Study design and participants
This was a prospective observational study conducted across Germany. Between February 2017 and December 2019, 2370 volunteers registered for the study by phone, email or on the study website. Of those, 464 could not be included as they did not meet the inclusion/exclusion criteria or did not sign written consent. Of the remaining 1906 participants, 20 from the e-bike group and seven from the bicycle group could not be evaluated because of missing data. Finally, 1250 e-bikers and 629 conventional cyclists were included in the analysis.
This study was carried out following the Declaration of Helsinki. The institutional ethics review board of Hannover Medical School approved the study (No 7237), and written informed consent was obtained before the inclusion of participants.
Recruitment of participants, inclusion and exclusion criteria
To recruit participants, we provided information material to local bicycle shops across Germany cooperating with the largest bike association in Germany (bike shopping cooperative (ZEG)). In addition, the study was advertised in print media and on a specially designed website (http://www.ebike-gesundheit.de/). According to the prestudy defined criteria, we included female and male volunteers aged 18 years or older who had their main residence in Germany. Exclusion criteria were orthopaedic, cardiovascular or other diseases restricting bicycle use or participants with no access to a smartphone or computer with internet access for data transmission. Competitive cyclists were also excluded from the study. Those interested in the study could register by phone, email or directly on the study website. During registration, the following were recorded: full name, gender, contact details, type of bicycle, date of bike purchase, and the response to questions concerning the exclusion criteria.
Group allocation
After registration for the study, eligible participants were sent a declaration of consent, information on data privacy and a medical history questionnaire by post. When meeting the study criteria and signing informed consent, volunteers were included according to their bike use in the e-bike group or the conventional bike group.
Questionnaires
We distributed questionnaires for the evaluation of the health-related quality of life (Short Form 36),17 for daily physical activity (Freiburger Physical Activity questionnaire),18 as well as a medical history questionnaire, a specially designed bicycle-specific user questionnaire and an accident documentation form (for more information on questionnaires see online supplemental information).
Observational period and procedures
After inclusion, participants started a consecutive 4-week observational period. All participants received an examination package consisting of the study-related questionnaires and an activity tracker (a smartwatch: Forerunner 35, Garmin, Garching, Germany) with a user manual and individual access data. The package also contained a sticker to attach to the bike to remind users to start and stop the tracking of cycling activities and a flyer with general safety information for cyclists in road traffic.
Participants were asked to record every bicycle ride by selecting and starting the bicycle profile on the smartwatch. Once started, the tracker records the riding time, the travelled distance (by GPS), and the heart rate (HR) via photoplethysmography. After stopping the cycling trip on the tracker, all activity data were saved on the tracker and transmitted to the manufacturer’s server (Garmin). Data were then extracted from the Garmin server, pseudonymised, depleted from GPS information about the exact location of the ride, and directly forwarded via an interface (API) to a server at Hannover Medical School, according to current privacy policy legislation. The resulting data were stored and analysed by the Institute of Biometry at Hannover Medical School.
The primary endpoint was the proportion of participants reaching the WHO recommendation for moderate to vigorous physical activity (MVPA) (≥150 min/week moderate intensity or ≥75 min/week vigorous intensity, or a combination of both) by cycling. Based on the recommendation of the American College of Sports Medicine (ACSM),19 moderate intensity was defined as an activity with a heart rate of 64–76% of the maximum heart rate (HRmax), and vigorous intensity as an activity with a heart rate above 77% of HRmax. The HRmax was calculated for each participant according to Whaley et al 1992,20 considering age, sex, smoking status, body weight and the resting heart rate of participants. Where not all parameters were available, the maximum heart rate was estimated by a simplified calculation (HRmax=208–0.7 x age).21 For each participant, recorded activity, moderate and vigorous intensity levels were determined at 1 s intervals. Overly long activities (>12 hours/day), very short tracked activities (<10 s) as well as activities with implausible heart rates or speed (mean heart rate ≤60 bpm or ≥200 bpm, mean speed ≤5 km/h or ≥40 km/h) were excluded from the analysis. According to the applicable WHO recommendations at the time of study initiation,22 a tracked activity was only counted as MVPA if the heart rate stayed above the lower threshold of the respective intensity level (moderate or vigorous) for at least 10 consecutive minutes. If the heart rate fell below the lower threshold for more than 1 min, we considered the preceding and subsequent physical activity to be separate activities. Vigorous activities counted double for the calculation of cycling-related MVPA minutes per week.
Statistical analysis
In the primary analysis, the difference between the study groups in reaching the success rates (cycling at least 150 min/week at MVPA) was tested by Χ²-test with a one-sided significance level of 2.5% and a non-inferiority margin of −7.5%. In addition, we performed a sensitivity analysis according to the ‘2020 WHO guidelines’23 that states that every MVPA activity counts (regardless of the criterion of at least 10 consecutive minutes). Subgroup analyses were performed for the following subgroups: sex (male/female), age (<53/≥ 53 years), comorbidities (yes/no), body mass index (<25/≥25 kg/m2), use of heart rate lowering drugs (yes/no), smoking status (yes/no), monthly net income and main purpose of use (every day use; commuting, leisure time, sports-related). Univariate binary logistic regression models were used to identify potential prognostic factors and confounders (p<0.1) influencing the success rate of reaching the physical activity target. In multiple binary logistic regression analyses, we used backward selection to drop independent variables with the highest p value until only those covariates and factors that were significantly associated with reaching the physical activity target remained in the model (p<0.05).
In secondary analyses, categorical and continuous outcomes (such as the average heart rates during cycling, the frequency of cycling (number of cycling trips per week)), and overall cycling time (all cycling activities independent of cycling intensity) were compared between the study groups with a Χ²-test and a two-sample t-test, respectively. Analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, North Carolina, USA) and R version 4.1.0 (R Foundation for Statistical Computing, Vienna, Austria). Data are given as absolute/relative frequencies per category or mean±SD
The sample size calculation was based on a previous feasibility study24 among workers from companies located in the Hannover area. The study showed that 26% of cyclists reached the WHO criteria for physical activity. With the anticipated 2:1 recruitment ratio, 1200 participants (800 e-bikers, 400 cyclists) needed to be enrolled to show non-inferiority of e-bikers compared with cyclists with a pragmatically justified non-inferiority margin of −7.5%, which was supposed to address the balance between the precision of the estimate and the ability to manage the trial. The one-sided significance level was set to 2.5% and the power to 80%. Another 200 participants were added to take account of possible dropouts, resulting in a total sample size of 1400 participants.