
Gender | ** | * | |||
Male (n = 943) | 47.4 | 32.8 | 32.2 | 26.0 | 19.2 |
Female (n = 1063) | 52.6 | 40.5 | 37.8 | 27.7 | 22.4 |
Race/Ethnicity | |||||
White, non-Hispanic (n = 1519) | 71.1 | 35.5 | 34.3 | 27.5 | 20.5 |
Black, non-Hispanic (n = 183) | 10.5 | 45.0 | 39.8 | 25.8 | 25.4 |
Hispanic (n = 181) | 11.6 | 35.3 | 32.7 | 22.9 | 18.3 |
Other, non-Hispanic (n = 123) | 6.8 | 41.8 | 40.9 | 29.8 | 22.6 |
Education | * | * | |||
High school or less (n = 661) | 40.1 | 35.2 | 32.9 | 23.6 | 20.9 |
Some college (n = 664) | 29.1 | 39.4 | 33.2 | 28.1 | 23.7 |
Bachelor's degree or higher (n = 681) | 30.8 | 36.7 | 39.8 | 30.1 | 18.3 |
Total annual household income | * | ||||
Less than $30,000 (n = 239) | 17.2 | 38.8 | 36.9 | 29.1 | 27.5 |
$30,000–$59,999 (n = 431) | 21.8 | 40.5 | 37.4 | 27.2 | 22.1 |
$60,000–$99,999 (n = 509) | 23.9 | 34.7 | 30.6 | 24.2 | 19.4 |
$100,000 or more (n = 827) | 37.2 | 35.2 | 35.9 | 27.5 | 18.2 |
Marital status | ** | * | ** | *** | |
Married or partnered (n = 1424) | 69.0 | 34.5 | 33.5 | 24.7 | 18.0 |
Not married or partnered (n = 582) | 31.0 | 42.1 | 38.8 | 31.9 | 27.4 |
Lives alone | |||||
Yes (n = 417) | 21.0 | 35.2 | 39.5 | 28.0 | 21.4 |
No (n = 1589) | 79.0 | 37.3 | 34.0 | 26.6 | 20.8 |
Physical health | ** | *** | *** | *** | |
Excellent/very good/good (n = 1716) | 84.5 | 35.1 | 33.1 | 22.8 | 16.1 |
Fair or Poor (n = 290) | 15.5 | 46.4 | 46.3 | 49.0 | 47.2 |
Felt isolated from others | *** | *** | *** | *** | |
Hardly ever (n = 1092) | 54.1 | 26.5 | 26.3 | 16.2 | 12.2 |
Some of the time/often (n = 914) | 45.9 | 49.1 | 45.5 | 39.5 | 31.1 |
- a Changes measured since the start of the pandemic in March 2020.
- Note: Significance based on Pearson's Chi-squared. * p < 0.05, **p < 0.01, ***p < 0.001. For columns 2 and 3, differences according to respondent characteristics are compared among those with reduced physical activity levels to those with no change or increased physical activity levels since the start of the pandemic. For instance, in column 3, characteristics such as age and gender are compared among those who since the start of the pandemic spent less time on their feet compared with those who had no change or spent more time on their feet. Similarly, for columns 4 and 5, differences in respondent characteristics are similarly compared among those with worsened physical function compared with those with no change or improvements in physical function since the start of the pandemic.
Changes in activity levels
A greater proportion of older (ages 65–80) compared with younger adults (ages 50–64) reported reductions in activity (39.5% vs. 35.1%) and daily time spent on their feet (38.0% vs. 33.2%) after the start of the pandemic (Table 1). Women also reported greater reductions than men for these respective activities (physical activity: 40.5% vs. 32.8%; time on feet: 37.8% vs. 32.2%). A greater proportion of non-Hispanic Blacks and non-Hispanic Other individuals, compared with non-Hispanic Whites, reported greater reductions in activity (45.0% and 41.8% vs. 35.5%) and daily time on feet (39.8% and 40.9% vs. 34.3%). Those with lower household income more often reported reductions in activity levels. Individuals who reported some of the time or often feeling socially isolated reported less activity (49.1%) and less time on their feet (45.5%), compared with 26.5% and 26.3% of those who hardly ever felt isolated.
Changes in physical function (physical conditioning and mobility)
A greater proportion of older (aged 65–80) compared with younger adults (aged 50–64) spent less daily time on their feet (38.0% vs. 33.2%), whereas similar proportions of older and younger adults reported worsened conditioning (27.7% vs. 26.4%) and worsened mobility (21.1% and 20.8%) (Table 1). Those with higher (Bachelor's degree or higher) compared with lower (High school or less) education levels more often reported worsened conditioning (30.1% vs. 23.6%) (Table 1). Individuals in lower-income households (<$30,000) compared with those in higher-income households (≥$100,000) also more often reported worsened conditioning (29.1% vs. 27.5%) and mobility (27.5% vs. 18.2%). More than twice as many individuals who were socially isolated, compared with those who hardly ever felt socially isolated, reported worsened conditioning (39.5% vs. 16.2%) and mobility (31.1% vs. 12.2%).
Those who became less, compared with more, active (55.2% vs. 9.2%, p < 0.001) and who spent less, compared with more, daily time on their feet (53.5% vs. 14.3%, p < 0.001) more often reported worsened conditioning (Table 2). Similar results were observed among adults aged 65–80 (Table S1).
Physical conditioning changea | Mobility changea | |||||||
---|---|---|---|---|---|---|---|---|
Improved | Worsened | No change | p | Improved | Worsened | No change | p | |
Total (n = 2006) | 11.2 | 27.1 | 67.1 | 9.5 | 21.0 | 69.5 | ||
Physical activityb | *** | *** | ||||||
More active (n = 264) | 52.7 | 9.2 | 38.2 | 43.3 | 7.3 | 49.4 | ||
Less active (n = 753) | 3.4 | 55.2 | 41.4 | 3.0 | 41.4 | 55.6 | ||
No change (n = 1003) | 6.3 | 10.9 | 82.9 | 5.6 | 9.3 | 85.1 | ||
Time spent on feetc | *** | *** | ||||||
More time (n = 241) | 46.7 | 14.3 | 39.0 | 39.4 | 12.6 | 48.0 | ||
Less time (n = 715) | 5.4 | 53.5 | 41.2 | 4.7 | 40.8 | 54.5 | ||
No change (n = 1062) | 7.1 | 12.2 | 80.7 | 6.1 | 9.4 | 84.5 | ||
Physical activity or time on feet decreased | *** | *** | ||||||
Yes (n = 938) | 5.5 | 49.8 | 44.7 | 4.8 | 37.2 | 58.0 | ||
No (n = 1082) | 16.1 | 7.5 | 76.4 | 13.6 | 6.9 | 79.5 |
- a Changes measured since the start of the pandemic in March 2020.
- b Moderate or vigorous physical activity.
- c Time spent on feet standing or walking.
- Note: Significance based on Pearson's Chi-squared. * p < 0.05, **p < 0.01, ***p < 0.001.
Falls and fear of falling
Roughly one-quarter of respondents reported one or more falls since March 2020 (Table 3). Those who spent less compared with more daily time on their feet more often reported one or more falls (27.9% vs. 25.8%, p = 0.02). Overall, 22.9% of respondents became more fearful, and 4.7% became less fearful, of falling since the start of the pandemic. Increases in fear of falling were more common for those who became less active, compared with more active, after the pandemic began (32% vs. 22%, p < 0.001); for those who spent less, compared with more, time on their feet (32% vs. 25%, p < 0.001); and for those with worsened, compared with improved, conditioning (42% vs. 16%, p < 0.001) or mobility (45% vs. 21%, p < 0.001). Increased fear of falling was also greater among those who felt a lack of companionship (30% vs. 16%, p < 0.001) or isolation from others (28% vs. 15%, p < 0.001). Similar results were observed among adults aged 65–80, with slightly higher percentages reporting one or more falls (28.3%) (Table S2).
Fall history | Fear of falling change | ||||||||
---|---|---|---|---|---|---|---|---|---|
n | 0 times | ≥1 time | p | n | More fearful | Less fearful | No change | p | |
Total | 2006 | 75.2 | 24.8 | 741 | 22.9 | 4.7 | 72.4 | ||
Physical activitya (n, %) | *** | ||||||||
More active | 263 | 74.6 | 25.4 | * | 92 | 22.3 | 15.4 | 62.3 | |
Less active | 746 | 72.0 | 28.0 | 334 | 31.5 | 4.2 | 64.2 | ||
No change | 997 | 77.6 | 22.4 | 315 | 14.0 | 2.2 | 83.8 | ||
Time spent on feetb (n, %) | * | *** | |||||||
More time | 240 | 70.9 | 18.8 | 92 | 24.9 | 14.0 | 61.1 | ||
Less time | 709 | 72.0 | 15.5 | 318 | 31.6 | 4.6 | 63.8 | ||
No change | 1057 | 78.2 | 13.5 | 330 | 13.9 | 2.2 | 84.0 | ||
Physical conditioning (n, %) | *** | *** | |||||||
Improved | 221 | 74.5 | 25.5 | 65 | 15.8 | 23.6 | 60.6 | ||
Worsened | 552 | 63.1 | 36.9 | 279 | 41.7 | 3.2 | 55.2 | ||
No change | 1233 | 80.5 | 19.5 | 396 | 11.1 | 2.8 | 86.0 | ||
Mobility (n, %) | *** | *** | |||||||
Improved | 184 | 73.9 | 26.1 | 56 | 21.3 | 23.0 | 55.7 | ||
Worsened | 416 | 57.4 | 42.6 | 240 | 45.1 | 4.2 | 50.7 | ||
No change | 1406 | 80.7 | 19.3 | 443 | 10.8 | 2.8 | 86.5 |
- a Moderate or vigorous physical activity.
- b Time spent on feet standing or walking.
- Note: Changes measured since the start of the pandemic in March 2020. Significance based on Pearson's Chi-squared. * p < 0.05, **p < 0.01, ***p < 0.001. Sample sizes for fall history and fear of falling change differ because only those 741 individuals who indicated any fear of falling were asked if their fear of falling changed since March 2020, the start of the pandemic.
Adjusted findings: changes in function
The risk of worsened conditioning was higher for individuals who reduced, compared with did not reduce, regular activity (adjusted risk ratio, ARR = 2.92, 95% CI: 2.38, 3.61); for those who spent less, compared with no change in, time on their feet (ARR = 1.95, 95% CI: 1.62, 2.34); and for socially isolated, compared with other, individuals (ARR = 1.51, 95% CI: 1.30, 1.74) (Figure 2A). For the worsened mobility outcome, increased risks were observed for each of reductions in regular activity (ARR = 2.49, 95% CI: 1.96, 3.16), less time on feet (ARR = 2.00, 95% CI: 1.60, 2.50), and social isolation (ARR = 1.53, 95% CI: 1.28, 1.84).
Adjusted findings: changes in fall-related outcomes
The risk of having one or more falls was greater (ARR = 1.70, 95% CI: 1.35, 2.15) for individuals with worsened, compared with not worsened, mobility and those who were socially isolated, compared with those hardly ever socially isolated (ARR = 1.23, 95% CI: 1.05, 1.45) (Figure 2B). Worsened conditioning was not associated with changes in the risk of a fall (ARR = 1.13, 95% CI: 0.90, 1.43).
Among individuals who reported fear of falling, each of worsened conditioning (ARR = 1.94, 95% CI: 1.26, 3.01) and worsened mobility (ARR = 2.02, 95% CI: 1.30, 3.13), but not social isolation (ARR = 1.29, 95% CI: 0.96, 1.74), were associated with increased fear of falling. For adults aged 65–80, the results were generally similar when compared with those for all participants, with several exceptions (Table S3). For adults aged 65–80, worsened mobility (ARR = 3.78, 95% CI: 1.89, 7.55) but neither of worsened conditioning (ARR = 1.28, 95% CI: 0.67, 2.44) and social isolation (ARR = 1.17, 95% CI: 0.84, 1.64) was associated with increased fear of falling (Table S3).
DISCUSSION
Since the start of the COVID-19 pandemic in the United States, one-third of older adults reported reduced activity levels. Declines in activity were associated with two to three-fold worsening of several self-reported measures of physical function. In turn, poorer function was associated with increases in the risk of falls and fear of falling. Social isolation was also associated with an increased risk of poorer function and worse fall-related outcomes. Together, the results suggest the pandemic was associated with changes in health and social activities among older adults, and that these changes appear to have led to physical deconditioning and increased risk of fall-related injuries.
Beyond infection rates, hospitalizations, and mortality, recent literature on the COVID-19 pandemic has observed excess mortality that likely reflects clinically under-addressed or untreated acute and long-term health issues.17 This study illustrates another concern, that of pronounced physical functioning decline among older adults, potentially associated with sustained periods of reduced physical and social activity during the pandemic. Beyond previously measured excess death and morbidity, this may presage worrisome population health disability and injury cascades in which deconditioning, functional loss, and injury created more disablement.
Observed changes could reflect common disability trends rather than effects from the pandemic. However, prior work observed that from 2019 to 2020 the proportions of adults aged 70 and older with mobility difficulties decreased only slightly, from 19.3% to 18.5%.18 Therefore, the findings are likely to reflect pandemic-related declines.
Risks for falls increase both with a history of previous falls and in the presence of fear of falling, as each can produce avoidant strategies such as using excessive caution getting around or avoidance of activity that can harm conditioning and mobility.19-21 These behaviors could increase short-term fall risk; they could also reduce risk by limiting opportunities for falls. In either case, fall risk may increase once regular activities resume, with important implications for older adults' health and health care costs.22 To the extent reduced activity levels mask shorter-term fall risk, and because falls are predictors for future disability, our study and others may understate the pandemic's longer-term impacts on older adults' functioning and safety.
Sedentary behavior and social isolation are strong predictors of morbidity and mortality, and short-term deconditioning due to these factors can lead to functional decline and injury risk.10, 11, 23, 24 The current findings suggest critical, if distinct, roles for these risk factors and confirm that short-term exposures to these risks (in the approximately year-long pandemic period we assessed) can be associated with sizeable impacts on conditioning and mobility. Reversing these deficits will, therefore, require interventions on a broad range of individual and environmental factors that require support from public health agencies, clinicians, and policy makers.6
Limitations
The study was subject to several potential limitations. First, it used a cross-sectional design. Although we could assess the retrospective evolution of specific factors, such as physical activity levels or mobility, when estimating relationships among factors, our study may be vulnerable to reverse causality bias. For instance, although reduced mobility was estimated to increase fall risk by 70%, it is plausible that individuals who reported having one or more falls subsequently experienced limited mobility. Still, the absolute levels of change toward poorer mobility and fall-related outcomes are reason for concern. Second, the results may reflect recall bias involving reports of activity levels, function, and falls. Survey respondents underreport falls, for instance.25 Third, measurement bias could be present if respondents at differing levels of social isolation or activity differentially reported key outcomes. However, reporting bias is diminished with shorter recall intervals anchored by a notable event, such as with the historic pandemic.26 Also, a roughly 1-year recall interval was used in contrast to other studies that used longer recall windows.27 Fourth, internet-based surveys are subject to bias if participants vary from nonrespondents in characteristics unaccounted for in model risk adjustment. However, survey participants without access to the internet were provided free internet service and a web-enabled device and were recruited using a probability-based methodology designed to be nationally representative. Therefore, findings should be broadly generalizable to the U.S. older adult population. Fifth, the results are not generalizable to the cognitively impaired older adult population, as individuals unable to complete the survey on their own or with the help of a caregiver due to cognitive impairment were unlikely to participate in the survey panel. Finally, the survey did not include individuals over the age of 80. Given greater disability and fall risks in this subpopulation, results may understate the proportion of U.S. older adults experiencing activity declines, deconditioning, and falls. An examination of such pandemic-related risks in this “oldest old” group is an important area for future research.
These limitations notwithstanding, these findings have important clinical and policy implications. First, reversing observed trends in older adults' health and social behaviors will require extraordinary efforts beyond routine clinical care. For example, clinicians may lack awareness, training, or adequate resources to address such behaviors. Gerontological models on disablement and injury prevention advise harnessing multisectoral resources in the community and healthcare organizations, and from policy makers.8, 28 Clinical assessments of function and fall risk are needed, but so are outreach and interventions from senior centers, public health nurses, and community-based care providers, including home health aides, physical and occupational therapists, social workers (for assessing home safety), and long-term care sites.
Policy makers should use the opportunity to address long-standing disincentives to address physical function and fall-related risks. One proposal entails a targeted Medicare falls benefit, including a detailed assessment by a physician to evaluate fall risk and tailored multifactorial fall prevention recommendations, an exercise program referral if requested by the physician, plus reimbursement for an additional follow-up visit.29
This proposal would expand care options beyond a “medically necessary services” policy that prioritizes care after, as opposed to before, an injury. For instance, Medicare will reimburse surgery and rehabilitation for a broken hip, but not routine geriatric assessments or timely replacement of durable medical equipment, such as a grab bar or bathing chair.30 Similarly, Medicare contractors are likely to reject payment for an occupational therapy evaluation if the therapist makes recommendations regarding non-covered equipment or for an assessment by a physiotherapist of balance issues not covered by a diagnosis code.30 Such determinations may discourage therapists from evaluating non-covered equipment and symptoms, and undermine the benefits of occupational and physical therapy for functioning and falls.31-33
A targeted benefit could also improve care integration across settings. Medicare reimburses hospital, rehabilitation, skilled nursing, and certain home health services through Part A, but reimburses outpatient physician and rehabilitation and some health services under Part B. Each Medicare Part has different incentives for providers (prospective versus fee-for-service payment) and beneficiary cost-sharing arrangements that pose challenges for care coordination. Moreover, to manage payments for these services, Medicare uses separate fiscal intermediaries, which have discretion in making local coverage determinations. Such complexity for providers and patients can hamstring prevention efforts to access needed interdisciplinary care (e.g., occupational therapists for home modifications,31-33 physiotherapists and physiologists for exercise programs and tests of functional mobility,34, 35 geriatricians and nurse practitioners for outpatient medical assessments,36 and pharmacists for medication regimen assessments37, 38).
Integrating Medicare services through dedicated payment streams and addressing regulatory barriers to service provision could follow prior work done in the Program for All-Inclusive Care for the Elderly (PACE) and Money Follows the Person models,39, 40 which combined funding sources to improve care coordination for Medicaid beneficiaries. Such program reforms have been long overdue and are particularly critical following a pandemic when millions of older adults' physical functioning and safety may have been compromised.
Private insurers might also consider coverage for falls prevention for adults less than 65 years old. Given heightened risks among the sample, including individuals aged 50–64, preventive efforts targeting occupational and physical pre-habilitation, durable medical equipment, and home safety modifications might prove cost-beneficial for insurers and beneficial to enrollees.
Absent such efforts, the scope of long-term care needs may largely fall on underfunded public programs and public health agencies as well as unpaid family caregivers, while policy efforts are instead directed to “headline” measures captured in the news media or health systems. Although mortality, excess deaths, and hospitalizations are critical measures of the pandemic's effects, longer-term effects on older adults' functioning and injury, including risks for disability and loss of independence, are also critical to the nation's health.
Conclusion
In a national survey, substantial decreases in activity levels, physical conditioning, and mobility from March 2020 to January 2021 were observed among U.S. older adults aged 50–80 years. An increase in population-wide fall risk was observed along with worsened physical functioning brought on by reduced physical activity levels. In all, the findings could imply rapid deconditioning associated with restricted physical and social activities during the pandemic and should be more closely examined and addressed through innovative clinical and policy changes that can help integrate interdisciplinary preventive care for at-risk older adults. Community interventions that facilitate physical activity, including safe areas for walking in parks and neighborhoods, and efforts to manage safety risks at home, particularly in lower-income areas, can help address threats to mobility among the most vulnerable older adults.
ACKNOWLEDGMENTS
I have listed everyone who contributed significantly to the work.
CONFLICT OF INTEREST
The National Poll on Healthy Aging is sponsored by AARP and Michigan Medicine, the academic medical center for the University of Michigan. Dr. Kullgren received support from the Department of Veterans Affairs, Veterans Health Administration, Health Services Research and Development Service. The other authors have no conflicts to report.
AUTHOR CONTRIBUTIONS
Study concept and design: Geoffrey J. Hoffman, Erica Solway, Dianne C. Singer, Jeffrey T. Kullgren, Preeti N. Malani. Acquisition of subjects and/or data: Dianne C. Singer, Matthias Kirch. Analysis and interpretation of data: Geoffrey J. Hoffman, Matthias Kirch.
SPONSOR'S ROLE
The funders had no role in the study design, data collection, management, and analysis, nor any participation in the preparation, review, and approval of the manuscript. The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs or the United States government.