Influence of a slow-start on overall performance and running kinematics during 6-h ultramarathon races

Abstract

The aim of this study was to describe the pacing during a 6-h ultramarathon (race 1) and to
investigate whether a slow-start affects performance, running kinematic changes, ratings
of perceived exertion (RPE) and fatigue (ROF) (race 2). After a critical speed test,
participants completed two 6-h ultramarathons. Race 1 (n = 16) was self-paced, whereas in
race 2 (n = 10), athletes performed the initial 36 min at speeds 18% below the mean speed
of the initial 36 min of race 1. In race 1, participants adopted an inverse sigmoid pacing.
Contact times increased after 1 h, and flight times decreased after 30 min (all P ≤ 0.009);
stride length reduced after 1 h 30 min (all P = 0.022), and stride frequency did not change.
Despite the lower speeds during the first 10% of race 2, and higher speeds at 50% and 90%,
performance remained unchanged (57.5 ± 10.2 vs. 56.3 ± 8.5 km; P = 0.298). However,
RPE and ROF were lowered for most of race 2 duration (all P < 0.001). For the comparison
of kinematic variables between races, data were normalised by absolute running speed at
each time point from 1 h onwards. No differences were found for any of the kinematic
variables. In conclusion, decreasing initial speed minimises RPE and ROF, but does not
necessarily affect performance. In addition, running kinematic changes do not seem to be
affected by pacing manipulation.

Full text

Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=tejs20
European Journal of Sport Science
ISSN: 1746-1391 (Print) 1536-7290 (Online) Journal homepage: https://www.tandfonline.com/loi/tejs20
Influence of a slow-start on overall performance
and running kinematics during 6-h ultramarathon
races
Guilherme G. Matta, Arthur H. Bossi, Guillaume Y. Millet, Pedro Lima, Jorge
P.de Lima & James G. Hopker
To cite this article: Guilherme G. Matta, Arthur H. Bossi, Guillaume Y. Millet, Pedro Lima, Jorge
P.de Lima & James G. Hopker (2019): Influence of a slow-start on overall performance and
running kinematics during 6-h ultramarathon races, European Journal of Sport Science, DOI:
10.1080/17461391.2019.1627422
To link to this article: https://doi.org/10.1080/17461391.2019.1627422
Accepted author version posted online: 03
Jun 2019.
Published online: 16 Jun 2019.
Submit your article to this journal
Article views: 66
View Crossmark data

ORIGINAL ARTICLE
Influence of a slow-start on overall performance and running kinematics
during 6-h ultramarathon races
GUILHERME G. MATTA
1
, ARTHUR H. BOSSI
2
, GUILLAUME Y. MILLET
3
,
PEDRO LIMA
1
, JORGE P. DE LIMA
1
, & JAMES G. HOPKER
2
1
Faculdade de Educação Física e Desportos, Universidade Federal de Juiz de Fora, Juiz de Fora, Brazil;
2
School of Sport and
Exercise Sciences, University of Kent, Chatham, UK &
3
Laboratoire Interuniversitaire de Biologie de la Motricité, Univ Lyon,
UJM-Saint-Etienne, Saint-Étienne, F-42023, EA 7424, France
Abstract
The aim of this study was to describe the pacing during a 6-h ultramarathon (race 1) and to investigate whether a slow-start
affects performance, running kinematic changes, ratings of perceived exertion (RPE) and fatigue (ROF) (race 2). After a
critical speed test, participants completed two 6-h ultramarathons. Race 1 (n= 16) was self-paced, whereas in race 2 (n=
10), athletes performed the initial 36 min at speeds 18% below the mean speed of the initial 36 min of race 1. In race 1,
participants adopted an inverse sigmoid pacing. Contact times increased after 1 h, and flight times decreased after 30 min
(all P≤.009); stride length reduced after 1 h 30 min (all P= .022), and stride frequency did not change. Despite the lower
speeds during the first 10% of race 2, and higher speeds at 50% and 90%, performance remained unchanged (57.5 ± 10.2
vs. 56.3 ± 8.5 km; P= .298). However, RPE and ROF were lowered for most of race 2 duration (all P< .001). For the
comparison of kinematic variables between races, data were normalised by absolute running speed at each time point from
1 h onwards. No differences were found for any of the kinematic variables. In conclusion, decreasing initial speed
minimises RPE and ROF, but does not necessarily affect performance. In addition, running kinematic changes do not
seem to be affected by pacing manipulation.
Keywords: Competitive behaviour, effort distribution, ultra-endurance, performance, biomechanics, running gait
Highlights
.6-h ultramarathon runners adopt sigmoid pacing, continuously increasing RPE and ROF throughout the race.
.A slow start in the first 36 min attenuates RPE and ROF development, although performance does not necessarily improve,
which may be dependent on runner’s motivation.
.Changes in running kinematics seem to reflect fluctuations in running speed rather than fatigue development.
Introduction
Pacing, as the distribution of work-rates during an
exercise, has been suggested to be crucial for athletes
aiming to achieve optimal racing outcomes (Abbiss &
Laursen, 2008). Therefore, studies have described
pacing during ultramarathons ranging from 100 to
161 km (Hoffman, 2014; Knechtle, Rosemann,
Zingg, Stiefel, & Rüst, 2015; Lambert, Dugas,
Kirkman, Mokone, & Waldeck, 2004;Parise&
Hoffman, 2011; Renfree, Crivoi do Carmo, &
Martin, 2016; Tan, Tan, & Bosch, 2016), and during
time-based 24-h ultramarathons (Bossi et al., 2017;
Takayama, Aoyagi, & Nabekura, 2016). Yet, none of
them has done so in a 6-h ultramarathon. Given that
6 h is considered to be the minimum duration for
ultramarathon races (Zaryski & Smith, 2005), it is
surprising the lack of specific pacing studies.
The best ultramarathon performances have been
associated with more even pacing, with conservative
initial running speeds, no matter the distance or dur-
ation (Bossi et al., 2017; Hoffman, 2014; Knechtle
et al., 2015; Lambert et al., 2004; Parise &
Hoffman, 2011; Renfree et al., 2016; Takayama
et al., 2016; Tan et al., 2016). Interestingly, our
recent study found an inverse correlation between
initial running speeds (normalised to the total race
© 2019 European College of Sport Science
Correspondence: Guilherme G. Matta Faculdade de Educação Física e Desportos, Universidade Federal de Juiz de Fora, Martelos, Juiz de
Fora/MG, Brazil. E-mail: guilhermegmatta@gmail.com
European Journal of Sport Science, 2019
https://doi.org/10.1080/17461391.2019.1627422

average) and overall performances during 24-h ultra-
marathons, suggesting athletes should perhaps start
conservatively to increase total distance covered
(Bossi et al., 2017). This hypothesis remains
untested. Pacing has been shown to be regulated
internally by the central nervous system (Konings &
Hettinga, 2018), and thus, ratings of fatigue (ROF)
(Micklewright, Gibson, Gladwell, & Al Salman,
2017) and/or perceived exertion (RPE) (Borg,
1982) may play a role in its regulation. Surprisingly,
these measures have not been used to investigate
the relationship between pacing and performance
during ultramarathon running.
Typically, runners experience long-lasting fatigue
during ultramarathons (Martin et al., 2010), which
is associated with several changes in running patterns
(Degache et al., 2013; Giovanelli, Taboga, & Lazzer,
2016; Morin, Samozino, & Millet, 2011; Vernillo
et al., 2014), presumably to avoid excessive muscle
damage (Millet, Hoffman, & Morin, 2012; Vernillo,
Millet, & Millet, 2017). The process of pacing optim-
isation should not disregard the impact changes in
running kinematics could have on performance.
Again, the ultramarathon literature lacks studies
investigating the influence of different types of
pacing on running kinematics.
The first aim of this study was to describe pacing,
ROF and RPE development, and running kinematic
changes during a 6-h ultramarathon race. We hypoth-
esised that positive pacing would be found, with con-
tinuously increasing ROF and RPE. Based on a
previous study (Giovanelli et al., 2016), we also
hypothesised that changes in running kinematics
would occurafter ∼4 h. The second aim of this
study was to investigate through an interventional
design whether a slow-start would affect ROF and
RPE development, running kinematic changes and
performance. We hypothesised a slow-start would
attenuate ROF and RPE development, possibly
affecting running kinematic changes and conse-
quently improving overall performance.
Methods
Participants
After providing written informed consent, sixteen
runners (4 women and 12 men; age: 38.6 ± 11.3
years, height: 1.74 ± 0.7 m, body mass: 71.5 ±
12.2 kg) were recruited to take part in the first part of
this study (descriptive analysis). All participants were
trained runners who had been training at least 6 h
per week and had completed at least one ultramarathon
race (i.e. ≥50 km) during the 6 months preceding the
data collection. Ten out of the sixteen initially recruited
(2 women and 8 men; age: 40.5 ± 11.0 years, height:
1.74 ± 0.8 m, body mass: 72.0 ± 13.5 kg) completed
the second part of the study (intervention). Six athletes
did not participate in the third session due to the devel-
opment of muscular injury before the race (n=3), or
scheduling conflicts (n=3). The university’shuman
research ethics committee approved the study in com-
pliance with the Declaration of Helsinki. All partici-
pants provided written informed consent.
Study design
Participants involved in the descriptive analysis only
were required to visit the testing location twice,
whereas a third visit was required for those also
involved in the intervention. In the first visit, anthro-
pometric measures, familiarisation trials and a critical
speed test were performed. In the following two visits,
participants completed two 6-h simulated ultramara-
thons on a 400-m athletics track, at the same time of
the day (8:00 am), but 4 weeks apart to enable
enough recovery (Gaudino, Martinent, Millet, &
Nicolas, 2019; Millet et al., 2011; Nicolas, Banizette,
& Millet, 2011). While both races were contested as a
mass-start event, the first race consisted of a self-
paced race, and the second consisted of manipulated
pacing during the first 36 min, followed by self-paced
race. Participants were not informed about the pur-
poses of the study until it was completed.
Visit 1 –familiarisation and determination of critical
speed
Firstly, participants had their height and body mass
measured. Subsequently, they received instructions
and familiarised themselves with the ROF scale
(Micklewright et al., 2017), the RPE scale (Borg,
1982), the total quality recovery scale (TQR; i.e. a
6–20 scale, based on RPE, which measures psycho-
physiological recovery) (Kentta & Hassmen, 1998),
and the motivation questionnaire (i.e. 14 statements
scored on a 5-point Likert scale that measure intrinsic
and success motivation; 0 = not at all and 4 = extre-
mely) (Matthews, Campbell, & Falconer, 2001). Par-
ticipants also performed three consecutive
countermovement jumps (CMJ) (Bosco, Luhtanen,
& Komi, 1983) as a familiarisation. They were
asked to jump as high as possible on a force plate
(Jump System Pro, CEFISE
®
, São Paulo, Brazil),
with hands on their hips (i.e. no arm-swing), and a
15-s rest between attempts. The average height of
the three jumps was recorded, as it has been shown
to be more sensitive than the highest jump to estimate
neuromuscular fatigue (Claudino et al., 2017).
To describe participants’aerobic capacity, a field-
based critical speed test was performed according to
Galbraith, Hopker, Jobson, and Passfield (2011).
2G.G. Matta et al.

This test was selected because it was more familiar to
our runners compared with laboratory tests, and also
because it has been recognised as a good predictor of
endurance performance (Galbraith, Hopker, Cardi-
nale, Cunniffe, & Passfield, 2014; Galbraith,
Hopker, Lelliott, Diddams, & Passfield, 2014).
Three time-trials of 3600, 2400 and 1200 m were
performed on a 400-m athletics track, interspersed
with 30-min rest periods. Participants completed a
standardised warm-up (i.e. 10-min jog at a self-
selected intensity) before the time-trials and were
instructed to complete each one as fast as possible.
They were not provided with elapsed time and each
run was hand-timed to the nearest second. RPE
and ROF were quantified at the end of each time-
trial for familiarisation purposes. To calculate critical
speed and D’(i.e. total distance covered above
critical speed until task failure), a linear regression
analysis was used after plotting time-trial distances
and respective times. The slope and y-intercept (i.e.
critical speed and D’, respectively) were used to
produce the model equation as d=(CS × t)+D
′,in
which d= distance (m), CS = critical speed (m·s
−1
)
and t = time (s).
Visit 2 and 3–6-h ultramarathon races
Race 1 consisted of a self-paced 6-h ultramarathon in
which runners started together and were free to adjust
their speed with the aim of achieving the greatest dis-
tance possible. Given the first 10% of a race seems
critical for overall performance (Bossi et al., 2017),
the distance covered by each athlete during the first
36 min was used to set speed targets for the first
36 min of race 2; i.e. athletes ran at constant speeds
18% slower. After the enforced-speed phase of race
2, athletes were allowed to run as desired. The 18%
target was selected after termination of race 1. We
decreased the initial speed for race 2 by matching it
to the overall speed of race 1 (i.e. an attempt to
produce an even pacing). Approximately one hour
before the start, participants were informed about
the speed manipulation and their individual targets
for each lap of the track (e.g. 10 km·h
−1
or 2 min 24
s). Two members of the research team, positioned
at the starting line, used chronometers to check if ath-
letes were running each lap at the intended speed –
providing them with feedback when necessary. To
analyse pacing, participants’mean running speed of
each 36-min interval were percentage-normalised to
their overall mean speed.
Before each race, ROF, success and intrinsic
motivation, TQR, mean CMJ height and body mass
were assessed to monitor runners’psychophysiologi-
cal state. CMJ jumps were used to estimate neuro-
muscular function before and after each race, as it
has been shown to have good reliability for the assess-
ment of fatigue after exercise trials (Lombard, Reid,
Pearson, & Lambert, 2017). Ambient temperature
(mercury thermometer INCOTERM
®
, Porto
Alegre, Brazil), relative humidity (thermo-
hygrometer MT-242, Minipa
®
, Joinville, Brazil)
and wind speed (anemometer GM8908 LCD,
Kkmoon
®
, Shenzhen, China) were measured at the
start and every 30 min. RPE and ROF were measured
every 12 min during the first 36 min, at 1 h and every
30 min thereafter. The official racing time, number of
laps and time spent in each lap were recorded by an
electronic-chip system (Speedway R220 RAIN
RFID, IMPINJ
®
,
Seattle, USA) attached to runners’
shoelace. Total distance covered in 6 h, and in the
first 36 min, were calculated as the sum of 400-m
laps completed in each duration plus the distance
covered during the incomplete lap. During the last
2 min before time marks, participants were required
to run while holding a small plastic cone with their
ID numbers –dropping the cones on the track at
the end of 36 min and 6 h, to obtain a measure of dis-
tance covered. Mean CMJ height and body mass
were reassessed ∼5 min after each race to quantify
changes in neuromuscular power and fatigue, as
well as the impact of the races on body water balance.
To analyse changes in running kinematics during
the races, a digital camera (Hero 4, GoPro
®
, San
Mateo, USA) operating at 120 Hz, with the fisheye
option deactivated to remove distortion effects, was
placed perpendicular to the running direction of the
participants, recording a 12 m-wide section of the
track. Five subsequent steps were analysed, with
Kinovea
®
0.8.15 software used to measure contact
and flight times. The mean values of both parameters
of 5 steps were then used to calculate stride frequency
and length according to the equations:
stride frequency =1
(contact time +flight time)
(1)
stride length =running speed
stride frequency (2)
Running speed was calculated based on time to
cover the 12-m section of the track. Running kin-
ematics were assessed during the first lap, and at
every subsequent 30-min time point, consistent
with Giovanelli et al. (2016). For the comparison of
kinematic variables between races, data were normal-
ised by individual athlete’s absolute running speed
(m·s
−1
) at each time point from 1 h. Participants
were requested to wear the same pair of shoes
during both races, and were not allowed to wear
Pacing and running kinematics during 6-h ultramarathon 3

calf compression sleeves to avoid running kinematic
alterations (Kerhervé, Samozino, et al., 2017).
During both races, the running direction around
the track was changed every hour once they reached
the starting line. Runners consumed food and/or bev-
erages ad libitum from a buffet provided by the
research team, or by themselves. They were
instructed to maintain their regular training and to
refrain from high-intensity and/or high-volume train-
ing in the 48 h preceding testing sessions. Runners
were also requested to report and replicate their
diet, as well as to abstain from caffeine, supplements
and alcohol in the last 24 h before races.
Statistical analysis
Results are presented as mean ± SD. In race 1, one-way
repeated-measures ANOVAs with planned contrasts
were performed to analyse pacing, ROF, RPE and
running kinematics. Running kinematics were reported
as percentage changes in relation to the first lap.
For the intervention, a paired t-test was performed
to compare race performances and variables assessed
only before each race (i.e. TQR, ROF, intrinsic and
success motivation). Two-way repeated measures
ANOVAs with Bonferroni pairwise comparisons
were used to assess differences between races in
pacing, ROF, RPE and running kinematics. We
focused on the main effect of the races and the inter-
action effects to avoid duplicate analyses. As pacing
data were percentage normalised, changes from one
race to another were assessed by the interaction
effect only. Partial eta squared (η
p
2
) or Cohen’sd
were calculated as effect sizes estimates. Two-way
repeated measures ANOVAs were also used to test
for differences in mean CMJ height and body mass
before and after each race. Data analysis was per-
formed using SPSS (23.0, IBM
®
, Armonk, USA),
with statistical significance set at P≤.05.
Results
Descriptive analysis
The critical speed and D’of the 16 athletes evaluated
in visit 1 was 4.0 ± 0.5 m·s
−1
and 125 ± 44 m,
respectively. The mean distance covered by athletes
in race 1 was 58.9 ± 9.4 km (2.73 ± 0.44 m·s
−1
; i.e.
68 ± 7% of critical speed). Overall analysis showed
runners adopted an inverse sigmoid pacing (F=
32.90, P< .001, η
p
2
= 0.69; Figure 1a), with the
highest running speeds during the first 50% of the
race, and slowing afterwards when compared to the
first 10% (P≤.005).
We found a main effect of time for RPE (F= 30.27,
P< .001, η
p
2
= 0.67) and ROF (F= 56.04, P< .001,
η
p
2
= 0.79). Both increased consistently throughout
the race (P≤.05; Figure 1b).
A main effect of time was found for contact time (F
=9.43,P<.001, η
p
2
=0.39;Figure 1c), flight time (F
=9.77, P< .001, η
p
2
=0.39; Figure 1d) and stride
length (F= 9.92, P<.001, η
p
2
= 0.40; Figure 1e), but
not for stride frequency (F=0.90, P=.45, η
p
2
=0.06;
Figure 1f). Contact times increased after 1 h (overall
change: +12%; all P≤.009) and flight times decreased
after 30 min (overall change: −34%; all P≤.001),
whereas stride length decreased after 1 h 30 min
(overall change: −13%; all P≤.022).
Intervention
The critical speed and D’of the 10 athletes involved in
both races was 3.9 ± 0.5 m·s
−1
and 120 ± 41 m. There
were no differences between races 1 and 2 (all
P≥.677) on the mean (range) temperature, relative
humidity and wind speed, respectively: 21.4°C (19.0
−25.0) vs. 21.3°C (17.0−25.5), 75.5% (53.0−100.0)
vs. 72.5% (46.0−100.0), 0.7 m·s
−1
(0.0−2.2) vs.
0.7 m·s
−1
(0.0−3.0). Before each race, there were no
significant differences in body mass, mean CMJ
height, TQR and ROF, but intrinsic and success
motivation were significantly lower before the second
race (Table 1). No interaction effects were evident
for body mass (F=0.77, P=.787, η
p
2
= 0.09) and
mean CMJ height (F=1.25, P=.293,η
p
2
=0.12).
Performance was not different between races (57.5
± 10.2 vs. 56.3 ± 8.5 km; t= 1.11, P= .298, d= 0.13;
2.66 ± 0.47 vs 2.61 ± 0.39 m·s
−1
), despite a differ-
ence in pacing (interaction effect: F= 3.78, P= .021,
η
p
2
= 0.30; Figure 2a). By design, pairwise compari-
sons showed the normalised running speed in race
2 was lower at 10% of race duration (P< .001). Con-
versely, normalised running speed was greater in race
2 at 50% (P< .001) and at 90% (P= .034) of race
duration.
We found a main effect of the race for both RPE
(F= 56.31, P< .001, η
p
2
= 0.86; Figure 2b) and
ROF (F= 27.81, P= .001, η
p
2
= 0.76; Figure 2c). An
interaction effect was also observed for both RPE
(F= 3.46, P< .001, η
p
2
= 0.28) and ROF (F= 2.30,
P= .010, η
p
2
= 0.20). Pairwise comparisons showed
that both parameters were lower in race 2, mainly
in the first half, but also at 5 h 30 min, and at 6 h
for RPE (P≤.05).
There were no significant main effects of the race
for normalised contact time (F= 0.68, P= .432, η
p
2
= 0.07), flight time (F= 0.48, P= .506, η
p
2
= 0.05),
stride length (F= 0.17, P= .688, η
p
2
= 0.02), and
stride frequency (F= 2.67, P= .137, η
p
2
= 0.23).
There were also no interaction effects for any of
the kinematic variables: contact time (F= 0.43,
4G.G. Matta et al.

P= .928, η
p
2
= 0.05; Figure 3a), flight time (F= 0.79,
P= .639, η
p
2
= 0.08; Figure 3b), stride length (F=
0.91, P= .532, η
p
2
= 0.09; Figure 3c) and stride fre-
quency (F= 0.55, P= .853, η
p
2
= 0.06; Figure 3d).
Discussion
The results of this study demonstrated that 6-h ultra-
marathon runners adopt high initial speeds for the
first 30% of the race, progressively decreasing speed
until ∼60%, and thereafter reaching a plateau. As
expected, RPE and ROF increased linearly through-
out the race, reaching near maximum values of ∼18
and ∼8, respectively. Early changes in running kin-
ematics were observed, contradicting our hypothesis
that changes would only be seen after ∼4 h. We also
hypothesised a slow-start intervention would
improve performance. A more even pacing indeed
lowered RPE and ROF, but overall performance
Figure 1. Mean ± SD participants’pacing (panel a), development of ratings of perceived exertion (RPE) and ratings of fatigue (ROF) (panel
b), and changes in contact time (panel c), flight time (panel d), stride length (panel e) and stride frequency (panel f) during the first race. Panel
a: ∗Difference from 10%; ‡Difference from 100% (P≤.05). Panel b: ∗Difference of RPE from the previous time-point; ‡Difference of ROF
from the previous time-point (P≤.05). Panels c, d, e: ∗Difference from the first lap (P≤.05).
Pacing and running kinematics during 6-h ultramarathon 5

was not affected, nor were changes in running kin-
ematics. Intrinsic and success motivation were
lower before the second race, potentially explaining
the lack of performance benefit.
Descriptive analysis: race 1
The distance achieved by participants during the first
race was 58.9 ± 9.4 km, similar to other 6-h investi-
gations, with mean distances varying from 56.2–
61.0 km (Akimov & Son’kin, 2012;Kerhervé,
McLean, Birkenhead, Parr, & Solomon, 2017; Woll-
seiffen et al., 2016). In our study, runners adopted
an inverse sigmoid pacing; i.e. the first 30% fast (rela-
tive to the mean running speed), decreasing until
∼60%, and then keeping a constant speed until the
end of the race (see Figure 1a). We have previously
attributed this terminology to the pacing of some
ultra-runners (Bossi et al., 2017), and it is interesting
to replicate Renfree’s findings (Renfree et al., 2016),
as this type of pacing has not been described in the
scientific literature (Abbiss & Laursen, 2008). More
often, our and other research groups have demon-
strated reverse J-shaped (Bossi et al., 2017; Takayama
et al., 2016; Tan et al., 2016) and positive pacing
(Knechtle et al., 2015; Lambert et al., 2004; Parise
& Hoffman, 2011; Tan et al., 2016) in ultramarathon
races. This differences in pacing most likely reflect the
variations in distance run, course elevation profile,
environmental conditions, athletes’running perform-
ance and race competitive dynamics (Abbiss &
Laursen, 2008; Konings & Hettinga, 2018). Although
an even pacing has been suggested to optimise per-
formance during prolonged exercises (Abbiss &
Laursen, 2008), it is rarely adopted in practice.
Previous studies have shown that pacing is mediated
by RPE (De Koning et al., 2011;Konings&Hettinga,
2018), displaying linear increases throughout a task, as
a function of the exercise time remaining. Accord-
ingly, our results corroborate the hypothesis that
RPE and ROF would increase continuously, reaching
near maximum values at the end of an ultramarathon
race. It has been suggested that pacing is regulated by a
complex relationship between the central nervous and
other physiological systems, and thus, RPE and ROF
may play a role, by ensuring catastrophic disturbance
to homeostasis does not occur (Abbiss, Peiffer,
Meeusen, & Skorski, 2015).
Previous studies that analysed changes in running
kinematics during ultramarathons have found
varying results (Degache et al., 2013; Giovanelli
et al., 2016; Morin et al., 2011; Schena et al., 2014;
Vernillo et al., 2014). Indeed, our results only par-
tially corroborate previous findings. Contact time
increased after 1 h (+7%) and flight time decreased
after 30 min (−34%), whereas both parameters
changed (contact time: +7.1% and flight time:
−29.0%) only after 4 h 30 min during another 6-h
ultramarathon (Giovanelli et al., 2016). Interestingly,
no changes in flight time were found after a 5-h hilly
running bout (Degache et al., 2013) and during a 24-
h treadmill run (Morin et al., 2011), whereas contact
time increased after the 5-h hilly running bout
(Degache et al., 2013) and in the latter parts of the
24-h treadmill run (Morin et al., 2011). Moreover,
stride length of our participants decreased (−13%)
after 1 h 30 min of running, whereas it decreased
(−5.1%) after 5 h in the study of Giovanelli et al.
(2016), and after 40 km of a 60-km race in the work
of Schena et al. (2014). Stride frequency did not
change in our study, corroborating the findings of
others (Giovanelli et al., 2016; Schena et al., 2014;
Vernillo et al., 2014), and suggesting that this might
be a robust parameter to progressive fatigue during
this type of ultramarathon race. Given that studies
analysed running kinematics during very different
running conditions, it is not possible to draw an
overall conclusion. Nevertheless, it has been hypoth-
esised that changes in running kinematics are associ-
ated with exercise-induced pain as a mechanism to
avoid excessive muscle damage (Millet et al., 2012;
Morin, Tomazin, Samozino, Edouard, & Millet,
2012).
Table 1. Comparison of the measures before and after each race.
Race 1 Race 2
Variable Pre Post P-value Cohens’dPre Post P-value Cohens’d
Body mass (kg) 72.3 ± 13.4 70.5 ± 13.2 0.011∗0.135 72.2 ± 14.1 70.5 ± 13.9 0.001
#
0.015
Mean CMJ height (cm) 25.5 ± 5.9 20.2 ± 5.5 0.049∗0.929 26.5 ± 6.0 22.6 ± 5.2 0.065 0.051
TQR (AU) 18.8 ± 2.2 - - - 18.1 ± 1.8 - 0.322 0.348
ROF (AU) 0.6 ± 0.8 - - - 1.0 ± 1.0 - 0.309 0.442
Intrinsic Motivation (AU) 26.0 ± 2.1 - - - 25.2 ± 1.6 - 0.037
†
0.429
Success Motivation (AU) 18.0 ± 4.9 - - - 16.5 ± 5.1 - 0.018
†
0.300
CMJ: countermovement jump; TQR: total quality recovery; ROF: ratings of fatigue; AU: arbitrary units.
∗Difference between pre vs. post in the first race.
#
Difference between pre vs. post in the second race.
†
Difference in variables measured only
before races. (P≤0.05).
6G.G. Matta et al.

Intervention: race 1 vs. race 2
This is the first study to manipulate pacing during a
simulated ultramarathon race. Runners were
required to complete the first 10% of race 2 at
speeds 18% slower than the self-paced race. This
conservative start led runners to run faster at both
50% and 90% in comparison to race 1 (see Figure
2a). Nevertheless, they achieved the same distance.
Before both ultramarathons, participants had
equivalent body mass, TQR, ROF and mean CMJ
height, suggesting they were at the same psychophy-
siological state. Moreover, weather conditions were
similar between races. Runners were however less
motivated before race 2, which may explain the lack
of performance improvement. All participants were
informed that an intervention would take place
before completing the motivation questionnaires,
although they were not given details until ∼1h
before the race. This may have played a role in the
pre-race motivation. It could also be that having less
athletes competing the second in comparison to the
first race (n= 10 vs. 16) may have affected their com-
petitiveness (Konings & Hettinga, 2018). Alterna-
tively, 4 weeks may have been insufficient to restore
athletes’motivation to perform such a demanding
task. Regardless, similar performance associated
with lower motivation could be viewed as a benefit.
Both RPE and ROF were consistently lower during
race 2 until approximately 50%, and at the penulti-
mate time point. Importantly, RPE was also lower
at 6 h. Had participants performed both trials at
their best, a performance benefit may have been
evident (Marcora & Staiano, 2010).
We sought to compare changes in running kin-
ematics during the race normalised to running speed
to avoid confounding effects. We found a similar
pattern in both races, despite differences in ROF.
This may suggest that running kinematics changes
are somewhat insensitive to the development of
fatigue, reflecting the speed athletes are able to
sustain. This is corroborated by Morin et al. (2012),
who analysed changes in running kinematics at 10
and 20 km·h
−1
, before and after a fatiguing protocol,
and found similar running gait despite decreases in
maximal force production of the lower limbs.
This study is not without limitations. We did not
randomise the order of the races. However, we
could not predict the self-paced strategy of our
runners considering the inconsistencies in the litera-
ture. An unsupervised 4-week period between races
may have affected our participants’running perform-
ance. However, they were instructed to maintain their
usual training programme, and so, we are confident
that any possible effects were minimal, given the
trained status of our runners. Moreover, the video-
based method of kinematic analysis may have been
insensitive to detect minor changes in the variables
analysed. Future studies using inertial measurement
units are therefore required to confirm our findings.
Finally, we did not control participants’calorie
Figure 2. Mean ± SD participants’pacing (panel a) and develop-
ment of ratings of perceived exertion (RPE, panel b) and ratings
of fatigue (ROF, panel c) throughout each 6-h ultramarathon
races (mean ± SD). ∗Difference between race 1 and 2 at the
time-point (P≤.05).
Pacing and running kinematics during 6-h ultramarathon 7

intake before and during races, although they were
requested to report and replicate their pre-race diet.
Conclusion
In 6-h ultramarathon races, runners adopt an inverse
sigmoid pacing while RPE and ROF increase linearly
throughout the race. Adopting a slow-start attenuates
the development of RPE and ROF, but does not
necessarily improve performance, as the relationship
between pacing and performance is likely dependent
on motivation, which we could not control for.
Changes in running kinematics can occur early in
an ultramarathon, associated with fluctuations in
racing speed, which suggests contact and flight
times, and stride length and frequency, are all some-
what insensitive to the development of fatigue.
Acknowledgements
A.H.B. is a CNPq (Conselho Nacional de Desenvol-
vimento Científico e Tecnológico) scholarship holder
[200700/2015-4]. We would like to thank Carolina
Campos for her invaluable support with race logistics,
Jefferson de Freitas for his helpful insights, and Fer-
nando Bussular, Daniel Lucas, Hudson Carvalho,
Vivian Caruso and Sylvia Miranda for their assistance
with data collection.
Disclosure statement
No potential conflict of interest was reported by the authors.
ORCID
Guilherme G. Matta http://orcid.org/0000-0001-
5622-5442
Arthur H. Bossi http://orcid.org/0000-0002-4098-
0192
Guillaume Y. Millet http://orcid.org/0000-0002-
6395-0762
Pedro Lima http://orcid.org/0000-0001-9967-8345
Jorge P. de Lima http://orcid.org/0000-0003-0073-
8673
James G. Hopker http://orcid.org/0000-0002-4786-
7037
Figure 3. Mean ± SD contact time (panel a), flight time (panel b), stride length (panel c) and stride frequency (panel d) normalised to running
speed during each race.
8G.G. Matta et al.

References
Abbiss, C. R., & Laursen, P. B. (2008). Describing and under-
standing pacing strategies during athletic competition. Sports
Medicine,38(3), 239–252.
Abbiss, C. R., Peiffer, J. J., Meeusen, R., & Skorski, S. (2015). Role
of ratings of perceived exertion during self-paced exercise: What
are we actually measuring? Sports Medicine,45(9), 1235–1243.
Akimov, E., & Son’kin, V. (2012). Physiological effects of an ultra-
marathon run. Human Physiology,38(6), 617–625.
Borg, G. A. (1982). Psychophysical bases of perceived exertion.
Medicine & Science in Sports & Exercise,14(5), 377–381.
Bosco,C.,Luhtanen,P.,&Komi,P.V.(1983). A simple method for
measurement of mechanical power in jumping. European Journal of
Applied Physiology and Occupational Physiology,50(2), 273–282.
Bossi, A. H., Matta, G. G., Millet, G. Y., Lima, P., Pertence, L. C.,
de Lima, J. P., & Hopker, J. G. (2017). Pacing strategy during
24-hour ultramarathon-distance running. International Journal
of Sports Physiology and Performance,12(5), 590–596.
Claudino,J.G.,Cronin,J.,Mezêncio,B.,McMaster,D.T.,
McGuigan, M., Tricoli, V., …Serrão, J. C. (2017). The counter-
movement jump to monitor neuromuscular status: A meta-analy-
sis. Journal af Science and Medicine in Sport,20(4), 397–402.
Degache, F., Guex, K., Fourchet, F., Morin, J. B., Millet, G. P.,
Tomazin, K., & Millet, G. Y. (2013). Changes in running
mechanics and spring-mass behaviour induced by a 5-hour
hilly running bout. Journal of Sports Sciences,31(3), 299–304.
De Koning, J. J., Foster, C., Bakkum, A., Kloppenburg, S., Thiel,
C., Joseph, T., …Porcari, J. P. (2011). Regulation of pacing
strategy during athletic competition. Plos One,6(1), e15863.
Galbraith, A., Hopker, J., Cardinale, M., Cunniffe, B., & Passfield,
L. (2014). A 1-year study of endurance runners: Training, lab-
oratory tests, and field tests. International Journal of Sports
Physiology and Performance,9(6), 1019–1025.
Galbraith, A., Hopker, J. G., Jobson, S. A., & Passfield, L. (2011).
A novel field test to determine critical speed. Journal of Sports
Medicine & Doping Studies,1(1), 1–4.
Galbraith, A., Hopker, J., Lelliott, S., Diddams, L., & Passfield, L.
(2014). A single-visit field test of critical speed. International
Journal of Sports Physiology and Performance,9(6), 931–935.
Gaudino, M., Martinent, G., Millet, G. Y., & Nicolas, M. (2019).
The time courses of runners’recovery-stress responses after a
mountain ultra-marathon: Do appraisals matter? European
Journal of Sport Science,1–9. doi:10.1080/17461391.2018.
1560507. [In press].
Giovanelli, N., Taboga, P., & Lazzer, S. (2016). Changes in running
mechanics during a six hours running race. International Journal
of Sports Physiology and Performance,12(5), 642–647.
Hoffman, M. D. (2014). Pacing by winners of a 161-km mountain
ultramarathon. International Journal of Sports Physiology and
Performance,9(6), 1054–1056.
Kentta, G., & Hassmen, P. (1998). Overtraining and recovery. A
conceptual model. Sports Medicine,26(1), 1–16.
Kerhervé, H. A., McLean, S., Birkenhead, K., Parr, D., &
Solomon, C. (2017). Influence of exercise duration on cardior-
espiratory responses, energy cost and tissue oxygenation within
a 6 h treadmill run. PeerJ,5, e3694.
Kerhervé, H. A., Samozino, P.,Descombe, F., Pinay, M., Millet, G.
Y., Pasqualini, M., & Rupp, T. (2017). Calf compression sleeves
change biomechanics but not performance and physiological
responses in trail running. Frontiers in Physiology,8, 247.
Knechtle, B., Rosemann, T., Zingg, M. A., Stiefel, M., & Rüst, C. A.
(2015). Pacing strategy in male elite and age group 100 km ultra-
marathoners. Open Access Journal of Sports Medicine,6,71–80.
Konings, M. J., & Hettinga, F. J. (2018). Pacing decision making in
sport and the effects of interpersonal competition: A critical
review. Sports Medicine,48(8), 1829–1843.
Lambert, M. I., Dugas, J. P., Kirkman, M. C., Mokone, G. G., &
Waldeck, M. R. (2004). Changes in running speeds in a 100 km
ultra-marathon race. Journal of Sports Science and Medicine,3(3),
167–173.
Lombard, W., Reid, S., Pearson, K., & Lambert, M. (2017).
Reliability of metrics associated with a counter-movement
jump performed on a force plate. Measurement in Physical
Education and Exercise Science,21(4), 235–243.
Marcora, S. M., & Staiano, W. (2010). The limit to exercise toler-
ance in humans: Mind over muscle? European Journal of Applied
Physiology,109(4), 763–770.
Martin, V., Kerhervé, H. A., Messonnier, L. A., Banfi, J.-C.,
Geyssant, A., Bonnefoy, R., …Millet, G. Y. (2010). Central
and peripheral contributions to neuromuscular fatigue
induced by a 24-h treadmill run. Journal of Applied Physiology,
108(5), 1224–1233.
Matthews, G., Campbell, S. E., & Falconer, S. (2001). Assessment
of motivational states in performance environments. Proceedings
of the Human Factors and Ergonomics Society Annual Meeting,45
(13), 906–910.
Micklewright, D., Gibson, A. S. C., Gladwell, V., & Al Salman, A.
(2017). Development and validity of the rating-of-fatigue scale.
Sports Medicine,47(11), 2375–2393.
Millet, G. Y., Hoffman, M. D., & Morin, J.-B. (2012). Sacrificing
economy to improve running performance—a reality in the
ultramarathon? Journal of Applied Physiology,113(3), 507–509.
Millet, G. Y., Tomazin, K., Verges, S., Vincent, C., Bonnefoy, R.,
Boisson, R.-C., …Martin, V. (2011). Neuromuscular conse-
quences of an extreme mountain ultra-marathon. Plos One,6
(2), e17059.
Morin, J.-B., Samozino, P., & Millet, G. Y. (2011). Changes in
running kinematics, kinetics, and spring-mass behavior over a
24-h run. Medicine & Science in Sports & Exercise,43(5), 829–836.
Morin, J.-B., Tomazin, K., Samozino, P., Edouard, P., & Millet,
G. Y. (2012). High-intensity sprint fatigue does not alter
constant-submaximal velocity running mechanics and
spring-mass behavior. European Journal of Applied Physiology,
112(4), 1419–1428.
Nicolas, M.,Banizette, M., & Millet, G. Y. (2011). Stress and recov-
ery states after a 24 h ultra-marathon race: A one-month follow-
up study. Psychology of Sport and Exercise,12(4), 368–374.
Parise, C. A., & Hoffman, M. D. (2011). Influence of temperature
and performance level on pacing a 161 km trail ultramarathon.
International Journal of Sports Physiology and Performance,6(2),
243–251.
Renfree, A., Crivoi do Carmo, E., & Martin, L. (2016). The influ-
ence of performance level, age and gender on pacing strategy
during a 100-km ultramarathon. European Journal of Sport
Science,16(4), 409–415.
Schena, F., Pellegrini, B., Tarperi, C., Calabria, E., Luca
Salvagno, G., & Capelli, C. (2014). Running economy during
a simulated 60-km trial. International Journal of Sports
Physiology and Performance,9(4), 604–609.
Takayama, F., Aoyagi, A., & Nabekura, Y. (2016). Pacing strategy
in a 24-hour ultramarathon race. International Journal of
Performance Analysis in Sport,16(2), 498–507.
Tan, P. L., Tan, F. H., & Bosch, A. N. (2016). Similarities and
differences in pacing patterns in a 161-km and 101-km ultra-
distance road race. The Journal of Strength & Conditioning
Research,30(8), 2145–2155.
Vernillo, G., Millet, G. P., & Millet, G. Y. (2017). Does the
running economy really increase after ultra-marathons?
Frontiers in Physiology,8, 783.
Vernillo, G., Savoldelli, A., Zignoli, A., Trabucchi, P., Pellegrini,
B., Millet, G. P., & Schena, F. (2014). Influence of the
world’s most challenging mountain ultra-marathon on energy
Pacing and running kinematics during 6-h ultramarathon 9

cost and running mechanics. European Journal of Applied
Physiology,114(5), 929–939.
Wollseiffen, P., Schneider, S., Martin, L. A., Kerhervé, H. A.,
Klein, T., & Solomon, C. (2016). The effect of 6 h
of running on brain activity, mood, and cognitive
performance. Experimental Brain Research,234(7), 1829–
1836.
Zaryski, C., & Smith, D. J. (2005). Training principles and issues
for ultra-endurance athletes. Current Sports Medicine Reports,4
(3), 165–170.
10 G.G. Matta et al.