Study design and participants
A two-arm, multicentre assessor-blind stratified randomised controlled trial among pre-frail and frail community-dwelling older adults was conducted between March 2019 and February 2020 in four senior activity centres located within the residential estates in Singapore (ClincalTrials.gov registration NCT04438876). To be eligible for the study, participants had to (a) be aged 55 years and older, (b) have low muscle strength, (c) ambulate without human assistance and have no other physical limitations to participation and adherence to exercise, and (d) be able to understand basic instructions. Low muscle strength was defined as handgrip strength (HGS) less than 26kg and 18kg in men and women respectively according to the Asian Working Group for Sarcopenia (AWGS) 2014 consensus (18). Participants were excluded if they (a) were currently enrolled in another study, (b) had any acute musculoskeletal injury or other contraindication to exercise, (c) were unable to participate in the full duration of the study, (d) were unwilling to participate if not assigned to the intervention group, and (e) were deemed not suitable to participate in exercise by a medical doctor. After initial eligibility screening by senior activity centre staff or research coordinator, a doctor examined each subject for medical clearance prior to exercise participation and excluded subjects who did not meet the medical criteria. Participants were then randomly allocated to either the control group (CG) or the intervention group (IG) with a 1:1 allocation ratio based on computerised block randomisation, stratified by their HGS. Ethical approval was obtained from the National Healthcare Group DSRB (2018/00593). All participants signed informed consent prior to participation in the study.
Intervention group
The IG underwent a 12-week structured FPT program, conducted by an exercise-science qualified trainer from a community service provider (Empower Ageing Limited; Singapore; http://www.empower.org.sg). Hourly sessions were held twice weekly at local senior activity centres. The exercise intervention comprised progressive power and balance exercises that targeted both upper and lower body muscles. The list of specific exercises for each respective session is presented in Table 1 and detailed elsewhere (19). Participants performed three sets of each power exercise, with 10-20 repetitions per set, and 12-60 repetitions per set for balance exercises. For the power training, body weight and/or resistance bands were used as resistance and participants were instructed to move as fast as they can during the concentric phase and slowly during the eccentric phase of the exercise movements. To ensure safety, blood pressure of participants was measured before and after each session with an automated sphygmomanometer (Omron HEM-7121, Omron, Kyoto, Japan). Arterial blood oxygen saturation and heart rate were measured at start, mid-point, and end of each exercise session with pulse oximetry (MD300C63, ChoiceMMed, Bristol, Pennsylvania, USA) to monitor exercise intensity levels. Participant with either (a) blood oxygen levels below 95% saturation; (b) high resting heart rate (>=90 beats per minute); (c) abnormal resting blood pressure (systolic blood pressure>=130 mmHg or diastolic blood pressure >=80 mmHg); (d) giddiness or (e) any form of discomfort did not initiate or continue with training session. Participants also rated their perceived exertion after each exercise set and were encouraged to alert the trainer if they felt any discomfort during the training. We monitored and recorded adverse events throughout the intervention program. In addition, participants also received monthly health education talks on nutrition and cognition, conducted by staff from Geriatric Education and Research Institute.
Control group
Participants in the CG could continue with the available exercise program at the respective senior activity centres. Senior activity centres typically had chairs arranged for participants who followed after daily video of stretching, aerobic and balance exercises produced by the Health Promotion Board (20). Participants then had snacks and socialise over group games provided at the centres. The CG was given an exercise manual with the list of exercises in the intervention programme. CG participants were also encouraged to attend the health education talks. Adherence of CG participants to centre activity was not monitored.
Measurements
Objective physical outcome assessments were conducted at baseline and 3-month follow-up by assessors who were blinded to participants’ group allocation.
Physical function assessment
Physical function assessments included HGS (21), knee extensor strength (KES) (22), timed up and go (TUG) (23), and the Short Physical Performance Battery (SPPB) (24). HGS was measured using the Jamar Plus+ Digital Hand Dynamometer (Patterson Medical, Evergreen Boulevard, Cedarburg, USA). KES was measured using a spring gauge strapped 10 cm above the ankle joint, and the highest of four readings (two trials per leg) recorded. For TUG, participants stood up from a chair, walked three metres and back, then sat back down. The test was performed twice, and the average time recorded. SPPB comprised three components: balance, gait speed and repeated chair stands. A composite score of 0–12 points was calculated (25), whereby higher scores indicate better functional performance.
Frailty status assessment
The frailty status of participants was assessed using Cardiovascular Health Study (CHS) Frailty Criteria, which characterises frailty based on five components: weakness, unintentional weight loss, slowness, exhaustion and low physical activity (17). Weakness was identified according the AWGS 2014 criteria as HGS less than 26kg and 18kg for men and women respectively (18). Unintentional weight loss was defined by either BMI less than 18.5kg/m2 or self-reported unintentional weight loss of at least 4.5 kg (10 pounds) in the last six months. Slowness was determined by 6 m walking speed with specified cut-offs based on gender and height. Exhaustion was self-reported through a 3-item questionnaire adapted from the SF-12 survey (26). The Longitudinal Aging Study of Amsterdam Physical Activity Questionnaire was administered to assess participants’ physical activity levels (27). Low physical activity was defined as energy expenditure less than 383 kcal/week and 270 kcal/week for men and women respectively. One point was given for presence of each component, and frailty status classification was defined as robust (0), pre-frail (1-2), and frail (3-5) (17).
Other measures
All participants answered a questionnaire on their baseline demographic information, smoking history, medical conditions, and comorbidities. Anthropometric measurements such as height, weight, body mass index (BMI) and weight circumference were taken during the baseline assessment session. The Mini-Mental State Examination was administered to assess cognitive function in participants. It was scored out of 30, with higher scores indicating better cognitive function (28).
Evaluation of program implementation
The implementation of the FPT intervention was evaluated using the RE-AIM framework (29), a model designed to appraise public health interventions. Present study employed four of the five dimensions specified in the framework: reach, effectiveness, adoption, and implementation. First, “reach” was calculated as the percentage of eligible participants who enrolled in the study. Second, “effectiveness” was assessed based on objective outcome assessments stated above, program attrition rate and participant experience. Participant experience was reported post-intervention in the IG only using a 9-item questionnaire administered by the research team. Participants responded by indicating the extent of agreement with the questionnaire items on a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). Third, “adoption” was determined as proportion of senior activity centres approached that agreed to implement the intervention. Fourth, “implementation” was assessed based on fidelity of intervention delivery at both provider and participant levels. At the provider-level, the extent of deviation between the trainer’s delivery and the intended exercise program was recorded by a research coordinator who observed every training session and recorded attendance. At the participant-level, implementation was evaluated by program attendance rate and proportion of the exercise program completed by participants. The intervention was deemed to have good adherence if a mean attendance rate of at least 80% was achieved. The dimension of “maintenance” was excluded as the present study was not designed to determine the longer term effects of the intervention. We monitored and recorded adverse events throughout the course of the program.
Sample size calculation
Based on a priori power analysis (G*Power 3.1.9.3) using a power of 0.90 and error probability of 0.05, a sample size of 50 particpants is required for each group to detect an assumed 20% difference in SPPB between IG and CG. In addition, with an assumption of 15% dropout rate, a sample size of 120 participants was initially targeted.
Statistical analysis
Independent samples t-tests and chi-squared tests were employed to examine differences in baseline measures between CG and IG for continuous and categorical variables respectively. Levene’s test was conducted to ensure no violation of equal variance assumption. Linear mixed-effect modelling was performed to examine changes in physical function and frailty status between baseline and 3-month follow-up across both groups. Primary independent variables in each model included treatment groups, time, as well as group x time interaction. The models also included random intercepts to account for correlations between repeated measures for each participant. Data analyses were performed based on the intention-to-treat (ITT) principle (30), and the maximum likelihood method was employed to impute missing values. Post-hoc pairwise comparisons were conducted to examine the main effect of time in respective groups. Statistical significance level was set at 0.05, and all analyses were performed using Statistical Package for the Social Science (SPSS Version 20.0, Chicago, IL). Threshold values were selected as standardised effect size (ES) of mean differences and deemed as 0.2, 0.6, 1.2 and 2.0 for small, moderate, large and very large, respectively (31).