Web-based exercise versus supervised exercise for decreasing visceral adipose tissue in older adults with central obesity: a randomized controlled trial

This study was a two-armed randomized controlled trial conducted in Umeå, Sweden, during January 2018 – November 2018. Participants were randomized 1:1 to an intervention group or a wait-list control group (ClinicalTrials.​gov registration no. NCT03450655). An overview and timeline of the study, follow-up assessments and delivery of interventions is presented in Fig. 1 Following randomization and baseline assessment, the intervention group received SE for 10 weeks at a university hospital research clinic, while the wait-list control group lived as usual. After this initial 10-week phase, both groups underwent follow-up assessment at the research clinic. The 10-week results of the trial have been published previously [27, 28]. Next, the wait-list control group underwent a 10-week wash-out period, after which they returned to the research clinic for another reassessment, which served as baseline in the wait-list control group’s intervention. The wait-list control group’s intervention was 10 weeks of WE, after which they returned for a final follow-up assessment. To compare the effects of the two interventions, changes in the outcome variables during weeks 0–10 in the SE group were compared to changes in the outcome variables during weeks 20–30 in the WE group

The methods, procedures, design and informed consent protocol for the present trial were reviewed and authorized by the regional research ethical review board of Umeå (Ref. 2017/521–31). The trial was conducted correspondingly to the World Medical Association’s Declaration of Helsinki and reporting of the trial was performed according to the CONSORT guidelines [29].

Participants were community-dwelling 70-year-old men and women recruited during January 2018 – February 2018 from the ongoing primary prevention study Health Ageing Initiative (HAI), described in detail previously [30]. In short, all 70-year-olds living in Umeå municipality are contacted using population registers and offered to participate in an extensive health survey. As part of the HAI protocol, several measurements are made including body composition, bone density, physical activity, blood pressure and blood markers among others. In addition, all participants are given tailored health recommendations and motivational interviewing with focus on physical activity and diet. So far, around 70% of all eligible participants have chosen to participate [30].

To be included in the present study, participants were required to be centrally obese defined as at least 1 kg and 2 kg of VAT mass for women and men respectively and also pass a resting electrocardiogram (ECG; Schiller Cardiovit CS-6/12; Schiller AG). These cutoff-values for VAT were based on previous dual-energy X-ray absorptiometry (DXA) measures from > 2600 participants in the HAI-study that were centrally obese defined by waist circumference [31] (> 88 cm for women and > 102 cm for men).

All individuals with conditions that contraindicated training or affected the ability to perform the training program were excluded. Specifically, these exclusion criteria were as follows: physical disabilities; myocardial infarction or stroke during the previous 12 months; blood pressure higher than 175/100 mmHg; heart conditions that may worsen through intense exercise e.g. angina pectoris.

Statistical power was calculated based on data on VAT mass from 1200 HAI-participants (606 women and 594 men) with central obesity, defined as above. The calculations showed that a total of 33 men or 45 women per group, with somewhere in between that range for mixed groups, would provide 80% power to detect a 20% reduction in VAT, with the alpha level set to 0.05. The analyses were performed separately for men and women due to the uneven VAT mass distribution between sexes. The calculations were performed using G*Power version 3.0.10 [32].

Following baseline assessment, all participants were randomized using 80 pre-prepared and shuffled opaque sealed envelopes. Enclosed were notes indicating “Exercise” or “Control” (40 each) where the former meant allocation to the SE group and the latter meant allocation to the WE group. MB and EL were in charge of the randomization and all envelopes were shuffled before each participant was given permission to draw an envelope to ensure allocation concealment.

Participants randomized to WE received a web-based weekly progressive interval training program which was recorded by MB and EL using a Nikon D5600 digital camera and edited using Corel VideoStudio X8 (Corel Corp. Ottawa, CA) before uploaded to a website (www.​healthyageingini​tiative.​com/​kondition). The training program was delivered in terms of 10 weekly videos of training sessions where each video represented one week of the intervention and would be viewed three times weekly. For the recording of these videos, a female and male participant from the SE group were the actors showcasing the program. Please see Additional file 1 for an example of the training videos. There was also a general informative video and instructional videos for all exercises included in the training program which were showcased by MB and EL. The general informative video included information regarding exercise and central obesity in relation to cardiovascular health; detailed description of the exercise program; and safety information. Specifically, participants were prompted to contact the National e-Health Service (1177 Healthcare Guide’s e-service Support and Treatment) if they perceived any discomfort. The exercises showcased in the instructional videos were dynamic body-weight exercises designed to allow individual progression and adjustments (Additional file 2). Specifically, participants could choose to perform an easier variation or a more challenging variation of each exercise. In addition, they were instructed to adjust the tempo and range of motion in the exercises to match their individual capacity. A detailed description of all exercise is available in Additional file 3. Work-to-rest ratio during the intervals was 40 s/ 20 s and the intervals were to be performed at a vigorous level of intensity, prescribed using a modified scale for rate of perceived exertion (RPE), based on the Borg CR10 scale [33]. Volume increments were applied simultaneously for all participants in terms of progressively adding the number of working sets throughout the 10 weeks of training, thus increasing the training duration. Initially, the duration of interval training was 18 min, with a gradual increment to a final maximum duration of 36 min. In the week before the start of the intervention, all participants in the WE group were invited to partake in one supervised session led by MB and EL with the aim of providing the participants with practical advice regarding performance of the exercise program and clarifying potential ambiguities. All participants were then given a suspension band that was required for a few of the exercises, and they were encouraged to use the bands for support of balance when necessary to reduce the risk of falling. Finally, AH presented the website and intervention material to all participants and provided point-by-point instructions in how to access and use the website and intervention material.

Participants randomized to the SE group performed the exact same training program at a university hospital research clinic in groups of 8–10 participants under supervision from two fitness instructors (MB and EL). Volume increments were identical to those in the WE group.

All assessments were performed during daytime, with the ambition of standardizing and matching the exact time-of-day of assessments as far as possible. The research nurses in charge of assessments were blinded at all occasions apart from the final follow-up where blinding was not possible. All participants were informed to refrain from exerting any intense physical activity or consuming alcohol the day prior to each assessment, while being in a fasted state for at least four hours. Participants self-reported medication, smoking, diabetes, and any history of myocardial infarction or stroke. Blood pressure was measured using the digital automatic blood pressure device Omron M6 Comfort HEM-7221-E (Omron Healthcare, Kyoto, Japan) after a 15-min rest with the subjects in a supine position. Two measurements were made whereof the lowest one was recorded. Data on daily physical activity were obtained from the HAI-study where the participants physical activity were objectively assessed during one week of registration using Actigraph GT3X+ accelerometers (Actigraph, Pensacola, FL, USA), described in depth elsewhere [30].

The primary outcome for this trial was the mean change in VAT mass (grams), measured using a Lunar iDXA device and the CoreScan application (GE Healthcare Lunar, Madison, WI, USA). The CoreScan application utilizes algorithms which distinguishes VAT from subcutaneous fat mass, thereby allowing quantification of VAT. Trials using iDXA CoreScan measurements have previously reported a test-retest precision error (RMSSD; root mean square standard deviation) of 41.4 g VAT in overweight people [34].

Data was inspected visually through histograms to assess normal distribution. Tests for significant differences between the groups prior to their interventions were performed using χ²-tests on categorical variables and t-tests for independent samples on continuous variables as data was deemed normally distributed. Within-group changes in the outcome variables were analyzed using t-tests for paired samples following that data was assumed normal in distribution. The adherence to WE was presented as median and interquartile range (IQR) and participant experiences based on the survey answers were presented as percentage.

The differences in changes in outcome variables from pre- to post intervention between the WE- and SE group were assessed using a series of analyses of covariance (ANCOVAs) with adjustments for baseline values. Effect sizes (ES) for between-group differences were calculated using Cohen’s δ [35]. All analyses were conducted on an intention-to-treat basis using SPSS software v25.0 (IBM Corp. Armonk, NY), with the significance level set at P ≤ 0.05 for all analyses.

From the 90 eligible individuals, seven individuals were absent during baseline assessments and another six individuals did not pass the resting ECG, resulting in exclusion. Eventually, N = 77 individuals were randomized to SE (n = 38) and WE (n = 39). Two participants in the SE group were unable to complete the trial and were excluded because of discomfort following training sessions (n = 1) and illness unrelated to the present intervention (n = 1). A total of five participants in the WE group withdrew from the study. Of these, three withdrew between week 0 and week 11, while the other two withdrew between week 21 and week 32. Consequently, 36 and 34 participants in the SE and WE group respectively completed the trial. A detailed figure of the participant flow is available in Fig. 2.

Participant characteristics are presented in Table 1. Both groups were similar on all variables except age and blood pressure. Mean VAT mass prior to the intervention was 2.10 ± 0.95 kg and 2.30 ± 0.82 kg for the WE group and the SE group respectively. Mean BMI prior to the intervention was 28.7 ± 3.5 kg/m² and 28.7 ± 3.1 kg/m². The sex distribution was similar in both groups (53% women in the WE group).

The WE intervention had no significant effect on the primary outcome VAT from pre- to post-intervention, although FM decreased by 450 g (95% confidence interval [CI], 37 to 836, P < 0.05), corresponding to a 1.4% decrease in FM, and BFP decreased from 39.0 ± 5.8% to 38.7 ± 6.0% (95% CI, 0.1 to 0.6, P < 0.05) (Table 2, Fig. 3). WE had no significant effect on the remaining outcomes (Table 2). The median adherence was 85% (IQR, 59–100%). Regarding experiences of the intervention, the questions asked (scale 1–5) at the follow-up assessment showed the following result: a) 97% rated the information prior to the intervention > 4 b) 94% rated the design of the exercise-videos > 4 c) 87% rated the sample of exercises > 4 d) 81% answered > 4 as for the probability that they would continue with this type of training.

Comparing the effects of the interventions, there was no statistically significant difference in decrease of VAT mass, although the SE group decreased VAT by 100 g (ES, 0.5, 95% CI, − 24 to 223) compared to the WE group. Moreover, the SE group decreased FM by 619 g (ES, 0.5, 95% CI, 22 to 1215, P < 0.05), compared to the WE group. No significant differences on the remaining outcomes were observed (Table 2, Fig. 3).