Sustained and attenuated effects of transitioning from supervised to self-managed resistance training in older adults with type 2 diabetes and sarcopenia: a randomized controlled trial

Abstract

Aims

This randomized controlled trial examined whether benefits achieved during supervised resistance training (SRT) were sustained after transitioning to self-managed home-based training in older adults with type 2 diabetes mellitus (T2DM) and sarcopenia.

Methods

In this single-center, parallel-group trial, adults aged ≥ 65 years with T2DM and sarcopenia were randomized 1:1 to a 24-week resistance-training intervention or control. The intervention comprised 12 weeks of SRT followed by 12 weeks of self-managed home-based training. Co-primary outcomes were skeletal muscle index (SMI), handgrip strength (HGS), five-times-sit-to-stand (5TSTS), and SARC-CalF score. Secondary outcomes were HbA_1c and SF-12 physical and mental component summary scores. Longitudinal changes were evaluated using generalized estimating equations.

Results

Sixty-two participants were randomized (median age, 72 years; 67.7% female). At 12 weeks, SRT significantly improved HGS, 5TSTS performance, and SARC-CalF score, whereas SMI showed no significant between-group difference. At 24 weeks, SMI showed a modest favorable change, but earlier improvements in HGS, 5TSTS performance, and SARC-CalF score were attenuated after transition to self-managed training. SF-12 scores showed supportive, favorable patterns through 24 weeks, whereas HbA_1c showed no clear intervention-related change.

Conclusions

In older adults with T2DM and sarcopenia, transition from SRT to self-managed home-based training produced outcome-specific effects. Functional gains diminished after supervision ended, whereas SMI showed a modest, delayed, favorable change. These findings suggest that hybrid support models incorporating periodic supervision, remote monitoring, automated reminders, and feedback-guided exercise logs may be needed to sustain adherence, exercise quality, and functional capacity.

Trial registration

Retrospectively registered at ClinicalTrials.gov (NCT07398495), February 23, 2026.

Introduction

Sarcopenia is an age-related syndrome characterized by progressive declines in muscle mass, strength, and physical function, and is associated with frailty, disability, hospitalization, mortality, and reduced quality of life in older adults [1–3]. Its clinical importance is further amplified in individuals with type 2 diabetes mellitus (T2DM), particularly in Asian populations with relatively lower baseline muscle mass [4]. Meta-analytic evidence suggests that approximately 17% of older adults with T2DM have sarcopenia [5]. Poor glycemic control, defined as glycated hemoglobin (HbA_1c) ≥ 8.0% (≥ 64 mmol/mol), may accelerate muscle loss through metabolic and inflammatory pathways [6, 7], whereas sarcopenia, in turn, may worsen insulin resistance and diabetes-related complications, thereby increasing the risk of functional decline, hospitalization, and mortality [8, 9]. Addressing this bidirectional relationship is therefore important for preserving muscle health and metabolic stability in older adults with T2DM. From a clinical perspective, preventing decline in muscle strength and physical function is particularly important because these outcomes are closely related to the ability to rise from a chair, mobility, fall risk, independence, and overall quality of life.

Resistance training (RT) is widely recognized as an effective strategy for preventing or attenuating sarcopenia in this population. Among available modalities, elastic-band RT is particularly suitable for older adults with T2DM because it is safe, practical, and easily implemented in both clinical and community settings [10, 11]. According to the frequency, intensity, time, and type (FITT) principle, effective RT programs for older adults generally involve 3–5 sessions per week, moderate and progressively increased intensity, session durations of approximately 20–60 min, and functional multi-joint exercises typically performed in at least three sets [12, 13]. Previous studies further suggest that improvements in balance may become evident after at least 8 weeks of training, whereas meaningful gains in lower-extremity strength and functional mobility usually require at least 12 weeks of consistent participation [11]. However, these functional adaptations may depend more on continued training intensity, progression, movement quality, and adherence than on changes in muscle mass alone.

Beyond exercise prescription, the mode of delivery is a critical determinant of intervention effectiveness and long-term sustainability. Supervised resistance training (SRT), delivered under professional guidance, may improve exercise fidelity and facilitate appropriate progression of training load. In individuals with T2DM, supervised exercise interventions have been associated with improvements in glycemic control, body weight, blood pressure, and lipid profiles, and may also enhance grip and upper-extremity muscle strength [14, 15]. Although both structured exercise and lifestyle behavior interventions can increase long-term physical activity levels in T2DM, the effects appear to be greater with supervised exercise [16]. However, fully supervised programs are often difficult to sustain in routine practice due to the demands on personnel and facilities and the associated costs. In addition, exercise participation often declines when individuals transition from structured supervision to home-based exercise, potentially compromising adherence and the maintenance of training benefits [12, 14]. This transition is clinically important because the withdrawal of supervision may disproportionately affect functional outcomes such as muscle strength, sit-to-stand performance, and mobility, which are central to patient independence and daily functioning.

To address these limitations, hybrid models that combine an initial supervised phase with a subsequent self-managed, home-based phase have gained increasing interest, as they may improve accessibility and long-term feasibility while preserving training-related benefits [17, 18]. Simple digital support strategies, such as QR-coded exercise videos and mobile messaging reminders, may further enhance adherence and real-world applicability in community-dwelling older adults [9]. Nevertheless, evidence remains limited regarding whether improvements in muscle mass, physical function, and quality of life can be maintained after transition from supervised to self-managed exercise, particularly in older adults with T2DM and sarcopenia. Importantly, it remains unclear whether self-managed exercise is sufficient to sustain the functional gains achieved during supervised training, or whether ongoing support is required to prevent their attenuation. This issue is clinically important because loss of benefit after withdrawal of supervision may reduce the long-term value of such interventions.

Therefore, this randomized controlled trial investigated whether and to what extent the benefits achieved during supervised resistance training were sustained after transitioning to a self-managed home-based program in older adults with T2DM and sarcopenia. Specifically, we evaluated the effects of a two-phase intervention consisting of 12 weeks of supervised resistance training followed by 12 weeks of self-managed exercise on skeletal muscle mass, muscle strength, physical function, glycemic control, and health-related quality of life.

Methods

Design

This single-center, randomized, controlled, parallel-group superiority trial with a 1:1 allocation ratio was conducted at a medical center in Taichung, central Taiwan. The study was reported in accordance with the Consolidated Standards of Reporting Trials (CONSORT) guidelines [19] (Supplementary Table S1). Participant recruitment was conducted between February and August 2025. The trial was retrospectively registered with ClinicalTrials.gov (NCT07398495) on 23 February 2026, after completion of participant recruitment.

Participants

Eligible participants were older adults aged ≥ 65 years with confirmed T2DM [20] and sarcopenia defined according to the 2019 Asian Working Group for Sarcopenia (AWGS) criteria [21]. Sarcopenia was defined as low muscle mass plus low muscle strength or impaired physical performance, and severe sarcopenia as impairment in all three domains. Exclusion criteria included major mobility limitation, cognitive impairment (Mini-Mental State Examination [MMSE] score < 24) [22], renal impairment (estimated glomerular filtration rate < 45 mL/min/1.73 m²) [23], diabetic foot, limb amputation, myocardial infarction, autonomic neuropathy, recent stroke, or any other condition precluding safe participation. Antidiabetic medication remained unchanged throughout the study. Participants were, therefore, clinically stable older adults who could safely engage in home-based resistance training.

Randomization, blinding, and sample size

Participants were randomly assigned in a 1:1 ratio to the supervised resistance training (SRT) group or control group (CG) using a computer-generated random sequence prepared by an independent researcher who was not involved in participant recruitment, intervention delivery, outcome assessment, or data analysis. No stratification or blocking was applied. Allocation concealment was ensured using 62 sequentially numbered, opaque, sealed envelopes, prepared by the independent researcher, which were opened sequentially only after participant enrollment and completion of the baseline assessment. The personnel responsible for enrollment and intervention assignment did not have access to the random allocation sequence before assignment. Owing to the nature of the exercise intervention, participants and training staff could not be blinded; however, outcome assessors and the data analyst remained blinded to group allocation throughout the study.

Sample size was estimated using R version 4.4.3, assuming a two-sided α of 0.05, 80% power, and a medium effect size (f = 0.25) based on previous research [23]. The calculation was based on a repeated-measures design and aimed to detect a group-by-time interaction. A minimum of 26 participants per group was required. After allowing for 10% attrition and maintaining balanced allocation, the target sample size was set at 62 participants, with 31 in each group. No interim analyses or formal stopping rules were planned. No subgroup, sensitivity, or post-hoc analyses were prespecified.

Intervention

At baseline, all participants received a standardized nutrition education booklet on sarcopenia prevention and management, including guidance on balanced intake across the six major food groups, adequate protein intake, dairy consumption, individualized caloric targets, and weekly dietary self-monitoring.

Participants in the CG received usual care, including standard advice on diet, medication management, and general physical activity, without a structured exercise program.

Participants in the SRT group received usual care plus a 24-week progressive elastic-band resistance training program consisting of a 12-week supervised phase followed by a 12-week self-managed home-based phase. The intervention was delivered by a licensed physical therapist with more than 20 years of experience in geriatric rehabilitation and exercise prescription for metabolic disease, who was trained in the study protocol to ensure fidelity.

Before starting the program, participants received standardized instruction on the exercise protocol, self-monitoring procedures, and safety precautions, together with diabetes self-management materials. They were instructed to measure resting blood pressure before each session and to defer exercise if systolic blood pressure was ≥ 180 mmHg or diastolic blood pressure was ≥ 100 mmHg [17]. They were also educated to recognize the symptoms of hypoglycemia and to stop exercising immediately if they occurred [24]. A structured exercise manual with QR-coded demonstration videos and a self-monitoring checklist was provided. The intervention was designed according to the FITT principle [25], and details are summarized in Table 1.

Table 1.

Supervised to self-managed program

Item/phase	Phase 1: Supervised Training (Weeks 1–12)	Phase 2: Self-Managed Exercise (Weeks 13–24)
Frequency	3–5 sessions /week	3–5 sessions /week
Intensity	Moderate to high (CR10 score 6–7) Sets/ reps: 3 sets of 8–10 reps	Maintain moderate to high (CR10 score 6–7) Sets/ reps: 3 sets of 8–10 reps
Time	30 min/session (8-min warm-up, 15-min main resistance training, 7-min cool-down)	30 min/session (8-min warm-up, 15-min main resistance training, 7-min cool-down)
Type	Resistance Band Training-4 multi-joint band exercises (video-guided): (1) Seated Resistance Band Knee Raise (Fig. S1) (2) Seated Resistance Band Overhead Raise (Fig. S2) (3) Resistance Band Sit-to-Stand (Fig. S3) (4) Seated Alternate Arm and Leg (Fig. S4)
Intervention	≥ 1 in-person PT session Weekly LINE delivery of QR-coded exercise videos	No weekly LINE delivery of QR-coded exercise videos Self-guided exercise supported by the manual and QR-coded videos
Monitor	Weekly telephone monitoring & manual checklist for adherence & adverse events	Self-monitor adherence & adverse events via manual checklist

FITT frequency, intensity, time, type, CR10 Scale category-ratio 10 scale

During phase 1 (weeks 1–12), participants performed home-based resistance training 3–5 times per week. Each session lasted approximately 30 min and included an 8-min warm-up, 15 min of resistance exercise, and a 7-min cool-down. The program comprised four multi-joint elastic-band exercises: seated resistance-band knee raises, seated resistance-band overhead raise, resistance-band sit-to-stand, and seated alternating arm-and-leg raise (Supplementary Figs. S1–S4). Each exercise was performed for three sets of 8–10 repetitions at a target intensity of 6–7 on the category-ratio 10 (CR10) scale. Participants attended at least one in-person session before training initiation, and adherence was supported through weekly telephone follow-up and LINE reminders. Progressive overload was applied by increasing band resistance when perceived exertion fell below 6 on the CR10 scale.

During phase 2 (weeks 13–24), participants continued the same exercise frequency, structure, and target intensity independently at home. Weekly telephone follow-up and reminders were discontinued. Exercise completion and adverse events were recorded using the self-monitoring checklist.

Exercise adherence was defined as achieving the prescribed training frequency (≥ 3 sessions/week). During phase 1, adherence was verified by telephone follow-up and checklist records; during phase 2, it was assessed using participant-completed checklists. Weeks without verifiable records were classified as non-adherent. Participants meeting the prescribed frequency in ≥ 75% of weeks were classified as adherent [26]. Adherence was evaluated separately for the supervised and self-managed phases to examine changes in exercise participation after transition to self-management.

Outcome measures

Baseline demographic, clinical, and lifestyle characteristics were assessed using standardized measures and validated questionnaires. Age, sex, body mass index, and T2DM duration were recorded at baseline. Frailty, cognitive function, dietary intake, and physical activity were assessed using the Kihon Checklist [27–29], MMSE [22], dietitian interview, and the Taiwanese self-administered short form of the International Physical Activity Questionnaire (MET-min/week) [30, 31], respectively.

The study endpoints and outcome hierarchy were defined before formal data analysis and before inspection of between-group outcome results. To clarify the hierarchy of outcomes, the primary muscle-related outcomes were skeletal muscle mass, muscle strength, physical performance, and the risk of sarcopenia. These domains were assessed using the skeletal muscle mass index (SMI), handgrip strength (HGS), the five-times sit-to-stand test (5TSTS), and the SARC-CalF score. SMI was measured using bioelectrical impedance analysis, HGS using a calibrated dynamometer, physical performance using the 5TSTS, and sarcopenia risk using SARC-CalF, which combines the SARC-F questionnaire with calf circumference [21]. Low muscle mass was defined as an SMI < 7.0 kg/m² in men and < 5.7 kg/m² in women; low muscle strength as HGS < 28 kg in men and < 18 kg in women; and impaired physical performance as > 12 s on the 5TSTS [21]. A SARC-CalF score ≥ 11 was used to indicate probable sarcopenia [21]. Primary muscle-related outcomes were assessed at baseline, 12 weeks, and 24 weeks.

Secondary outcomes were glycemic control and health-related quality of life. Glycemic control was assessed using HbA_1c, measured with National Glycohemoglobin Standardization Program-certified methods and classified according to American Diabetes Association criteria [32, 33]. Health-related quality of life was assessed using the 12-item Short Form Health Survey (SF-12), from which the physical component summary (PCS) and mental component summary (MCS) scores were derived [34, 35]. Secondary outcomes were assessed at baseline, 12 weeks, and 24 weeks. Lower HbA_1c values indicate better glycemic control, whereas higher PCS and MCS scores indicate better physical and mental health-related quality of life, respectively.

Harms assessment

Adverse events were defined as any unfavorable symptom, injury, hypoglycemic event, cardiovascular symptom, fall, or other unintended effect occurring during or after exercise sessions. Harms were systematically assessed using participant self-monitoring checklists. During the supervised phase, adverse events were reviewed during follow-up contacts, whereas during the self-managed phase, they were assessed from checklist records. All reported adverse events were recorded and summarized by study group.

Statistical analysis

All analyses were conducted according to the intention-to-treat principle, whereby all randomized participants were included and analyzed according to their originally assigned groups. Continuous variables are presented as mean ± standard deviation or median with interquartile range, as appropriate, and categorical variables are presented as counts and percentages. Baseline characteristics were summarized by treatment group and compared using independent-samples t tests or Mann–Whitney U tests for continuous variables and Pearson’s chi-square or Fisher’s exact tests for categorical variables.

The statistical analysis plan, including the outcome hierarchy, generalized estimating equation (GEE) model specification, working correlation structure, and approach to multiplicity, was defined before formal data analysis and before inspection of between-group outcome results. SMI, HGS, 5TSTS performance, and SARC-CalF score were specified as protocol-defined co-primary muscle-related outcomes. HbA_1c, the SF-12 (PCS), and SF-12 (MCS) were specified as secondary outcomes.

Longitudinal between-group effects for primary and secondary outcomes were analyzed using GEE models with a normal distribution and identity link function. Time was treated as a categorical variable, and each model included group, time, and the group × time interaction, with participant ID specified as the clustering variable. A first-order autoregressive working correlation structure was selected because outcomes were repeatedly measured at ordered time points. The group × time interaction was used to evaluate whether changes over time differed between the supervised resistance training (SRT) and control group (CG) [36].

.

For the four protocol-defined co-primary muscle-related outcomes, the primary inferential family consisted of the four GEE group × time interaction tests, one for each co-primary outcome. To account for multiplicity across these co-primary outcomes, statistical significance for the primary interaction tests was evaluated using a Bonferroni-adjusted two-sided α level of 0.0125 (0.05/4). For secondary outcomes, GEE interaction results were considered supportive and were evaluated using a two-sided α level of 0.05 [36].

Time-specific between-group differences were estimated from model-based estimated marginal means and are presented as adjusted mean differences between the SRT and CG groups at each time point. Pairwise comparisons between groups at each time point were adjusted using the Bonferroni method and are presented as supportive analyses to aid interpretation of the timing and magnitude of between-group differences. Differences are presented as SRT minus CG [36].

All available repeated measurements were included in the GEE analyses. Because all randomized participants completed the outcome assessments, no outcome data were missing, and no imputation was performed. Adverse events were summarized descriptively by group, and no formal hypothesis testing for harms was performed. No subgroup analyses, sensitivity analyses, or other post hoc analyses were prespecified or performed. Analyses were performed using IBM SPSS Statistics version 23.0.

Results

The study flow is shown in Fig. 1. Between February and August 2025, 91 older adults with T2DM were screened according to the AWGS 2019 criteria. Of these, 29 were excluded: 17 declined participation, and 12 did not meet the eligibility criteria. The remaining 62 participants with sarcopenia or severe sarcopenia were randomly assigned to the CG (n = 31) or SRT group (n = 31). All participants completed the 24-week study and were included in the intention-to-treat analysis. No participants were lost to follow-up, and no primary or secondary outcome data were missing.

Fig. 1.

CONSORT flow diagram

Baseline characteristics are summarized in Table 2. The groups were well balanced, with no significant differences in demographic, lifestyle, or clinical variables (all P > 0.05). Overall, the median age was 72.0 years (IQR 69.0–76.8), and 67.7% of participants were female. Diabetes-related characteristics, nutritional indicators, BMI, SMI, HGS, 5TSTS performance, SARC-CalF score, physical activity level, and frailty status were comparable between groups.

Table 2.

Baseline clinical characteristics of participants

	Control group	SRT group	All	P
N	31	31	62
Age (years)	72.0 (69.0–76.5)	72.0 (70.0–76.5)	72.0 (69.0–76.8)	0.72
T2DM duration (years)	17.1 ± 8.6	17.3 ± 8.4	17.2 ± 8.5	0.89
BMI (kg/m2)	22.3 ± 3.6	22.7 ± 2.9	22.5 ± 3.3	0.57
SMI (kg)	5.5 (5.3–6.1)	5.6 (5.5–6.0)	5.6 (5.3–6.1)	0.31
HGS (kg)	16.6 (14.3–20.7)	18.4 (15.8–23.2)	17.8 (15.1–22.4)	0.19
5TSTS (sec)	14.0 (10.6–16.9)	16.0 (13.4–18.9)	14.3 (12.2–18.5)	0.10
SARC-CalF (score)	12.0 (11.0–13.0)	12.0 (11.0–13.0)	12.0 (11.0–13.0)	0.75
KCL (score)	5.0 (2.0–10.0)	3.0 (1.5–10.0)	4.0 (2.0–10.0)	0.72
HbA1c (%)	6.9 (6.2–7.6)	6.7 (6.3–7.1)	6.8 (6.2–7.3)	0.60
MMSE (score)	29.0 (27.0–30.0)	29.0 (27.5–30.5)	29.0 (27.0–30.0)	0.80
Energy (kcal/day)	1341 (1182.0-1528.3)	1426 (1339.3-1559.8)	1412 (1244.7-1554.3)	0.49
Protein intake(g/day)	52.0 (44.0-63.8)	53.0 (47.3–63.5)	52.8 (46.5–64.1)	0.62
Carbohydrate (g/day)	151.0 (131.3–173.0)	164.0 (147.0–183.9)	160.3 (137.1–178.9)	0.67
Fat intake (g/day)	58.6 ± 14.3	60.4 ± 10.9	59.5 ± 12.6	0.42
Family history (%)	15 (48.4)	16 (51.6)	31 (50.0)	1
Sex, n (%)				0.79
Males	11 (35.5)	9 (29.0)	20 (32.3)
Females	20 (64.5)	22 (71.0)	42 (67.7)
Frailty, n (%)				0.66
No	13 (41.9)	17 (54.8)	30 (48.4)
Yes	18 (58.1)	14 (45.2)	32 (51.6)
Living with family, n (%)				1
No	3 (9.7)	3 (9.7)	6 (9.7)
Yes	28 (90.3)	28 (90.3)	56 (90.3)
Education level, n (%)				1
≦High school graduate	25 (80.6)	26 (83.9)	51(82.3)
≧College graduate	6 (19.4)	5 (16.1)	11 (17.7)
Income/month, n (%)				1
< 20,000	22 (71.0)	22 (71.0)	44 (71.0)
≧ 20,000	9 (29.0)	9 (29.0)	18 (29.0)
IPAQ-SS level, n (%)				0.86
Low	14 (45.2)	15 (48.4)	29 (46.8)
Moderate	12 (38.7)	10 (32.3)	22 (35.5)
High	5 (16.1)	6 (19.4)	11 (17.7)

Data are presented as mean ± SD^a, median (interquartile range)^b, or n (%)^c

Statistical significance was defined as p < 0.05

CG control group, SRT supervised resistance training, BMI body mass index, SMI skeletal muscle mass index, HGS handgrip strength, 5TSTS 5-time sit-to-stand test, SARC-CalF scale SARC-F (strength, assistance with walking, rising from a chair, climbing stairs, and falls) questionnaire, combined with calf circumference, KCL kihon checklist, MMSE mini-mental state examination, HbA_1c glycated hemoglobin, IPAQ-SS Taiwanese version of the International Physical Activity Questionnaire Self-Administered Short Version

^aAnalyzed using an independent-samples t-test for normally distributed continuous variables

^bAnalyzed using the Mann–Whitney U test for non-normally distributed continuous variables

^cAnalyzed using Pearson’s chi-square test or Fisher’s exact test for categorical variables

Across the 24-week intervention, the mean exercise adherence rate in the SRT group was 75.1 ± 11.4%. Adherence differed between phases, with higher adherence during the supervised phase than during the self-managed home-based phase. Specifically, mean adherence was 85.0 ± 8.6% during the supervised phase and declined to 65.3 ± 20.0% during the self-managed phase. Using the predefined adherence threshold of ≥ 75%, 29/31 participants (93.5%) were classified as adherent during the supervised phase, whereas only 12/31 participants (38.7%) remained adherent during the self-managed phase. This decline was considered clinically relevant because it coincided with attenuation of several functional outcomes at 24 weeks. No adverse events or unintended effects were reported.

Primary outcome

Model-based estimates of the primary muscle-related outcomes are shown in Table 3. At baseline, no statistically significant between-group differences were observed in SMI, HGS, 5TSTS performance, or SARC-CalF score.

Table 3.

Model-based estimates of primary muscle-related outcomes from generalized estimating equation analyses

Outcome	Time	SRT(n = 31)	CG (n = 31)	Adjusted between-group difference a (95% CI)	GEE group × time interaction b
		Mean ± SD	Mean ± SD		Difference in change from baseline β (95% CI)	P value
SMI, kg/m²	Baseline	5.79 ± 0.63	5.60 ± 0.78	0.18 (− 0.34 to 0.71)	−
	12 weeks	5.82 ± 0.63	5.59 ± 0.77	0.23 (− 0.29 to 0.75)	0.05(0.01 to 0.08)	0.02
	24 weeks	6.12 ± 0.62	5.75 ± 0.72	0.38 (0.05 to 0.70)	0.20(0.05 to 0.34)	0.01
HGS, kg	Baseline	20.16 ± 7.04	17.81 ± 6.76	2.35 (− 2.71 to 7.41)	−
	12 weeks	24.43 ± 7.21	16.97 ± 6.64	7.46 (2.41 to 12.51)	5.11(3.30 to 6.92)	< 0.001
	24 weeks	22.21 ± 6.81	18.98 ± 6.82	3.23 (− 1.77 to 8.23)	0.88(− 1.60 to 3.35)	0.49
5TSTS, sec	Baseline	16.22 ± 4.95	14.05 ± 5.30	2.17 (− 1.59 to 5.93)	−
	12 weeks	10.44 ± 4.25	15.11 ± 3.92	−4.67 (− 7.67 to − 1.67)	− 6.84(− 8.93 to − 4.75)	< 0.001
	24 weeks	12.77 ± 6.42	12.05 ± 3.53	0.71 (− 3.09 to 4.51)	− 1.46(− 4.10 to 1.16)	0.27
SARC-CalF	Baseline	12.26 ± 1.50	12.13 ± 1.63	0.13 (− 1.02 to 1.28)	−
	12 weeks	9.00 ± 2.54	13.19 ± 1.52	−4.19 (− 5.73 to − 2.66)	− 4.32(− 5.44 to − 3.21)	< 0.001
	24 weeks	9.87 ± 3.21	9.26 ± 3.71	0.61 (− 1.93 to 3.16)	0.48(− 1.46 to 2.43)	0.63

SMI skeletal muscle mass index, HGS handgrip strength, 5TSTS 5-time sit-to-stand test, SARC-CalF scale SARC-F (strength, assistance with walking, rising from a chair, climbing stairs, and falls) questionnaire, combined with calf circumference, SD standard deviation, CI confidence interval

^aTime-specific between-group differences were estimated from model-based estimated marginal means and represent the adjusted mean difference between the SRT and CG groups at each time point. Pairwise comparisons were adjusted using the Bonferroni method. Differences are presented as SRT minus CG

^bThe GEE group × time interaction estimate represents the between-group difference in change from baseline at each follow-up time point. For the four co-primary muscle-related outcomes, statistical significance was evaluated using a Bonferroni-adjusted α level of 0.0125 (0.05/4)

The primary longitudinal inference was based on GEE group × time interactions, interpreted using a Bonferroni-adjusted α level of 0.0125 for the four co-primary muscle-related outcomes. For SMI, the interaction did not meet this threshold at 12 weeks [β = 0.05 kg/m²; 95% CI, 0.01 to 0.08; P = 0.02], but reached statistical significance at 24 weeks [β = 0.20 kg/m²; 95% CI, 0.05 to 0.34; P = 0.01]. The adjusted between-group difference at 24 weeks also favored the SRT group [0.38 kg/m²; 95% CI, 0.05 to 0.70], although the magnitude was modest. At 12 weeks, significant group × time interactions favored the SRT group for HGS [β = 5.11 kg; 95% CI, 3.30 to 6.92], 5TSTS performance [β = −6.84 s; 95% CI, − 8.93 to − 4.75], and SARC-CalF score [β = −4.32; 95% CI, − 5.44 to − 3.21; all P < 0.001]. The corresponding time-specific adjusted between-group differences showed the same favorable pattern. By 24 weeks, neither the interaction effects nor the adjusted between-group differences for these outcomes remained statistically significant.

Overall, SRT improved muscle strength, lower-limb performance, and sarcopenia-related screening scores after the supervised phase, but these benefits attenuated after transition to self-managed training. In contrast, SMI showed a modest, favorable change that became evident at 24 weeks.

Secondary outcomes

Model-based estimates of the secondary outcomes are shown in Table 4. At baseline, no statistically significant between-group differences were observed in HbA_1c, SF-12 PCS, or SF-12 MCS.

Table 4.

Model-based estimates of secondary outcomes from generalized estimating equation analyses

Outcome	Time	SRT(n = 31)	CG(n = 31)	Adjusted between-group differencea (95% CI)	GEE group × time interaction b
		Mean ± SD	Mean ± SD		Difference in change from baseline β (95% CI)	P value
HbA1c, %	Baseline	6.77 ± 0.96	6.88 ± 0.98	−0.11 (− 0.82 to 0.60)	−
	12 weeks	6.74 ± 0.91	6.90 ± 0.86	−0.16 (− 0.81 to 0.49)	-0.05(-0.34 to 0.24)	0.73
	24 weeks	6.78 ± 0.84	7.08 ± 1.18	−0.30 (− 1.05 to 0.45)	-0.19(-0.57 to 0.19)	0.32
SF-12(PCS)	Baseline	27.74 ± 3.14	28.53 ± 3.03	−0.79 (− 3.05 to 1.47)	−
	12 weeks	37.52 ± 4.26	25.61 ± 3.15	11.92 (9.17 to 14.67)	12.71(10.45 to 14.97)	< 0.001
	24 weeks	36.99 ± 4.97	24.91 ± 2.80	12.08 (9.13 to 15.04)	12.87(10.48 to 15.27)	< 0.001
SF-12(MCS)	Baseline	29.29 ± 3.17	29.83 ± 3.20	−0.54 (− 2.88 to 1.80)	−
	12 weeks	38.14 ± 4.48	25.79 ± 2.23	12.35 (9.76 to 14.95)	12.90(10.32 to 15.50)	< 0.001
	24 weeks	37.56 ± 4.98	24.81 ± 1.95	12.75 (9.98 to 15.52)	13.30(10.80 to 15.80)	< 0.001

HbA_1c glycated hemoglobin, SF-12 12-item short-form health survey, PCS physical component summary, MCS mental component summary; SD standard deviation, CI confidence interval

^aTime-specific between-group differences were estimated from model-based estimated marginal means and represent the adjusted mean difference between the SRT and CG groups at each time point. Pairwise comparisons were adjusted using the Bonferroni method. Differences are presented as SRT minus CG

^bThe GEE group × time interaction estimate represents the between-group difference in change from baseline. Statistical significance was evaluated at α = 0.05

Supportive longitudinal analyses for secondary outcomes were based on GEE group × time interactions, interpreted using a two-sided α level of 0.05. For HbA_1c, neither the group × time interactions nor the time-specific adjusted between-group differences were significant at 12 or 24 weeks.

For SF-12 PCS, significant group × time interactions favored the SRT group at both 12 weeks [β = 12.71; 95% CI, 10.45 to 14.97; P < 0.001] and 24 weeks [β = 12.87; 95% CI, 10.48 to 15.27; P < 0.001]. For SF-12 MCS, significant group × time interactions similarly favored the SRT group at 12 weeks [β = 12.90; 95% CI, 10.32 to 15.50; P < 0.001] and 24 weeks [β = 13.30; 95% CI, 10.80 to 15.80; P < 0.001]. The corresponding time-specific adjusted between-group differences showed the same favorable pattern for both SF-12 PCS and MCS through 24 weeks.

Overall, HbA_1c showed no clear intervention-related change, whereas SF-12 PCS and MCS showed favorable supportive patterns in the SRT group through 24 weeks. Because these were secondary outcomes, the SF-12 findings should be interpreted as supportive rather than confirmatory.

Discussion

This 24-week randomized controlled trial showed outcome-specific patterns after transitioning from supervised to self-managed resistance training in older adults with T2DM and sarcopenia. Improvements in HGS, 5TSTS performance, and SARC-CalF scores were observed after the supervised phase, but these functional gains were not fully sustained by 24 weeks. In contrast, SMI showed a modest favorable change at 24 weeks, whereas HbA_1c showed no clear intervention-related change. SF-12 PCS and MCS showed favorable patterns through 24 weeks, but these secondary outcomes should be interpreted as supportive rather than confirmatory. Together, these findings suggest that supervised resistance training may produce short-term functional benefits, whereas self-managed training alone may be insufficient to preserve improvements in muscle strength and physical performance after direct supervision ends.

The decline in functional outcomes after transition to self-managed training should be interpreted within a behavioral–exercise framework. This pattern may reflect the combined effects of reduced exercise dose, insufficient progression, and weaker behavioral reinforcement during the home-based phase. The proportion of participants achieving the predefined adherence threshold of ≥ 75% decreased from 93.5% during supervised training to 38.7% during self-managed training, suggesting lower total exercise exposure after supervision was withdrawn. Because adherence was assessed using self-reported checklists, actual training exposure may have been lower than recorded. In addition, the absence of supervision may have limited progressive overload, adjustments in intensity, correction of movement quality, and timely feedback. Reduced monitoring may also have weakened accountability, exercise confidence, and long-term habit formation. These factors are particularly relevant because HGS and 5TSTS performance depend on continued neuromuscular adaptation and may therefore be more sensitive than skeletal muscle mass to reductions in training stimulus [13, 25]. This interpretation is consistent with evidence that supervised resistance training generally produces greater improvements in strength, function, lean mass, and health-related quality of life than unsupervised or minimally supervised home-based exercise [37–39].

The modestly favorable SMI pattern observed at 24 weeks should be interpreted cautiously. Older adults with T2DM are vulnerable to accelerated muscle loss because of insulin resistance, chronic inflammation, metabolic dysregulation, and neuromuscular impairment [8, 9, 40]. Our findings are consistent with evidence that progressive resistance training can increase muscle mass and reduce the severity of sarcopenia, including in individuals with T2DM [10, 41, 42]. However, the relatively favorable SMI pattern, together with the loss of sustained improvement in HGS, 5TSTS performance, and SARC-CalF scores at 24 weeks, suggests a partial dissociation between structural muscle adaptation and functional performance. This pattern is biologically plausible because strength and physical function depend not only on muscle mass but also on muscle quality, motor-unit recruitment, neuromuscular coordination, movement velocity, balance, and task-specific movement execution. In patients with T2DM, myosteatosis, an indicator of impaired muscle quality, has been associated with lower thigh muscle strength [43], and low muscle strength rather than low muscle mass has been associated with cardiovascular autonomic neuropathy [44]. These findings support the view that muscle function and diabetes-related functional risk are not fully explained by muscle mass alone. Thus, the observed SMI pattern should be viewed as a modest structural benefit rather than evidence that functional gains were retained. This interpretation is particularly important because established minimal clinically important differences for SMI remain insufficiently defined in older adults with T2DM and sarcopenia.

The favorable SF-12 PCS and MCS patterns suggest that the intervention may have influenced health-related quality of life beyond musculoskeletal outcomes. During the supervised phase, direct guidance may have enhanced exercise confidence, self-efficacy, and familiarity with the program; during the self-managed phase, the manual and QR-coded videos may have supported behavioral continuity. These psychological and behavioral mechanisms are relevant in older adults with chronic disease and are consistent with evidence from supervised home-based telerehabilitation and tele-exercise interventions in patients with T2DM [17, 45]. Nevertheless, because SF-12 PCS and MCS were secondary outcomes, these findings should be interpreted as supportive rather than confirmatory. Although a difference of approximately 3 points has been suggested as a minimally important difference for interpreting group mean differences in SF-12 physical and mental component summary scores [46], this threshold has not been specifically validated in older adults with T2DM and sarcopenia undergoing low-intensity elastic-band resistance training. Therefore, although the observed changes in PCS and MCS exceeded this commonly cited reference value, their practical significance in this specific population remains uncertain. These improvements may also partly reflect increased contact, attention, self-efficacy, and participants’ expectations, rather than the physiological effects of resistance training alone [17, 18, 45]. Future studies should establish population-specific minimally important differences for older adults with T2DM and sarcopenia.

No clear intervention-related improvement in HbA_1c was observed over 24 weeks. This may reflect the cohort’s clinical stability, relatively good baseline glycemic control, and stable antidiabetic medication use. Although resistance training can improve HbA_1c in individuals with T2DM, such effects are generally more evident in those with poorer baseline glycemic control or in higher-intensity exercise trials [14, 47, 48]. In older adults with T2DM and sarcopenia, preserving musculoskeletal health and limiting functional decline may remain clinically relevant even without short-term glycemic improvement. Given the absence of serious exercise-related adverse events, these findings support a cautious interpretation of the intervention as a potentially feasible approach for selected musculoskeletal and quality-of-life benefits, rather than as an intervention primarily targeting glycemic control.

From a clinical perspective, elastic-band resistance training is practical, low-cost, and adaptable to home settings, making it a feasible exercise strategy for older adults. The strengths of this study include its randomized controlled design, its focus on older adults with T2DM and sarcopenia, and a pragmatic two-phase intervention that reflects the real-world transition from supervised to self-managed exercise. However, the decline in several functional outcomes after supervision was withdrawn suggests that home-based training may require additional implementation strategies to maintain exercise dose, progression, and quality. Future programs could combine home-based resistance training with periodic booster sessions, individualized progression plans, remote monitoring, automated reminders, adherence tracking, feedback-based exercise logs, and recorded exercise videos. These strategies may help preserve chair-rise ability, mobility, fall-risk reduction, and independence while maintaining the feasibility and flexibility of home-based exercise.

Several limitations should be acknowledged. First, the single-center design and modest sample size may limit generalizability. Because participants were clinically stable older adults who were able to perform home-based resistance training, the findings may not apply to frailer patients or those with advanced diabetic complications, major mobility limitations, unstable cardiovascular disease, severe renal impairment, or other contraindications to exercise. Second, adherence during the self-managed phase was assessed using self-reported checklists, which may have introduced recall or social desirability bias and overestimated actual training exposure. Training intensity, progression, and exercise quality were also not objectively monitored, limiting the ability to determine whether the decline in functional effects resulted from reduced dose, insufficient progression, poorer movement quality, or their combined effects. Third, dietary intake was not monitored longitudinally, and potential changes in nutrition or lifestyle behaviors could not be fully accounted for. Larger multicenter studies with longer follow-up and objective monitoring of adherence, exercise quality, dietary intake, and lifestyle factors are needed to clarify the long-term clinical value of this two-phase resistance training approach.

Conclusions

Transitioning from supervised resistance training to self-managed home-based exercise showed outcome-specific effects in older adults with T2DM and sarcopenia. Strength and functional gains improved after the supervised phase but attenuated after supervision was withdrawn. In contrast, SMI showed a modest favorable change at 24 weeks, while SF-12 PCS and MCS showed supportive favorable patterns; HbA_1c showed no clear intervention-related change. These findings suggest that self-managed home-based training alone may be insufficient to sustain functional benefits. Hybrid support models may benefit from more intensive integration of periodic supervision, remote monitoring, automated reminders, and feedback-driven exercise logs to sustain adherence, optimize exercise quality, and preserve functional capacity.

Author contributions

Concept and design: Y.C.W. Acquisition, analysis, or interpretation of data: Y.C.W., W.C.L., C.N.H., and all authors. Critical review of the manuscript for important intellectual content: all authors.

Funding

Open access funding provided by Chung Shan Medical University/Chung Shan Medical University Hospital. This work was supported by the Taiwanese Association of Diabetes Educators and Chung Shan Medical University Hospital Research grant (no. CSH-2024-A-035).

Data availability

The datasets generated and analyzed in the current study are available from the corresponding author upon reasonable request.

Declarations

Conflict of interest

The authors declare no conflicts of interest related to this work.

Ethics statement

This study was approved by the Institutional Review Board of Chung Shan Medical University Hospital (approval no. CS2-23149) and was registered with Clinicaltrials.gov (NCT07398495). This study was conducted in accordance with the Declaration of Helsinki of 1964 and its subsequent amendments, or with comparable ethical standards. All participants provided written informed consent.

Consent

Written informed consent was obtained from all participants, who were informed of the study procedures and their right to withdraw at any time.