Bourne et al.

International Journal of Behavioral Nutrition and Physical Activity

(2018) 15:116
https://doi.org/10.1186/s12966-018-0751-8

REVIEW Open Access

Health benefits of electrically-assisted

cycling: a systematic review
Jessica E. Bourne1,2* , Sarah Sauchelli2, Rachel Perry2, Angie Page1,2, Sam Leary2, Clare England1,2
and Ashley R. Cooper1,2

Abstract
Background: Electrically assisted bicycles (e-bikes) have been highlighted as a method of active travel that could
overcome some of the commonly reported barriers to cycle commuting. The objective of this systematic review
was to assess the health benefits associated with e-cycling.
Method: A systematic literature review of studies examining physical activity, cardiorespiratory, metabolic and
psychological outcomes associated with e-cycling. Where possible these outcomes were compared to those from
conventional cycling and walking. Seven electronic databases, clinical trial registers, grey literature and reference
lists were searched up to November 2017. Hand searching occurred until June 2018. Experimental or observational
studies examining the impact of e-cycling on physical activity and/or health outcomes of interest were included. E-
bikes used must have pedals and require pedalling for electric assistance to be provided.
Results: Seventeen studies (11 acute experiments, 6 longitudinal interventions) were identified involving a total of
300 participants. There was moderate evidence that e-cycling provided physical activity of at least moderate
intensity, which was lower than the intensity elicited during conventional cycling, but higher than that during
walking. There was also moderate evidence that e-cycling can improve cardiorespiratory fitness in physically
inactive individuals. Evidence of the impact of e-cycling on metabolic and psychological health outcomes was
inconclusive. Longitudinal evidence was compromised by weak study design and quality.
Conclusion: E-cycling can contribute to meeting physical activity recommendations and increasing physical fitness.
As such, e-bikes offer a potential alternative to conventional cycling. Future research should examine the long-term
health impacts of e-cycling using rigorous research designs.
Keywords: Electrically-assisted bicycle, E-bike, Physical activity, Health

Background associated with a lower risk of cardiovascular disease inci-

Given the high rates of global physical inactivity [1] a grow- dence and mortality [4]. The greatest gains in health out-
ing body of research has focused on the potential of active comes from active commuting are reported in the least
travel to increase physical activity behaviour and potentially active individuals [5, 6].
lead to population health benefits. Engagement in active Travel is an essential part of everyday life for most
travel, specifically commuting, has been shown to be pre- people, and the adoption of active travel represents an effi-
dictive of a lower BMI [2] and reduced risk of diabetes diag- cient way to increase daily physical activity. For example,
nosis [3]. A recent prospective study reported that active Falconer and colleagues [2] found that active commuting
commuting, involving cycling, was associated with a lower was associated with an additional 73 weekly minutes of
risk of all-cause mortality and cancer incidence and mortal- moderate to vigorous physical activity in men and 105
ity [4]. In addition, commuting by bicycle or on foot was weekly minutes in women with type 2 diabetes, compared
to those commuting using motorised transport. With half
* Correspondence: [email protected] of all car journeys in the UK being between 1 and 5 miles
1
Centre for Exercise, Nutrition and Health Sciences, School of Policy Studies,
University of Bristol, 8 Priory Road, Bristol BS8 1TZ, UK
in length [7], the substitution of many car journeys by
2
NIHR Bristol Biomedical Research Centre, University Hospitals Bristol NHS walking and/or cycling may be an achievable aim.
Foundation Trust and University of Bristol, Bristol, UK

© The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Bourne et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:116 Page 2 of 15

Due to a growing body of evidence, the UK National In- guidelines outlined by the Preferred Reporting Items for Sys-
stitute of Health and Care Excellence (NICE) now endorse tematic Reviews and Meta-Analyses guidelines [15].
active travel, with a particular focus on commuting, as a
feasible method to incorporate physical activity into daily Search strategy
life [8]. However, rates of active commuting are low The following databases were searched from their inception
[9].Common barriers to cycle commuting include the phys- to November 2017: PsychINFO, MEDLINE and Embase (via
ical constraints associated with hilly terrain, poor physical Ovid), ISI Web of Science, CINAHL complete, SPORTDiscus
fitness, lack of time and the distance to work [10]. and Scopus. Search terms were ‘pedelec’, ‘e-bike’, ‘electrically
Electrically assisted bicycles (e-bikes) have been highlighted assisted bicycle’, ‘electrically assisted cycle’, ‘electrically assisted
as an alternative method of active travel that could overcome bike’, ‘pedal-assist’, ‘electric bicycle’, ‘electric bike’, ‘electric cycle’, ‘elec-
some of the commonly reported barriers to cycle commuting tric mobility’ (see Additional file 1 for example). Reference lists
[11]. The term e-bike includes a range of designs including from all selected articles were hand-searched for relevant stud-
throttle-controlled bikes which do not require the rider to ies. OpenGrey and Google Scholar (first 20-pages) were
pedal and electrically assisted bikes which provide electrical searched using the term ‘electrically-assisted bicycle’.
assistance only when the rider is pedalling, through sensors Hand-searching occurred until June 2018.
which detect pedalling speed and force [11]. It is through
pedalling that electrically-assisted cycling may serve to in- Inclusion criteria and selection process
crease physical activity. With lower motor power and max- Studies were eligible for inclusion if they met the follow-
imum speeds compared to throttle-controlled e-bikes, ing criteria:
electrically-assisted bikes are legally classified as bicycles.
[11]. For this review the term e-bike will be used exclusively 1) participants: adults ≥18 years of age,
to refer to electrically-assisted bicycles which require the 2) electrically-assisted bicycle must have pedals and be
rider to pedal. operated by the individual, with assistance available
In recent years e-bikes have become commonplace in Euro- from an electric motor
pean countries [11] with projected global sales of 47.6 million 3) at least one of the following outcomes; objective
by the end of 2018 [12]. E-bikes are increasingly used for both measure of physical activity intensity whilst e-
leisure and commuting purposes [13]. The assistance provided cycling (e.g., metabolic equivalents, energy expend-
has been reported to motivate novice cyclists and increase the iture), cardiorespiratory, metabolic or quality of life
likelihood that these individuals will continue to cycle in the fu- (as a measure of psychological health),
ture [10]. Given the increasing interest in e-bikes, and their use 4) type of study: experimental or observational studies.
for active travel, there is a need to understand their potential to
promote physical activity of a sufficient intensity to gain clinical Studies could be published or unpublished in any lan-
benefit (i.e., moderate-to-vigorous intensity [14]) and to exam- guage. For articles in a language other than English the title
ine their impact on broader health outcomes. Such research is and abstract were translated using Google Translate. If full
required to inform relevant health economic assessments and text screening was required, the article was translated by an
public health policy. To date, there has been no systematic re- individual fluent in the language. Studies were excluded if
view on the physical activity intensity and health outcomes as- they reported using bicycles that did not require the individ-
sociated with e-cycling. As such the aims of this systematic ual to pedal to provide power, were review articles or com-
review are to answer the following research questions: mentary pieces, and/or used self-reported measures of
physical activity. Title and abstract screening was conducted
1. What is the intensity of physical activity associated by two reviewers independently (J.E.B. and S.S.). There was a
with riding an e-bike? 93% agreement between reviewers on title and abstract
2. Does use of an e-bike lead to changes in health out- screening. Full texts were screened by the two reviewers in-
comes including cardiorespiratory, metabolic or dependently and any discrepancies were discussed.
psychological outcomes?
3. Do physiological responses to riding an e-bike differ Quality assessment and strength of the evidence
to those generated by other modes of active trans- The quality of included studies was assessed using the
portation (i.e. walking and conventional cycling)? Quality Assessment Tool for Quantitative Studies (EPHPP;
[16]). The tool appraises studies on six components; 1) se-
lection bias, 2) study design, 3) control of confounders, 4)
Methods blinding, 5) reliability and validity of data collection
A review protocol was registered at the PROSPERO database: methods and 6) withdrawals and dropouts. Each compo-
Registration number CRD42018086544 (http://www.crd.york.a- nent was rated as; strong, moderate or weak for each study
c.uk/prospero). This review was conducted according to the based on outcomes of interest.
Bourne et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:116 Page 3 of 15

A global rating for each study was then determined randomized (Table 1). Nine studies were conducted in
based on the criteria; 1) strong when no weak ratings were Europe and two in the USA. Sample sizes ranged from 3
reported, 2) moderate when one weak rating was reported, to 22 with a total of 147 participants. Participants were
and 3) weak when two or more components were rated as aged between 20 and 70. Three studies recruited physic-
weak. This tool has been used in a previous review exam- ally inactive individuals [18–20] and one study included in-
ining the impact of cycling on health [6]. The blinding dividuals with coronary artery disease [21]. Six studies
component was not included in the overall study rating as compared e-cycling to conventional cycling [18, 21–25]
participants are unable to be blinded to condition alloca- and five compared e-cycling with assist to riding an e-bike
tion following randomisation in physical activity interven- without assistance [19, 20, 26–28]. Two studies included
tions. The overall strength of the evidence was assessed walking as a comparator [18, 23].
based on previously specified best evidence synthesis cri- Rest periods between conditions ranged from 2-min to 1
teria [17] (Additional file 2). month and distance ridden from 3.54 to 27 km. Nine stud-
ies were conducted in a natural setting with topography
Data extraction and synthesis ranging from flat to elevations between 33.5 and 260 m.
Members of the review team (J.E.B and either S.S. or Four studies specifically examined the impact of topog-
A.R.C) independently extracted data for each study. raphy on physiological outcomes by separating rides into
Quality assessment was confirmed by a fourth reviewer different topographical sections (Additional file 3). Four
(R.P.). Data were extracted using an adapted version of a studies required participants to stop and go during rides to
Cochrane Data Extraction Form, which was piloted prior to simulate typical riding conditions [20, 26] or delivering mail
use. Discrepancies regarding data extraction were resolved [24, 25]. In seven studies participants were instructed to
through discussion between reviewers. Data extracted ride at a self-selected pace.
included study design, characteristics of participants,
outcomes measured, and results. Due to the heterogeneity Longitudinal studies
of study design and outcomes reported, a meta-analysis Six studies examined the longitudinal impact of
was not deemed appropriate. Data were synthesized and e-cycling, using a variety of study designs (Table 1).
presented narratively. The effect of the intervention on All studies were conducted in high income countries
physical activity and health outcomes for each study was including Belgium, Switzerland, Norway, UK (n = 2)
summarized based on reported statistical significance and and the USA. Sample sizes ranged from 20 to 32,
effect size, both within group (pre-post) and between group with a total of 153 participants. Most participants
where possible, or by examining means or medians when were between 30 and 50 years of age. Four studies
no hypothesis testing was conducted. recruited physically inactive individuals [13, 29–31].
One study included individuals with type 2 diabetes
Results [32] and for one study the health status of individ-
A total of 4399 articles were identified through initial uals was unclear [33]. Interventions ranged from
searches (Fig. 1). After removing duplicates 2894 titles 4-weeks to 8-months in length. One study included
and abstracts were screened, resulting in 119 studies published data from mid-point of the intervention,
which underwent full text screening for inclusion. but not post intervention [33]. Three studies pro-
Sixteen articles met the criteria for inclusion plus one vided participants with guidelines on minimum rid-
included after author contact. Eleven studies assessed the ing requirements, all of which specified riding the
acute response to e-cycling (i.e., one bout of e-cycling), e-bike for commuting purposes at least three times
and six examined the longitudinal effect of e-cycling (i.e., per week [13, 29, 30].
more than one bout of e-cycling, including pre-post mea-
surements). Reasons for exclusion included no measure of Physical activity intensity
specified outcomes, study not related to e-bikes, studies Studies reported a range of outcomes related to
focused on the engineering of e-bikes, qualitative studies physical activity intensity. Given the heterogeneity
or not presenting original research. Three studies were between studies regarding route length and topog-
identified through clinicaltrials.gov but were excluded for raphy, mean values and/or percent of maximum
the following reasons: 1) data not published, 2) currently values during conditions are reported to enable
recruiting, 3) authors were not reachable. comparison between studies. Physiological outcomes
reported within the manuscript include oxygen up-
Study characteristics take, metabolic equivalents,1 energy expenditure per
Acute studies minute, heart rate and power output (Table 2).
Eleven studies examined the acute physiological impact Additional outcomes are reported in Additional file
of e-cycling using cross over designs, five of which were 4 and Additional file 5.
Bourne et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:116 Page 4 of 15

Fig. 1 Flow chart of literature search

Oxygen uptake Metabolic equivalents (METs)

Eight studies reported oxygen uptake [18, 19, 21–23, Nine studies reported mean estimated METs while
25–27]. Riding an e-bike led to a relative mean oxy- riding an e-bike at a self-selected intensity [13, 18–
gen uptake of 14.7 to 29 ml/min/kg or 51 to 74% of 23, 26, 27], which ranged from 4.9 to 8.3 METs.
maximum oxygen uptake. E-cycling required lower Overall, e-cycling led to a lower mean MET score
oxygen uptake than conventional cycling (19.3 to 37 than conventional cycling or e-cycling without assist-
ml/min/kg) or e-cycling with no assistance (22.9 to ance. However, the significance of the difference is
23.4 ml/min/kg), with statistically significant differ- inconclusive. One study reported a difference in
ences reported in four studies, one of which re- mean METs between walking and e-cycling only dur-
ported an effect size of 1.73 [19]. Walking elicited ing uphill sections [23], while another study reported
lower oxygen uptake compared to self-selected no difference between walking and e-cycling over
e-cycling [23] and e-cycling on low assist [18]. varied terrain [18].
Bourne et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:116 Page 5 of 15

Table 1 Summary of included studies

First author, Study design Participants; Clinical status Exposure Length of intervention Ride characteristics
year, country gender conditions Distance (km), Topography, Distinctive
(%female); features, Ride instructions
Age, years
(mean, SD);
BMI, kg/m2
(mean, SD)
Acute studies
Bernsten, Randomized N = 8, 25% Active adults E-bike vs. CB Trials conducted on Route 1: 8.1 km, flat route
2017, cross over Age (Mdn, IQR): (4 conditions, same day, 2-min break Route 2: 7.1 km, one hill climbed twice
Norway [22] 39(3) hilly vs. flat between trials 130 m elevation gain. Self-selected
BMI (Mdn, IQR): terrain) intensity
24(7)
Gojanovic, Non- N = 18, 33.33% Inactive adults E-bike LA vs. Trials conducted over Biking: 5.1 km, 178 m elevation gain,
2011, randomized Age: 35.7 (±9.7) E-bike HA vs. 2-days. 30-min break average gradient 3.4% Instructed to ride
Switzerland cross over BMI: 24.0 (±3.3) CB vs. between trials con- at comfortable pace maintaining 60 rpm
[18] walking ducted on same day Walking: 1.7 km, uphill, 110 m elevation
gain, average grade 6.5%
Hansen, Randomized N = 17, 13% Coronary E-bikes LA Trials conducted on 10 km, 102 m elevation change
2017, cross over Age: 64 (±7) artery disease vs. E-bike HA separate days (3–4 days No traffic or stop and go points
Belgium [21] vs. CB between) Instructed to cycle at self-selected pace
on prespecified mode
La Salle, Randomized N = 12, 50% Active adults E-bike pedal Trials conducted in 3.54 km, hill 0.64 km 11% gradient
2017, USA cross over Age: M = 25(±1), with cycling assist vs. E- same day. Average Seven pedestrian crossings participants
[26] F = 22(±1) experience bike NA time between trials 12- required to dismount and walk. Self-
Body Fat %: M min selected pace
= 16.8(±1.9), F =
23.4 (±3.3)
Langford, Non- N = 17, 35% Adults, part of E-bike vs. CB Trials conducted on 4.4 km, 1.6 km downhill (− 33.2 m), 1.8 km
2017, USA randomized Age: < 20 yrs. = e-bike sharing vs. Walking separate days flat (− 0.3 m), 1.0 km uphill (+ 33.5 m). Self-
[23] cross over 3, 20-30 yrs. = 10, system (minimum 24h rest) selected pace
31-40 yrs. = 2, >
50 yrs. = 2
BMI: M = 26.1, F
= 23.1
Louis, 2012, Randomized N = 20 (10 T, 10 Highly active E-bike NA Trials conducted on Completed on indoor trainer.
France [27] cross-over UT) adults (T) vs. E-bike LA same day. 5-min breaks Instructed to pedal at specified mode for
Age: T = 38.7 Recreationally vs. E-bike HA between trials total of 45-min at pre-specified speeds:
(±14.8); UT 28.9 active adults 15-min at 16 km/hr., 21 km/hr. and free
(±6.3) (UT) speed totaling 45-min.
BMI: T = 22
(±1.1), UT = 22.2
(±3.7)
Meyer, 2014, Non- N = 3, 0% Active adults, E-bike pedal Trials conducted on 27 km track divided in 5 sections
Germany randomized Age: 25, 25, 27 recreational assist vs. E- separate days, 1-day
[28] cross over Weight (Kg): 74, cyclists bike no apart.
71, 79 assist
Simons, Non- N = 12, 50% 42% inactive E-bike NA Trials conducted in 4.3 km, flat route, two stop and go section
2009, randomized Age: 52.2 (8.7), adults vs. E-bike LA same day. One-hour participants required to dismount and
Netherlands cross over range 32–60 58% vs. E-bike HA rest between trials. restart. Self-selected pace on pre-specified
[20] BMI: 24.5 (2.6) recreationally intensity
active adults
Sperlich, Randomized N = 8, 100% Inactive adults E-bike pedal Trials conducted in 1.9 km × 5 = 9.5 km, 200 m uphill 1, 5.9%,
2012, cross over Age: 38(±15) assist vs. E- same day. One-hour 700 m downhill, 300 m uphill 2, 5.8%, 700
Germany BMI: 25.3 (±2.1) bike no rest between trials. m flat. Self-selected pace and gear
[19] assist
Theurel, Non- N = 22, 18% Active postal E-bike vs. CB Trials conducted on Postal route, one group completed rides
2011, France randomized female workers same weekday, 1- in residential neighbourhood, the other
[24] cross over Age: M = month apart completed the ride in downtown location
41(±11), F =
34(±9)
Weight (Kg): M
= 68(±18), F =
76(±10)
Bourne et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:116 Page 6 of 15

Table 1 Summary of included studies (Continued)

First author, Study design Participants; Clinical status Exposure Length of intervention Ride characteristics
year, country gender conditions Distance (km), Topography, Distinctive
(%female); features, Ride instructions
Age, years
(mean, SD);
BMI, kg/m2
(mean, SD)
Theurel, Non- N = 10, 50% Active adults E-bike vs. CB Trials separated by 1 30-min of intermittent cycling on inside
2012, France randomized female week track alternating cycling of 10 sec duration
[25] cross over Age: F = 30 and recovery of 20 sec. Aimed to
(±12), M = 35 complete 60 m in 10 sec (average speed
(±14) = 21.6 km/hr)
Longitudinal studies
Cooper, Single group N = 20 (report Type 2 One group Up to 5 months E-bike training provided. Provision of e-
2018, UK [32] feasibility on 18) Diabetes e-bike bike for up to 5-months. Support for
Age: 58.1 (±7.9) mechanical issues provided. No instruc-
BMI: 30.2 (4.4) tion on how or when to ride bike
De Geus, Non- N = 24, 46% Inactive adults E-bike vs. Control = 4 weeks Instructed to ride e-bike at least three
a
2013, randomized Age: M = 47(±7) Control E-bike = 6 weeks times per week to commute to and from
Belgium [29] cross over F = 43(±6) work
BMI: M = 27.0
(±2.8), F = 24.7
(±4.6)
Hochsmann, Pilot N = 32, 13% Inactive adults E-bike vs. CB 4 weeks Instructed to use bike for active commute
2017, randomized Age, (Mdn, IQR): to work on at least 3-days per week, over
Switzerland controlled F = 35(34–45), 6 km. Self-selected pace
[30] trial M = 43(38–45)
BMI, (Mdn, IQR)
E-bike = 29
(27,31), regular
bike = 28 (26,29)
Malnes, Single group N = 25, 72% Inactive adults One group Up to 8 months 3 sites: 2 provided e-bikes for up to 8-
2016, pilot Age: 42(±12) e-bike months, 1 e-bike up to 3-months.
Norway [31] BMI: M = Instructed to use bike as desired. In 2-
25.4(±12.3), F = centres if e-bikes not used they were
28.7(±15.8) withdrawn from participant. Group was
separated into high and low fitness
groups based on baseline testing
Page, 2017, Non- N = 31, 80% Unclear E-bike Data reported mid-way No instructions on how to ride bike, full
UK [33] randomized Age Range: 21- commuting into intervention – 2 roadside assistance provided.
two group 55 years vs. passive months
commuting
Peterman, Single group N = 21, 70% (of Inactive adults One group 4 weeks Instructed to ride e-bike at least 3 days
2016, USA 20 in analysis) e-bike per week for at least 40-min for
[13] Age: 41.5 commuting
(±11.5).
T trained (engage in endurance sport at least 4 times per week), UT untrained (moderately active but less than 4× per week), Inactive <150 min/week of moderate
to vigorous physical activity, Active ≥150 min/week of moderate to vigorous physical activity a report as sedentary but do not specifically measure moderate to
vigorous physical activity, F female, M male, NA no assistance, LA low assistance, HA high assistance, CB conventional bike

Energy expenditure per minute energy expenditure per minute while riding an
Four studies assessed energy expenditure per minute e-bike ranged from 4.9 to 6.5 kcal/min.
[13, 23, 24, 27]. On an indoor trainer, energy ex-
penditure per minute was lower on an e-bike with
assistance (high or low) compared to an e-bike with- Heart rate
out assistance in physically active adults [27]. In out- Twelve studies reported heart rate while e-cycling [13,
door trials two studies reported no difference in 18–20, 23–28, 30, 32]. During e-cycling the percentage
energy expenditure per minute between e-cycling of maximum heart rate ranged from 67.1 to 79.1. Over-
and conventional cycling, though mean values were all, mean heart rate while riding an e-bike was lower
consistently lower for e-cycling [23, 24]. Absolute than riding a conventional bike or an e-bike with no
Bourne et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:116 Page 7 of 15

Table 2 Physical activity intensity outcomes of interest measured during rides*

Study Outcomes Results; mean (SD)
E-bike Comparison Comparison Comparison Significance testing, p value
1 2 3
Bernsten, 2017 (Median, IQR) E-bike CB
[22] a
Percentage VO2max 51 (27) 58 (28) NC
Measured METs 8.5 (3.1) 10.9 (2.7) NC
Estimated METs 6.9 (2.1) 8.4 (1.8) NC
Cooper, 2018 [32] E-bike Walking
Mean HR 125.2 (18.1) 107.6 (15.8) NC
Men 121.2 (17.2) 103.2 (14.1) NC
Women 132.6 (18.9) 116.5 (16.9) NC
Percentage HR max 74.7 64.3 NC
Gojanovic, 2011 E-bike HA E-bike LA CB Walking
[18]
Mean absolute VO2peak 1.50 (.038) 1.79 (0.46) 2.00 (0.44) 1.6 (0.34) < 0.001 overall, <.05, all comparisons except HA vs.
Walk (>.05)
Percentage VO2peak 54.9 (11) 65.7 (8.1) 72.8 (6.4) 59 (9.1) < 0.001 overall, <.05, all comparisons except HA vs.
Walk (>.05)
Mean estimated METs 6.1 (1.4) 7.3 (1.0) 8.2 (1.3) 6.5 (0.8) < 0.001 overall, <.05, all comparisons except HA vs.
Walk (>.05)
Mean HR 138.4 (18) 149 (17.7) 157.0 (11.2) 132.7 (17.4) < 0.001 overall, <.05, all comparisons except HA vs.
Walk (>.05)
Percentage HR max 74.5 (8.7) 80.3 (8.7) 84.6 (5.2) 71.5 (9.2) < 0.001 overall, <.05, all comparisons except HA vs.
Walk (>.05)
Hansen, 2017 [21] E-bike HA E-Bike LA CB
Mean absolute VO2 1.72 (0.54) 1.89 (0.62) 1.85 (0.52) .02 overall, .04 LA vs. HA, > .05 CB vs. LA, CB vs. HA
Percentage VO2peak 68 (7.1) 74 (6.2) 73 (4.6) .01 overall, .03 LA vs. HA, > .05 CB vs. LA, CB vs. HA
Mean estimated METs 6 (1.8) 6.6 (2) 6.4 (1.6) .02 overall; .027 HA vs. LA; >.05, CB vs LA, CB vs. HA
Hochsmann, 2017 (Median, IQR) E-bike CB
[30]
Percentage HR max+ 74.9 (67.4, 73.3 (67.7, NC
82.8) 78.2)
Langford, 2017 E-bike CB Walking
[23] a,c
Mean relative VO2 16.95 (5.17) 19.32 (5.47) 15.12 (5.35) NC
Mean relative EE per 0.08 (0.03) 0.10 (0.02) 0.07 (0.03) NC
minute
Mean estimated METs 5.1 5.8 4.5 NC
Mean HR 121.35 127.45 115.25 NC
(17.04) (18.17) (14.41)
Mean power output 63.28 (22.89) 73.13 (35.79) NA NC
La Salle, 2017 [26] E-bike CB
a
Mean absolute VO2 2.3 (0.1) 2.5 (0.1) .45
Percentage VO2max 66.4 (2.6) 68 (2.8) NR
Mean estimated METs 8.3 (0.5) 8.5 (0.6) .65
Mean HR 147 (5) 149 (5) .064
Percentage HR max 79.1 (2.4) 80.4 (2.6) NR
Mean power output 115 (11) 128 (11) .38
b
Louis, 2012 [27] Trained E-bike HA E-bike LA E-bike NA
Mean relative VO2 14.7 (2.0) 19.5 (2.4) 22.9 (2.2) < .05, all comparisons
Mean estimated METs 4.2 (0.6) 5.6 (0.7) 6.5 (0.6) < .05, all comparisons
Mean absolute EE per 5.1 (0.8) 7.6 (0.8) 7.8 (0.5) < .05, all comparisons
minute
Bourne et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:116 Page 8 of 15

Table 2 Physical activity intensity outcomes of interest measured during rides* (Continued)
Study Outcomes Results; mean (SD)
E-bike Comparison Comparison Comparison Significance testing, p value
1 2 3
Mean HR 77.7 (11) 89.4 (10.2) 92.8 (11.6) < .05, all comparisons
Mean power output 47.3 (9.1) 83.6 (4.0) 104.2 (4.2) < .05, all comparisons
Untrained E-bike HA E-bike LA E-bike NA
Mean relative VO2 15.0 (2.0) 21.7 (4.2) 23.4 (3.6) < .05, all comparisons
Mean estimated METs 4.3(0.6) 6.2 (1.2) 6.7 (1.0) < .05, all comparisons
Mean absolute EE per 4.9 (0.8) 6.7 (0.8) 7.5 (0.9) < .05, all comparisons
minute
Mean HR 96.8 (16.8) 116.8 (21.7) 116.7 (16.2) < .05, all comparisons
Mean power output 40.0 (7.1) 79.8 (4.8) 99.9 (6.9) < .05, all comparisons
Meyer 2014 [28] a E-bike E-bike NA
Mean HR 94.71 131.31 NC
Peterman, 2016 E-bike
[13]
Mean estimate METs 4.9 (1.2
Mean absolute EE per 6.5 (1.9)
minute
Percentage HR max 72.1 (5.4)
Simons, 2009 [20] E-bike HA E-bike LA E-bike NA
Mean estimated METs 5.2 (1.4) 5.7 (1.2) 6.1 (1.6) <.05 HA and NA, >.05 HA vs. LA, LA vs. NA
Mean HR 112.4 (22.9) 116.2 (22.4) 123.8 (23.2) <.05 NA vs. HA; NA vs. LA, >.05 HA vs. LA
Percentage HR max 6 7.1 (14.1) 69.3 (13.5) 73.9 (14.5) <.05 NA vs. HA; NA vs. LA, >.05 HA vs. LA
Mean absolute power 94.2 (29.2) 101.8 (24.8) 118.2 (30.9) <.05 All comparisons
Sperlich, 2012 [19] E-bike CB
a
Mean relative VO2 18 (3.8) 25.5 (4.8) <.05, ES = 1.73
Mean absolute VO2 1.33 (0.35) 1.77 (0.43) < .05, ES = 1.12
Mean estimated METs 5.2 (1.7) 7.1 (1.4) <.05, ES = 1.22
Mean HR 105 (20) 133 (19) <.05, ES = 1.53
Mean absolute power 363 (23) 415 (28) <.05, ES = 2.02
Theurel, 2011 [24] E-bike CB
Mean absolute EE per 5.6 (1.3) 5.9 (1.8) NR
minute
Mean HR NR NR .02, 3% lower with e-bike
Theurel, 2012 [25] E-bike CB
Mean relative VO2 29 (5) 37 (5) < .001
Mean HR 136 (23) 167 (17) <.001
*
Given the difference in the cycle routes conducted mean values or percentage of maximum for outcomes related to physical activity intensity are
reported (e.g., Mean VO2peak, mean heart rate, mean energy expenditure). For additional physical activity related outcomes reported in the studies
see Additional file 4
+
reported for only a subsample of the group (n = 5 e-bikes, n = 4 conventional bike)
EE energy expenditure, HR heart rate, METs metabolic equivalent, VO2 volume of oxygen, VO2 oxygen intake value; VO2max highest oxygen intake value
attainable for an individual, VO2peak the highest oxygen intake value obtained on a specific test, CB conventional bike, HA high assistance, LA low
assistance, NA no assistance
ES effect size measured as Cohen’s d, NC not conducted, NR not reported
Relative VO2, VO2max and VO2peak measured as ml/min/kg; Absolute VO2, VO2max and VO2peak measured in l/min; Mean absolute energy expenditure
measured in kcal/min; Mean relative energy expenditure measured in kcal/kg/min; Mean heart rate measured in beats per minute (bpm); Mean power
output measured in Watts, Estimated METs measured using assumption that resting energy expenditure (i.e.,1 MET) = 3.5 ml/kg/min; Measured METs
measured through assessed individual resting energy expenditure
a
Results are reported to total cycle routes. Studies separated results for different route topography. See Additional file 3 for details on different
cycling topography; b Participants completed same activity at three different speeds, self-selected speed reported; c Total sample analyses not
conducted, see Additional file 3 for analyses between ride segments
Bourne et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:116 Page 9 of 15

assistance. Heart rate showed a trend towards being women respectively [29]. Gender differences were also re-
lower while walking compared to e-cycling [18, 23, 32]. ported in maximum power output with women reporting
lower increases in maximum power than men following a
Power output 6-week and 5-month intervention [29, 32].
Five studies assessed power output during conditions
[19, 20, 23, 26, 27]. Mean power output was lower while Health outcomes
riding an e-bike compared to a conventional bike or Three studies examined the impact of e-cycling on health out-
e-cycling with no assistance. Riding an e-bike on high comes beyond fitness (Table 3), for which the outcomes
assistance compared to low assistance led to significantly assessed were heterogeneous. After 4-weeks of e-cycling there
lower power outputs. were no changes in systolic or diastolic blood pressure at rest
Overall, e-cycling was performed at a moderate intensity, [13, 30]. There was no evidence of a difference in blood pres-
but the intensity was lower than during conventional cycling. sure whilst cycling between conventional cycling and e-cycling
Most studies reported significant differences in the associ- [30]. Peterman and colleagues [13] reported no changes in in-
ated outcomes between e-cycling and conventional cycling. sulin resistance or lipid profiles following 4-weeks of e-cycling.
However, one study found no differences in physiological However, a significant reduction in 2-h post plasma glucose
markers of intensity between e-cycling and conventional cyc- concentration was reported. No changes were reported in the
ling [26]. While the evidence is limited, e-cycling appears to one study examining quality of life following 8 weeks of
be performed at a greater intensity than walking. e-cycling [33].

Impact of topography Quality assessment and quality of the evidence

Five studies directly compared the impact of e-cycling in The global rating of acute studies yielded six moderate and
varying topographies (Additional file 3). The energy cost five weak ratings according to the EPHPP tool (Table 4). Ten
during e-cycling and conventional cycling uphill ranged studies were rated as weak for representativeness of the target
from 5.2 to 6.8 and 7.2 to 8.5 METs respectively. This dif- population, often due to a failure to report how participants
ference was statistically significant in the three studies that were recruited. Methods of assessment were rated as strong.
conducted hypothesis testing. Examination of means and The repeated nature of conditions ensured the control of con-
medians suggested that energy expenditure (METs) during founders, therefore yielding a strong rating. Overall there was
downhill and flat sections were lower while e-cycling com- moderate evidence that e-cycling could lead to physical activ-
pared to conventional cycling, but that this difference in en- ity at an intensity associated with beneficial health outcomes
ergy cost was less distinct than during uphill sections. [14]. A global rating of strong was given to one longitudinal
Across all studies, greater elevation gains in routes led to study, moderate was given to four studies and weak to one
higher energy cost for both e-cycling and conventional cyc- study. There was moderate evidence that e-cycling could lead
ling compared to flat routes or those conducted indoors. to increased fitness. The evidence related to the impact of
Differences in heart rate between e-cycling and conven- e-cycling on additional health outcomes was inconclusive.
tional cycling appear to be greater during uphill sections,
except for one study [19] that reported similar differences Discussion
in heart rate between cycling conditions across all The aim of the current review was to assess the intensity
topographies. of physical activity when riding an e-bike, and to exam-
ine the physiological and psychological outcomes associ-
Physical fitness ated with e-cycling. Where possible these outcomes
A pilot randomized control trial of physically inactive individ- were compared to traditional methods of active travel
uals reported an increase in peak oxygen uptake (VO2peak) of (i.e., walking and cycling). Eleven acute and six longitu-
10% following 4-weeks of e-cycling compared to a 6% increase dinal studies were identified. There was moderate evi-
following 4-weeks of conventional cycling [30] (Table 3). In a dence that e-cycling provides moderate intensity
similar population, using a single-group quasi-experimental physical activity in both physically active and inactive in-
design, one study reported an 8% increase in VO2peak follow- dividuals. Furthermore, there was moderate evidence
ing 4-weeks of e-cyling [13] and another reported a 7.7% in- that e-cycling positively impacted cardiorespiratory fit-
crease in VO2peak following 3-months of e-cycling [31]. When ness in physically inactive individuals. The impact of
separated into low and high fitness groups a significant in- e-cycling on health outcomes beyond physical fitness
crease in VO2peak was reported only in individuals with low was inconclusive given the sparsity of current research.
levels of fitness, with a 9.6% increase compared to a 1.5% in-
crease in high fitness individuals [31]. Gender differences were Quality of the evidence
reported in one study following 6-weeks of e-cycling with a 2 The quality of all studies, bar one [30], was weak to
and 7% increase in VO2peak in physically inactive men and moderate. These ratings should be viewed with caution
Bourne et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:116 Page 10 of 15

Table 3 Results of longitudinal intervention studies

Study Outcomes Results, mean, SD (95% CI)
Intervention Control Significance, p-value
Pre Post Pre Post
E-bike
Cooper, 2018 [32] Max absolute power 157.5 (55.7) 174.3 (70.8) NC
Men 182.1 (51.5) 206.2 (64.9) NC
Women 118.9 (38.9) 124.3 (49.0) NC
E-bike NE Within groups
De Geus, 2013 [29] Absolute VO2peak
Men 2.56 (0.35) 2.61 (0.38) 2.62 (0.46) 2.56 (0.35) >.0.025 E-bike, NE
Women 1.94 (0.37) 2.07 (0.41) 1.91 (0.35) 1.94 (0.37) >.0.025 E-bike, NE
Relative VO2peak
Men 30.2 (4.3) 30.7 (5.6) 30.8 (4.9) 30.2 (4.3) >.0.025 E-bike, NE
Women 30.0 (6.0) 32.3 (6.5) 29.4 (5.1) 30.0 (6.0) >.0.025 E-bike, NE
Absolute max power
Men 169.5 (19.9) 192.1 (28.7) 173.8 (27.1) 169.5 (19.9) <.0.025 E-bike, >.0.025 NE
Women 130.9 (21.6) 145.9 (24.8) 131.1 (21.7) 130.9 (21.6) <.0.025 E-bike, >.0.025 NE
Relative max power
Men 2.00 (0.28) 2.30 (0.40) 2.05 (0.35) 2.00 (0.28) <.0.025 E-bike, >.0.025 NE
Women 2.03 (0.41) 2.30 (0.55) 2.04 (0.43) 2.03 (0.41) <.0.025 E-bike, >.0.025 NE
E-bike CB
Hochsmann, 2017 [30] Relative VO2peak 35.7 (5.8) 39.3 (8.3) 36.4 (7.3) 38.6 (6.2) 0.327, 1.4 (− 1.4–4.1)+
Relative power output 2.9 (0.6) 3.2 (0.6) 3 (0.5) 3.3 (0.5) 0.995, 0.0 (− 0.1–0.1)+
Resting HR 64.7 (6.5) 65.1 (7.6) 68.8 (8.8) 65.5 (10.6) 0.505, 2.0 (−4.2–8.2) +
HR at 100 W max text 113.4 (9.2) 111.5 (7.7) 113.4 (15.9) 109.2 (14.2) 0.219, 2.4 (− 1.5–6.2) +
SBP at rest 125.9 (13.8) 124.1 (11.3) 127.3 (10.6) 123.1 (12.4) 0.538, 2.0 (−4.5–8.5) +
DBP at rest 82.4 (8.5) 82.1 (8.2) 87.7 (8) 84.5 (8.8) 0.625, 1.2 (−3.9–6.3) +
SBP @ 100 W 174.1 (22.9) 160.3 (21.2) 160.8 (20) 150.4 (18.5) 0.93, −0.4 (−9.4–8.7) +

DBP @ 100 W 86.2 (8.3) 81.9 (6.5) 88 (7.1) 84 (8.1) 0.709, −1.1 (−7.5–5.2) +
E-bike
Malnes, 2016 [31] Relative VO2peak 34.1 (31.6, 36.7) 36.5 (34.4, 38.6) <.001
Relative VO2peak, % gain 7.7 (4.3, 11.1)
High Fitness 1.5 (−5.6, 8.6) 0.626
Low Fitness 9.6 (5.9, 13.3) <.05
Peak HR 181 (175, 187) 180 (174, 186) 0.429
E-bike commute Passive commute
Page, 2017 [33] QOL (baseline and week 8) 38.00 (3.86) 39.67 (4.47) 29.63 (6.57) 35.71 (5.59) >.05 E-bike, Passive commute
OQL (week 4) 38.84 (4.16) 32.67 (6.08) <.01, ES = 0.28
E-bike
Peterman, 2016 [13] Absolute VO2max 2.21 (0.48) 2.39 (0.52) <.05
MVPA 28.1 (17.5) 29.0 (20.2) >.05
MVPA10+ 11.7 (14.3) 13.0 (15.2) >.05
Absolute max power 165.1 (37.1) 189.3 (38.3) <.05
Fasting glucose 4.99 (0.52) 5.02 (0.47) >.05
2 h post plasma glucose 5.53 (1.18) 5.03 (0.91) <.05
Bourne et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:116 Page 11 of 15

Table 3 Results of longitudinal intervention studies (Continued)

Study Outcomes Results, mean, SD (95% CI)
Intervention Control Significance, p-value
Pre Post Pre Post
HOMA 2.46 (0.95) 2.55 (0.82) >.05
Total cholesterol 3.90 (0.87) 3.92 (0.79) >.05
LDL 2.33 (0.8) 2.34 (0.71) >.05
HDL 1.21 (0.24) 1.18 (0.22) >.05
Triglycerides 0.95 (0.42) 0.91 (0.27) >.05
MAP 84.6 (10.5) 83.2 (9.4) >.05
SBP 110.0 (12.4) 109.1 (10.9) >.05
DBP 67.7 (8.8) 67.0 (8.0) >.05
+
difference between groups, 95% CI, ES = effect size
Distance (total and weekly) measured in kilometres; Duration (total and weekly) measured in minutes
NE no activity, CB conventional bike
SBP systolic blood pressure, DBP diastolic blood pressure, MAP mean arterial blood pressure, QOL quality of life, LDL low density lipo-protein, HDL high density
lipo-protein, HOMA measure of insulin sensitivity using homeostatic model assessment, MVPA moderate to vigorous physical activity, MVPA10+ moderate to
vigorous physical activity of bout of 10-min or greater, W watts
VO2max = highest oxygen value attainable for an individual, VO2peak = the highest oxygen intake value obtained on a specific test
Relative VO2max and VO2peak measured as ml/min/kg; Absolute VO2max and VO2peak measured in l/min Mean energy expenditure measured in kcal/min; Mean heart
rate or peak heart rate measured in beats per minute (bpm); Mean absolute max power measured in Watts, Mean relative power measured in watts/kg; glucose,
cholesterol, LDL, HDL, Triglycerides measured in mmol/L; blood pressure measured in millimeter of mercury (mmHg), MVPA and MVPA10+ measured in minutes
per day

Table 4 Quality assessment of included studies according to the Effective Public Health Practice Project tool
Study Component rating Global
ratinga
Selection Bias Design Confounders Blinding Methods Drop-outs
Acute studies
Bernsten [22] Weak Strong Strong Weak Strong Strong Moderate
Gojanovic [18] Weak Moderate Strong Weak Strong Strong Moderate
Hansen [21] Moderate Strong Strong Weak Strong Strong Moderate
Langford [23] Weak Moderate Strong Weak Strong Moderate Moderate
La Salle [26] Weak Strong Strong Weak Strong Strong Moderate
Louis [27] Weak Strong Strong Weak Strong Weak Weak
Meyer [28] Weak Weak Strong Weak Strong Weak Weak
Simons [20] Weak Moderate Strong Weak Strong Strong Moderate
Sperlich Weak Strong Strong Weak Strong Weak Weak
Theurel, 2011 [24] Weak Weak Strong Weak Strong Weak Weak
Theurel, 2012 [25] Weak Weak Strong Weak Strong Weak Weak
Longitudinal studies
Cooper [32] Moderate Moderate Strong Weak Strong Moderate Moderate
De Geus [29] Weak Moderate Strong Weak Strong Moderate Moderate
Hochsmann [30] Moderate Strong Strong Weak Strong Strong Strong
Malnes [31] Weak Moderate Strong Weak Strong Strong Moderate
Page [33] Moderate Weak Weak Weak Strong Weak Weak
Peterman [13] Weak Moderate Strong Weak Strong Moderate Moderate
a
Strong = no weak component rating; moderate = one weak component rating; weak = two or more weak component ratings
Note: blinding was not included in the overall global rating calculation
Bourne et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:116 Page 12 of 15

as the purpose of physiological studies, such as the acute uptake than e-cycling across all topographies, though
experiments reported here, is to explore a specific event significant MET differences were only reported during
in a controlled environment with less focus on obtaining uphill sections, with e-cycling expending more energy
representative samples. As such, many studies did not re- than walking. The few studies conducted suggest e-cycling
port how participants were recruited, leading to a weak rating is performed at a higher intensity than walking, however,
for the selection bias component of the assessment. Study de- more studies are needed to confirm these trends.
sign, control of confounders and methods of assessment are In relation to conventional cycling, this review suggests
often considered more crucial in these designs, all of which that e-cycling elicits lower physiological markers of inten-
were strong in the acute studies reported here. Furthermore, sity than conventional cycling, however the strength of this
while blinding is often unachievable in physical activity inter- finding depends on the physiological assessment measure
ventions, the use of objective methodology limits the impact and route topography. Overall, mean percent of VO2max/
of research bias on the outcomes. peak is similar between conventional cycling and e-cycling
Regarding longitudinal studies, methods of data collection ranging from 58 to 74% and 51 to 73% respectively. Studies
were consistently strong, but with large variation in representa- examining active commuting on conventional bikes have
tiveness, design and reporting of withdrawals and dropouts. reported similar mean percent of VO2max in healthy adults
Confounders were considered in the context of differences be- ranging from 57 to 79% [6, 37]. However, mean relative
tween groups and were therefore rated as strong if studies used oxygen uptake is lower during e-cycling compared to con-
a single-group design. One pilot randomized control trial was ventional cycling or e-cycling without assistance. Similarly,
conducted and was rated as strong [30]. Overall, there was a means and medians of estimated METs are consistently
lack of high-quality longitudinal intervention-based research in- higher during conventional cycling or e-cycling without as-
cluding pre-post measures examining the impact of e-cycling sistance compared to assisted e-cycling, with values ranging
on physiological and psychological health outcomes. from 6.1 to 8.5 and 4.9 to 8.3 respectively, though the sig-
nificance of the differences varied across studies.
The impact of e-cycling on physical activity intensity La Salle and colleagues [26] reported similar MET
To accrue health benefits, The American College of Sports values between e-cycling and conventional cycling. How-
Medicine recommend healthy adults engage in ever, the values reported were substantially higher than
moderate-to-vigorous physical activity for 150-min per those reported in other studies, with mean estimated
week [14]. Moderate intensity activity is classified as three METs of 8.3 and 8.5 for e-cycling and conventional cyc-
to six metabolic equivalents (METs) and vigorous intensity ling respectively. Participant demographics may have
activity at six METs or above. The current review suggests accounted for these differences, since participants were
that e-cycling, even while using a high assistance mode, younger and had previous cycling experience. These par-
provides physical activity of at least moderate intensity on a ticipants may have had higher aerobic capacity and
variety of terrain, including downhill. Furthermore, therefore self-selected a higher intensity activity level at
e-cycling can elicit vigorous activity during uphill riding which to complete the conditions. This is likely given
[18] and during rides with highly varied terrain [18, 26]. that the relative intensity of activity is similar in studies
Interestingly, Bernsten and colleagues [22] reported that of e-cycling in physically inactive individuals [13, 18–20,
mean estimated METs were lower than mean measured 30, 32]. When given the choice to self-select pace and
METs during e-cycling. Estimated METs have been sug- intensity individuals may select a similar physiological
gested to overestimate resting energy expenditure, thereby intensity across activities regardless of the mechanical
underestimating activity energy expenditure [34]. As such, assistance, thereby resulting in similar physiological out-
the mean estimated METs reported in this review provide a comes. In support of this, when individuals were re-
conservative estimate of exercise intensity. quired to maintain a cycling cadence of 60 revolutions
Relative physiological outcomes further suggest that per minute throughout a condition, there were signifi-
e-cycling is performed at a moderate intensity with the cant differences in oxygen uptake and heart rate between
percent of maximum heart rate ranging from 67.1 to e-bikes and conventional bikes [18] compared to studies
79.1 and the percent of VO2peak/max ranging from 51 to in which individuals were able to self-selected their inten-
75. These values exceed the hypothesised minimum in- sity [21, 22, 26]. Similarly, when instructed to complete
tensity thresholds required for improvements in cardio- 60-meters of riding in 10-sec for a total of 30-min the
respiratory fitness in healthy adults [14, 35, 36]. reported relative VO2max was 29 ml/min/kg for e-cycling
and 37 ml/min/kg for conventional cycling [25]. This
E-cycling vs. traditional active transportation suggests that performing the same amount of work
Three studies compared e-cycling to walking [18, 23, 32] requires more effort on a conventional bike than an
of which one compared the two modes on the same e-bike, but that human beings reduce the amount of work
route [23]. In this study walking led to lower oxygen conducted on a conventional bike, through choosing a
Bourne et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:116 Page 13 of 15

slower speed, to account for the increase in expended novel in the context of e-cycling and conventional cycling. In
effort. the same study, no other metabolic changes were reported.
In hilly terrain, where there is less opportunity to adjust ef- Similar null effects on metabolic outcomes were reported in
fort levels to produce comparable intensity levels, the differ- two systematic reviews on conventional cycling [37, 40].
ences between conventional cycling and e-cycling may
become more pronounced, with e-cycling requiring lower in- E-cycling for public health?
tensity activity, as found in studies comprised of routes with Overall e-cycling can elicit at least moderate intensity phys-
hilly features [18, 23]. This suggests that e-bikes are less sen- ical activity. However, total energy expenditure when riding
sitive to environmental factors such as topography. There- an e-bike is lower than when riding a conventional bike or
fore, physiological measures of intensity are lower on the walking over the same distance, given the reduced amount
e-bike than those reported on a conventional bike during up- of time taken to complete a ride on an e-bike. Consequently,
hill riding. The reduced intensity required during uphill rid- if e-cycling were to replace journeys made by walking or
ing when using an e-bike is one of the leading arguments for conventional cycling, individuals would have to ride for lon-
the promotion of e-bikes as an alternative mode of active ger for comparable weekly energy expenditure. However,
transportation. e-cycling is associated with lower ratings of perceived exer-
tion than conventional cycling [23, 26], potentially enabling
E-cycling and health people to ride more frequently or for a longer duration. This
In the current review three studies provided weekly possibility is supported by Hendriksen and colleagues [41],
e-cycling goals for physically inactive individuals in the who reported that individuals in the Netherlands commuted
context of active commuting [13, 29, 30]. Two of these 50% further with an e-bike than on a conventional bike.
studies reported increases in VO2peak and maximum Findings reported here suggest that e-cycling may be
power output following 4-weeks of e-cycling [13, 30]. In suitable for individuals with compromised health. Han-
contrast de Geus and colleagues [10] reported no changes sen and colleagues [21] showed that e-cycling elicited
in VO2peak following a 6-week intervention, though differ- moderate intensity activity in older, obese individuals re-
ences in maximum power output were seen. Differences covering from surgery due to coronary artery disease,
between studies could be due to distance cycled. Specific- while Cooper and colleagues [32] reported that e-cycling
ally, both Hochsmann [30] and Peterman and colleagues was feasible for middle-aged, overweight individuals with
[13] reported cycling distances of 70 km and 69.4 km per type 2 diabetes mellitus.
week respectively, compared to 54.3 km per week reported Overall, while there is a trend towards increased fitness fol-
by de Geus [10]. The two studies reporting significant in- lowing engagement in e-cycling interventions, more interven-
creases in fitness also described self-selected riding inten- tion research of a longer duration is required before the
sities of between 72.1 and 74.9% of maximum heart rate long-term impact of e-cycling on health can be determined.
(within the moderate intensity zone [13, 30] with an aver- Fifty percent of the longitudinal studies in this review were ap-
age of 205 min (±43.3) of e-cycling per week [13]. This proximately 1-month in length. This may not be enough time
suggests that e-cycling can contribute to meeting weekly to see changes in body composition and some metabolic out-
physical activity guidelines. comes. Longer trials with larger samples sizes should be con-
Without the provision of e-cycling goals, single group ducted with a focus on including a range of health outcomes
studies with physically inactive individuals reported in- in addition to cardiorespiratory fitness. These trials should
creases in maximal power output of 7 to 10% over 3–8 utilize randomized controlled designs and clearly report their
months, despite lower average distance travelled than target population, recruitment process and dropouts and/or
other studies [31, 32]. Fitness benefits were greatest in in- withdrawals. Interventions should also be conducted in clinical
dividuals classified as having low fitness [31], similar to populations where physical activity is compromised. In
findings with conventional cycling [6]. These results sug- addition, more research is needed to understand the impact of
gest that in the absence of specific goals (i.e., under free e-cycling on health based on sex or fitness level.
living conditions), participants engage in e-cycling and this It is also important to consider the negative outcomes asso-
e-cycling can contribute to improvements in fitness. ciated with e-cycling when assessing their potential utilization
Beyond cardiorespiratory fitness, there is a lack of research for health promotion. In the USA, e-bike users reported feel-
examining the impact of e-cycling on physiological or psycho- ing safer riding their e-bike than a conventional bike, stating
logical health outcomes, limiting our ability to draw conclu- that the e-bike helped them to avoid crashes due to their sta-
sions. Peterman and colleagues [13] reported a decrease in bility, powerful brakes and the acceleration to avoid incidents
2-h plasma glucose during an oral glucose tolerance test after and keep up with traffic. However, riders reported cycling fas-
4-weeks of e-cycling. This finding is in line with studies that ter on an e-bike than a conventional bike and felt that other
have examined the impact of exercise on 2-h post exercise road users misjudged their speed leading to potentially dan-
glucose concentrations in obese individuals [38, 39] but is gerous situations [42]. In the Netherlands data suggest that,
Bourne et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:116 Page 14 of 15

after controlling for age, gender and amount of cycling, use of health initiatives to promote e-cycling to improve popula-
an e-bike was associated with an increased risk of being in- tion health.
volved in a crash compared to conventional cycling [43]. The
severity of these crashes was not significantly different from Endnote
1
conventional cycling [43]. More context specific research is The MET is an expression of energy cost and is cal-
required to enable a risk-benefit assessment of engaging spe- culated from rest where 1 MET is estimated to equal
cifically in e-cycling. Nevertheless, e-cyclists would be well 3.5 ml/kg/min
advised to be appropriately trained and use safety equipment
to minimize risk. Additional files

Additional file 1: Example search strategy. (DOCX 12 kb)

Strengths and limitations
Additional file 2: Description of overall strength of evidence criteria.
This is the first review to examine the physical activity inten- (DOCX 12 kb)
sity, cardiorespiratory, metabolic and psychological outcomes Additional file 3: Outcomes of interest by route topography for acute
associated with e-cycling. This review used two pragmatic experimental and quasi-experimental studies. (DOCX 23 kb)
tools to assess the quality of studies and to provide an overall Additional file 4: Additional physical activity outcomes measured in
rating of the evidence. These tools provided an overall repre- acute studies. (DOCX 28 kb)

sentation of the strength of research evidence related to Additional file 5: Additional physical activity outcomes measured in
longitudinal studies. (DOCX 22 kb)
e-cycling and health. Limitations of this review include the
fact that some published studies may not have been identi-
Acknowledgements
fied. However, our systematic and broad search strategy The authors would like to thank Sarah Koch and Kylie Gobereau for their
makes this unlikely. It is more likely that we did not identify help in translating study manuscripts.
eligible unpublished studies or those published in an alterna-
Funding
tive language to English. Sample sizes used in studies were This review was supported by the NIHR Bristol Nutrition Biomedical Research
small and sample size calculations were rarely reported. Centre.
Therefore, caution should be taken when interpreting the
Availability of data and materials
statistical significance of evidence. Given the heterogeneity Not applicable.
in outcome measurement we were unable to quantify the
effects of e-cycling on outcomes of interest using Authors’ contributions
JEB conceptualized the review. JEB, RP, ARC and AP contributed to design
meta-analyses. In addition, focus on quality of life as a psy-
and search strategy. JEB, SST, ARC and RP contribute to screening, data
chological outcome may have meant studies examining extraction and quality assessment. JEB drafted the full manuscript. All authors
psychological outcomes such as depression or anxiety were read and approved the final manuscript.
excluded.
Ethics approval and consent to participate
Not applicable.

Conclusion Consent for publication

The composite results of the 17 studies included in this Not applicable.
novel systematic review provide moderate evidence that
Competing interests
e-cycling elicits activity at an intensity high enough to pro- The authors declare that they have no competing interests.
mote some positive health outcomes. E-cycling leads to re-
duced activity volume and intensity over the same distance
Publisher’s Note
compared to conventional cycling. Therefore, e-cycling re- Springer Nature remains neutral with regard to jurisdictional claims in
quires more frequent and longer rides to accrue compar- published maps and institutional affiliations.
able health benefits. However, given that most individuals
Received: 8 August 2018 Accepted: 8 November 2018
travel by car to work [44] e-cycling offers a physically active
alternative to the largely sedentary behaviour associated
with motorized commuting. Furthermore, longitudinal References
1. World Health Organization. Physical activity fact sheet; 2018. Available from:
studies suggest, with moderate confidence, that e-cycling http://www.who.int/news-room/fact-sheets/detail/physical-activity.
can lead to increases in cardiorespiratory fitness. Longer 2. Falconer CL, Cooper AR, Flint E. Patterns and correlates of active
and higher-quality intervention studies, with transparent commuting in adults with type 2 diabetes: cross-sectional evidence from
UK Biobank. BMJ Open. 2017;7:10.
reporting, are needed to develop a strong evidence-based 3. Laverty AA, Mindell JS, Webb EA, Millett C. Active travel to work and
understanding of the impact of e-cycling on cardiorespira- cardiovascular risk factors in the United Kingdom. Am J Prev Med. 2013;
tory health and to explore the impact of e-cycling on meta- 45(3):282–8.
4. Celis-Morales CA, Lyall DM, Welsh P, Anderson J, Steell L, Guo Y, et al.
bolic and psychological outcomes. This will extend the Association between active commuting and incident cardiovascular disease,
current body of knowledge and provide guidance on public cancer, and mortality: prospective cohort study. BMJ. 2017;357:j1456.
Bourne et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:116 Page 15 of 15

5. Kelly P, Kahlmeier S, Gotschi T, Orsini N, Richards J, Roberts N, et al. Systematic Fitness in Overweight Adults: A 4-Week Randomized Pilot Study.
review and meta-analysis of reduction in all-cause mortality from walking and Clinical Journal of Sport Medicine. 2017;Publish Ahead of Print.
cycling and shape of dose response relationship. Int J Behav Nutr Phys Act. 31. Malnes L, Lobben S, Bere E, Tjelta LI, Kristoffersen M, Mildestvedt T, et al.
2014;11:132. How access to an E-bike affects bicycle use and cardiopulmonary fitness in
6. Oja P, Titze S, Bauman A, de Geus B, Krenn P, Reger-Nash B, et al. Health benefits inactive Norwegian adults: A pilot study. [Master’s Thesis]. In press 2016.
of cycling: a systematic review. Scand J Med Sci Sports. 2011;21(4):496–509. 32. Cooper A, Tibbitts B, England C, Procter D, Searle A, Sebire S, et al. The
7. Department for Transport. Road Use Statistics: Great Britain 2016 2016 [cited 2018 potential of electric bicycles to improve the health of people with type 2
5th March]. Available from: https://assets.publishing.service.gov.uk/government/ diabetes: a feasibility study. Diabet Med. 2018.
uploads/system/uploads/attachment_data/file/514912/road-use-statistics.pdf. 33. Page NC, Nilsson VO. Active commuting: workplace health promotion for
8. National Institute for Health and Care Excellence. Physical activity: walking improved employee well-being and organizational behavior. Front Psychol.
and cycling 2012 [Available from: https://www.nice.org.uk/guidance/ph41/ 2017;7:1994.
resources/physical-activity-walking-and-cycling-pdf-1996352901061. 34. Byrne NM, Hills AP, Hunter GR, Weinsier RL, Schutz Y. Metabolic equivalent:
9. Department for Transport. Walking and Cycling Statistics, England: 2016: one size does not fit all. J Appl Physiol. 2005;99(3):1112–9.
Statistical Release 2018 [Available from: https://assets.publishing.service.gov. 35. Swain DP, Franklin BA. VO2 reserve and the minimal intensity for improving
uk/government/uploads/system/uploads/attachment_data/file/674503/ cardiorespiratory fitness. Med Sci Sports Exerc. 2002;34(1):152–7.
walking-and-cycling-statistics-england-2016.pdf. 36. Garber CE, Blissmer B, Deschenes MR, Franklin BA, Lamonte MJ, Lee I-M, et
10. De Geus B, Hendriksen I. Cycling for transport, physical activity and health: al. Quantity and quality of exercise for developing and maintaining
What about Pedelecs? 2015. cardiorespiratory, musculoskeletal, and Neuromotor fitness in apparently
11. Fishman E, Cherry C. E-bikes in the mainstream: reviewing a decade of healthy adults: guidance for prescribing exercise. Med Sci Sports Exerc.
research. Transp Rev. 2016;36(1):72–91. 2011;43(7):1334–59.
12. MacArthur J, Dill J, Person M. Electric bikes in North America. Transp Res 37. De Geus B, Van Hoof E, Aerts I, Meeusen R. Cycling to work: influence on
Rec: J Transp Res Board. 2014;2468:123–30. indexes of health in untrained men and women in Flanders. Coronary heart
13. Peterman JE, Morris KL, Kram R, Byrnes WC. Pedelecs as a physically active disease and quality of life. Scand J Med Sci Sports. 2008;18(4):498–510.
transportation mode. Eur J Appl Physiol. 2016;116(8):1565–73. 38. Rynders CA, Weltman JY, Jiang B, Breton M, Patrie J, Barrett EJ, et al. Effects of
14. Haskell WL, Lee I-M, Pate RR, Powell KE, Blair SN, Franklin BA, et al. Physical exercise intensity on postprandial improvement in glucose disposal and insulin
activity and public health: updated recommendation for adults from the sensitivity in Prediabetic adults. J Clin Endocrinol Metab. 2014;99(1):220–8.
American College of Sports Medicine and the American Heart Association. 39. Little JP, Jung ME, Wright AE, Wright W, Manders RJF. Effects of high-
Med Sci Sports Exerc. 2007;39(8):1423–34. intensity interval exercise versus continuous moderate-intensity exercise on
15. Moher D, Liberati A, Tetzlaff J, Altman DG, Altman D, Antes G, et al. postprandial glycemic control assessed by continuous glucose monitoring
Preferred reporting items for systematic reviews and meta-analyses: the in obese adults. Appl Physiol Nutr Metab. 2014;39(7):835–41.
PRISMA statement. PLoS Med. 2009;6:e1000097. 40. Oja P, Mänttäri A, Heinonen A, Kukkonen-Harjula K, Laukkanen R, Pasanen
16. Tools NCCfMa. Dictionary for the effective public health practice project M, et al. Physiological effects of walking and cycling to work. Scand J Med
quality assessment tool for quantitative studies. 2008 [Available from: Sci Sports. 1991;1(3):151–7.
https://merst.ca/ephpp/. 41. Hendriksen I, Engbers L, Schrijver J, Gijlswijk RV, Weltevreden J, Wilting J.
17. De Bourdeaudhuij I, Van Cauwenberghe E, Spittaels H, Oppert JM, Rostami Elektrisch Fietsen : marktonderzoek en verkenning toekomstmogelijkheden.
C, Brug J, et al. School-based interventions promoting both physical activity Leiden: TNO; 2008 2008-01-01. Report No.: 9789059862876.
and healthy eating in Europe: a systematic review within the HOPE project. 42. MacArthur J, Cherry C, Harpool M, Daniel S. Electric Boost: Insights from a
Obes Rev. 2011;12(3):205–16. National E-bike Owner Survey. Portland, OR: Transportation Research and
18. Gojanovic B, Welker J, Iglesias K, Daucourt C, Gremion G. Electric bicycles as Education Center (TREC); 2018.
a new active transportation modality to promote health. Med Sci Sports 43. Schepers JP, Fishman E, den Hertog P, Wolt KK, Schwab AL. The safety of
Exerc. 2011;43(11):2204–10. electrically assisted bicycles compared to classic bicycles. Accid Anal Prev.
19. Sperlich B, Zinner C, Hebert-Losier K, Born DP, Holmberg HC. Biomechanical, 2014;73:174–80.
cardiorespiratory, metabolic and perceived responses to electrically assisted 44. Bassett DRJ, Pucher J, Buehler R, Thompson DL, Crouter SE. Walking, cycling,
cycling. Eur J Appl Physiol. 2012;112(12):4015–25. and obesity rates in Europe, North America, and Australia. J Phys Act Health.
20. Simons M, Van Es E, Hendriksen I. Electrically assisted cycling: a new mode for 2008;5(6):785–814.
meeting physical activity guidelines? Med Sci Sports Exerc. 2009;41(11):2097–102.
21. Hansen D, Soors A, Deluyker V, Frederix I, Dendale P. Electrical support during
outdoor cycling in patients with coronary artery disease: impact on exercise
intensity, volume and perception of effort. Acta Cardiol. 2017;73:1–8.
22. Berntsen S, Malnes L, Langaker A, Bere E. Physical activity when riding an
electric assisted bicycle. Int J Behav Nutr Phys Act. 2017;14(1):55.
23. Langford BC, Cherry CR, Bassett DR, Fitzhugh EC, Dhakal N. Comparing
physical activity of pedal-assist electric bikes with walking and conventional
bicycles. J Transp Health. 2017;6:463–73.
24. Theurel J, Lepers R. Réponses cardiovasculaires lors de la distribution du courrier
postal avec un vélo à assistance électrique. Mov Sport Sci. 2011/2;73:33–37.
25. Theurel J, Theurel A, Lepers R. Physiological and cognitive responses when
riding an electrically assisted bicycle versus a classical bicycle. Ergonomics.
2012;55(7):773–81.
26. La Salle DT, Shute R, Heesch M, Slivka D. Demands of simulated commuting
using an electrically assisted bicycle. Int J Exerc Sci. 2017;10(3):454–64.
27. Louis J, Brisswalter J, Morio C, Barla C, Temprado JJ. The electrically assisted
bicycle: an alternative way to promote physical activity. Am J Phys Med
Rehabil. 2012;91(11):931–40.
28. Meyer D, Steffan M, Senner V. Impact of electrical assistance on
physiological parameters during cycling. Procedia Eng. 2014;72:150–5.
29. de Geus B, Kempenaers F, Lataire P, Meeusen R. Influence of electrically
assisted cycling on physiological parameters in untrained subjects. Eur J
Sport Sci. 2013;13(3):290–4.
30. Höchsmann C, Meister S, Gehrig D, Gordon E, Li Y, Nussbaumer M,
et al. Effect of E-Bike Versus Bike Commuting on Cardiorespiratory