Exercise Snacks Are Feasible to Perform in the Real World and Improve Physical Capacity for Adults Living With Non‐Insulin Treated Type 2 Diabetes: A Randomised Trial Authors: Fiona J Babir, Alexis Marcotte‐Chénard, Roderick E Sandilands, Kaja Falkenhain, Noah Mulkewich, Hashim Islam, Seth F McCarthy, Douglas L Richards, Kenneth Madden, Joel Singer, Michael C Riddell, Mary E Jung, Martin J Gibala, Jonathan P Little Published: 2026 ABSTRACT ABSTRACTAimsTo investigate the feasibility and preliminary efficacy of a 12‐week remotely‐delivered exercise snacks (ES) intervention in adults with type 2 diabetes.Material and MethodsInsufficiently active adults with type 2 diabetes (N = 69; 46 females; mean age ± SD: 58 ± 11 years) were randomised to an ES or mobility/stretching comparator group (CON), which involved 4 × 1‐min bouts of either vigorous or low intensity exercise, respectively, on ≥ 5 days/week. The primary outcome was feasibility based on adherence. Secondary outcomes included exercise enjoyment (1–7 scale), rating of perceived exertion (RPE; 0–10 scale), heart rate (HR), haemoglobin A1c (HbA1c), blood biomarkers of cardiometabolic health, 30‐s sit‐to‐stand capacity, grip strength, estimated maximal oxygen uptake, and anthropometrics.ResultsWeekly adherence (estimated marginal mean [95% confidence interval]: 18 bouts [16–21] for both groups; p = 0.99) and total enjoyment (ES: 4.5 [4.1–4.8] vs. CON: 4.3 [4.0–4.7]; p = 0.64) were high and not different between groups. Despite higher RPE (5.7 [5.4–6.1]) and peak HR (73 [70–77] % of age‐predicted HR maximum) in ES versus CON (2.0 [1.7–2.4] and 61 [58–64] % of age‐predicted HR maximum, respectively) (all p < 0.001), there were no between‐group differences in the change in any secondary outcome (all p > 0.05) except for greater sit‐to‐stand capacity in ES after training (between‐group effect estimate [95% confidence interval]: 1.9 repetitions [0.3–3.4]; p = 0.02).ConclusionsExercise snacks were feasible to perform in the real world and improved sit‐to‐stand capacity to a greater extent than CON in adults living with type 2 diabetes. Trial Registration: ClinicalTrials.gov identifier: NCT06407245Keywords: clinical trial, exercise intervention, randomised trial, real‐world evidence, type 2 diabetes Introduction People living with type 2 diabetes (T2D) are advised to accumulate ≥ 150 min of moderate‐to‐vigorous intensity aerobic physical activity per week [1, 2] to improve glycaemic management [3]. Recent estimates suggest that only ~16% of women and ~26% of men with T2D meet this guideline and there is a high prevalence of sedentary behaviour [2]. Given the health risks associated with inactivity, innovative strategies that enhance physical activity engagement and interrupt sedentary behaviour in people living with T2D are needed [1, 4]. “Exercise snacks”—brief (≤ 1 min) bouts of vigorous (aiming for RPE ≥ 7/10, ≥ 76% heart rate maximum or ≥ 60% heart rate reserve) effort performed intermittently throughout the day [5, 6, 7, 8]—represent a time‐efficient method that alleviates the need for specialised equipment or dedicated exercise space [5, 6, 9]. Exercise snacks pose as an accessible and practical approach to encourage physical activity and break up sedentary time compared to traditional exercise training approaches [5, 6, 8] with recent studies demonstrating their feasibility in middle‐aged and older adults [10, 11, 12, 13]. But the feasibility of exercise snacks in real‐world settings among adults living with T2D is unknown. The purpose of this randomised trial was to assess the feasibility and preliminary efficacy of a 12‐week exercise snacks (ES) intervention compared to a mobility/stretching exercise comparator (CON) condition in people living with T2D. The primary outcome was feasibility based on intervention adherence. Secondary outcomes included haemoglobin A1c (HbA1c) and other indices of cardiometabolic health and physical capacity. Materials and Methods Study Overview, Sample Size, and Participants This two‐site randomised trial was conducted between April 2024 and August 2025. Prior approval was obtained from the Hamilton Integrated Research Ethics Board (16834) and the University of British Columbia Clinical Research Ethics Board (H21‐03417) with pre‐registered on ClinicalTrials.gov (NCT06407245). Figure 1 summarises the study design. Although there are no formal sample size guidelines for a feasibility trial, an a priori sample size estimation was performed to help ground the recruitment goals. This revealed that 31 participants per group were required to detect a medium effect size (Cohen's f = 0.25, d = 0.50) for a within‐between interaction with 90% power, a conservative repeated measures correlation of r = 0.5, and a two‐tailed α set at 0.05 (G*Power 3.1). This sample size compares favourably with a previous audit of pilot and feasibility trials that reported medians of n = 36 and n = 30 per arm, respectively [14]. To account for ~15% loss to follow‐up, we aimed to recruit ~40 participants per group (N = 80 total) from the areas surrounding each study site primarily through paid advertisements by a third‐party service (Wayturn; Mariefred, Sweden). Participant eligibility criteria are presented in Table 1. Study design schematic. Baseline and post testing measures included anthropometrics, questionnaires, fasted blood sampling, grip strength, 30‐s sit‐to‐stand, and the YMCA submaximal predictive fitness test. Following these baseline measures, participants were randomised to the exercise snacks or the placebo, exercise stretching, comparator group before completing the exercise familiarisation where heart rate was measured using the Polar chest strap at both sites and the Fitbit Charge 6 simultaneously at the McMaster site. During the 12‐week intervention, participants received six phone call check‐ins. After the 12‐week intervention, participants returned to their respective study site for post‐testing measurements. TABLE 1Participant inclusion and exclusion criteria.Criteria for inclusion30–75 years old, physician‐diagnosed type 2 diabetes, performing < 150 min of moderate‐to‐vigorous intensity aerobic exercise per week, body mass index (BMI) between 18.5–40 kg/m2, no changes to prescribed medications for at least 6 months, able to maintain current medication doses, able to maintain current physical activity patterns, HbA1c ≤ 8.5% (≤ 69 mmol/mol), access to a computer, tablet, or smartphone, able to travel to a study site for laboratory testing visits, able to read, write, and understand English, anticipated internet access throughout the intervention, and cleared to participate in exercise based on the Canadian Society for Exercise Physiology Get Active Questionnaire [15].Criteria for exclusionTaking ≥ 4 glucose‐lowering medications, taking insulin, taking beta‐blockers, taking ≥ 3 prescribed medications for the prevention of cardiovascular disease (e.g., statins, antihypertensives), episode of severe hypoglycaemia in the past 6 months, cigarette smoker, chronic musculoskeletal condition that would prevent participation in exercise, recent (within the last 2 years) cardiovascular event that prevents participation in exercise, angina upon exertion (self‐reported and screened for using survey questions based on those in the Rose Cardiovascular Questionnaire V1 Section A [16]), uncontrolled hypertension or an atypical blood pressure or pulse rate at rest or during exercise as determined by a physician, surgical procedure scheduled during the intervention that would prevent exercise participation, current diagnosis of a cardiac or pulmonary disease that would prevent exercise participation, a psychiatric disorder that could prevent completion of the study procedures or visits, donated > 0.5 L of blood within the last 4 weeks, following an extreme diet (e.g., very low carbohydrate/calorie, ketogenic), taking dietary/nutritional supplements that impact glucose control (e.g., exogenous ketones), having diabetic ulcers, peripheral vascular disease, or diabetic neuropathy that would prevent participation in exercise, undergoing dialysis treatment, participating in another clinical trial that would interfere with the study procedures, and pregnant or planning to become pregnant during the intervention. Participant inclusion and exclusion criteria. Study Protocol Screening Interested individuals completed a screening call to discuss eligibility criteria. If deemed eligible based on self‐reported responses, an in‐person baseline testing visit was scheduled. Baseline Testing and Exercise Familiarisation Individuals arrived after a ≥ 8‐h fast and provided signed informed consent before baseline measures were collected. Participants were instructed to refrain from taking glucose‐lowering medication on the morning of the baseline visit. Height, body mass, and waist circumference were measured. Following this, a fasted venous blood sample was obtained via venipuncture and collected into ethylene‐diaminetetraacetic acid (EDTA) tubes. HbA1c was measured using a point‐of‐care multi‐assay analyser (Afinion 2, Abbott Laboratories, Chicago, Illinois, USA) from whole blood. Plasma was obtained by centrifugation at 2500 g for 15 min at 4°C and stored at −80°C prior to batch analyses. Participants were offered a snack ≥ 15 min before the next testing procedures. Participants completed an entry survey, which included questions based on demographics, health, and the Godin‐Shephard Leisure‐Time Physical Activity Questionnaire [17]. Participants then completed a maximal grip strength test involving 2–3 attempts on each hand using a hand dynamometer (Lafayette Instrument Professional Grip Dynamometer, Lafayette, Indiana, USA). Maximum grip strength was the sum of the highest attempt from each hand. Participants then performed a sit‐to‐stand test by repeating as many sit‐to‐stand transitions as possible within 30 s from a chair with no armrests. Participants then completed the YMCA submaximal exercise test [18, 19] on a cycle ergometer (McMaster: Lode Excalibur Sport V 2.0; Lode BV, Groningen, The Netherlands and UBC: SCIFIT ISO1000R; SCIFIT Systems Inc., Tulsa, Oklahoma, USA) to estimate maximal oxygen uptake (VO2max). The test started at 25 watts (W) for 3 min and the heart rate (HR) response to this stage determined the workloads for the subsequent 3‐min stages [19]. The aim was for participants to complete two stages that elicit a steady‐state HR between 110 bpm and 85% of their age‐predicted HR maximum (APHRM; 220‐age). HR was measured using the chest worn Polar H7 Heart Rate Sensor (Polar, Finland) connected to the Polar Flow app. Participants were then randomised (stratified by study site and sex) to the ES or CON group by a study investigator through the Research Electronic Data Capture (REDCap) system (Vanderbilt University, Nashville, Tennessee, USA). The randomisation schedule was created by a third‐party researcher and loaded into REDCap in advance. After randomisation, participants completed 1–2 familiarisation exercise bouts while HR was measured via the Polar chest strap. At the McMaster site, HR was also simultaneously measured with a wrist‐worn Fitbit Charge 6 (Fitbit, San Francisco, California, USA). Participants were asked to rate the session RPE (0–10 Session RPE scale [20]) after completing each exercise bout. HR and RPE were averaged if the participant completed two familiarisation exercises. Exercise Intervention Participants completed the 12‐week remote intervention using a custom smartphone app (Seven Movements, Canmore, Alberta, Canada), which delivered daily push notifications and two rotating instructional exercise videos. Participants determined the timing of push notifications and could adjust them throughout the intervention in accordance with their schedules and convenience. Participants were asked to perform ≥ 20 isolated exercise bouts per week (prescribed as 4 × 1‐min bouts on ≥ 5 days per week) and log each bout in the app. After each bout, participants recorded their RPE [20] and exercise enjoyment (Exercise Enjoyment Scale [21, 22]) on the app. Participants were given a Fitbit Charge 6 to wear throughout the intervention to measure HR responses to each exercise bout. The only difference between the two intervention groups was the types of exercises prescribed. The ES group was assigned higher‐intensity bodyweight exercises (e.g., squats, lunges, step‐ups, etc.) and were instructed to adjust the pace/effort of the exercises to achieve an RPE of ≥ 7, corresponding to “very hard” [20]. The CON group was assigned lower‐intensity stretches and mobility movements. A full list of exercises prescribed is listed in Supporting Information File 1. Six phone call check‐ins were scheduled throughout the intervention: once a week during the first 4 weeks and then once more during both weeks 5–8 and 9–12. The phone calls helped to promote intervention adherence by checking in on exercise progress for the week (participants were asked to self‐report the number of bouts performed over the past week) and providing support with any issues encountered with the intervention. For some participants, exercises were removed from the program based on preferences and/or limitations discussed before the start of the intervention or during the phone call check‐ins. Post‐Intervention Testing Twelve weeks after starting the intervention, participants returned to the laboratory for follow‐up testing. Baseline testing measures were repeated and participants completed an exit survey, which included questions based on health and the Godin‐Shephard Leisure‐Time Physical Activity Questionnaire [17]. Outcomes Intervention Feasibility: Adherence and Enjoyment App‐based adherence during weeks 1–12 was the primary feasibility outcome. Feasibility was defined by the research team a priori as > 70% of participants completing at least 67% (two‐thirds) of the exercise bouts (i.e., > 13 weekly bouts) on at least 8 out of 12 weeks. Rates of recruitment, drop‐out, retention, exercise enjoyment, and other adherence metrics were used as additional indicators of intervention feasibility. The other markers of adherence included: (1) the total number of bouts completed on the app; (2) the total number of bouts recorded on the Fitbit; (3) the weekly average number of bouts recorded on the Fitbit during weeks 1–12; (4) the weekly average number of self‐reported exercise bouts completed on 6 weeks of the intervention (reported during the phone call check‐ins). Total app‐based adherence during weeks 1–12 was also classified as high (> 70% = > 168 bouts), moderate (50%–69% = 168–120 bouts), or low (< 50% = < 120 bouts). The total number of app or Fitbit recorded bouts were not included for participants who did not complete the entire intervention period. Exercise Intensity The average RPE was calculated as the mean of all app‐recorded RPE ratings during the intervention as an additional feasibility outcome. Second‐by‐second HR files and activity logs recorded by the Fitbit were downloaded for all participants who registered on Fitabase (n = 44) (San Diego, California, USA). Activity logs were cleaned in Python according to the following parameters: manually recorded, 30–180 s in duration, recorded between the training start date and post‐testing date. The time stamp of each activity enabled identification of peak HR for each activity log with corresponding HR data. The number of Fitbit activity logs was used to calculate Fitbit weekly and total adherence. Cardiorespiratory Fitness and Leisure Score Index VO2max was estimated (see Supporting Information File 2) [19] based on a predictive equation using the HR achieved during two stages of the YMCA test where the steady‐state HR was between 110 bpm and 85% of APHRM (VO2max ≥ 110 bpm). Due to the low number of participants who completed a full valid test with two stages in the desired HR range, we explored the estimated VO2max with the same equation method [19] but without the lower HR limit parameter of 110 bpm (VO2max no lower HR limit). We also explored the submaximal HR response for participants who completed the first 25 W workload stage to assess a potential training response to a standardised workload. The Godin‐Shephard leisure score index calculation is described in Supporting Information File 3. Godin‐Shephard responses were removed from analysis as input errors if the number of mild/light reported bouts per week was > 60 or if the number of moderate reported bouts per week surpassed eligibility criteria activity level. Plasma Analyses Fasted plasma samples were analysed for glucose (Human Glucose Assay Kit, Chrystal Chem), insulin (Insulin Elisa Kit, Chrystal Chem), and cytokines (interleukin‐6, interleukin‐10, tumour necrosis factor‐alpha; high‐sensitivity VPlex assay, MesoScale Discovery). Statistical Methods Baseline characteristics are presented as mean ± standard deviation for continuous variables or N (%) for categorical variables. The pre‐registered primary outcome of intervention feasibility was assessed descriptively as > 70% of participants completing at least 67% of the exercise bouts (i.e., > 13 weekly bouts) on at least 8 out of 12 weeks. Adherence and intensity‐related outcomes assessed across the intervention period (e.g., mean/total number of bouts, enjoyment, RPE, and HR) or during the exercise familiarisation (e.g., RPE and HR) were analysed using a linear model with a fixed effect for group assignment and accounting for stratified allocation factors (i.e., sex and site). Preliminary efficacy outcomes collected at baseline and follow‐up were analysed using constrained longitudinal data analysis via a linear mixed model with fixed effects for timepoints, the interaction between timepoint and intervention group, stratified allocation factors (i.e., sex and site), and a random effect for participant. Model assumptions of normality and linearity were assessed visually via inspection of diagnostic plots. Models were run using log‐transformed outcome variables if departure from model assumptions were observed. Data were analysed on an intention‐to‐treat basis, and no statistical imputations were used to replace missing data. Data are presented as model‐derived estimated marginal means and within‐ and between‐group effect estimates with corresponding 95% confidence intervals (CI) and p‐values. Data analyses were conducted in R (version 4.3.2) and statistical significance was established at a two‐sided ɑ = 0.05, with no adjustments applied to secondary and exploratory outcomes. Results Participant Recruitment and Retention The participant CONSORT flowchart is displayed in Figure 2. In total, 358 individuals expressed interest in the study through Wayturn and the other listed modes with N = 82 participants recruited (n = 42 at McMaster and n = 40 at UBC). Baseline measures revealed that seven individuals did not qualify to participate despite self‐reporting meeting the eligibility during screening, resulting in 75 participants randomised (ES: n = 35 vs. CON: n = 40). After randomisation, six participants were also found to be ineligible based on observations made or information disclosed during baseline testing, resulting in 69 participants (ES: n = 31 vs. CON: n = 38) starting the study intervention and included in the participant baseline characteristics (Table 2). Additional health and demographic information are presented in Supporting Information File 4 and data separated by sex are displayed in Supporting Information File 5. Of the 69 participants who started the intervention, 63 completed the study and attended follow‐up testing, representing a 91% retention rate (63/69). Participant CONSORT flowchart. TABLE 2Participant baseline characteristics.TotalESCON N 693138Female/Male, N (%)46/23 (67/33)21/10 (68/32)25/13 (66/34)Age, year58 (11)58 (11)58 (11)AnthropometricsBody mass, kg84.7 (16.5)89.4 (18.3)80.9 (14.1)Body mass index, kg/m2 30.1 (5.4)31.4 (5.5)28.9 (4.9)Waist circumference, cm102 (16)105 (19)99 (12)FitnessVO2max ≥ 110 bpm, mL/kg/min25.9 (6.4)24.7 (7.9)27.2 (4.4)VO2max no lower HR limit, mL/kg/min23.4 (5.6)23.3 (5.8)23.5 (5.4)Submaximal HR at 25 W, bpm97 (13)99 (12)96 (13)Grip strength, kg74.3 (25.2)72.7 (24.0)75.4 (26.5)Sit‐to‐stand 30 s, repetitions13 (4)13 (3)14 (4)Metabolic healthHbA1c, %6.6 (0.8)6.6 (0.7)6.6 (0.9)Glucose, mmol/L6.5 (1.4)6.3 (0.9)6.6 (1.7)Insulin, mU/L10.0 (7.0)10.5 (6.9)10.0 (8.0)Tumour necrosis factor‐alpha, pg/mL1.0 (0.3)0.9 (0.4)1.0 (0.3)Interleukin‐6, pg/mL0.6 (0.8)0.7 (1.1)0.6 (0.5)Interleukin‐10, pg/mL0.2 (0.2)0.2 (0.1)0.2 (0.3)Physical activityGodin‐Shephard, leisure score index19 (18)18 (19)20 (17) Note: Data presented as mean (SD) unless otherwise specified. Baseline values were based on n = 69 unless otherwise noted. Waist circumference: n = 66; VO2max ≥ 110 bpm: n = 19; VO2max no lower HR limit: n = 44; submaximal HR at 25 W: n = 61; grip strength: n = 53; sit‐to‐stand: n = 68; glucose, insulin, tumour necrosis factor‐alpha, interleukin‐6, and interleukin‐10: n = 62; Godin‐Shephard: n = 68.Abbreviations: CON = stretching comparator group; ES = exercise snacks group; HbA1c = haemoglobin A1c; HR = heart rate; VO2max = estimated maximal oxygen uptake; W = watts. Participant baseline characteristics. Note: Data presented as mean (SD) unless otherwise specified. Baseline values were based on n = 69 unless otherwise noted. Waist circumference: n = 66; VO2max ≥ 110 bpm: n = 19; VO2max no lower HR limit: n = 44; submaximal HR at 25 W: n = 61; grip strength: n = 53; sit‐to‐stand: n = 68; glucose, insulin, tumour necrosis factor‐alpha, interleukin‐6, and interleukin‐10: n = 62; Godin‐Shephard: n = 68. Abbreviations: CON = stretching comparator group; ES = exercise snacks group; HbA1c = haemoglobin A1c; HR = heart rate; VO2max = estimated maximal oxygen uptake; W = watts. Feasibility: Adherence and Enjoyment App‐based adherence showed that 77% (48/62) of participants completed > 13 exercise bouts (i.e., > 67% of the weekly exercise prescription) on ≥ 8 of the 12 intervention weeks. Weekly app‐based adherence is displayed in Figure 3. The mean weekly number of exercise bouts recorded on the app was not different between ES and CON (estimated marginal means [95% CI]: 18 bouts [16 to 21] in both groups; p = 0.99; n = 64). Exploratory analyses across weeks demonstrated no significant group x week interaction (p = 0.142) and no main effect of group (p = 0.821) but a main effect of time (p < 0.001, Supporting Information File 6). Total exercise bouts completed on the app were not different between ES (230 bouts [199 to 262]) and CON (231 bouts [202 to 260]) (p = 0.96; n = 62). Based on the total bouts recorded on the app between weeks 1–12, 49 participants (~79%) fell into the high adherence category (> 70% = > 168 bouts), 8 participants (~13%) had moderate adherence (50%–69% = 168–120 bouts), and 5 participants (~8%) had low adherence (< 50% = < 120 bouts). The total (ES: 190 bouts [145 to 235] vs. CON: 165 bouts [128 to 203]; p = 0.36; n = 43) and weekly (ES: 16 bouts [12 to 20] vs. CON: 13 bouts [10 to 16]; p = 0.25; n = 44) number of exercise bouts recorded on the Fitbit were also not different between groups. The weekly average number of exercise bouts self‐reported during the phone call check‐ins was also not different between groups (ES: 20 bouts [19 to 22] vs. CON: 20 bouts [19 to 21]; p = 0.86; n = 65). Exercise enjoyment, on a scale of 1–7, were not different between groups (ES: 4.5 [4.1 to 4.8] vs. CON: 4.3 [4.0 to 4.7]; p = 0.64; n = 64). Weekly app‐based exercise adherence for ES (panel A) and CON (panel B). Line and error bars shown for each week display the mean ± standard deviation. The three horizontal dashed/dotted lines across the graph represent (1) the prescribed weekly number of exercise bouts (y = 20), (2) the breakpoint between high (> 70%) and moderate (50%–69%) adherence (y = 14), and (3) the breakpoint between moderate and low (< 50%) adherence (y = 10). ES = exercise snacks group, CON = stretching comparator group. Exercise Intensity: Heart Rate and RPE During familiarisation, RPE was 6.3 (5.8 to 6.7) in the ES group vs. 1.6 (1.0 to 2.1) in the CON group with a between‐group difference of 4.7 (4.0 to 5.4) (p < 0.001; n = 45). Peak HR measured with the Polar chest strap during the familiarisation was 133 bpm (127 to 139) or 81% (78 to 84) of APHRM in the ES group compared to 99 bpm (94 to 104) or 62% (59 to 64) of APHRM in the CON group (p < 0.001; n = 63). Peak HR measured with the Fitbit during the exercise familiarisation was 132 bpm (125 to 140) or 80% (76 to 84) of APHRM in the ES group compared to 94 bpm (88 to 101) or 57% (53 to 61) of APHRM in the CON group (p < 0.001; n = 32). During the intervention, RPE was higher in the ES group (5.7 [5.4 to 6.1]) compared to the CON group (2.0 [1.7 to 2.4]) with a between‐group difference of 3.7 (3.2 to 4.2) (p < 0.001; n = 64). Peak HR recorded with the Fitbit during the intervention was higher in the ES group (121 bpm [113 to 128]) vs. the CON group (97 bpm [91 to 104]) (p < 0.001; n = 41). This corresponded to 73% (70 to 77) of APHRM in the ES group vs. 61% (58 to 64) in the CON group (p < 0.001; n = 41). RPE and HR during weeks 1–12 of the intervention are displayed in Figure 4. Weekly exercise intensity during the 12‐week intervention. Peak heart rate (HR) responses recorded with the Fitbit for exercise bouts shown as percentages of age‐predicted HR maximum (APHRM) (left y‐axis) and session rating of perceived exertion (RPE) reported on the app after completing an exercise bout (right y‐axis). ES = exercise snacks group, CON = stretching comparator group. ES RPE are based on n = 25–29, ES HR are based on n = 14–18, CON RPE are based on n = 27–35, and CON HR are based on n = 18–22 depending on the number of individuals who recorded RPE and HR during each week of the intervention. Efficacy Outcome Measures Within‐ and between‐group effect estimates and 95% CI for all clinical and physical outcomes are presented in Table 3. There were no between‐group differences in the change in any clinical efficacy measure including HbA1c, fasting glucose, fasting insulin, plasma cytokines, body mass, BMI, or waist circumference in response to the intervention (p > 0.05). There was an increase in 30‐s sit‐to‐stand capacity in the ES and CON group (within‐group effect estimates [95% CI]: 4.4 repetitions [3.2 to 5.7] vs. 2.6 repetitions [1.5 to 3.6], respectively; both p < 0.001), resulting in a between‐group difference indicating a greater increase in the ES group (1.9 repetitions [0.3 to 3.4]; p = 0.02). There was a within‐group reduction in waist circumference in the ES group from baseline to follow‐up (−1.6 cm [−3.2 to −0.1]; p = 0.04). There were no between‐group differences observed in the change in the other physical parameters measured including grip strength, estimated VO2max, or HR responses to submaximal exercise (p > 0.05). There were also within‐group changes observed for the Godin‐Shephard leisure score index in both ES (14 [8 to 20]; p < 0.001) and CON (7 [2 to 13]; p = 0.01), with no between‐group differences (p > 0.05). TABLE 3Effect estimates for changes in outcomes measures.Change in ESChange in CONDifference between groupsAnthropometricsBody mass, kg−0.7 (−1.5 to 0.1), p = 0.09−0.7 (−1.4 to 0.0), p = 0.060.0 (−1.0 to 1.1), P = 0.96Body mass index, kg/m2 −0.2 (−0.5 to 0.1), p = 0.15−0.2 (−0.5 to 0.0), p = 0.090.0 (−0.4 to 0.4), p = 0.91Waist circumference, cm−1.6 (−3.2 to −0.1), p = 0.04−0.2 (−1.6 to 1.2), p = 0.75−1.4 (−3.5 to 0.7), p = 0.18FitnessVO2max ≥ 110 bpm, mL/kg/min−1.0 (−5.0 to 3.0), p = 0.60−1.5 (−5.8 to 2.8), p = 0.470.5 (−5.1 to 6.1), p = 0.85VO2max no lower HR limit, mL/kg/min0.1 (−1.7 to 1.9), p = 0.91−0.9 (−3.1 to 1.3), p = 0.431.0 (−1.7 to 3.7), p = 0.47Submaximal HR at 25 W, bpm−3.2 (−6.9 to 0.6), p = 0.10−3.1 (−6.8 to 0.5), p = 0.090.0 (−5.0 to 4.9), p = 0.99Grip strength, kg2.0 (−0.6 to 4.7), p = 0.130.3 (−2.1 to 2.6), p = 0.831.8 (−1.7 to 5.3), p = 0.31Sit‐to‐stand 30 s, repetitions4.4 (3.2 to 5.7), p < 0.0012.6 (1.5 to 3.6), p < 0.0011.9 (0.3 to 3.4), p = 0.02Metabolic healthHbA1c, %−0.1 (−0.3 to 0.1), p = 0.230.0 (−0.2 to 0.1), p = 0.62−0.1 (−0.3 to 0.2), p = 0.57Glucose, mmol/L−0.1 (−0.6 to 0.3), p = 0.620.1 (−0.3 to 0.5), p = 0.69−0.2 (−0.8 to 0.4), p = 0.51Insulin, mU/L a 5 (−16 to 31), p = 0.681 (−17 to 24), p = 0.903 (−23 to 38), p = 0.82Tumour necrosis factor‐alpha, pg/mL 1 0 (−7 to 8), p = 0.94−5 (−11 to 1), p = 0.096 (−3 to 16), p = 0.22Interleukin‐6, pg/mL a 4 (−16 to 29), p = 0.702 (−17 to 24), p = 0.873 (−22 to 35), P = 0.85Interleukin‐10, pg/mL a −4 (−23 to 20), P = 0.69−17 (−33 to 2), p = 0.0815 (−13 to 53), p = 0.32Physical activityGodin‐Shephard, leisure score index14 (8 to 20), p < 0.0017 (2 to 13), p = 0.017 (−1 to 14), p = 0.09 Note: All data are presented as (within‐ or between‐group) effect estimates (95% confidence interval) with corresponding p‐values. Effect estimates are based on intention‐to‐treat analyses. Data analysed via constrained longitudinal data analysis using a linear mixed model with fixed effects for timepoints, the interaction between timepoint and intervention group, stratified allocation factors (sex and site), and a random effect for participant. Post testing values were based on n = 63 unless otherwise noted; n = 1 from the ES group experienced a high blood pressure episode at home near the end of the intervention which impacted the collection of certain post testing measurements for this participant. Waist circumference: n = 54; VO2max ≥ 110 bpm: n = 15; VO2max no lower HR limit: n = 35 (n = 1 removed from the dataset because calculated VO2max was outside of reasonable physiologic range); submaximal HR at 25 W: n = 58; grip strength: n = 46; sit‐to‐stand: n = 61; HbA1c: n = 60; glucose, insulin, tumour necrosis factor‐alpha, interleukin‐6, and interleukin‐10: n = 58.Abbreviations: CON = stretching comparator group; ES = exercise snacks group; HbA1c = haemoglobin A1c; HR = heart rate; VO2max = estimated maximal oxygen uptake; W = watts. a Log‐transformed, interpret as percent change. Effect estimates for changes in outcomes measures. Note: All data are presented as (within‐ or between‐group) effect estimates (95% confidence interval) with corresponding p‐values. Effect estimates are based on intention‐to‐treat analyses. Data analysed via constrained longitudinal data analysis using a linear mixed model with fixed effects for timepoints, the interaction between timepoint and intervention group, stratified allocation factors (sex and site), and a random effect for participant. Post testing values were based on n = 63 unless otherwise noted; n = 1 from the ES group experienced a high blood pressure episode at home near the end of the intervention which impacted the collection of certain post testing measurements for this participant. Waist circumference: n = 54; VO2max ≥ 110 bpm: n = 15; VO2max no lower HR limit: n = 35 (n = 1 removed from the dataset because calculated VO2max was outside of reasonable physiologic range); submaximal HR at 25 W: n = 58; grip strength: n = 46; sit‐to‐stand: n = 61; HbA1c: n = 60; glucose, insulin, tumour necrosis factor‐alpha, interleukin‐6, and interleukin‐10: n = 58. Abbreviations: CON = stretching comparator group; ES = exercise snacks group; HbA1c = haemoglobin A1c; HR = heart rate; VO2max = estimated maximal oxygen uptake; W = watts. Log‐transformed, interpret as percent change. Discussion The primary finding of the present study was that a 12‐week remotely delivered, exercise snacks intervention was feasible for insufficiently active adults living with T2D to perform in a real‐world setting. We observed that 77% of participants performed > 13 exercise bouts on ≥ 8 of the 12 intervention weeks, which exceeded our pre‐defined threshold of > 70% adherence. We also found a greater increase in 30‐s sit‐to‐stand capacity in the ES group compared to the CON group and a within‐group reduction in waist circumference in the ES group after training, suggesting improved physical capacity and body composition with vigorous exercise snacks. However, other clinical efficacy markers including HbA1c did not differ between groups, likely reflecting the samples' well‐controlled baseline glycaemia. To our knowledge, this is the first trial to demonstrate the feasibility of a real‐world exercise snacks intervention in those living with T2D. Total app‐based adherence revealed that 79% of participants were high adherers to the intervention, completing > 168 exercise bouts across 12 weeks. Consistent with our findings, previous home‐based exercise snacks interventions in non‐T2D participants support high exercise adherence and feasibility [10, 11, 12, 13]. In the current study, exercise enjoyment was between 4 and 5 out of 7 representing “moderate” to “quite a bit” on the Exercise Enjoyment Scale [21, 22], aligning with previously reported exercise snacks studies performed in lab [23] and real‐world [13] settings. Furthermore, reported exercise intensity was higher than previous real‐world exercise snacks interventions, averaging ~6/10 versus ≤ 4/10 [10, 13] which aligns with previous lab‐based exercise snacks interventions, where ratings were between ~5 and 7/10 [23, 24, 25]. Although not captured in all participants, the 1‐min exercise snacks were associated with a peak HR response of ~73% of APHRM, suggesting that participants approached common vigorous‐intensity thresholds of > 76% HR maximum [26]. Together, these findings support the feasibility of a remotely delivered app‐based exercise snacks intervention to increase short bouts of vigorous activity in people living with T2D. We did not observe any between‐group differences in the change of HbA1c, blood markers of cardiometabolic health, body mass, BMI, or waist circumference. Although a previous highly controlled lab‐based study showed that hourly stair‐climbing exercise snacks improved insulin sensitivity over a 9‐h period in adults with overweight or obesity [27], markers of metabolic health did not differ in this 12‐week intervention. It is possible that acute metabolic changes to exercise snacks do not translate to chronic adaptations or that the small dose of prescribed exercise in the current study is not sufficient to elicit improvements in glycaemic regulation. However, our sample had well‐controlled glucose at baseline (mean HbA1c of ~6.6% and mean fasting glucose of 6.5 mmol/L) which may have limited the ability to see improvements. Similarly, we did not observe between‐group differences in the change in physical outcome measures (predicted VO2max, submaximal exercise HR responses, and grip strength). We did, however, see a greater improvement in 30‐s sit‐to‐stand capacity in the ES group, which is a key indicator of lower body strength and fall risk in older adults [28, 29]. The modest duration of the prescribed exercise may explain the absence of change in secondary outcomes. However, sit‐to‐stand capacity may be particularly responsive to bodyweight style exercise snacks used here, thus introducing an efficacious way to improve physical capacity in this population. Strengths of this study include the real‐world nature of the intervention, the autonomy given to participants to select between two daily exercise choices, and the use of push notifications and phone call check‐ins to promote adherence. This study was not without limitations. It is possible that the different adherence measurement techniques used may not have fully captured all the exercise bouts completed by the participants. It also appears that despite high adherence and enjoyment, participants may have had difficulty performing exercise snacks at a vigorous enough intensity when unsupervised in the real world, at least based on Fitbit HR responses. The increase in self‐reported physical activity in both groups may introduce an additional factor that could have reduced any between‐group differences in secondary outcomes. The lack of change in metabolic markers may also be attributed to the relatively short (12 week) duration of the intervention. Additionally, we did not assess follow‐up adherence beyond the 12‐week intervention. We also found that the use of the submaximal YMCA predictive fitness test for estimating VO2max performed poorly within this study population, with only ~30% of participants completing a valid test according to the established protocol [18, 19]. Future work should investigate more suitable fitness testing options for this population. In conclusion, 12 weeks of exercise snacks were feasible to perform in a real‐world setting and improved sit‐to‐stand capacity in insufficiently active individuals with T2D. Continued research is needed to determine whether variations in real world exercise snack prescriptions can yield clinically meaningful improvements in health related outcomes. Funding This work was supported by Diabetes Canada (OG‐3‐22‐5623‐JL). Conflicts of Interest M.J.G. is an advisor to and holds equity in Longevity League Ltd., a US‐based company whose services in part relate to exercise. J.P.L. is Chief Scientific Officer of the Institute for Personalized Therapeutic Nutrition (a Canadian registered charity) and holds founders' shares in Metabolic Insights Inc., a company developing non‐invasive metabolic monitoring devices. The other authors declare no conflicts of interest. Supporting information Supporting Information: File 1 List of exercises prescribed in the exercise snacks (ES) and placebo, stretching exercise comparator (CON) group. Supporting Information: File 2 Equations used to predict VO2max from leg ergometry [1]. Supporting Information: File 3 Godin‐Shephard leisure score index. Supporting Information: File 4 Participant health and demographic information. Supporting Information: File 5 Data separated by sex. Supporting Information: File 6 Exercise bouts per week for both Exercise Snacks and Low‐intensity placebo groups. Data presented as estimated marginal means (95% CI) with p‐values derived from LMM. Acknowledgements This project was supported by a Diabetes Canada Operating Grant: End Diabetes 2022 Award. F.J.B. is supported by a Canadian Institutes of Health Research Canadian Graduate Scholarship—Doctoral. A.M.‐C. was supported by the Canadian Institutes of Health Research, Fonds de recherche du Québec—Santé, and Michael Smith Health Research BC postdoctoral fellowship. K.F. is supported by a postdoctoral fellowship from the American Heart Association (25POST1365227). J.P.L. is University of British Columbia Okanagan Principal's Research Chair in Metabolism. Babir F. J., Marcotte‐Chénard A., Sandilands R. E., et al., “Exercise Snacks Are Feasible to Perform in the Real World and Improve Physical Capacity for Adults Living With Non‐Insulin Treated Type 2 Diabetes: A Randomised Trial,” Diabetes, Obesity and Metabolism 28, no. 10 (2026): 9294–9303, 10.1111/dom.71052. Handling Editor: Johan Jendle Data Availability Statement Upon reasonable request to the corresponding author, data can be made available. References Associated Data Supplementary Materials Supporting Information: File 1 List of exercises prescribed in the exercise snacks (ES) and placebo, stretching exercise comparator (CON) group. Supporting Information: File 2 Equations used to predict VO2max from leg ergometry [1]. Supporting Information: File 3 Godin‐Shephard leisure score index. Supporting Information: File 4 Participant health and demographic information. Supporting Information: File 5 Data separated by sex. Supporting Information: File 6 Exercise bouts per week for both Exercise Snacks and Low‐intensity placebo groups. Data presented as estimated marginal means (95% CI) with p‐values derived from LMM. Data Availability Statement Upon reasonable request to the corresponding author, data can be made available.