Abstract
Introduction:
Low physical activity is a criterion of phenotypic frailty defined as an increased state of vulnerability to adverse health outcomes. Whether disengagement from daily all-purpose physical activity is prospectively associated with frailty, and possibly modified by chronic inflammation—a pathway often underlying frailty—remains unexplored.
Methods:
Using STURDY (Study to Understand Fall Reduction and Vitamin D in You) data from 477 robust/prefrail adults (mean age=76±5 years; 42% women), we examined whether accelerometer patterns (activity counts/day, active minutes/day, and activity fragmentation [broken accumulation]) were associated with incident frailty using Cox proportional-hazard regression. Baseline interactions between each accelerometer metric and markers of inflammation that include interleukin-6 (IL-6), C-reactive protein (CRP), and tumor necrosis factor-alpha receptor 1 (TNF-aR1) were also examined.
Results:
Over an average of 1.3 years, 42 (9%) participants developed frailty. In Cox regression models adjusted for demographics, medical conditions, and device wear days, every 30 min/day higher baseline active time, 100,000 more activity counts/day, and 1% lower activity fragmentation was associated with a 16% (p=0.003), 13% (p=0.001), and 8% (p<0.001) lower risk of frailty, respectively. No interactions between accelerometer metrics and either baseline IL-6, CRP, TNF-aR1 were detected (interaction p>0.06 for all).
Conclusions:
Among older adults who are either robust or prefrail, constricted patterns of daily physical activity (i.e., lower total activity minutes and counts, and higher activity fragmentation) were prospectively associated with higher risk of frailty, but not modified by frailty-related chronic inflammation. Additional studies, particularly trials, are needed to understand if this association is causal.
Keywords: ACCELEROMETRY, IMMUNE FUNCTION, AGING, DEBILITY, OLDER ADULTS
INTRODUCTION
Growing evidence suggests physical activity is an effective intervention to combat frailty(1)—a medical syndrome defining an increased risk to stressors that adversely affect health(2). Physical activity upregulates the immune, circulatory, endocrine, musculoskeletal, and nervous systems—systems thought to dynamically degrade together and manifest as frailty(2).
One of the most common measures of frailty, the physical frailty phenotype, is partly defined by low amounts of self-reported physical activity as a criterion(3). This self-reported measure captures participation in exercise-like activities (e.g., moderate intensity activities, walking for exercise)(4). Yet in older adults, most physical activity performed throughout the day is spent in light intensity physical activity or sedentary inactivity(5). Surveillance of physical activity performed throughout the day, not just time spent in exercise, makes it possible to detect subtle alterations in everyday movement indicative of frailty risk(6).
Current evidence linking physical activity to frailty risk focuses on the total amount of physical activity(7-9). However, total physical activity misses information on physical activity fragmentation, i.e., whether activity bouts are sustained (longer bouts) or more fragmented (shorter bouts) throughout the day(10). Both total physical activity and activity fragmentation have been associated with physical functioning(11), morbidity(12-14), mental health(15,16), and mortality(17), and cross-sectionally associated with frailty(18). Yet importantly there is a lack of longitudinal evidence on the association of daily physical activity accumulation patterns with frailty incidence(9).
Immune dysregulation is one of the most recognized pathways underlying frailty(2). Specifically, high levels of interleukin-6 (IL-6), C-reactive protein (CRP) and tumor necrosis factor-alpha (TNF-a) are commonly observed among individuals experiencing “inflammaging”—a state associated with a higher risk of disability, morbidity, and mortality(19). These inflammatory markers are inversely associated with physical activity(20,21), primarily within the scope of exercise (volitional physical activity intended for health benefits)(22). Whether these markers of “inflammaging” modify the relationship between patterns of everyday physical activity and frailty incidence remains under explored.
The first aim of this study was to examine how objectively measured patterns of daily physical activity relate to frailty incidence. We hypothesized that constricted patterns of daily physical activity (i.e., lower total activity and higher activity fragmentation) are associated with higher risk of frailty among older adults who are either robust or prefrail. The second, more exploratory, aim of this study was to examine whether markers of chronic inflammation play a moderating role between physical activity and frailty incidence. Though this aim was exploratory, we hypothesized that any association of physical activity constriction with increased risk of frailty onset would be amplified by higher levels of baseline inflammation.
METHODS
Study Design and Population
Baseline accelerometer and repeated frailty data were collected in the STURDY (Study to Understand Fall Reduction and Vitamin D in You) trial, conducted from October 2015 to May 2019. STURDY was a 2-stage Bayesian, response-adaptive, randomized trial that documented that high-dose vitamin D supplementation did not prevent falls in older adults with elevated fall risk(23). Eligibility criteria included being ≥70 years old, a serum 25-hydroxyvitamin D level of 10-29 ng/mL, and self-reporting at least one fall in the past year or self-reporting a fear of falling. A total of 688 participants were enrolled at 2 community-based research units (Hagerstown or Woodlawn, Maryland), each at approximately 39° latitude. Of the 688 participants, 645 (94%) had sufficient accelerometry data (≥3 valid days) for inclusion in the analysis. One participant had missing frailty measured at baseline and 71 participants with frailty at baseline were excluded. Those with missing baseline IL-6 (n=20), CRP (n=10), and tumor necrosis factor-alpha receptor 1 (TNF-aR1) (n=3) and one participant with an extremely high IL-6 level (45 pg/mL) were excluded. After excluding participants with no longitudinal frailty data (n=62), the final analytic sample was 477 participants. A Johns Hopkins University institutional review board approved the trial protocol, and each participant provided written informed consent.
Accelerometer Variables
Each participant was instructed to wear a wrist-worn Actigraph GT9X monitor (Actigraph, Pensacola, FL) for 7 consecutive days with a 24-hour wear protocol except during swimming or bathing activities lasting >30 minutes. The monitor was set at a sampling rate of 80 Hz and positioned on the non-dominant wrist. After the 7-day collection period, each participant was instructed to mail the Actigraph monitor back to the clinical center via a pre-addressed, pre-paid envelope.
When a monitor was returned, data were downloaded and processed into activity counts (unitless quantities of movement) over 1-minute epochs using the low-frequency extension filter through the ActiLife software (version 6.13.4). Non-wear time was identified and removed using a 90-minute threshold of consecutively occurring zero values, based on Choi and colleagues’ algorithm(24). A valid accelerometer wear day was defined as having <10% missing data(15). Periods between 11:00pm-5:00am, representing sleep, were removed from valid wear time data.
For each participant, five accelerometer variables were derived, using conventional thresholds to define categories(18):
Total activity counts/day was calculated as the mean of total activity counts on valid wear days
Active minutes/day was calculated as the mean of total active minutes (threshold of ≥1,853 counts/minute(25)) on valid wear days
Sedentary minutes/day was calculated as the mean of total sedentary minutes (threshold of <1,853 counts/minute) on valid wear days
Activity fragmentation was calculated as the reciprocal of the average active bout length (i.e., length of consecutively occurring active minutes) on valid wear days
Sedentary fragmentation was calculated as the reciprocal of the average sedentary bout length (i.e., length of consecutively occurring sedentary minutes) on valid wear days
For the primary analyses, each of the five accelerometer variables was treated continuously. Potential for nonlinear relationship with frailty onset risk was assessed using lowess smooth plots; relationships appeared adequately linear, and consequently relationships were so modeled in multivariable analyses. In the sensitivity analyses, each variable was treated categorically by separating each metric into “low”, “medium”, and “high” tertiles.
Inflammatory Markers
Non-fasting blood (serum) was drawn at baseline. Aliquots were stored, using customary secure procedures, at −70° C at both field centers. For this analysis, the following soluble protein biomarkers were measured: IL-6, CRP, and TNF-aR1. IL-6 was measured using a solid-phase immunometric assay on the IMMULITE 2000 (Siemens Healthineers). Assay detection was conducted via an enzyme-labeled, sequential assay in which a substrate was converted to the chemiluminescent product. The total imprecision of this method was acceptable as indicated by a 4-6% between-run coefficient of variation. High-sensitivity CRP (hsCRP) was measured using a Vista 1500 (Siemens Healthineers) by immunonephelometry in which polystyrene particles coated with monoclonal antibodies against human CRP aggregate with this protein in samples. The between-run imprecision (coefficient of variation) was typically 4 to 6%. TNF-aR1 was measured using a quantitative sandwich enzyme immunoassay (R&D Systems, Inc) where assay detection was colorimetric. Total imprecision of the runs was acceptable as indicated by a 3-4% between-run coefficient of variation.
Physical Frailty Outcome
Frailty was assessed at four time-points: baseline and 3-, 12-, and 24-month follow-up visits. Using the physical frailty phenotype algorithm, a robust, prefrail, or frail status was defined as having 0, 1-2, or ≥3 frailty components, respectively(3). The five components were defined as follows:
Weight loss was defined as body mass index (BMI) <18.5 kg/m or >5% body weight unintentionally lost in the past year (baseline assessment by interview; follow-up assessments by weight measurement and interview).
Self-reported fatigue or exhaustion was defined as unusually tired or unusually weak all or most of the time or energy level rating ≤3 (of 10).
Slowness was defined as ≤0.65 m/s if height ≤173 cm or ≤0.76 m/s if height >173 cm for males and ≤0.65 m/s if height ≤159 cm or ≤0.76 m/s if height >159 cm for females. The faster of the two self-paced 4m walking tests during the Short Physical Performance Battery (SPPB) was used for measuring walking speed(26).
Low physical activity was defined as <128 kcal for males or <90 kcal for females of physical activity energy expenditure per week, measured from questions about walking for exercise, doing moderately strenuous household chores, doing moderately strenuous outdoor chores, dancing, bowling, and/or participating in a regular exercise program in the past 2 weeks(4).
Weakness was defined as grip strength ≤29 kg if BMI≤24 kg/m2, ≤30kg if BMI is 24.1-26 kg/m2, ≤30 kg if BMI is 26.1-28 kg/m2, or ≤32 kg if BMI>28 kg/m2 for males, and ≤17kg if BMI ≤23 kg/m2, ≤17.3 kg if BMI is 23.1-26 kg/m2, ≤18 kg if BMI is 26.1-29 kg/m2, or ≤21 kg if BMI>29 kg/m2 for females, using the maximal measure in the dominant hand from three trials measured with a hand-held dynamometer.
Baseline Covariates
Age, sex, and race/ethnicity were ascertained via self-report. Weight (kg) was measured using a digital scale and height (cm) using a wall-mounted stadiometer. BMI was calculated as kg/m2. SPPB consisted of three tests, each scored from 0-4 (balance, walking and chair rises)(26). The SPPB score was calculated as the sum of these tests (ranging from 0-12), where higher scores indicate better physical performance. Participants were asked if a physician told him/her of the following medical conditions: cancer, heart disease, high cholesterol, high blood pressure, stroke, peripheral vascular disease, chronic obstructive pulmonary disease, diabetes, kidney disease, liver disease, connective tissue disease, arthritis, Parkinson’s disease, and multiple sclerosis. A morbidity index score was derived by summing the number of medical conditions. Randomized intervention assignment to 200, 1000, 2000, or 4000 IU/day of vitamin D3 (cholecalciferol) oral supplementation and the number of valid device wear days were also treated as covariates.
Statistical Approach
Participant characteristics were described for the overall sample and by physical activity tertiles, using means and standard deviations for continuous variables and frequencies and percentages for categorical variables. Baseline accelerometer variables by frailty status were reported using means and standard deviations.
Multivariable discrete-time Cox regression models were constructed to estimate the association between each accelerometer variable (predictor) and incidence of first-time frailty, with the Efron method for handling ties. Separate models for each accelerometer variable were constructed due to collinearity between the accelerometer metrics (see Supplemental Table 1, SDC 1, Pearson correlations between accelerometer metrics at baseline). The covariates described above were successively added to each model. To assess the potential role of inflammatory markers as moderators, interactions between each accelerometer variable and each inflammatory marker were assessed (e.g., total activity counts x IL-6) in separate models. Proportional hazard assumptions were tested using Schoenfeld Residuals.
Given that baseline frailty level strongly relates to frailty incidence and relates to physical activity, additional models adjusted for the number of frailty criteria at baseline were constructed as a sensitivity analysis.
Statistical significance was set at p<0.05 and determined using two-tailed hypothesis testing. All statistical analyses were conducted using Stata software (version 16.1; Stata Corporation, College Station, TX).
RESULTS
Baseline participant characteristics
Mean age was 77 (SD= 5) years, 41% were women, and 82% self-identified as White among 477 frailty-free participants (Table 1). Mean BMI was 30 (6) kg/m2 and mean number of morbid conditions was 4 (2). Mean IL-6 was 5.2 (3.2) pg/mL, mean CRP was 2.4 (2.7) mg/L, and mean TNF-aR1 was 1,991.1 (652.5) pg/mL. Accelerometer wear showed high compliance, with mean wear days of 6.7 and a mean of 1437 wear minutes/valid day. Participants in higher activity tertiles tended to be younger, women, and White (see Table 1). Robust participants appeared to be slightly younger, women, White, have higher mean physical performance, and slightly lower mean levels of IL-6, CRP, and TNF-aR1 compared to those with prefrailty (see Supplemental Table 2, SDC 1, Baseline participant characteristics by frailty level).
Table 1.
Baseline participant characteristics
| Total N=477 |
Low activity n=159 |
Medium activity n=159 |
High activity n=159 |
|
|---|---|---|---|---|
| Age (years), M (SD) | 76.5 (5.2) | 77.3 (5.8) | 76.5 (5.1) | 75.7 (4.6) |
| Women, % (n) | 41.1 (196) | 27.0 (43) | 42.8 (68) | 53.5 (85) |
| White, % (n) | 82.4 (393) | 81.8 (130) | 80.5 (128) | 84.9 (135) |
| Frailty status, % (n) | ||||
| Robust | 35.6 (170) | 28.3 (45) | 36.5 (58) | 42.1 (67) |
| Prefrail | 64.4 (307) | 71.7 (114) | 63.5 (101) | 57.9 (92) |
| Randomization, % (n) | ||||
| 200 IU/day (control) | 49.7 (237) | 43.4 (69) | 53.5 (85) | 52.2 (83) |
| 1000 IU/day | 28.5 (136) | 30.8 (49) | 23.3 (37) | 31.5 (50) |
| 2000 IU/day | 10.9 (52) | 12.0 (19) | 12.0 (19) | 8.8 (14) |
| 4000 IU/day | 10.9 (52) | 13.8 (22) | 11.3 (18) | 7.6 (12) |
| Body mass index (kg/m2), M (SD) | 30.4 (5.9) | 30.9 (5.9) | 30.6 (5.9) | 29.6 (5.8) |
| SPPB score, M (SD) | 9.1 (2.2) | 9.1 (2.0) | 9.1 (2.3) | 9.2 (2.2) |
| Number of morbid conditions, M (SD) | 3.9 (2.0) | 4.1 (2.0) | 3.8 (1.9) | 3.7 (2.0) |
| IL-6 (pg/mL), M (SD) | 5.2 (3.2) | 5.5 (3.7) | 5.4 (3.3) | 4.8 (2.5) |
| hsCRP (mg/L), M (SD) | 2.4 (2.7) | 2.3 (2.6) | 2.7 (3.0) | 2.1 (2.3) |
| TNF-aR1 (pg/mL), M (SD) | 1991.1 (652.5) | 2141.8 (737.5) | 1954.9 (580.7) | 1876.5 (603.5) |
| Number of days with valid accelerometer data, M (SD) | 6.7 (0.7) | 6.6 (0.7) | 6.7 (0.6) | 6.7 (0.7) |
| Valid wear minutes per day, M (SD) | 1436.9 (8.6) | 1436.7 (8.1) | 1437.1 (8.6) | 1436.9 (9.3) |
Notes: Tertiles of time spent active define low, medium, and high activity groups; M: mean; SD: standard deviation; SPPB score: Short Physical Performance Battery score (0-12; higher scores indicate better physical performance); IL-6: Interleukin-6; hsCRP: high sensitivity C-reactive protein; TNF-aR1: Tumor necrosis factor-alpha receptor 1
Baseline physical activity characteristics
At baseline, the 477 participants averaged 6 hours/day of active time, 12 hours/day of sedentary time, 1,806,300 activity counts/day, 25% activity fragmentation, and 13% sedentary fragmentation (Table 2). Further, 64% were prefrail and appeared to have slightly degraded, if not similar, activity/sedentary patterns of daily physical activity compared to their robust counterparts.
Table 2.
Baseline accelerometer metrics across frailty status
| Total N=477 |
Robust n=170 |
Prefrail n=307 |
|
|---|---|---|---|
| Active time (hours/day), M (SD) | 6.2 (1.7) | 6.4 (1.7) | 6.0 (1.7) |
| Sedentary time (hours/day), M (SD) | 11.8 (1.7) | 11.5 (1.7) | 12.0 (1.7) |
| Total activity (counts/day), M (SD) | 1,806,300.9 (536,740.0) | 1,907,857.4 (551,115.4) | 1,750,064.4 (521,048.6) |
| Activity fragmentation (%), M (SD) | 25.0 (6.5) | 24.4 (6.7) | 25.3 (6.5) |
| Sedentary fragmentation (%), M (SD) | 12.8 (3.5) | 13.3 (3.3) | 12.5 (3.5) |
Notes: M: mean; SD: standard deviation
Physical activity and frailty incidence
Over 2 years (average follow-up of 1.3 years), there were 42 first-time events of frailty. For every 30 minutes/day more spent in active time, there was an associated 16% lower risk of frailty (hazard ratio [HR]=0.84; 95% confidence interval [CI]: 0.75-0.93; p=0.001) (Figure 1, Model 2). Inversely, each 30 minutes more of sedentary minutes/day was associated with 20% higher frailty risk (HR=1.20; 95% CI:1.08-1.33; p=0.001) (Figure 1, Model 2). Every 100,000 higher activity count/day was associated with 13% lower frailty risk (HR=0.87; 95% CI:0.80-0.93; p<0.001). Every 1% higher activity fragmentation was associated with 8% higher frailty risk (HR=1.08; 95% CI:1.04-1.12; p<0.001). Sedentary fragmentation was not associated with lower frailty risk (p=0.07). All results remained robust after additionally adjusting for IL-6, hsCRP, and TNF-aR1 (Figure 1, Model 3).
Figure 1. Natural log of first-time incident frailty hazard ratio across baseline accelerometer metrics, n=477.
Model 1: adjusted for age, sex, race, and treatment assignment
Model 2: adjusted for model 1 covariates + body mass index (kg/m2), Short Physical Performance Battery score (0-12; higher scores indicate higher performance), number of morbid conditions, and valid wear days
Model 3: adjusted for model 2 covariates + interleukin-6, C-reactive protein, and tumor necrosis factor-alpha receptor 1
In sensitivity analyses that categorized each accelerometer metric into tertiles, there were no significant associations between groups by active time, activity count, sedentary time, and sedentary fragmentation and frailty incidence (Table 3, global p>0.08 for all). Differences by activity fragmentation and frailty incidence were significant (global p=0.006). Specifically, those with high activity fragmentation had 4.37 times higher frailty risk compared to the low activity fragmentation (95% CI: 1.71-11.21; p=0.002). Those with high fragmentation (i.e., those in the low activity tertile) had an average activity fragmentation of 30% (SD=7) compared to 20% (4) activity fragmentation among those with the low fragmentation in the high activity tertile group (see Supplemental Table 3, SDC 1, Baseline accelerometry metrics by activity tertiles).
Table 3.
First-time incident frailty hazard ratio across tertiles of accelerometer metrics at baseline (sensitivity analysis)
| Reference: Robust or Prefrail status | ||||
|---|---|---|---|---|
| HR (95% CI) | ||||
| Model 1 | Global test p-value |
Model 2 | Global test p-value |
|
| Active time (min/day) | ||||
| -Low | 2.23 (0.94, 5.30) | 0.193 | 2.14 (0.90, 5.11) | 0.224 |
| -Medium | 1.66 (0.72, 3.80) | 1.49 (0.64, 3.44) | ||
| -High | Reference | Reference | ||
| Sedentary time (min/day) | ||||
| -Low | Reference | 0.193 | Reference | 0.224 |
| -Medium | 1.66 (0.72, 3.80) | 1.49 (0.64, 3.44) | ||
| -High | 2.23 (0.94, 5.30) | 2.14 (0.90, 5.11) | ||
| Total activity (counts/day) | ||||
| -Low | 2.76 (1.14, 6.69)* | 0.079 | 2.63 (1.06, 6.53)* | 0.112 |
| -Medium | 1.96 (0.81, 4.74) | 1.81 (0.74, 4.45) | ||
| -High | Reference | Reference | ||
| Activity fragmentation (%) | ||||
| -Low | Reference | 0.008 | Reference | 0.006 |
| -Medium | 2.27 (0.89, 5.77) | 2.15 (0.83, 5.62) | ||
| -High | 4.08 (1.66, 10.00)** | 4.37 (1.71, 11.21)** | ||
| Sedentary fragmentation (%) | ||||
| -Low | 1.19 (0.57, 2.46) | 0.191 | 1.22 (0.59, 2.50) | 0.239 |
| -Medium | 0.55 (0.24, 1.27) | 0.59 (0.26, 1.37) | ||
| -High | Reference | Reference | ||
Model 1: adjusted for age, sex, race, and treatment assignment
Model 2: adjusted for model 1 covariates + body mass index (kg/m2), Short Physical Performance Battery score (0-12; higher scores indicate higher performance), number of morbid conditions, and valid wear days
p<0.05
p<0.01
p<0.001
The proportional hazard assumption for all models was met (global goodness-of-fit test p>0.14). However, the continuous sedentary fragmentation predictor variable appeared to violate the proportional hazard assumption (p<0.03). In sensitivity analyses additionally adjusting for number of baseline frailty criteria, all results remained similar.
Inflammatory marker by physical activity interaction and frailty incidence
When examining baseline interactions between each accelerometer metric and inflammatory marker (see Supplemental Table 4, SDC 1, Interaction of continuous accelerometer metrics and inflammatory biomarkers at baseline), there were no significant findings for interactions with IL-6 (interaction p>0.46 for all accelerometer metrics), hsCRP (interaction p>0.06 for all accelerometer metrics), and TNF-aR1 (interaction p>0.15 for all accelerometer metrics). Significant main effects of either baseline hsCRP (HR=1.36; 95% CI: 1.00-1.84; p=0.049) and TNF-aR1 (HR=5.87; 95% CI: 2.04-16.87; p=0.001), but not IL-6 (p=0.37), on frailty risk were observed in models without the physical activity variables and adjusting for demographics and treatment assignment (see Supplemental Table 5, Model 1, SDC 1, First-time incident frailty hazard ratio across log-transformed inflammation biomarkers at baseline). Only the baseline TNF-aR1 main effect remained statistically significant after full covariate adjustment (HR=3.89; 95% CI: 1.26-12.06; p=0.02) (see Supplemental Table 5, Model 2, SDC 1, First-time incident frailty hazard ratio across log-transformed inflammation biomarkers at baseline).
When examining the association of each inflammatory marker on frailty incidence, there was no association between IL-6 or hsCRP and frailty incidence (p>0.25 for all baseline accelerometer models). There was a positive association between baseline TNF-aR1 and frailty incidence in the sedentary fragmentation model (HR=3.53; 95% CI: 1.04-10.87; p=0.04), but this association was not detected in models with the other metrics (p>0.05; Figure 1, Model 3).
DISCUSSION
In this study that enrolled older adults at high risk for falls, certain features of everyday physical activity/movement were associated with risk of incident frailty. Specifically, our study showed that objectively measured markers of daily physical activity that represent the amount, intensity, and the degree to which activity is broken up during the day (e.g., activity fragmentation) were associated with frailty incidence. Interestingly, all observed associations were not modified by frailty-related markers of inflammation. Further research is needed to determine whether measures of everyday movement hold potential as intervention targets to prevent frailty onset.
Our results showed an inverse association between everyday movement, in the form of sitting less and moving more often as measured continuously and frailty incidence. This is important because everyday movement performed by older adults largely consists of low-intensity activity(27,28). Though our results generally support previous cross-sectional findings between physical activity and frailty(6,29-32), two observational studies suggest that only high, but not low, intensity daily activity is associated with attenuated frailty progression(8,33). Differences in measuring light-intensity physical activity in those studies compared to ours might explain the differences in average times spent in lower intensity levels of physical activity (e.g., self-reported single question on the intensity of leisure-time physical activity(33) or uniaxial accelerometry set a lower frequency epoch collection of steps/day that resulted in a mean of 55 minutes/day compared to this study’s sample mean of 6 hours/day)(8)).
Our findings indicate that sedentary behavior reduction might serve as an intervention target for frailty prevention. In 2018, the effects of a physical activity program tailored to older adults, with the goal of reaching 150 minutes/week of moderate-intensity activity, on frailty was tested(34). The physical activity program, which largely consisted of walking but also included strength, balance, and flexibility, was not associated with frailty reduction over 24 months. Although, the intervention was associated with lower frailty incidence within 6 months of the intervention, these effects were not robust to sensitivity analyses. There was lower adherence at 24 months (66% session attendance) compared to 6 months (76% session attendance), possibly explaining the null findings at 24 months. Lower adherence might be explained through difficulty in transitioning from supervised (center-based) to unsupervised (home-based) exercise participation(35), a feature built into the physical activity intervention(34,36). Our observational findings support this hypothesis, suggesting that higher daily activity at lower intensities, perhaps achieved by decreasing at-home sitting-like behaviors, might be a manageable alternative pathway, or complement to meeting current physical activity guidelines(37) that potentially combats frailty. Practical, home-based interventions to reduce sedentary behavior have yet to be tested in trials.
Our findings suggest that daily movement performed in a less continuous manner (i.e., more broken up) might be an important indicator of frailty risk. Further, findings support and extend upon a 2021 publication that showed a cross-sectional association between accelerometer-derived activity fragmentation and phenotypic frailty(18) by highlighting alterations of normal daily movement (not just lacking exercise) as a possible precursor to frailty onset. These findings indicate that in older adults, higher activity fragmentation might represent a form of endurance measured in everyday lifestyle activity. This low-order endurance to perform daily lifestyle activity, that is typically at light intensities, could signal shared dysregulation of underlying mechanisms that are postulated to drive frailty(2), such as impairments affecting metabolic, musculoskeletal, hormonal, stress, immune, and circulatory systems. These same pathways also connect to fatigue(38) and fitness levels(39) which could explain increasingly fragmented activity patterns(10) indicative of frailty. While calculating the total amount of activity is important(40), particularly as an indicator of frailty, total active time only represented 34% of awake time in this sample. Further, our findings point to ≥30% activity fragmentation as a potential clinical threshold indicative of impending frailty risk—a similar threshold observed with mortality(17). Therefore, the manner in which activity is accrued during awake time (continuous versus fragmented pattern) could also help in identifying subtle alterations of underlying physiological systems driving movement.
This study did not observe an association between sedentary fragmentation (broken up sedentary time) and decreased frailty risk as originally hypothesized. Five possible reasons exist to explain our observation: 1) two years or less was not enough time between measurements of sedentary fragmentation and frailty to detect a relationship(41); 2) the relationship between sedentary fragmentation and frailty occurs earlier than the mean age of 77 years for this study’s sample(42); 3) there might be reverse causation where older frail adults exhibit subsequent lower sedentary fragmentation levels(18); 4) there was insufficient heterogeneity in sedentary fragmentation in this study sample; or 5) no true relationship exists.
Our study found that chronic inflammation, in the form of IL-6, hsCRP and TNF-aR1, did not alter the detected associations between physical activity patterns and frailty incidence. This is inconsistent with our original hypothesis. The most likely reason is that study participants have high inflammation levels more reflective of disease and multimorbidity than phenotypic frailty, despite possible overlap(43). Inflammation levels in our sample were similar to the 1,018 participants in another study who were aged 60 years and older and living with the same average number of medical conditions(44). Also, it is possible there was insufficient heterogeneity in the inflammatory markers of interest among this sample with high BMI and inadequate statistical power to detect associations with the low number of frailty events. Whether a specific threshold of chronic inflammation exists, before multimorbidity and related inflammation, that modifies the relationship between inactivity and frailty risk warrants further investigation.
Our study has limitations. First, there is no universal definition of frailty, but this study uses one of the most common definitions in research(3). Second, generalizability might be limited due to the sample population being from a fall prevention study of older adults with low vitamin D levels and at higher risk of falls. However, low vitamin D is extremely common among older persons. Third, accelerometry does not measure the activity type (e.g., sitting versus standing) but excels at collecting detailed information on duration, intensity, and frequency of activity and inactivity. Fourth, there is the potential for collinearity among the different PA metrics. Fifth, the number of incident frailty events was low (9%) elevating the possibility of type 1 error when exploring interactions. Sixth, there is potential for reverse causation bias with 64% of the baseline sample being prefrail. Study strengths include a large sample of at-risk older adults, the employment of a longitudinal design to examine frailty incidence, and the measurement of free-living physical activity using accelerometry.
CONCLUSIONS
In conclusion, among older adults who are either robust or prefrail, constricted patterns of daily physical activity (i.e., lower total activity minutes and counts, and higher activity fragmentation) were prospectively associated with higher risk of frailty. Additional studies, particularly trials, are needed to understand if this association is causal.
Supplementary Material
SDC 1: STURDY accel inflam and frailty incidence_R2_SDC.docx
Supplemental Table 1. Pearson correlations between accelerometer metrics at baseline
Supplemental Table 2. Baseline participant characteristics by frailty level
Supplemental Table 3. Baseline accelerometry metrics by activity tertiles
Supplemental Table 4. Interaction of continuous accelerometer metrics and inflammatory biomarkers at baseline
Supplemental Table 5. First-time incident frailty hazard ratio across log-transformed inflammation biomarkers (per logged unit) at baseline
Acknowledgements
We thank all the participants and study staff for their contributions to the STURDY trial.
Conflict of Interest and Funding Source:
Outside of submitted work, Dr. Michos reports advisory boards with Amarin, AstraZeneca, Bayer, Boehringer Ingelheim, Esperion, Novartis, Novo Nordisk, and Pfizer. The other authors have no disclosures to report. The results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation. The results of the present study do not constitute endorsement by the American College of Sports Medicine. Main findings were presented at the annual 2021 Gerontological Society of America. STURDY was funded by the National Institute on Aging (U01AG047837) with support from the Office of Dietary Supplements, the Mid-Atlantic Nutrition Obesity Research Center (P30DK072488), and the Johns Hopkins Institute for Clinical and Translation Research (JHICTR; UL1TR003098). Dr. Wanigatunga was supported by the Johns Hopkins Older Americans Independence Center (JHOAIC; P30AG059298) and the Johns Hopkins Alzheimer’s Disease Resource Center for Minority Aging Research (P30AG021334). Dr. Bandeen-Roche was supported by the JHICTR and the JHOAIC.
REFERENCES
- 1.Oliveira JS, Pinheiro MB, Fairhall N, et al. Evidence on physical activity and the prevention of frailty and sarcopenia among older people: a systematic review to inform the World Health Organization physical activity guidelines. J Phys Act Health. 2020;17(12):1247–58. [DOI] [PubMed] [Google Scholar]
- 2.Fried LP, Cohen AA, Xue QL, Walston J, Bandeen-Roche K, Varadhan R. The physical frailty syndrome as a transition from homeostatic symphony to cacophony. Nat Aging. 2021;1(1):36–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Fried LP, Tangen CM, Walston J, et al. Frailty in older adults: evidence for a phenotype. J Gerontol A Biol Sci Med Sci. 2001;56(3):M146–56. [DOI] [PubMed] [Google Scholar]
- 4.Richardson MT, Leon AS, Jacobs DR, Ainsworth BE, Serfass R. Comprehensive evaluation of the Minnesota leisure time physical activity questionnaire. J Clin Epidemiol. 1994;47(3):271–81. [DOI] [PubMed] [Google Scholar]
- 5.Troiano RP, Berrigan D, Dodd KW, Mâsse LC, Tilert T, Mcdowell M. Physical activity in the United States measured by accelerometer. Med Sci Sports Exerc. 2008;40(1):181–8. [DOI] [PubMed] [Google Scholar]
- 6.Huisingh-Scheetz M, Wroblewski K, Kocherginsky M, et al. The relationship between physical activity and frailty among U.S. older adults based on hourly accelerometry data. J Gerontol A Biol Sci Med Sci. 2018;73(5):622–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Zhang X, Tan SS, Franse CB, et al. Longitudinal association between physical activity and frailty among community-dwelling older adults. J Am Geriatr Soc. 2020;68(7):1484–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Yuki A, Otsuka R, Tange C, et al. Daily physical activity predicts frailty development among community-dwelling older Japanese adults. J Am Med Dir Assoc. 2019;20(8):1032–6. [DOI] [PubMed] [Google Scholar]
- 9.Kehler DS, Theou O. The impact of physical activity and sedentary behaviors on frailty levels. Mech Ageing Dev. 2019;180:29–41. [DOI] [PubMed] [Google Scholar]
- 10.Wanigatunga AA, Ferrucci L, Schrack JA. Physical activity fragmentation as a potential phenotype of accelerated aging. Oncotarget. 2019;10(8):807–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Schrack JA, Kuo PL, Wanigatunga AA, et al. Active-to-sedentary behavior transitions, fatigability, and physical functioning in older adults. J Gerontol A Biol Sci Med Sci. 2019;74(4):560–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Wanigatunga AA, Gresham GK, Kuo PL, et al. Contrasting characteristics of daily physical activity in older adults by cancer history. Cancer. 2018;124(24):4692–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Cai Y, Schrack JA, Wang H, et al. Visual impairment and objectively measured physical activity in middle-aged and older adults. J Gerontol A Biol Sci Med Sci. 2021;76(12):2194–203. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Palmberg L, Rantalainen T, Rantakokko M, et al. The associations of activity fragmentation with physical and mental fatigability among community-dwelling 75-, 80-, and 85-year-old people. J Gerontol A Biol Sci Med Sci. 2020;75(9):e103–10. [DOI] [PubMed] [Google Scholar]
- 15.Wanigatunga AA, Wang H, An Y, et al. Association between brain volumes and patterns of physical activity in community-dwelling older adults. J Gerontol A Biol Sci Med Sci. 2021;76(8):1504–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.del Pozo Cruz B, del Pozo-Cruz J. Associations between activity fragmentation and subjective memory complaints in middle-aged and older adults. Exp Gerontol. 2021;148:111288. [DOI] [PubMed] [Google Scholar]
- 17.Wanigatunga AA, Di J, Zipunnikov V, Urbanek JK, et al. Association of total daily physical activity and fragmented physical activity with mortality in older adults. JAMA Netw Open. 2019;2(10):e1912352. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Wanigatunga AA, Cai Y, Urbanek JK, et al. Objectively measured patterns of daily physical activity and phenotypic frailty. J Gerontol A Biol Sci Med Sci. 2022;77(9):1882–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Ferrucci L, Fabbri E. Inflammageing: chronic inflammation in ageing, cardiovascular disease, and frailty. Nat Rev Cardiol. 2018;15(9):505–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Nicklas BJ, Beavers DP, Mihalko SL, Miller GD, Loeser RF, Messier SP. Relationship of objectively-measured habitual physical activity to chronic inflammation and fatigue in middle-aged and older adults. J Gerontol A Biol Sci Med Sci. 2016;71(11):1437–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Miles MP, Wilson S, Yeoman CJ. Physical activity and inflammation phenotype conversion. J Clin Exerc Physiol. 2019;8(2):64–73. [Google Scholar]
- 22.Nimmo MA, Leggate M, Viana JL, King JA. The effect of physical activity on mediators of inflammation. Diabetes Obes Metab. 2013;15 Suppl 3:51–60. [DOI] [PubMed] [Google Scholar]
- 23.Appel LJ, Michos ED, Mitchell CM, et al. The effects of four doses of vitamin D supplements on falls in older adults. Ann Intern Med. 2021;174(2):145–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Choi L, Liu Z, Matthews CE, Buchowski MS. Validation of accelerometer wear and nonwear time classification algorithm. Med Sci Sports Exerc. 2011;43(2):357–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Koster A, Shiroma EJ, Caserotti P, et al. Comparison of sedentary estimates between activPAL and hip- and wrist-worn ActiGraph. Med Sci Sports Exerc. 2016;48(8):1514–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Guralnik JM, Simonsick EM, Ferrucci L, et al. A short physical performance battery assessing lower extremity function: association with self-reported disability and prediction of mortality and nursing home admission. J Gerontol. 1994;49(2):M85–94. [DOI] [PubMed] [Google Scholar]
- 27.Wanigatunga AA, Tudor-Locke C, Axtell RS, et al. Effects of a long-term physical activity program on activity patterns in older adults. Med Sci Sports Exerc. 2017;49(11):2167–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Buman MP, Hekler EB, Haskell WL, et al. Objective light-intensity physical activity associations with rated health in older adults. Am J Epidemiol. 2010;172(10):1155–65. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Blodgett J, Theou O, Kirkland S, Andreou P, Rockwood K. The association between sedentary behaviour, moderate–vigorous physical activity and frailty in NHANES cohorts. Maturitas. 2015;80(2):187–91. [DOI] [PubMed] [Google Scholar]
- 30.Kehler DS, Theou O. The impact of physical activity and sedentary behaviors on frailty levels. Mech Ageing Dev. 2019;180:29–41. [DOI] [PubMed] [Google Scholar]
- 31.Mañas A, del Pozo-Cruz B, Rodríguez-Gómez I, et al. Can physical activity offset the detrimental consequences of sedentary time on frailty? A moderation analysis in 749 older adults measured with accelerometers. J Am Med Dir Assoc. 2019;20(5):634–38.e1. [DOI] [PubMed] [Google Scholar]
- 32.Nagai K, Tamaki K, Kusunoki H, et al. Isotemporal substitution of sedentary time with physical activity and its associations with frailty status. Clin Interv Aging. 2018;13:1831–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Rogers NT, Marshall A, Roberts CH, Demakakos P, Steptoe A, Scholes S. Physical activity and trajectories of frailty among older adults: Evidence from the English Longitudinal Study of Ageing. PLoS One. 2017;12(2):e0170878. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Trombetti A, Hars M, Hsu FC, et al. Effect of physical activity on frailty. Ann Intern Med. 2018;168(5):309–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Schutzer KA, Graves BS. Barriers and motivations to exercise in older adults. Prev Med. 2004;39(5):1056–61. [DOI] [PubMed] [Google Scholar]
- 36.Pahor M, Guralnik JM, Ambrosius WT, et al. Effect of structured physical activity on prevention of major mobility disability in older adults: rhe LIFE study randomized clinical trial. JAMA. 2014;311(23):2387. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Piercy KL, Troiano RP, Ballard RM, et al. The Physical Activity Guidelines for Americans. JAMA. 2018;320(19):2020–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Matura LA, Malone S, Jaime-Lara R, Riegel B. A systematic review of biological mechanisms of fatigue in chronic illness. Biol Res Nurs. 2018;20(4):410–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Silverman MN, Deuster PA. Biological mechanisms underlying the role of physical fitness in health and resilience. Interface Focus. 2014;4(5):20140040. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Schrack JA, Zipunnikov V, Goldsmith J, et al. Assessing the “physical cliff”: detailed quantification of age-related differences in daily patterns of physical activity. J Gerontol A Biol Sci Med Sci. 2014;69(8):973–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Mañas A, del Pozo-Cruz B, Rodríguez-Gómez I, et al. Breaking sedentary time predicts future frailty in inactive older adults: a cross-lagged panel model. J Gerontol A Biol Sci Med Sci. 2021;76(5):893–900. [DOI] [PubMed] [Google Scholar]
- 42.Godin J, Blodgett JM, Rockwood K, Theou O. Replacing Sedentary time with light or moderate–vigorous physical activity across levels of frailty. J Aging Phys Act. 2020;28(1):18–23. [DOI] [PubMed] [Google Scholar]
- 43.Rockwood K, Mitnitski A. Frailty in relation to the accumulation of deficits. J Gerontol A Biol Sci Med Sci. 2007;62(7):722–7. [DOI] [PubMed] [Google Scholar]
- 44.Fabbri E, An Y, Zoli M, et al. Aging and the burden of multimorbidity: associations with inflammatory and anabolic hormonal biomarkers. J Gerontol A Biol Sci Med Sci. 2015;70(1):63–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
SDC 1: STURDY accel inflam and frailty incidence_R2_SDC.docx
Supplemental Table 1. Pearson correlations between accelerometer metrics at baseline
Supplemental Table 2. Baseline participant characteristics by frailty level
Supplemental Table 3. Baseline accelerometry metrics by activity tertiles
Supplemental Table 4. Interaction of continuous accelerometer metrics and inflammatory biomarkers at baseline
Supplemental Table 5. First-time incident frailty hazard ratio across log-transformed inflammation biomarkers (per logged unit) at baseline