Evaluation of Light Physical Activity Measured by Accelerometry and Mobility Disability During a 6-Year Follow-up in Older Women

Key Points Question Is there an association between light-intensity physical activity and incident mobility disability among older women? Findings In this cohort study of 5735 postmenopausal, community-dwelling women without mobility limitation, the risk of incident mobility disability over 6 years of follow-up was 22%, 40%, and 40% lower for women in the second, third, and fourth quartiles of daily mean light-intensity physical activity, respectively, compared with the lowest quartile. Meaning These results suggest that recommendations to increase light-intensity physical activity have the potential to improve prospects for preserving mobility among older women.


Introduction
Adults older than 65 years in the US are projected to reach 78 million people by 2035, outnumbering children under 18 years for the first time in US history. 1 The aging of the US population is, in part, driven by increasing life expectancy, with women of all races/ethnicities holding a longevity advantage of almost 5 years over men. 2,3 Even after adjusting for women's longevity advantage, women have higher rates of morbidities, disabilities, and health care costs than men. [4][5][6] Mobility disability, defined by the US Centers for Disease Control and Prevention as serious difficulty walking or climbing stairs, is the leading form of disability in the US and a fundamental quality-of-life issue affecting 27% of adults older than 65 years. 3,5,7 Higher rates of mobility disability are observed among older women, with 24% unable to walk 2 to 3 blocks compared with 14% of older men. 3 Older adults with mobility disability on average experience more hospitalizations and spend more than $2700 more in annual health care costs compared with those without mobility disability. 8 Nearly all the evidence to date regarding physical activity (PA) and mobility, including that in the 2018 PA guidelines, has focused on higher-intensity PA. 9 The Lifestyle Interventions and Independence for Elders (LIFE) study, a randomized clinical trial among older adults with existing physical limitations, showed that a program of structured moderate-intensity PA increased lower extremity physical functioning and reduced incident mobility disability, 10 providing evidence that higher-intensity PA can prevent mobility disability. Whether light-intensity physical activity (LPA), such as casual walking or engaging in most daily life activities, reduces the risk of mobility disability is unknown and is among the major evidence gaps in the fields of aging and PA. 9 Recent studies have found an association between accelerometer-measured LPA and decreased mortality and incident cardiovascular disease. 11,12 Studying LPA is particularly important for older adults, because moderate to vigorous-intensity PA (MVPA) is increasingly difficult to perform, and engagement in MVPA declines with increasing age. 13 A main reason for the lack of evidence related to LPA is because it is poorly measured by self-reporting, with a correlation to accelerometer-measured LPA of between 0.04 and 0.16. 14 To our knowledge, this is the first study of accelerometer-measured LPA and incident mobility disability among community-dwelling older adults with intact mobility. The objective of this study was to investigate the association of accelerometer-measured LPA and incident mobility disability over 6 years of follow-up in a large diverse cohort of older women.

Study Participants
Between 1993 and 1998 the Women's Health Initiative (WHI) enrolled postmenopausal women, aged 50 to 79 years, from 40 clinical sites across the US. 15,16 The present cohort study includes women who participated in the Objectively Measured Physical Activity and Cardiovascular Health (OPACH) study, a longitudinal cohort and ancillary study of the WHI. In the OPACH study, consent was obtained from 7058 ambulatory community-dwelling women aged 63 years or older from March Our primary analytic sample of all incident mobility disability included 5735 women from the original OPACH cohort (eFigure 1 in the Supplement). Participants were excluded for death before receipt of an accelerometer (n = 10), not returning an accelerometer (n = 327) or returning the accelerometer with no usable data (n = 232), no days with 10 or more waking hours of wear time (n = 107), mobility disability at baseline (n = 602), or lack of follow-up for mobility status (n = 45). In the secondary analyses examining only persistent mobility disability, 418 women with incident mobility disability and reported recovery were excluded to provide a population at risk throughout follow-up (n = 5317). The protocol for the OPACH study was approved by the Fred Hutchinson Cancer Research Center institutional review board, and all women provided informed consent either in writing or orally by telephone. This report followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Mobility Disability
The definition of mobility was consistent with previously published studies in older adults as selfreported ability to walk 1 block and up a flight of stairs during a typical day. 18 Mobility status was collected annually via mail or telephone from the OPACH study baseline through March 31, 2018.
Incident mobility disability was defined as the first report of inability to walk 1 block or up a flight of stairs. Persistent mobility disability, defined as incident mobility loss persisting through the end of follow-up, was a secondary outcome, representing a more severe form of mobility loss with a stronger association with mortality. 19

PA Measures
Accelerometer data, measured between March 2012 and April 2014, using the ActiGraph GT3X+, were originally collected at 30 Hz and then integrated to 15-second epochs using the normalfrequency filter within ActiLife version 6 software (ActiGraph Corp). Accelerometer nonwear periods were identified and removed using the Choi algorithm. 17,20 Sleep time was removed using reported in-bed and out-of-bed times from sleep logs. Missing bedtimes were imputed using participantspecific mean times or, if all data were missing, the OPACH population mean.
LPA includes all movement requiring energy expenditure between 1.6 and 2.9 metabolic equivalents. Examples of LPA include washing and drying dishes, gardening, and walking at a pace of about 1.5 mph, such as while shopping. Accelerometers were calibrated specifically for older women in a separate laboratory study of 200 women from the same OPACH population. 21 LPA was computed as the mean number of hours per day with sufficient movement that the vector magnitude counts per 15-second epoch were between 19 and 518. 21 MVPA, defined as activity requiring 3.0 or more metabolic equivalents, was measured as the mean number of minutes per day with vector magnitude counts per 15-second epoch of at least 519. As recommended, PA measures were computed using data from days with 10 or more hours of awake wear time. 22  (excellent/very good, good, or fair/poor), and short physical performance battery (SPPB) score (0-9 [low function] or 10-12 [high function]). 23

Statistical Analysis
To address differences in sleep time or nonwear time, PA measures were adjusted for awake wear time using the residuals method. 24 Descriptive statistics were computed by quartile of LPA.
Differences in characteristics across LPA categories were tested for statistical significance using analysis of variance for continuous variables and Pearson χ 2 for categorical variables.
Hazard ratios (HRs) and 95% CIs for incident mobility disability (all and persistent) were determined using Cox proportional hazards regression models by quartiles of LPA in successively adjusted models. Person-years for each participant were calculated from OPACH baseline (first day of accelerometer wear) until first reported incident mobility disability, end of follow-up, or death, whichever occurred first. The proportional hazards assumption was assessed based on graphical inspection of survival curves. The minimally adjusted model (model 1) adjusted for age, race/ ethnicity, and education. A confounder-adjusted model (model 2) further adjusted for health factors and behaviors associated with LPA and mobility disability (smoking, alcohol use, multimorbidities, and self-rated health). Model 3 adjusted for all model 2 covariates and MVPA.
To evaluate the consistency of associations among older women at higher vs lower risk of incident mobility loss, we conducted stratified analyses using the confounder-adjusted Cox regression model (model 2) by younger vs older age (median, 79 years; range, 63-97 years), race/ ethnicity (White, Black, and Hispanic/Latina), high vs low MVPA (median, 45 min/d; range, 0-351 min/day), BMI (<30.0 vs Ն30.0), and high vs low SPPB score (0-9 vs Ն10). Interaction terms were added to the multivariable-adjusted Cox regression model to test for potential effect modification at α = .10, with age, MVPA, BMI, and SPPB score included as continuous terms.
The dose-response association of continuous LPA with incident mobility disability was then visualized using plots of the confounder-adjusted Cox regression model (model 2) and MVPA model (model 3) with restricted cubic-spline functions with 3 knots. Linearity of the dose-response association was tested using Wald tests. For reporting, HRs were calculated at 4, 5, 6, and 7 mean hours of LPA per day relative to the tenth percentile of LPA (3.3 hours per day).
All CIs were calculated at 95% and with an α of .05 unless otherwise specified. All analyses were conducted from August 2018 to May 2019, using RStudio version 1.1.463 for Mac (R Foundation) with the survival (survival analysis) and rms (regression modeling strategies) packages. All P values were 2-sided, and P < .05 was considered significant.

Results
A total of 5735 participants were included for primary analysis of all incident mobility disability (mean    (Figure 1  To address possible residual confounding by poor health status, sensitivity analyses re-examined models for LPA and incident mobility disability after excluding women from the analytical sample who died within 1 year of follow-up, who reported fair or poor self-rated health, or who had 2 or more major chronic conditions. In no case were results meaningfully changed (eTable 1, eTable 2, and eTable 3 in the Supplement).

Discussion
To our knowledge, this is the first study to examine the association of objectively measured LPA with incident mobility disability in a general population of older adults with intact mobility. Lower risk of incident mobility disability associated with increasing LPA was observed starting in the second lowest quartile of LPA. Women in the highest quartile of LPA had a 40% lower risk of incident mobility disability and a 48% lower risk of persistent mobility disability compared with women in the lowest quartile of LPA after adjustment for demographic and confounding factors. Women with a BMI less than 30.0 had a stronger beneficial association than those with a BMI greater than or equal to 30.0 kg/m 2 . Results suggest that high levels of LPA are not necessary to reduce the risk of mobility  disability, because no further reductions were observed over approximately 5 hours of LPA per day.
Overall, the interpretation of LPA and incident mobility disability was robust to consideration of MVPA as a possible confounder or effect modifier. The association of LPA with health outcomes was identified as a gap in current knowledge by the 2018 Physical Activity Guidelines Advisory Committee. 9 To our knowledge, few studies have examined LPA and mobility among older adults. 25 The LIFE Study reported a 15% lower risk of incident mobility disability associated with the addition of an increment of 30 minutes per day in LPA among highly sedentary adults aged 70 to 89 years with existing physical limitations. 26 A study among community-dwelling adults aged 49 years or older with or at risk for knee osteoarthritis found a 42% lower risk of incident disability for the highest vs lowest quartile of objectively measured LPA, similar to the magnitude in our study. 27 Although these studies are comparable in their use of accelerometer-measured LPA and focus on disability and health among older adults, our study is distinct in several ways. We focused on older women, who bear a disproportionate burden of mobility disability. Our examination of incident mobility disability among women with intact mobility at baseline supports LPA as potentially beneficial to all older women, not just those with existing physical limitations. Furthermore, because of our focus on mobility disability rather than poor health or a broader definition of physical function, our findings provide strong evidence for the translation of LPA into medical practice as a method for preserving mobility.
The beneficial association of LPA was stronger for women with lower BMI. Obesity in older women is strongly associated with poorer lower body function and walking ability. 28 Within the WHI, obesity was associated with very high odds of incident mobility disability among women who survived to 85 years. 29 Moreover, in the National Institutes of Health-AARP Diet and Health Study cohort and the Health, Aging and Body Composition study, beneficial associations between physical activity and mobility disability were weaker among adults with higher BMI levels. 30,31 Taken together, these studies suggest that the potential favorable influence of PA on mobility disability may not overcome the increased risk related to higher BMI. However, older adults with excess weight can still potentially benefit from increased PA. In the Look AHEAD (Action for Health in Diabetes) trial, a nutritional and PA intervention among adults with excess weight and type 2 diabetes, individuals in the intervention arm experienced significant weight loss, improved physical fitness, and improved physical function. 32 The physiological mechanisms that lead to mobility disability are not fully understood. In a systematic review and meta-analysis of activities of daily living among older adults, activities of daily living dependency was strongly associated with sarcopenia, low muscle strength, and low physical performance. 33 Previous studies have found increased PA to be a protective factor against the development of sarcopenia among older adults. 34 Additionally, trials among frail older adults have produced generally positive results of PA interventions with improvements in muscle strength, balance, and mobility, although the strength of these associations is inconsistent. 35 Light-intensity PA is characterized by lower-required energy expenditure compared with MVPA but still requires engagement of muscle groups. Although few studies have been conducted on the physiological mechanisms associated with mobility disability and specifically LPA, positive associations have been found between LPA and muscle strength in the leg 36 and bone mineral density, 37 suggesting that LPA can provide similar physiological benefits as other forms of PA.

Strengths and Limitations
The OPACH study included a large cohort of postmenopausal women from multiple sites, races/ ethnicities, and socioeconomic backgrounds. This diversity increases the generalizability of our findings. This study also had minimal loss to follow-up, and the use of accelerometers allowed us to objectively measure LPA. We were able to capture PA through a reliable measure, and the cut points for intensity were based on a calibration study among this specific population. 21 Although this study

JAMA Network Open | Geriatrics
Light Physical Activity and Mobility Disability During 6-Year Follow-up in Older Women focused on older women, studies among men have found beneficial associations between LPA and all-cause mortality, suggesting the potential health benefits of LPA are likely not limited to women. 38 This study also has some limitations. Mobility disability status was based on self-report, a method generally considered less reliable compared with objective measures. However, selfreported abilities to walk and climb stairs provide valuable information on the participants' evaluation of their own physical limitations in usual life circumstances. Furthermore, our definition and collection method are comparable to those of the Behavioral Risk Factor Surveillance System, a telephone-based survey used for US statistics. 5 Annual ascertainment of mobility is a potential limitation as shorter duration incident mobility disability with recovery in less than 1 year may not have been captured. Longer-term and persistent mobility disability is of greater interest given the effects on independence and mortality. Light-intensity PA was measured over a single 7-day baseline wear period, and we were unable to examine repeated measures. Nonetheless, the baseline measure most removed from the outcome reduces the possibility of reverse causation.