Study outline

This explorative study was part of another study investigating the proof of concept of a biological age model following a 14-week lifestyle intervention in people with overweight and obesity (unpublished). The study was carried out in the spring of 2022 (n = 46) and 2023 (n = 27) at “Ubberup Højskole” (UBH) in Denmark. UBH is a privately run Folk high school that offers a 14-week course on healthy living and lifestyle changes, where students live and voluntarily participate in daily lectures and activities at the school. The school typically enrolls around 70–80 students per course and most of the students have overweight or obesity. The study inclusion criteria were age above 18 years and a body mass index between 25 and 50 kg·m–2. The exclusion criteria were previous or current cardiovascular disease, pregnancy, and conditions preventing the performance of an exercise test to exhaustion. Participants received oral and written information about the study and the associated risks before written consent was obtained. This was an explorative study and not a confirmatory study of a well-defined hypothesis, therefore a sample size calculation was not conducted. The plan was to include as many participants as possible. The study adhered to the principles of the Helsinki Declaration and was approved by the Science Ethical Committee of the Greater Region of Copenhagen, Denmark (2022: H-19073643) and (2023: H-23012857). The intervention of the study is registered at ClinicalTrials.gov (NCT04279366).

Lifestyle intervention

The course provides lectures on physical activity, health and nutrition, personal development, creative expression, and social studies and amounts to between 16 and 21 h of education per week. Furthermore, different physical activities, such as volleyball, swimming, cycling, walking, etc. are organized and provided daily. Food is also provided by the school with 4–6 daily meals and a particular emphasis is placed on healthy good-tasting food that is inspiring and stimulating to all the senses. It is recommended by the school that 50% of the plate should be vegetables, 25% starchy food, and 25% protein-containing foods. The students are provided with the opportunity to follow an individual diet, which guides them in portion size and structured eating times. It is the student’s own choice and responsibility to follow lectures, participate in activities, and/or follow the food guidance during the 14-week course.

Overall experimental design

Testing was performed over three consecutive days in the first week and the last week of the 14-week course at UBH. Participants arrived in groups of five every 1.5 h from 07.00 to 11.30 following an overnight fast for measurement of anthropometry, body composition with bioimpedance (Tanita MC-780MA P, Tanita Corporation, Tokyo, Japan), blood pressure in supine position in triplicates separated by two minutes (Boso Medicus Control, BOSCH + SOHN GmbH u. Co. KG, Jungingen, Germany), estimation of V̇O2peak using SCG and completion of the International Physical Activity Questionnaire (IPAQ) short form for assessment of self-reported physical activity. Data from the body composition and blood pressure measurements will be presented as part of the biological age study (unpublished). The participants then arrived in the same groups of five on the same day for exercise testing occurring between 12.30 and 17.00 h.

Estimation of V̇O2peak using Seismocardiography

The SCG V̇O2peak estimation procedures were performed as previously described [20]. Non-invasive cardiac vibrations on the sternum were recorded at supine rest with a small medical device (Seismofit®, VentriJect Aps, Hellerup, Denmark) containing a three-axis digital output accelerometer. The Seismofit device was placed with double adhesive tape on the sternum two centimeters proximal to the processus Xiphoideus. The Seismofit device is connected to a smartphone app with a cloud-based solution for SCG signal processing and algorithm-derived eV̇O2peak using before-entered age, sex, weight, and height of the participant in the app. The entire measurement lasts approximately three minutes (40-second SCG recording, 2 min data transferring and signal processing). The app then provides the estimated V̇O2peak score together with a resting heart rate, however, the recorded SCG signal is not available. The 4.6 version of the SCG V̇O2peak estimation model was used in this study (Schmidt et al., major revision in NPJ Cardiovascular Health 2024). Schmidt et al. will describe the development of this model version and it is presented in Table 2. When the influence of the SCG in the V̇O2peak model in the 4.6 database is assessed by comparing a regression model using only age, sex, weight, and height as variables with the model including SCG, the overall total explained variance increase from 0.65 to 0.76% in the model with SCG and the standard error of estimate (SEE), and mean absolute percentage error (MAPE) decrease from 6.1 ml·min–1·kg–1 to 5.0 ml·min–1·kg–1 and from 11.8% to 8.9%, respectively.

Estimation of V̇O2peak using other non-exercise equations

Two other non-exercise V̇O2peak equations were chosen to compare the performance of the SCG V̇O2peak estimation model (Table 2). The equation by Schembre and Riebe [21] was used because this equation includes self-reported activity level obtained from IPAQ which was also obtained in the present study as a part of the biological age design. The equation by Myers et al. [22] was used as this is considered the best reference equation that uses the same demographic variables used in the SCG equation. In addition, the regression model without SCG and only demographics from the 4.6 database (No SCG eV̇O2peak) was applied to investigate the influence of SCG within the model in this population.

V̇O2peak exercise testing

A graded exercise test protocol for determination of V̇O2peak was performed on a stationary cycle ergometer (Monark 839E, Monark Exercise AB, Vansbro, Sweden) with pulmonary gas exchange rates measured continuously using a mixing chamber and online equipment (Quark CPET, Cosmed, Rome, Italy). Pulmonary gas exchange measurements were obtained using 30-s running average with sampling every 10-s automatically by the software (Omnia, Cosmed, Rome, Italy). Two identical setups of cycle ergometers and online equipment were used simultaneously to keep up with the flow of participants. The applied exercise protocol was a 5-min warm-up at 30 W for women and 50 W for men, followed by increments of 20 W and 25 W every minute until voluntary exhaustion for women and men, respectively. Three experienced physiologists performed the testing and evaluation if the participant was at voluntary exhaustion. The participants were verbally familiarised with the test beforehand and encouraged throughout the entire test to exert their absolute best. The V̇O2peak was determined as the highest V̇O2 measured during 30 consecutive seconds. According to often-used test validity criteria, 69% of baseline and 65% of follow-up tests showed a VO2-plateau, defined as less than 2.1 ml·min–1·kg–1 with increasing workload. Only three baseline tests and three follow-up tests did not meet a VO2-plateau or at least two out of three secondary criteria (respiratory exchange ratio >1.1, within 10 bpm of age-predicted maximal heart rate, and 18≥ on the Borg 6–20 rate of perceived exertion scale) but these are included in the data analysis. The cycle ergometers and online equipment were calibrated according to the manufacturer’s instructions using a compressed gas mixture (5% CO2 and 16% O2) for the gas analyzers and a 3 L calibration syringe for the digital flowmeter in the mixing chamber setup, in the morning and after every other test. Participants were measured with the same equipment at baseline and at the follow-up test.

Statistics

Data were normally distributed and presented as means ± 95% confidence intervals (CI) unless otherwise stated. The agreement and estimation error between measured and estimated V̇O2peak were analyzed using Bland-Altman plot (BA-plot) with 95% limits of agreement (LoA) [23], Standard Error of Estimate (SEE), and Mean Absolute Percentage Error (MAPE). Pearson product-moment correlation coefficient (Pearson’s r) was used to evaluate the relationship between V̇O2peak and eV̇O2peak. SEE was calculated using: \(=\surd \sum \frac^ }\right)}^}\), with Y representing V̇O2peak and Y’ representing eV̇O2peak values. To evaluate the applicability of the eV̇O2peak equations, the agreement in classifying participants into V̇O2peak tertile groups was assessed at baseline and follow-up by Cohen’s κ coefficients and the 95% CI and interpreted as previously described [24]. The participants were divided into V̇O2peak tertiles based on age and sex using the Fitness Registry and the Importance of Exercise National Database (FRIEND) [25] and classified as having “lower” V̇O2peak if they were below the 33rd percentile, “mean” V̇O2peak if they were between the 33rd and 66th percentile, and “higher” V̇O2peak if they were above the 66th percentile. For participants completing the intervention, differences between baseline and follow-up were analyzed with a student’s paired t-test, except for V̇O2peak (absolute and relative) which was also analyzed across methods (eV̇O2peak and V̇O2peak) by two-way ANOVA repeated measures. Results from the two-way ANOVA are presented as mean difference and [95% CI of difference], with the intervention: follow-up (B)—baseline (A), method: measured (1)—estimated (2), interaction; (B1-A1)—(B2-A2). A significance level of α = 0.05 was applied. The performance of the eV̇O2peak equations in the detection of changes in V̇O2peak was based on Pearson’s r and BA-plot between changes. Absolute eV̇O2peak was found by multiplying the body mass of the participant to the estimate. Statistical analyses were performed, and figures were constructed in GraphPad Prism 10.1.1 (Software Inc., Boston, Massachusetts, USA) and Microsoft Excel (Microsoft Corporation, Redmond, Washington, USA).