Abstract
Background
Limited evidence suggests that physical activity may prevent frailty and associated negative outcomes in older adults. Definitive data from large, long-term, randomized trials are lacking.
Objective
To determine whether a long-term structured moderate-intensity physical activity (PA) program is associated with lower risk of frailty and whether frailty status alters the effect of PA on the reduction in major mobility disability (MMD) risk.
Design
Multicenter, single-blind, randomized trial.
Setting
Eight centers in the United States.
Participants
1635 community-dwelling adults, age 70–89 years with functional limitations.
Interventions
A structured moderate-intensity PA program incorporating aerobic, resistance, and flexibility activities or a health education program consisting of workshops and stretching exercises.
Measurements
Frailty, as defined by the Study of Osteoporotic Fractures (SOF) index, at baseline, 6, 12, and 24 months, and MMD, defined as the inability to walk 400 m, for up to 3.5 years.
Results
Over 24 months of follow-up, the risk of frailty (n=1623) was not significantly different in the PA group versus the health education group (adjusted prevalence difference, −0.021 [95%CI, −0.049 to 0.007]). Among the three criteria of the SOF index, the PA intervention was associated with improvement in the inability to rise from a chair criterion (adjusted prevalence difference, −0.050 [CI, −0.081 to −0.020]). Baseline frailty status did not modify the effect of PA on reducing incident MMD (P for interaction=0.91).
Limitation
Frailty status was not an entry criterion nor a randomization stratum.
Conclusions
A structured, moderate-intensity PA program was not associated with a reduced risk of frailty over two years among sedentary community-dwelling older adults. The beneficial effect of PA on incidence of MMD did not differ between frail and non-frail individuals.
Primary Funding Source
National Institute on Aging, National Institutes of Health
Keywords: Aging, Frailty, Physical activity, Disability
INTRODUCTION
Frailty, a state of increased vulnerability to stressor events, confers high risk of surgical complications, morbidity, disability and mortality in older adults (1–9). There is presently no universally accepted consensus definition of frailty (7,10). However, a widely used definition has been proposed by Ensrud and colleagues (2) using data from the Study of Osteoporotic Fractures (SOF): the SOF frailty index. This measure defines frailty based on 2 or more of the following criteria: inability to rise from a chair five times without using arms, self-reported reduced energy level, and weight loss.
Emerging evidence suggests that exercise-based interventions may improve physical functioning and prevent disability in frail older individuals (11–13). Yet, to date, no large randomized trial has examined whether long-term physical activity (PA) can reduce the risk of frailty over an extended period of follow-up, and prevent associated mobility disability. The main findings from the Lifestyle Interventions and Independence for Elders (LIFE) trial showed that a structured, moderate-intensity PA program reduced major mobility disability (MMD) over 2.7 years among older adults at risk of disability (14). As frailty status was not examined in the primary LIFE study findings, it remains unknown whether long-term PA may also prevent MMD in frail older individuals.
To address the limitations of prior research, we conducted a secondary analysis of data from the LIFE trial to specifically evaluate the effect of PA on frailty and MMD. The objectives of our analyses were to determine whether i) a long-term structured, moderate-intensity PA program is associated with the risk of frailty, as defined by the SOF frailty index, and ii) if frailty status at baseline modifies the reduction of MMD observed with PA.
METHODS
Trial Design
The LIFE study was a multicenter, single-blind, parallel group, randomized trial conducted across the USA between February 2010 and December 2013, and designed to compare a long-term PA program with a health education (HE) program on the incidence of MMD (ClinicalTrials.gov Identifier: NCT01072500). This article presents the results of a secondary analysis that was not prespecified in the study protocol. The rationale, design, and methods of the LIFE study have been presented in detail elsewhere (14–16). The study protocol was approved by the institutional review boards at all participating sites. Written informed consent was obtained from all participants.
Participants
Participants have been previously described (14,16). Briefly, participants were eligible for the study if they were (i) between the ages of 70 and 89, (ii) at high risk for mobility disability (i.e., Short Physical Performance Battery (SPPB) score ≤9) (17), (iii) able to walk 400 m in ≤15 minutes unassisted, and (iv) sedentary.
Randomization and Interventions
Participants were randomized with a 1:1 allocation to the PA or HE program with stratification by sex and field center. Each participant was followed until the last participant randomized completed the 24-month visit. The intervention duration ranged from approximately 2 to 3.5 years.
The PA intervention involved a combination of walking (up to 150 min/wk), strength, balance, and flexibility exercises, as previously described (15). The HE program included workshops emphasizing educational topics relevant for older adults (15).
Outcome Measures
Participants were assessed every 6 months and assessment staff (nurses or project coordinators) was blinded to the intervention assignment.
Frailty
Frailty status was determined at baseline, 6, 12 and 24 months using the SOF frailty index (2). The inability to rise from a chair five times without using arms was obtained from the chair rise test component of the SPPB (17). Self-reported reduced energy level was defined by using the following statement of the health-related quality of life (HRQL) questionnaire: “During the past week, how often have you felt full of energy?”. The criterion was considered as present if the participants answered “Some of the time”, “A little bit of the time”, or “None of the time”. The criterion weight loss was based on weight measurements and considered as present if body weight loss was ≥4.55 kg or ≥5% during the last 12 months, or ≥2.275 kg or ≥2.5% during the last 6 months, with the exception of the baseline visit. As no objective information was available at baseline regarding weight loss, the criterion was considered present at baseline if the participant reported a loss of appetite on the HRQL questionnaire. Subjects were considered “frail” if at least two of the three criteria were fulfilled.
Major mobility disability
MMD was defined as the inability to complete a 400-m walk and persistent mobility disability (PMD) was defined as having two consecutive MMD assessments or MMD followed by death (14,15).
Statistical Analysis
Baseline characteristics were summarized by randomization group and frailty status using mean and standard deviation, or percentages. The “intent-to-treat (ITT)” approach was used as the primary analysis where participants were grouped according to their randomization assignment. All eligible randomized participants were included in the analyses except the incidence analysis where those who had frailty at baseline were excluded.
The difference in cumulative incidence of first frailty between the two intervention groups at each visit among non-frail participants at baseline were analyzed using Poisson regression (poisson in Stata) adjusting for gender and center, with logarithm of time as an offset. No competing risk analysis was considered. The marginal standardization approach (margins in Stata) (18) was used to calculate the 95% confidence interval (CI) of the difference in cumulative incidence. Sensitivity analyses were also performed. To correct for the potential imbalance in the number of frail participants by group at baseline caused by the secondary analysis design, we weighted by the probability of randomization calculated from a logistic regression using the inverse probability weighting (IPW) approach (19). Also, we explored the impact of incomplete follow-up using the same approach (Appendix 1).
The differences in prevalence between the two intervention groups for frailty and each criterion of frailty among the whole population were analyzed using generalized estimating equation (GEE) models with the logit link function, binomial distribution, and exchangeable working correlation (xtgee in Stata). The marginal standardization approach was used to calculate the CI of the prevalence difference. In these models, baseline outcome measure was retained in the outcome vector, and time was treated as continuous. Unadjusted and adjusted models were fitted. Centered time (time minus mean of time), time2, and time3 were entered into the model as covariates in addition to gender, field center, and intervention. Sensitivity analyses were performed using IPW analysis (weight equivalent to the probability of remaining in the study calculated using a logistic regression). Transition models (20) and complier average causal effect (CACE) analysis were also performed (Appendix 1) (21,22).
The ordinal logistic regression with clustered sandwich estimator was used to assess the effect of the intervention on the number of frailty criteria with the same covariates specified in the models (ologit in Stata), with also the interactions between time variable(s) and intervention entered into the models. The sensitivity analyses detailed above were also performed (Appendix 1).
The cumulative incidence curves for the first post-randomization occurrence of MMD and PMD when considering death as a competing risk were plotted by intervention and baseline frailty groups (proc lifetest in SAS). Event time was defined as the time from randomization to the initial endpoint or death and censoring time as the time from randomization to the last assessment. The effects of the intervention and baseline frailty status on the first post-randomization occurrence of MMD and PMD were analyzed using Cox proportional hazards models considering death as a competing risk (proc phreg in SAS). Gender and field center were treated as stratifying variables for the baseline hazard. Intervention and frailty status and their interaction were included in the model.
A two-sided P value less than or equal to 0.05 was considered statistically significant. Statistical analyses were performed in SAS (SAS Institute) version 9.4 (TS1M3), and in Stata/IC (StataCorp) version 12.1.
Role of Funding Source
The NIH sponsor was a voting member of the Steering Committee, which approved the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, and approval of the manuscript; and decision to submit the manuscript for publication.
RESULTS
Of the 14,831 individuals who were screened, 1635 were randomized (Appendix Figure 1). Overall, participants’ mean (SD) age was 78.9 (5.2) and 67.2% were women. Baseline characteristics were similar in the two randomization groups (14,16). At baseline, 12 participants had no frailty data and 19.7% (319/1623) of the participants were frail (Table 1). The mean (SD) number of frailty criteria was 0.9 (0.8) (Appendix Table 1). Intervention adherence over 24 months in the PA group did not differ between participants classified as frail and not frail at enrollment (Appendix Tables 2 and 3). By month 24, 97 (5.9%) participants dropped out the study, including 53 deaths.
Table 1.
Baseline Characteristics of Participants by Randomization Group and Frailty Status*
| Physical Activity (n=812)
|
Health Education (n=811)
|
||||||
|---|---|---|---|---|---|---|---|
| Frail (n=159) | Not frail (n=653) | All | Frail (n=160) | Not frail (n=651) | All | ||
| Mean (SD) age, years | 79.5 (5.7) | 78.5 (5.1) | 78.7 (5.2) | 80.0 (5.3) | 78.8 (5.2) | 79.1 (5.2) | |
| Female sex | 114 (71.7%) | 431 (66.0%) | 545 (67.1%) | 118(73.8%) | 428 (65.7%) | 546 (67.3%) | |
| Ethnicity/race | |||||||
| White | 110 (69.2%) | 492 (75.3%) | 602 (74.1%) | 128 (80.0%) | 502 (77.1%) | 630 (77.7%) | |
| African American | 37 (23.3%) | 124 (19.0%) | 161 (19.8%) | 25 (15.6%) | 100 (15.4%) | 125 (15.4%) | |
| Other | 12 (7.5%) | 37 (5.7%) | 49 (6.0%) | 7 (4.4%) | 49 (7.5%) | 56 (6.9%) | |
| Live alone | 81 (50.9%) | 306 (46.9%) | 387 (47.4%) | 90 (56.3%) | 326 (50.1%) | 416 (51.3%) | |
| Education | |||||||
| Post graduate | 38 (23.9%) | 153 (23.5%) | 191 (23.6%) | 42 (26.6%) | 165 (25.4%) | 207 (25.7%) | |
| Mean (SD) BMI, kg/m2 | 30.4 (6.1) | 30.0 (5.6) | 30.1 (5.7) | 29.3 (6.4) | 30.6 (6.2) | 30.3 (6.2) | |
| Mean (SD) 3MSE, score | 90.9 (5.6) | 91.7 (5.4) | 91.6 (5.5) | 91.5 (5.7) | 91.7 (5.3) | 91.6 (5.3) | |
| Mean (SD) CES-D, score | 14.2 (9.7) | 7.0 (6.4) | 8.4 (7.7) | 14.0 (9.2) | 7.6 (7.0) | 8.8 (7.9) | |
| Mean (SD) quality of life, score | 0.6 (0.1) | 0.6 (0.1) | 0.6 (0.1) | 0.6 (0.1) | 0.6 (0.1) | 0.6 (0.1) | |
| Mean (SD) albumin, g/dl | 4.1 (0.3) | 4.2 (0.3) | 4.1 (0.3) | 4.1 (0.3) | 4.2 (0.2) | 4.1 (0.2) | |
| Mean (SD) creatinine clearance, ml/min/1.73 m2 | 70.1 (29.0) | 69.6 (24.6) | 69.7 (25.5) | 67.1 (25.8) | 72.4 (26.4) | 71.3 (26.4) | |
| Mean (SD) number of chronic conditions | 1.8 (1.1) | 1.8 (1.2) | 1.8 (1.1) | 2.0 (1.2) | 1.8 (1.1) | 1.8 (1.2) | |
| Mean (SD) accelerometry of moderate physical activity, min/wk† | 149.9 (117.1) | 201.3 (159.4) | 191.7 (153.7) | 165.7 (166.6) | 210.4 (188.9) | 201.8 (185.6) | |
| Mean (SD) SPPB, score | 6.5 (1.9) | 7.7 (1.4) | 7.4 (1.6) | 6.3 (1.8) | 7.6 (1.4) | 7.3 (1.6) | |
| SPPB >= 8 | 60 (37.7%) | 403 (61.7%) | 463 (57.0%) | 53 (33.1%) | 384 (59.0%) | 437 (53.9%) | |
| Inability to rise from a chair one time without using the arms) | 43 (27.0%) | 33 (5.1%) | 76 (9.4% | 63 (39.4%) | 36 (5.5%) | 99 (12.2%) | |
| Mean (SD) chair stand time, s | 16.4 (5.9) | 16.8 (4.8) | 16.7 (5.0) | 18.8 (12.5) | 16.8 (5.0) | 17.0 (6.3) | |
| Mean (SD) handgrip strength, kg | 21.9 (8.0) | 25.5 (10.5) | 24.8 (10.2) | 22.4 (9.0) | 24.9 (9.8) | 24.5 (9.7) | |
| Mean (SD) 400-m gait speed, m/s | 0.8 (0.2) | 0.8 (0.2) | 0.8 (0.2) | 0.8 (0.2) | 0.8 (0.2) | 0.8 (0.2) | |
| Criteria for the SOF frailty index: | |||||||
| Inability to rise from a chair five times without using the arms | 67 (42.1%) | 48 (7.4%) | 115 (14.2%) | 87 (54.4%) | 51 (7.8%) | 138 (17.0%) | |
| Weight loss‡ | 111 (69.8%) | 38 (5.8%) | 149 (18.3%) | 112 (70.0%) | 37 (5.7%) | 149 (18.4%) | |
| Reduced energy level§ | 152 (95.6%) | 263 (40.3%) | 415 (51.1%) | 150 (93.8%) | 277 (42.5%) | 427 (52.7%) | |
| Frail according to the SOF frailty index | 159 (100%) | 0 (0%) | 159 (19.6%) | 160 (100%) | 0 (0%) | 160 (19.7%) | |
| Mean (SD) number of SOF frailty index criteria | 2.1 (0.3) | 0.5 (0.5) | 0.8 (0.8) | 2.2 (0.4) | 0.6 (0.5) | 0.9 (0.8) | |
| Number of SOF frailty index criteria | |||||||
| 0 | 0 (0%) | 304 (46.6%) | 304 (37.4%) | 0 (0%) | 286 (43.9%) | 286 (35.3%) | |
| 1 | 0 (0%) | 349 (53.4%) | 349 (43.0%) | 0 (0%) | 365 (56.1%) | 365 (45.0%) | |
| 2 | 147 (92.5%) | 0 (0%) | 147 (18.1%) | 131 (81.9%) | 0 (0%) | 131 (16.2%) | |
| 3 | 12 (7.6%) | 0 (0%) | 12 (1.5%) | 29 (18.1%) | 0 (0%) | 29 (3.6%) | |
BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); 3MSE, modified mini-mental state examination; CES-D, center for epidemiologic studies depression scale; SPPB, short physical performance battery; SOF, Study of Osteoporotic Fractures.
Frailty was defined according to SOF frailty index. Values are numbers (percentages) unless stated otherwise.
Defined based on the 760 counts/min cut point.
The criterion weight loss was based on weight measurements, with the exception of the baseline visit: the criterion was considered as present at baseline if the participant reported a loss of appetite on the health-related quality of life questionnaire.
The reduced energy level criterion was defined by using the following statement of the health-related quality of life questionnaire: “During the past week how often have you felt full of energy”.
Among non-frail participants at baseline, the cumulative incidence of first frailty was lower in the PA group than the HE group at 6-month (adjusted difference, −0.05 [CI, −0.09 to −0.01]; P=0.013) and 24-month follow-up (−0.06 [CI, −0.13 to −0.001]; P=0.048) (Table 2). The difference at 6-month remained significant in IPW analyses.
Table 2.
Cumulative Incidence of First Frailty Among Non-Frail Participants at Baseline According to Randomization Group*
| Physical Activity (n=653) |
Health Education (n=651) |
Difference between Physical Activity and Health Education† (difference, [95% CI], P Value) |
|||
|---|---|---|---|---|---|
| Assessment time | n (count/person year) [95% CI |
n (count/person year) [95% CI |
ITT Model | IPW Model 1‡ | IPW Model 2§ |
|
|
|
|
|||
| 6 Month | 69 (0.25) [0.20 to 0.30] |
99 (0.34) [0.30 to 0.39] |
−0.05 [−0.09 to −0.01] 0.013 |
−0.04 [−0.08 to −0.005] 0.029 |
−0.04 [−0.08 to −0.002] 0.041 |
| 12 Month | 120 (0.23) [0.20 to 0.26] |
133 (0.25) [0.22 to 0.28] |
−0.02 [−0.08 to 0.03] 0.38 |
−0.01 [−0.07 to 0.04] 0.59 |
−0.02 [−0.07 to 0.04] 0.54 |
| 24 Month | 168 (0.18) [0.16 to 0.20] |
186 (0.19) [0.17 to 0.21] |
−0.06 [−0.13 to −0.001] 0.048 |
−0.05 [−0.11 to 0.01] 0. 099 |
−0.05 [−0.12 to 0.01] 0.110 |
SOF, Study of Osteoporotic Fractures; ITT, intent-to-treat; IPW, inverse probability weighting.
Frailty was defined according to the SOF index.
Test based on Poisson regression after adjusting for gender and clinical sites. Logarithm of follow-up time was used as an offset.
Weighted for the inverse probability of randomization to the intervention group among non-frail participants to account for the potential imbalance in the number of frail participants by group at baseline caused by the secondary analysis design.
Weighted for the inverse probability of incomplete follow-up to account for loss to follow-up and adjusted for the probability of randomization to the intervention group among non-frail participants (treated it as a propensity score) to account for baseline imbalance.
The adjusted prevalence difference for the risk of frailty was not significant in both the ITT and IPW analyses (−0.021 [CI, −0.049 to 0.007]; P=0.148 in ITT analysis) (Table 3; Appendix Table 1). Conditional on the frailty status at the previous visit in transition models, the prevalence of frailty was lower in the PA group than the HE group (ITT unadjusted prevalence difference, −0.021 [CI, −0.042 to −0.0003]; P=0.047) though the adjusted estimate and IPW transition models were not significant (Appendix Table 4). Taking into account adherence with the interventions, the CACE estimates of the reduction in frailty with PA compared to HE were significant in all models (Appendix Table 5).
Table 3.
Risk of Frailty over 24 months among all Study Participants, According to Randomization Group*
| Average Prevalence Over 24 Months‖ | Unadjusted prevalence difference [95% CI]§ | P Value | Adjusted prevale difference [95% CI]§ | P Value | ||
|---|---|---|---|---|---|---|
|
|
||||||
| Physical activity | Health education (Reference) | |||||
| ITT Model† | 19.1% | 20.8% | −0.021 −0.049 to 0.008] | 0.156 | −0.021 [−0.049 to 0.007] | 0.148 |
| IPW Model†‡ | −0.015 [−0.045 to 0.014] | 0.30 | −0.016 [−0.045 to 0.014] | 0.30 | ||
SOF, Study of Osteoporotic Fractures; ITT, intent-to-treat; IPW, inverse probability weighting.
Frailty was defined according to the SOF index.
Gender and field center (both used to stratify randomization), intervention, time, time2, and time3 included in the generalized estimating equation (GEE) models (logit link, binomial distribution). Baseline outcome in the outcome vector.
Weighted for the inverse probability of remaining in the study.
Marginal standardization was used to obtain the prevalence difference and its 95% CI.
Marginal standardization was used to obtain the prevalence difference and its 95% CI.
Average prevalence over time calculated as the sum of prevalence from each visit, including baseline, in the raw data divided by 4.
The inability to rise from a chair five times was the only frailty criterion affected by the intervention over the 24-month follow-up in both ITT and IPW analyses (adjusted prevalence difference in ITT analysis, −0.050 [CI, −0.081 to −0.020]; P=0.001) (Appendix Table 6). The prevalence of inability to rise from a chair was 2.8% to 5.8% lower in the PA group than the HE group across assessment visits (Appendix Table 1). The mean number of frailty criteria was generally lower in the PA group than the HE group over time (Appendix Table 1). The risk of getting SOF frailty criteria in the PA group decreased over time when compared to the HE group (P for the interaction between time and randomization arm = 0.033 in ITT analysis; Table 4). The results for the transition models showed that participants in the PA group had lower odds of having higher number of frailty criteria compared to the HE group in both the ITT and IPW analyses after adjusting for criteria at the previous visit (OR, 0.88 [CI, 0.78 to 0.98]; P=0.019 in ITT analysis; Appendix Table 7). The CACE result was not stable (data not shown).
Table 4.
Association between the Number of Frailty Criteria and Randomization Group over 24 Months in All Study Participants*
| Average Mean Number of Frailty Criteria Over 24 Month‖ | Unadjusted OR [95% CI] |
P Value | Adjusted OR [95% CI] |
P Value | P Value for interaction time*randomization arm§ | ||
|---|---|---|---|---|---|---|---|
|
|
|||||||
| Physical activity | Health education (Reference) | ||||||
| ITT Model† | 0.86±0.76 | 0.92±0.78 | 0.88 [0.76 to 1.01] | 0.075 | 0.87 [0.75 to 1.01] | 0.072 | 0.033 |
| IPW Model†‡ | 0.89 [0.77 to 1.03] | 0.130 | 0.89 [0.77 to 1.03] | 0.127 | 0.042 | ||
SOF, Study of Osteoporotic Fractures; ITT, intent-to-treat; IPW, inverse probability weighting.
Frailty was defined according to the SOF index.
Gender and field center (both used to stratify randomization), intervention, time, time2, and time3 included in the ordinal logistic regression models. Baseline outcome was retained in the outcome vector. The ORs from ordinal logistic regression models with clustered sandwich estimators were presented.
Weighted for the inverse probability of remaining in the study.
With additionally the interaction time*randomization arm entered into the models.
Average mean number (± standard deviation) of frailty criteria over time calculated as the sum of mean number from each visit, including baseline, in the raw data divided by 4.
Among the subgroup of frail individuals, MMD was experienced by 67/159 (42%) PA participants and 78/160 (49%) HE participants, while 37/159 (23%) and 45/160 (28%) experienced PMD, respectively. Overall, baseline frailty status did not modify the effect of PA on reducing incident MMD (P for interaction=0.91) and PMD (P for interaction=0.64) (Figure 1).
Figure 1.
Cumulative incidences of a) major mobility disability and b) persistent mobility disability by baseline frailty and intervention groups*.
*Frailty was defined according to the SOF index.
SOF, Study of Osteoporotic Fractures.
DISCUSSION
This study demonstrated that 24 months of a structured, moderate-intensity PA program was not associated with a reduction in the overall risk of frailty in older adults, but it was associated with improvement in the inability to rise from a chair criterion of the SOF index. Beneficial effects of the physical intervention on incidence and persistence of MMD were not influenced by frailty status.
Data from the LIFE-pilot study suggested that PA was associated with a reduction in frailty prevalence, as measured by the Fried frailty index (9,23). This measure includes the level of PA as a frailty criterion, and results were due to increased PA behavior while the other criteria of frailty were not modified, suggesting that PA may not influence frailty status (23). Also, the small number of frail individuals may have limited the statistical power. Another randomized trial examined whether a multifactorial intervention which includes physical exercise could reduce frailty in participants who met the Fried frailty criteria and showed that the 12-month intervention reduced frailty by 14.7% (24). Because physical exercise was a component of the intervention, it was not possible to isolate its specific effect from the other components. In the current study, our analysis used the SOF frailty index, which does not include PA as a criterion. We showed that the PA program was not associated with a reduced risk of frailty, but did have a beneficial effect on the chair rise criterion of the SOF index, by reducing the proportion of participants unable to get up from a chair five times without arms. Although we observed a robust reduction in the incidence of frailty at 6-month and in the number of frailty criteria, the effect of intervention on frailty prevalence was not consistent across different analytic approaches.
The interactions between frailty status and randomization arm were not significant for MMD outcome, suggesting that the effect of intervention was not different according to the frailty status. The results suggest the potential value of engaging frail individuals in such structured PA programs, given the important benefit they may gather. The effect of exercise on disability among frail individuals was examined by Daniels et al. (12) in a systematic review including studies where participants had at least one physical frailty indicator but not based on a validated frailty definition (25). Their results suggested that PA may reduce disability, but were not confirmed by another meta-analysis using a more stringent definition of frailty (11). The main problem of previous studies in the field was the heterogeneity of the definition of frailty applied, often not based on validated criteria.
The study has important strengths, including i) a large sample with well-defined frailty status who have typically been excluded from randomized trials of physical activity, ii) extended intervention and follow-up periods, and iii) a high retention rate. However, our findings should be interpreted in light of several limitations. First, the inclusion criteria of the LIFE study may limit the generalizability of the findings. Second, even if each frailty criterion was evaluated, no information was available for weight loss at baseline and this item was replaced by the loss of appetite. Third, this was a secondary analysis not prespecified in the protocol and frailty status was not an entry criterion nor a randomization stratum. Thus, findings should be confirmed in other studies. Finally, we were unable to determine which of the components of the PA intervention were instrumental to the reduction in frailty status.
In conclusion, a structured, moderate-intensity PA program compared with a HE program was not associated with a reduction in overall frailty status over two years among sedentary community-dwelling older adults. On the other hand, the beneficial effect on incidence and persistence of MMD was not altered by frailty status. These findings highlight the feasibility and the importance of effective long-term community-based PA programs for frail and non-frail older adults.
Acknowledgments
GRANT SUPPORT
The Lifestyle Interventions and Independence for Elders Study was funded by cooperative agreement UO1AG22376 from the National Institutes of Health (NIH) and National Institute on Aging; supplement 3U01AG022376-05A2S from the National Heart, Lung, and Blood Institute; and was sponsored in part by the Intramural Research Program. The research is partially supported by the Claude D. Pepper Older Americans Independence Centers at the University of Florida (1 P30 AG028740), Wake Forest University (1 P30 AG21332),Tufts University (1P30AG031679), University of Pittsburgh (P30 AG024827), and Yale University (P30AG021342) and the NIH/NCRR CTSA at Stanford University (UL1 RR025744), at University of Florida (U54RR025208) and at Yale University (UL1 TR000142). Tufts University is also supported by the Boston Rehabilitation Outcomes Center (1R24HD065688-01A1). LIFE investigators are also partially supported by the following: Dr Thomas Gill (Yale University) is the recipient of an Academic Leadership Award (K07AG3587) from the National Institute on Aging. Dr Carlos Fragoso (Spirometry Reading Center, Yale University) is the recipient of a Career Development Award from the Department of Veterans Affairs. Dr Roger Fielding (Tufts University) is partially supported by the US Department of Agriculture, under agreement #58-1950-4-003.
Dr. Fielding reports receiving grant funding from the Dairy Research Institute, Department of Defense, Nestle, Regeneron Pharmaceuticals, Biophytis, Amazentis, and Unilever; consulting for Dairy Management, Eli Lilly, Essentient, Merck, Nestle, and Regeneron; and serving on the board for Aging in Motion, Ammonett, Cytokinetics, Myosyntax, Nestle, and Segterra.
Appendix: Research Investigators for the LIFE Study
Administrative Coordinating Center, University of Florida, Gainesville, FL: Marco Pahor, MD—Principal Investigator of the LIFE Study; Jack M. Guralnik, MD, PhD—Coinvestigator of the LIFE Study (University of Maryland School of Medicine, Baltimore, MD); Christiaan Leeuwenburgh, PhD; Connie Caudle; Lauren Crump, MPH; Latonia Holmes; Jocelyn Lee, PhD; Ching-ju Lu, MPH. Data Management, Analysis and Quality Control Center: Wake Forest University, Winston Salem, NC: Michael E. Miller, PhD—DMAQC Principal Investigator; Mark A. Espeland, PhD—DMAQC Coinvestigator; Walter T. Ambrosius, PhD; William Applegate, MD; Daniel P. Beavers, PhD, MS; Robert P. Byington, PhD, MPH, FAHA; Delilah Cook, CCRP; Curt D. Furberg, MD, PhD; Lea N. Harvin, BS; Leora Henkin, MPH, Med; John Hepler, MA; Fang-Chi Hsu, PhD; Laura Lovato, MS; Wesley Roberson, BSBA; Julia Rushing, BSPH, MStat; Scott Rushing, BS; Cynthia L. Stowe, MPM; Michael P. Walkup, MS; Don Hire, BS; W. Jack Rejeski, PhD; Jeffrey A. Katula, PhD, MA; Peter H. Brubaker, PhD; Shannon L. Mihalko, PhD; Janine M. Jennings, PhD; National Institutes of Health, Bethesda, MD; Evan C. Hadley, MD (National Institute on Aging); Sergei Romashkan, MD, PhD (National Institute on Aging); Kushang V. Patel, PhD (National Institute on Aging); National Heart, Lung and Blood Institute, Bethesda, MD; Denise Bonds, MD, MPH. Field Centers: Northwestern University, Chicago, IL: Mary M. McDermott, MD—Field Center Principal Investigator; Bonnie Spring, PhD—Field Center Coinvestigator; Joshua Hauser, MD—Field Center Coinvestigator; Diana Kerwin, MD—Field Center Coinvestigator; Kathryn Domanchuk, BS; Rex Graff, MS; Alvito Rego, MA. Pennington Biomedical Research Center, Baton Rouge, LA: Timothy S. Church, MD, PhD, MPH—Field Center Principal Investigator; Steven N. Blair, PED (University of South Carolina); Valerie H. Myers, PhD; Ron Monce, PA-C; Nathan E. Britt, NP; Melissa Nauta Harris, BS; Ami Parks McGucken, MPA, BS; Ruben Rodarte, MBA, MS, BS; Heidi K. Millet, MPA, BS; Catrine Tudor-Locke, PhD, FACSM; Ben P. Butitta, BS; Sheletta G. Donatto, MS, RD, LDN, CDE; Shannon H. Cocreham, BS; Stanford University, Palo Alto, CA; Abby C. King, PhD—Field Center Principal Investigator; Cynthia M. Castro, PhD; William L. Haskell, PhD; Randall S. Stafford, MD, PhD; Leslie A. Pruitt, PhD; Kathy Berra, MSN, NP-C, FAAN; Veronica Yank, MD; Tufts University, Boston, MA; Roger A. Fielding, PhD—Field Center Principal Investigator; Miriam E. Nelson, PhD—Field Center Coinvestigator; Sara C. Folta, PhD—Field Center Coinvestigator; Edward M. Phillips, MD; Christine K. Liu, MD; Erica C. McDavitt, MS; Kieran F. Reid, PhD, MPH; Won S. Kim, BS; Vince E. Beard, BS; University of Florida, Gainesville, FL; Todd M. Manini, PhD—Field Center Principal Investigator; Marco Pahor, MD—Field Center Coinvestigator; Stephen D. Anton, PhD; Susan Nayfield, MD; Thomas W. Buford, PhD; Michael Marsiske, PhD; Bhanuprasad D. Sandesara, MD; Jeffrey D. Knaggs, BS; Megan S. Lorow, BS; William C. Marena, MT, CCRC; Irina Korytov, MD; Holly L. Morris, MSN, RN, CCRC (Brooks Rehabilitation Clinical Research Center, Jacksonville, FL); Margo Fitch, PT (Brooks Rehabilitation Clinical Research Center, Jacksonville, FL); Floris F. Singletary, MS, CCC-SLP (Brooks Rehabilitation Clinical Research Center, Jacksonville, FL); Jackie Causer, BSH, RN (Brooks Rehabilitation Clinical Research Center, Jacksonville, FL); Katie A. Radcliff, MA (Brooks Rehabilitation Clinical Research Center, Jacksonville, FL); University of Pittsburgh, Pittsburgh, PA; Anne B. Newman, MD, MPH—Field Center Principal Investigator; Stephanie A. Studenski, MD, MPH—Field Center Coinvestigator; Bret H. Goodpaster, PhD; Nancy W. Glynn, PhD; Oscar Lopez, MD; Neelesh K. Nadkarni, MD, PhD; Kathy Williams, RN, BSEd, MHSA; Mark A. Newman, PhD; George Grove, MS; Janet T. Bonk, MPH, RN; Jennifer Rush, MPH; Piera Kost, BA (deceased); Diane G. Ives, MPH; Wake Forest University, Winston Salem, NC; Stephen B. Kritchevsky, PhD—Field Center Principal Investigator; Anthony P. Marsh, PhD—Field Center Coinvestigator; Tina E. Brinkley, PhD; Jamehl S. Demons, MD; Kaycee M. Sink, MD, MAS; Kimberly Kennedy, BA, CCRC; Rachel Shertzer-Skinner, MA, CCRC; Abbie Wrights, MS; Rose Fries, RN, CCRC; Deborah Barr, MA, RHEd, CHES; Yale University, New Haven, CT; Thomas M. Gill, MD—Field Center Principal Investigator; Robert S. Axtell, PhD, FACSM—Field Center Coinvestigator (Southern Connecticut State University, Exercise Science Department); Susan S. Kashaf, MD, MPH (VA Connecticut Healthcare System); Nathalie de Rekeneire, MD, MS; Joanne M. McGloin, MDiv, MS, MBA; Karen C. Wu, RN; Denise M. Shepard, RN, MBA; Barbara Fennelly, MA, RN; Lynne P. Iannone, MS, CCRP; Raeleen Mautner, PhD; Theresa Sweeney Barnett, MS, APRN; Sean N. Halpin, MA; Matthew J. Brennan, MA; Julie A. Bugaj, MS; Maria A. Zenoni, MS; Bridget M. Mignosa, AS. Cognition Coordinating Center, Wake Forest University, Winston Salem, NC; Jeff Williamson, MD, MHS—Center Principal Investigator; Kaycee M Sink, MD, MAS—Center Coinvestigator; Hugh C. Hendrie, MB, ChB, DSc (Indiana University); Stephen R. Rapp, PhD; Joe Verghese, MB, BS (Albert Einstein College of Medicine of Yeshiva University); Nancy Woolard; Mark Espeland, PhD; Janine Jennings, PhD; Electrocardiogram Reading Center, University of Florida, Gainesville, FL; Carl J. Pepine MD, MACC; Mario Ariet, PhD; Eileen Handberg, PhD, ARNP; Daniel Deluca, BS; James Hill, MD, MS, FACC; Anita Szady, MD. Spirometry Reading Center, Yale University, New Haven, CT; Geoffrey L. Chupp, MD; Gail M. Flynn, RCP, CRFT; Thomas M. Gill, MD; John L. Hankinson, PhD (Hankinson Consulting, Inc.); Carlos A. Vaz Fragoso, MD; Cost Effectiveness Analysis Center; Erik J. Groessl, PhD (University of California, San Diego and VA San Diego Healthcare System); Robert M. Kaplan, PhD (Office of Behavioral and Social Sciences Research, National Institutes of Health).
Appendix 1: Supplementary Details of Sensitivity Analyses
Sensitivity analyses were performed using inverse probability weighting (IPW) analysis, transition models, and complier average causal effect (CACE) analysis.
First, to investigate the effect of loss to follow-up, we used the IPW approach. A weight equivalent to the probability of remaining in the study at 24 months was assigned to each participant on the basis of age, gender, race, education, number of chronic diseases, live alone, field center, baseline SPPB and baseline 400-m gait speed and calculated using a logistic regression. For the analysis in Table 2, when both probabilities of randomization (defined as the probability of randomization to the intervention group among non-frail participants and calculated using a logistic regression) and remaining in the study were considered, we could not include both probabilities as weights in the model. Therefore the probability of randomization was treated as a propensity score and adjusted for as a covariate in the model, and the probability of remaining in the study was treated as a weight in the model. Propensity score is defined as the probability of treatment assignment conditional on observed baseline covariates (26). Four propensity score methods are usually used for removing the confounding effects when estimating the treatment effect on outcomes: propensity score matching, stratification on the propensity score, IPW using the propensity score, and covariate adjustment using the propensity score (27). IPW and covariate adjustment were used in this study. The covariates we used to calculate the probability of randomization were the same as those we used to calculate the probability of remaining in the study.
Second, in transition models, we modeled the conditional distribution of the outcome measure at any follow-up visit given the outcome measure at the previous visit, assuming the first-order Markov chain model. Transition models using the generalized estimating equation (GEE) models were used (20) to study the association between SOF frailty and intervention over 24 months. Gender and field center (both used to stratify randomization), intervention, time, time2, time3, and outcome at the previous visit were included in the models. For this regression setting, we model the transition probability as a function of covariates under the special case there are no interactions between the previous outcome measure and the covariates. The interpretation of the intervention effect is slightly different from the effect estimated from the other models that we presented. It is the intervention effect after adjusting for covariates and the previous outcome measure. We did not adjust for the previous outcome measure in the other models. Note that conditioning on the history of previous outcome measure may lead to attenuation of the intervention effect. For number of SOF frailty criteria, ordinal logistic regression with clustered sandwich estimator was used. Same covariates were adjusted in the models.
Third, the CACE analysis was used to take into account the intervention adherence using an instrumental variable approach (21, 22). Assuming the randomization effect on the outcome is mediated by the adherence to the intervention and the same proportion of participants in the groups would not have adhered to the intervention if they had been offered it, randomization was treated as an instrumental variable. The CACE analysis was performed as a longitudinal data analysis. Baseline outcome was retained in the outcome vector. For SOF frailty index, a binary outcome, an instrumental variable probit model (ivprobit command in Stata) was used. Gender, field centers, and continuous time including time, time2, and time3 were adjusted for in the models. A Wald test of the exogeneity of the instrumental variables was provided. Exogeneity is defined as no correlation between covariate (e.g., adherence) and error term. In order for the instrumental variable model to be valid, the covariate needs to be exogenous. Because the covariate can be endogenous (correlated with the error term), we would like to replace the covariate by a “proxy” variable, known as an instrumental variable (e.g., randomization groups), which is independent of the error term. There are two conditions for a valid instrument. The first condition is instrument relevance which defines as the correlation between instrumental variable and covariate is not equal to 0. The second condition is instrument exogeneity. This test is provided in Stata. If the test is significant, we reject the null hypothesis of no endogeneity. If the test is not significant, we do not reject the null hypothesis, so it may not be necessary to use an instrumental variable analysis. For number of SOF frailty criteria, the analysis was explored in its continuous form with estimation using the instrumental variable regression (ivregress command with 2-stage least squares in Stata). The same covariates listed for the binary outcome analysis were included in the model except that time2 and time3 were not adjusted.
To provide more detailed description of the adherence measure used in the CACE analysis, the mean number of intervention sessions attended and due by randomization groups are presented in Appendix Table 2. The median for the attendance percentage after excluding medical leaves throughout the whole follow-up period was 0.71 (average = 0.63) in the physical activity group and the median for the attendance percentage was 0.82 (average = 0.72) in the Health Education group. In CACE analysis, adherence was treated as a binary variable using the median as a cut-off point (≥ median vs. < median).
This is the prepublication, author-produced version of a manuscript accepted for publication in Annals of Internal Medicine. This version does not include post-acceptance editing and formatting. The American College of Physicians, the publisher of Annals of Internal Medicine, is not responsible for the content or presentation of the author-produced accepted version of the manuscript or any version that a third party derives from it. Readers who wish to access the definitive published version of this manuscript and any ancillary material related to this manuscript (e.g., correspondence, corrections, editorials, linked articles) should go to Annals.org or to the print issue in which the article appears. Those who cite this manuscript should cite the published version, as it is the official version of record.
Appendix Figure 1.
Flow of participants through the trial.
*SOF frailty was assessed over 24 months.
†Discontinuation of the intervention was operationalized as participants who did not attend at least 1 intervention session during their last 6-months of follow-up before the last planned follow-up visit date. Deaths and intervention withdrawals are included in these numbers.
SPPB, short physical performance battery; SOF, Study of Osteoporotic Fractures; MMD, major mobility disability.
Appendix Table 1.
SOF Frailty Index and Its Individual Components by Randomization Group and Baseline Frailty Status. Values are Numbers (Percentages) unless Stated Otherwise
| Physical Activity (n=812)
|
Health Education (n=811)
|
|||||
|---|---|---|---|---|---|---|
| Frail (n=159) | Not frail (n=653) | All | Frail (n=160) | Not frail (n=651) | All | |
| Prevalence of SOF frailty index | ||||||
| Baseline | 159 (100%) | 0 (0%) | 159 (19.6%) | 160 (100%) | 0 (0%) | 160 (19.7%) |
| 6 Month | 66 (48.2%) | 69 (11.5%) | 135 (18.3%) | 76 (53.2%) | 99 (16.1%) | 175 (23.0%) |
| 12 Month | 50 (40.0%) | 95 (17.2%) | 145 (21.5%) | 59 (44.4%) | 84 (14.6%) | 143 (20.2%) |
| 24 Month | 40 (35.1%) | 71 (13.2%) | 111 (17.0%) | 50 (40.7%) | 88 (15.9%) | 138 (20.4%) |
| Inability to rise from a chair five times without using the arms | ||||||
| Baseline | 67 (42.1%) | 48 (7.4%) | 115 (14.2%) | 87 (54.4%) | 51 (7.8%) | 138 (17.0%) |
| 6 Month | 47 (32.6%) | 45 (7.3%) | 92 (12.0%) | 57 (38.0%) | 82 (13.0%) | 139 (17.8%) |
| 12 Month | 48 (34.0%) | 57 (9.5%) | 105 (14.2%) | 67 (45.9%) | 79 (12.9%) | 146 (19.2%) |
| 24 Month | 44 (35.5%) | 76 (13.3%) | 120 (17.3%) | 64 (48.5%) | 99 (17.2%) | 163 (23.1%) |
| Weight loss* | ||||||
| Baseline | 111 (69.8%) | 38 (5.8%) | 149 (18.3%) | 112 (70.0%) | 37 (5.7%) | 149 (18.4%) |
| 6 Month | 44 (31.2%) | 172 (28.0%) | 216 (28.6%) | 46 (31.1%) | 150 (23.9%) | 196 (25.3%) |
| 12 Month | 38 (30.2%) | 158 (28.2%) | 196 (28.6%) | 38 (28.1%) | 133 (22.9%) | 171 (23.8%) |
| 24 Month | 21 (16.7%) | 67 (12.0%) | 88 (12.8%) | 23 (17.0%) | 88 (15.1%) | 111 (15.5%) |
| Reduced energy level† | ||||||
| Baseline | 152 (95.6%) | 263 (40.3%) | 415 (51.1%) | 150 (93.8%) | 277 (42.5%) | 427 (52.7%) |
| 6 Month | 139 (95.2%) | 243 (39.5%) | 382 (50.2%) | 139 (93.3%) | 268 (42.7%) | 407 (52.4%) |
| 12 Month | 101 (70.6%) | 273 (44.9%) | 374 (49.8%) | 99 (67.8%) | 279 (44.8%) | 378 (49.2%) |
| 24 Month | 99 (73.9%) | 275 (47.1%) | 374 (52.1%) | 100 (72.5%) | 310 (52.0%) | 410 (55.9%) |
| Mean (SD) number of SOF frailty index criteria | ||||||
| Baseline | 2.08 (0.26) | 0.53 (0.50) | 0.84 (0.77) | 2.18 (0.39) | 0.56 (0.50) | 0.88 (0.80) |
| 6 Month | 1.57 (0.67) | 0.74 (0.66) | 0.89 (0.74) | 1.61 (0.69) | 0.79 (0.73) | 0.95 (0.79) |
| 12 Month | 1.32 (0.78) | 0.81 (0.71) | 0.90 (0.75) | 1.42 (0.77) | 0.80 (0.71) | 0.92 (0.77) |
| 24 Month | 1.28 (0.72) | 0.71 (0.73) | 0.81 (0.76) | 1.37 (0.77) | 0.84 (0.72) | 0.94 (0.76) |
| Number of SOF frailty index criteria | ||||||
| Baseline | ||||||
| 0 | 0 (0%) | 304 (46.6%) | 304 (37.4%) | 0 (0%) | 286 (43.9%) | 286 (35.3%) |
| 1 | 0 (0%) | 349 (53.4%) | 349 (43.0%) | 0 (0%) | 365 (56.1%) | 365 (45.0%) |
| 2 | 147 (92.5%) | 0 (0%) | 147 (18.1%) | 131 (81.9%) | 0 (0%) | 131 (16.2%) |
| 3 | 12 (7.5%) | 0 (0%) | 12 (1.5%) | 29 (18.1%) | 0 (0%) | 29 (3.6%) |
| 6 Month | ||||||
| 0 | 1 (0.7%) | 228 (37.9%) | 229 (31.0%) | 3 (2.1%) | 235 (38.1%) | 238 (31.3%) |
| 1 | 70 (51.1%) | 305 (50.7%) | 375 (50.7%) | 64 (44.8%) | 283 (45.9%) | 347 (45.7%) |
| 2 | 53 (38.7%) | 66 (11.0%) | 119 (16.1%) | 62 (43.4%) | 91 (14.7%) | 153 (20.1%) |
| 3 | 13 (9.5%) | 3 (0.5%) | 16 (2.2%) | 14 (9.8%) | 8 (1.3%) | 22 (2.9%) |
| 12 Month | ||||||
| 0 | 17 (13.6%) | 199 (36.1%) | 216 (32.0%) | 13 (9.8%) | 208 (36.2%) | 221 (31.3%) |
| 1 | 58 (46.4%) | 257 (46.6%) | 315 (46.6%) | 61 (45.9%) | 282 (49.1%) | 343 (48.5%) |
| 2 | 43 (34.4%) | 94 (17.1%) | 137 (20.3%) | 49 (36.8%) | 76 (13.2%) | 125 (17.7%) |
| 3 | 7 (5.6%) | 1 (0.2%) | 8 (1.2%) | 10 (7.5%) | 8 (1.4%) | 18 (2.5%) |
| 24 Month | ||||||
| 0 | 13 (11.4%) | 234 (43.4%) | 247 (37.8%) | 13 (10.6%) | 184 (33.3%) | 197 (29.2%) |
| 1 | 61 (53.5%) | 234 (43.4%) | 295 (45.2%) | 60 (48.8%) | 280 (50.7%) | 340 (50.4%) |
| 2 | 35 (30.7%) | 63 (11.7%) | 98 (15.0%) | 41 (33.3%) | 79 (14.3%) | 120 (17.8%) |
| 3 | 5 (4.4%) | 8 (1.5%) | 13 (2.0%) | 9 (7.3%) | 9 (1.6%) | 18 (2.7%) |
SOF, Study of Osteoporotic Fractures.
The criterion weight loss was based on weight measurements, with the exception of the baseline visit: the criterion was considered as present at baseline if the participant reported a loss of appetite on the health-related quality of life questionnaire.
The reduced energy level criterion was defined by using the following statement of the health-related quality of life questionnaire: “During the past week how often have you felt full of energy”.
Appendix Table 2.
Intervention Sessions Attended
| Physical Activity (n=806)
|
Health Education (n=807)
|
|||||
|---|---|---|---|---|---|---|
| Excluding Sessions Not Attended Due to Medical Leaves
|
Including All Scheduled Sessions
|
|
||||
| Mean Number of Sessions Attended/Mean Number of Sessions Due* | Mean Percentage ± SD (median, Q3)* | Mean Number of Sessions Attended/Mean Number of Sessions Due† | Mean Percentage ± SD (median, Q3)* | Mean Number of Sessions Attended/Mean Number of Sessions Due† | Mean Percentage ± SD (median, Q3)* | |
|
|
|
|
||||
| 6 Month | 31.6/41.1 | 0.76 ± 0.22 (0.82, 0.91) |
31.6/43.9 | 0.72 ± 0.25 (0.81, 0.91) |
17.2/22.3 | 0.77 ± 0.23 (0.84, 0.92) |
| 12 Month | 61.0/83.6 | 0.72 ± 0.23 (0.79, 0.88) |
60.9/92.0 | 0.66 ± 0.27 (0.75, 0.87) |
23.4/30.2 | 0.77 ± 0.23 (0.85, 0.93) |
| 24 Month | 110.6/163.0 | 0.66 ± 0.25 (0.74, 0.85) |
110.5/185.1 | 0.59 ± 0.28 (0.68, 0.82) |
32.4/42.9 | 0.75 ± 0.24 (0.84, 0.91) |
| All | 140.4/219.7 | 0.63 ± 0.27 (0.71, 0.83) |
140.3/254.6 | 0.55 ± 0.29 (0.65, 0.79) |
39.0/53.9 | 0.72 ± 0.25 (0.82, 0.90) |
The calculation excluded medical leaves. Only the physical activity group had a count of medical leaves. Note that the mean number of sessions attended when medical leaves were excluded is not exactly the same as that when all scheduled sessions were included. When the extended leave sessions were excluded, participants who did not contribute any data were excluded. Thus the denominators in the calculations are different.
The calculation used all scheduled sessions.
Appendix Table 3.
Intervention Sessions Attended by Baseline SOF Frailty Status
| Physical Activity (n=806)
|
Health Education (n=807)
|
||||||||
|---|---|---|---|---|---|---|---|---|---|
| Excluding Sessions Not Attended Due to Medical Leaves
|
Including All Scheduled Sessions
|
|
|||||||
| Mean Percentage of Sessions Attended (Median)* | Mean Percentage of Sessions Attended (Median)* | P Value | Mean Percentage of Sessions Attended (Median)† | Mean Percentage of Sessions Attended (Median)† | P Value | Mean Percentage of Sessions Attended (Median)† | Mean Percentage of Sessions Attended (Median)† | P Value | |
|
|
|
|
|||||||
| 6 Month | 0.74 (0.79) | 0.76 (0.83) | 0.23 | 0.69 (0.78) | 0.73 (0.82) | 0.22 | 0.73 (0.81) | 0.79 (0.85) | 0.005 |
| 12 Month | 0.70 (0.75) | 0.72 (0.79) | 0.21 | 0.63 (0.72) | 0.67 (0.75) | 0.142 | 0.73 (0.81) | 0.79 (0.86) | 0.002 |
| 24 Month | 0.64 (0.72) | 0.67 (0.74) | 0.28 | 0.56 (0.64) | 0.60 (0.69) | 0.133 | 0.70 (0.80) | 0.77 (0.85) | <0.001 |
| All | 0.62 (0.70) | 0.64 (0.72) | 0.59 | 0.52 (0.59) | 0.56 (0.65) | 0.27 | 0.66 (0.73) | 0.74 (0.83) | <0.001 |
SOF, Study of Osteoporotic Fractures.
The calculation excluded medical leaves. Only the physical activity group had a count of medical leaves.
The calculation used all scheduled sessions.
Appendix Table 4.
Association between SOF Frailty and Randomization Group over 24 Months in All Study Participants using Transition Models
| Physical Activity vs. Health Education | Average Prevalence Over 24 Months‖ | Unadjusted prevalence difference [95% CI] |
P Value | Adjusted prevalence difference [95% CI] |
P Value | |
|---|---|---|---|---|---|---|
|
|
||||||
| Physical activity | Health education (Reference) |
|||||
| ITT Model* | 19.1% | 20.8% | −0.021 [−0.042 to −0.0003] | 0.047 | −0.020 [−0.041 to 0.0005] | 0.055 |
| IPW Model*† | −0.018 [−0.039 to 0.003] | 0.099 | −0.017 [−0.038 to 0.004] | 0.119 | ||
SOF, Study of Osteoporotic Fractures; ITT, intent-to-treat; IPW, inverse probability weighting.
Gender and field center (both used to stratify randomization), intervention, time, time2, time3, and outcome at the previous visit included in the transition models. The prevalence differences from the generalized estimating equation (GEE) models (logit link, binomial distribution) were presented.
Weighted for the inverse probability of remaining in the study.
Average prevalence over time calculated as the sum of prevalence from each visit, including baseline, in the raw data divided by 4.
Appendix Table 5.
Association between SOF Frailty and Randomization Group over 24 Months in All Study Participants Taking Adherence with Intervention into Account*
| Physical Activity vs. Health Educ | Average marginal effect‖ [95% CI] |
P Value | P value for test of exogeneity |
|---|---|---|---|
|
|
|||
| Excluding Sessions Not Attended Due to Medical Leaves† | |||
| CACE§ | −0.60 (−0.70 to −0.51) | <0.001 | 0.411 |
| CACE+ IPW§¥ | −0.59 (−0.78 to −0.39) | <0.001 | 0.416 |
| Including All Scheduled Sessions‡ | |||
| CACE§ | −0.57 (−0.75 to −0.38) | <0.001 | 0.229 |
| CACE+ IPW§¥ | −0.54 (−0.89 to −0.19) | 0.028 | 0.352 |
CACE = complier average causal effect; IPW, inverse probability weighting.
Take adherence to intervention into account. Adherence to intervention is defined as the attendance percentage larger or equal to median (0.71 in physical activity after excluding medical leaves, 0.65 in physical activity while including all sessions, and 0.82 in health education). Non-adherence to intervention is defined as the attendance percentage smaller than median.
Adherence percentage calculated based on scheduled sessions excluding medical leaves.
Adherence percentage calculated based on all scheduled sessions without excluding medical leaves.
Gender and field center (both used to stratify randomization), intervention, time, time2, and time3 included in the models. Baseline outcome was in the outcome vector.
Weighted for the inverse probability of remaining in the study.
Average marginal effect of the difference of probability of SOF frailty.
Appendix Table 6.
Association between Each SOF Frailty Criterion and Randomization Group over 24 Months in All Study Participants
| Physical Activity vs. Health Education | Unadjusted prevalence difference [95% CI]‡ | P Value | Adjusted prevalence difference [95% CI]‡ | P Value |
|---|---|---|---|---|
|
|
||||
| Reduced energy level | ||||
| ITT Model* | −0.018 [−0.058 to 0.021] | 0.37 | −0.018 [−0.057 to 0.021] | 0.36 |
| IPW Model*† | −0.012 [−0.052 to 0.028] | 0.56 | −0.012 [−0.051 to 0.028] | 0.56 |
| Weight loss | ||||
| ITT Model* | 0.011 −0.016 to 0.037] | 0.43 | 0.013 [−0.012 to 0.039] | 0.30 |
| IPW Model*† | 0.010 [−0.016 to 0.037] | 0.45 | 0.013 [−0.014 to 0.040] | 0.35 |
| Inability to rise from a chair five times without using the arms | ||||
| ITT Model* | −-0.050 [−0.081 to −0.020] | 0.001 | −0.050 [−0.081 to −0.020] | 0.001 |
| IPW Model*† | −0.047 [−0.080 to −0.015] | 0.004 | −0.048 [−0.079 to −0.016] | 0.003 |
SOF, Study of Osteoporotic Fractures; ITT, intent-to-treat; IPW, inverse probability weighting.
Gender and field center (both used to stratify randomization), intervention, time, time2, and time3 included in the generalized estimating equation (GEE) models (logit link, binomial distribution). Baseline outcome was retained in the outcome vector.
Weighted for the inverse probability of remaining in the study.
Marginal standardization was used to obtain the prevalence difference and its 95% CI.
Appendix Table 7.
Association between the Number of SOF Frailty Criteria and Randomization Group over 24 Months in All Study Participants using Transition Models
| Physical Activity vs. Health Education | Unadjusted OR [95% CI] |
P Value | Adjusted OR [95% CI] |
P Value |
|---|---|---|---|---|
|
|
||||
| ITT Model* | 0.88 [0.78 to 0.98] | 0.021 | 0.88 [0.78 to 0.98] | 0.019 |
| IPW Model*† | 0.89 [0.79 to 0.99] | 0.038 | 0.89 [0.79 to 0.99] | 0.038 |
SOF, Study of Osteoporotic Fractures; ITT, intent-to-treat; IPW, inverse probability weighting.
Gender and field center (both used to stratify randomization), intervention, time, time2, time3, and outcome at the previous visit included in the transition models. The ORs from ordinal logistic regression models with clustered sandwich estimators were presented.
Weighted for the inverse probability of remaining in the study.
CURRENT MAILING ADDRESSES
Andrea Trombetti, MD, Division of Bone Diseases, Department of Internal Medicine Specialties, Geneva University Hospitals and Faculty of Medicine, Rue Gabrielle-Perret-Gentil 4, CH-1211 Geneva 14, Switzerland;
Mélany Hars, PhD, Division of Bone Diseases, Department of Internal Medicine Specialties, Geneva University Hospitals and Faculty of Medicine, Rue Gabrielle-Perret-Gentil 4, CH-1211 Geneva 14, Switzerland;
Fang-Chi Hsu, PhD, Department of Biostatistical Sciences, Wake Forest School of Medicine, Medical Center Blvd., Winston-Salem, NC 27157;
Kieran F. Reid, PhD, Nutrition, Exercise Physiology, and Sarcopenia Laboratory, Jean Mayer USDA Human Nutrition Research Center on Aging, Tufts University, 711 Washington Street, Boston, MA 02111;
Timothy S. Church, MD, PhD, 6400 Perkins Road, Baton Rouge, LA 75080;
Thomas M. Gill, MD, Yale School of Medicine, Adler Geriatric Center, 874 Howard Avenue, New Haven, CT 06510;
Abby C. King, PhD, Stanford University School of Medicine, 259 Campus Drive, HRP Redwood Building, Rm T221, Stanford, CA 94305-5405;
Christine K. Liu, MD, Boston University School of Medicine, 88 East Newton Street, Robinson 2, Boston, MA 02118;
Todd M. Manini, PhD, Institute on Aging, Department of Aging and Geriatric Research, PO Box 112610, Gainesville, FL 32611-0107;
Mary M. McDermott, MD, Department of Medicine and Preventive Medicine, Northwestern University, Feinberg School of Medicine, Galter Room 18-200, 675 N Saint Clair, Chicago, IL 60611;
Anne B. Newman, MD, Graduate School of Public Health, University of Pittsburgh, A527 Crabtree Hall, 130 DeSoto Street, Pittsburgh, PA 15261;
W. Jack Rejeski, PhD, Department of Health & Exercise Science, Box 7868, Wake Forest University, Winston-Salem, NC 27109;
Jack M. Guralnik, MD, PhD, University of Maryland School of Medicine,Department of Epidemiology and Public Health,Division of Gerontology, 660 West Redwood Street, Room 204, Baltimore, Maryland 21201 Marco Pahor, MD, Department of Aging and Geriatric Research, College of Medicine, University of Florida, 2004 Mowry Road, Gainesville, FL 32611;
Roger A. Fielding, PhD, Nutrition, Exercise Physiology, and Sarcopenia Laboratory, Jean Mayer USDA Human Nutrition Research Center on Aging, Tufts University, 711 Washington Street, Boston, MA 02111
AUTHOR CONTRIBUTIONS
Conception and design: A. Trombetti, M. Hars, F.C. Hsu, K.F. Reid, T.S. Church, T.M. Gill, A.C. King, C.K. Liu, T.M. Manini, M.M. Mac Dermott, A.B. Newman, W.J. Rejeski, J.M. Guralnik, M. Pahor, R.A. Fielding. Analysis and interpretation of the data: A. Trombetti, M.Hars, F.C. Hsu, K.F. Reid, T.S. Church, T.M. Gill, A.C. King, C.K. Liu, T.M. Manini, M.M. Mac Dermott, A.B. Newman, W.J. Rejeski, J.M. Guralnik, M. Pahor, R.A. Fielding.
Drafting of the article: A. Trombetti, M. Hars, K.F. Reid, R.A. Fielding.
Critical revision of the article for important intellectual content: A. Trombetti, M. Hars, F.C. Hsu, K.F. Reid,
T.S. Church, T.M. Gill, A.C. King, C.K. Liu, T.M. Manini, M.M. Mac Dermott, A.B. Newman, W.J. Rejeski, J.M. Guralnik, M. Pahor, R.A. Fielding.
Final approval of the article: A. Trombetti, M. Hars, F.C. Hsu, K.F. Reid, T.S. Church, T.M. Gill, A.C. King, C.K. Liu, T.M. Manini, M.M. Mac Dermott, A.B. Newman, W.J. Rejeski, J.M. Guralnik, M. Pahor, R.A. Fielding.
Provision of study materials or patients: K.F. Reid, T.S. Church, T.M. Gill, A.C. King, C.K. Liu, T.M. Manini, M.M. Mac Dermott, A.B. Newman, W.J. Rejeski, J.M. Guralnik, M. Pahor, R.A. Fielding. Statistical expertise: F.C. Hsu.
Administrative, technical or logistic support: K.F. Reid, T.S. Church, T.M. Gill, A.C. King, C.K. Liu, T.M. Manini, M.M. Mac Dermott, A.B. Newman, W.J. Rejeski, J.M. Guralnik, M. Pahor, R.A. Fielding. Collection and assembly of data: A. Trombetti, M. Hars, F.C. Hsu, K.F. Reid, R.A. Fielding. The authors had full access to all the study data, take responsibility for the accuracy or integrity of the analysis, and had authority over manuscript preparation and the decision to submit the manuscript for publication. All authors approved the manuscript and agree to adhere to all terms outlined in Annals of Internal Medicine information for authors including terms for copyright.
DISCLOSURES
Authors not named here have disclosed no conflicts of interest.
REPRODUCIBLE RESEARCH STATEMENT
Study protocol: Available at www.thelifestudy.org/public/index.cfm. Statistical code and data set: Available to approved persons through agreement with the LIFE study steering committee (e-mail, Roger.Fielding@tufts.edu).
Footnotes
This is the prepublication, author-produced version of a manuscript accepted for publication in Annals of Internal Medicine. This version does not include post-acceptance editing and formatting. The American College of Physicians, the publisher of Annals of Internal Medicine, is not responsible for the content or presentation of the author-produced accepted version of the manuscript or any version that a third party derives from it. Readers who wish to access the definitive published version of this manuscript and any ancillary material related to this manuscript (e.g., correspondence, corrections, editorials, linked articles) should go to Annals.org or to the print issue in which the article appears. Those who cite this manuscript should cite the published version, as it is the official version of record.
DISCLAIMER
The content of this article is solely the responsibility of the authors and do not necessarily represent the official view of the US Department of Agriculture.
ETHICAL APPROVAL
The LIFE trial was overseen by ethics committees at all eights participating institutions, by the coordinating center, and by a data and safety monitoring board. Each institution obtained human subjects committee approval, and informed consent was given by all participants.
References
- 1.Chang SF, Lin PL. Frail phenotype and mortality prediction: A systematic review and metaanalysis of prospective cohort studies. Int J Nurs Stud. 2015;52(8):1362–1374. doi: 10.1016/j.ijnurstu.2015.04.005. [DOI] [PubMed] [Google Scholar]
- 2.Ensrud KE, Ewing SK, Taylor BC, et al. Comparison of 2 frailty indexes for prediction of falls, disability, fractures, and death in older women. Arch Intern Med. 2008;168(4):382–9. doi: 10.1001/archinternmed.2007.113. [DOI] [PubMed] [Google Scholar]
- 3.Boyd CM, Xue QL, Simpson CF, Guralnik JM, Fried LP. Frailty, hospitalization, and progression of disability in a cohort of disabled older women. Am J Med. 2005;118(11):1225–31. doi: 10.1016/j.amjmed.2005.01.062. [DOI] [PubMed] [Google Scholar]
- 4.Bandeen-Roche K, Xue QL, Ferrucci L, et al. Phenotype of frailty: characterization in the women’s health and aging studies. J Gerontol A Biol Sci Med Sci. 2006;61(3):262–6. doi: 10.1093/gerona/61.3.262. [DOI] [PubMed] [Google Scholar]
- 5.Woods NF, LaCroix AZ, Gray SL, et al. Frailty: emergence and consequences in women aged 65 and older in the Women’s Health Initiative Observational Study. J Am Geriatr Soc. 2005;53(8):1321–30. doi: 10.1111/j.1532-5415.2005.53405.x. [DOI] [PubMed] [Google Scholar]
- 6.Bilotta C, Nicolini P, Case A, Pina G, Rossi S, Vergani C. Frailty syndrome diagnosed according to the Study of Osteoporotic Fractures (SOF) criteria and adverse health outcomes among community-dwelling older outpatients in Italy. A one-year prospective cohort study. Arch Gerontol Geriatr. 2012;54(2):e23–8. doi: 10.1016/j.archger.2011.06.037. [DOI] [PubMed] [Google Scholar]
- 7.Clegg A, Young J, Iliffe S, Rikkert MO, Rockwood K. Frailty in elderly people. Lancet. 2013;381(9868):752–62. doi: 10.1016/S0140-6736(12)62167-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Morley JE, Vellas B, van Kan GA, et al. Frailty consensus: a call to action. J Am Med Dir Assoc. 2013;14(6):392–7. doi: 10.1016/j.jamda.2013.03.022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.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: 10.1093/gerona/56.3.m146. [DOI] [PubMed] [Google Scholar]
- 10.Xue QL. The frailty syndrome: definition and natural history. Clin Geriatr Med. 2011;27(1):1–15. doi: 10.1016/j.cger.2010.08.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Gine-Garriga M, Roque-Figuls M, Coll-Planas L, Sitja-Rabert M, Salva A. Physical exercise interventions for improving performance-based measures of physical function in community-dwelling, frail older adults: a systematic review and meta-analysis. Arch Phys Med Rehabil. 2014;95(4):753–769 e3. doi: 10.1016/j.apmr.2013.11.007. [DOI] [PubMed] [Google Scholar]
- 12.Daniels R, van Rossum E, de Witte L, Kempen GI, van den Heuvel W. Interventions to prevent disability in frail community-dwelling elderly: a systematic review. BMC Health Serv Res. 2008;8:278. doi: 10.1186/1472-6963-8-278. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.de Labra C, Guimaraes-Pinheiro C, Maseda A, Lorenzo T, Millan-Calenti JC. Effects of physical exercise interventions in frail older adults: a systematic review of randomized controlled trials. BMC Geriatr. 2015;15:154. doi: 10.1186/s12877-015-0155-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Pahor M, Guralnik JM, Ambrosius WT, et al. Effect of structured physical activity on prevention of major mobility disability in older adults: the LIFE study randomized clinical trial. JAMA. 2014;311(23):2387–96. doi: 10.1001/jama.2014.5616. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Fielding RA, Rejeski WJ, Blair S, et al. The Lifestyle Interventions and Independence for Elders Study: design and methods. J Gerontol A Biol Sci Med Sci. 2011;66(11):1226–37. doi: 10.1093/gerona/glr123. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Marsh AP, Lovato LC, Glynn NW, et al. Lifestyle interventions and independence for elders study: recruitment and baseline characteristics. J Gerontol A Biol Sci Med Sci. 2013;68(12):1549–58. doi: 10.1093/gerona/glt064. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Guralnik JM, Ferrucci L, Simonsick EM, Salive ME, Wallace RB. Lower-extremity function in persons over the age of 70 years as a predictor of subsequent disability. N Engl J Med. 1995;332(9):556–61. doi: 10.1056/NEJM199503023320902. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Muller CJ, MacLehose RF. Estimating predicted probabilities from logistic regression: different methods correspond to different target populations. Int J Epidemiol. 2014;43(3):962–70. doi: 10.1093/ije/dyu029. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Hernan MA, Hernandez-Diaz S, Robins JM. Randomized trials analyzed as observational studies. Ann Intern Med. 2013;159(8):560–2. doi: 10.7326/0003-4819-159-8-201310150-00709. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Diggle PJ, Heagerty P, Liang KY, Zeger SL, et al., editors. Analysis of longitudinal data. 2nd. Oxford: Oxford University Press; 2002. [Google Scholar]
- 21.Shrier I, Steele RJ, Verhagen E, Herbert R, Riddell CA, Kaufman JS. Beyond intention to treat: what is the right question? Clin Trials. 2014;11:28–37. doi: 10.1177/1740774513504151. [DOI] [PubMed] [Google Scholar]
- 22.Ten Have TR, Normand SL, Marcus SM, Brown CH, Lavori P, Duan N. Intent-to-treat vs. non-intent-to-treat analyses under treatment non-adherence in mental health randomized trials. Psychiatr Ann. 2008;38(12):772–783. doi: 10.3928/00485713-20081201-10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Cesari M, Vellas B, Hsu FC, et al. A physical activity intervention to treat the frailty syndrome in older persons - Results from the LIFE-P Study. J Gerontol A Biol Sci Med Sci. 2015;70(2):216–22. doi: 10.1093/gerona/glu099. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Cameron ID, Fairhall N, Langron C, et al. A multifactorial interdisciplinary intervention reduces frailty in older people: randomized trial. BMC medicine. 2013;11:65. doi: 10.1186/1741-7015-11-65. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Ferrucci L, Guralnik JM, Studenski S, Fried LP, Cutler GB, Jr, Walston JD. Designing randomized, controlled trials aimed at preventing or delaying functional decline and disability in frail, older persons: a consensus report. J Am Geriatr Soc. 2004;52(4):625–34. doi: 10.1111/j.1532-5415.2004.52174.x. [DOI] [PubMed] [Google Scholar]
- 26.Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika. 1983;70(1):41–55. [Google Scholar]
- 27.Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res. 2011;46(3):399–424. doi: 10.1080/00273171.2011.568786. [DOI] [PMC free article] [PubMed] [Google Scholar]