See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/325880246

The Effectiveness of Resisted Sled Training (RST) for Sprint Performance: A

Systematic Review and Meta-analysis

Article  in  Sports Medicine · June 2018

DOI: 10.1007/s40279-018-0947-8

CITATIONS READS

8 1,842

4 authors:

Pedro E. Alcaraz Jorge Carlos-Vivas

Universidad Católica San Antonio de Murcia Universidad Católica de Murcia (UCAM)
122 PUBLICATIONS   880 CITATIONS    15 PUBLICATIONS   41 CITATIONS   

SEE PROFILE SEE PROFILE

Oribi Bruno Alejandro Martínez-Rodríguez

Universidad Católica San Antonio de Murcia University of Alicante
3 PUBLICATIONS   9 CITATIONS    46 PUBLICATIONS   126 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Have been wondering about the use of Sled Towing or Pulling Training for Speed? Exciting news! View project

Promoting the shift sedentary Lifestyle towards active Ageing- LIFEAGE View project

All content following this page was uploaded by Oribi Bruno on 01 July 2018.

The user has requested enhancement of the downloaded file.

Sports Medicine
https://doi.org/10.1007/s40279-018-0947-8

SYSTEMATIC REVIEW

The Effectiveness of Resisted Sled Training (RST) for Sprint

Performance: A Systematic Review and Meta‑analysis
Pedro E. Alcaraz1,2 · Jorge Carlos‑Vivas1 · Bruno O. Oponjuru1 · Alejandro Martínez‑Rodríguez3

© Springer International Publishing AG, part of Springer Nature 2018

Abstract
Background  Sprinting is key in the development and final results of competitions in a range of sport disciplines, both indi-
vidual (e.g., athletics) and team sports. Resisted sled training (RST) might provide an effective training method to improve
sprinting, in both the acceleration and the maximum-velocity phases. However, substantial discrepancies exist in the lit-
erature regarding the influence of training status and sled load prescription in relation to the specific components of sprint
performance to be developed and the phase of sprint.
Objectives  Our objectives were to review the state of the current literature on intervention studies that have analyzed the
effects of RST on sprint performance in both the acceleration and the maximum-velocity phases in healthy athletes and to
establish which RST load characteristics produce the largest improvements in sprint performance.
Methods  We performed a literature search in PubMed, SPORTDiscus, and Web of Science up to and including 9 January
2018. Peer-reviewed studies were included if they met all the following eligibility criteria: (1) published in a scientific jour-
nal; (2) original experimental and longitudinal study; (3) participants were at least recreationally active and towed or pulled
the sled while running at maximum intensity; (4) RST was one of the main training methods used; (5) studies identified the
load of the sled, distance covered, and sprint time and/or sprint velocity for both baseline and post-training results; (6) sprint
performance was measured using timing gates, radar gun, or stopwatch; (7) published in the English language; and (8) had
a quality assessment score > 6 points.
Results  A total of 2376 articles were found. After filtering procedures, only 13 studies were included in this meta-analysis.
In the included studies, 32 RST groups and 15 control groups were analyzed for sprint time in the different phases and full
sprint. Significant improvements were found between baseline and post-training in sprint performance in the acceleration
phase (effect size [ES] 0.61; p = 0.0001; standardized mean difference [SMD] 0.57; 95% confidence interval [CI] − 0.85 to
− 0.28) and full sprint (ES 0.36; p = 0.009; SMD 0.38; 95% CI − 0.67 to − 0.10). However, non-significant improvements
were observed between pre- and post-test in sprint time in the maximum-velocity phase (ES 0.27; p = 0.25; SMD 0.18; 95%
CI − 0.49 to 0.13). Furthermore, studies that included a control group found a non-significant improvement in participants
in the RST group compared with the control group, independent of the analyzed phase.
Conclusions  RST is an effective method to improve sprint performance, specifically in the early acceleration phase. However,
it cannot be said that this method is more effective than the same training without overload. The effect of RST is greatest in
recreationally active or trained men who practice team sports such as football or rugby. Moreover, the intensity (load) is not
a determinant of sprint performance improvement, but the recommended volume is > 160 m per session, and approximately
2680 m per week, with a training frequency of two to three times per week, for at least 6 weeks. Finally, rigid surfaces appear
to enhance the effect of RST on sprint performance.

* Pedro E. Alcaraz
[email protected]
* Alejandro Martínez‑Rodríguez
[email protected]
Extended author information available on the last page of the article

Vol.:(0123456789)
 P. E. Alcaraz et al.

since different factors affect performance in each phase; in

Key Points  addition, not all sports include both phases [6]. Therefore,
training protocols to develop each of these phases must also
Resisted sled training (RST) is an effective training differ. For example, the main muscle groups involved in the
method for the development of sprint performance, acceleration phase are the ankle, knee, and hip extensors [7].
specifically in the early acceleration phase (≤ 10 m), Given the limited time available to produce force during a
independent of participant and load characteristics. sprint, the main manifestation of force is explosive, defined
However, the effects are larger in male, trained, and as the development of maximal force in minimal time, or
team sports athletes. Conversely, this method is not more rate of force development (RFD) [8]. An athlete’s position,
effective than the unresisted sprint in improving sprint with a slight body lean, relates to the horizontal net ground
performance. reaction force (GRF), which is paramount to accelerate the
RST has a small effect on performance in the maximum- body forward [9], with higher contact times (CTs) [9, 10]
velocity phase (≥ 15-m flying sprints at maximum and ratio of forces (horizontal:vertical GRF), when com-
intensity using a run-in distance of ≥ 10 m) and/or when pared with the maximum-velocity phase [9]. In contrast,
performing sprints ≥ 20 m. the main muscle groups involved in the maximum-velocity
phase are the hip and ankle extensors, which contribute to
There is no optimal load for RST, and the load should greater vertical GRF (vGRF) [11]. Weyand and colleagues
be adapted according to the desired objective. However, [12, 13] stated that top speed was mechanically limited by
when replicating the demands of sprinting (i.e., the maximal vGRF, presenting a strong positive relationship
movement pattern, load, and movement velocity) using between top speed and the average vGRF applied during
sled towing with a slight overload, the load must never the first half of the stance period [14]. Therefore, it seems
be > 20% of body mass. that, together with RFD, maximum (relative) strength may
play an important role in this phase. It should be noted that
the maximal-velocity phase includes an asymmetrical pro-
1 Introduction duction of force and the RFD is very high [15] as a result of
much shorter ground CT than observed in the acceleration
Sprinting is recognized as the fastest mode of unaided phase [9, 16, 17]. In addition, the erect stance-phase posture
human locomotion. It is an important action that humans that sprinters adopt in this phase likely contributes to the
have employed since prehistoric times, not only as a determi- stiffness required to decelerate the limb and body relatively
nant of survival but also as a key activity in the development quickly after the instant of foot–ground impact [14], all in
and final results of competitions in a range of sport disci- order to produce the maximum possible mechanical power
plines, both individual (e.g., athletics) and team sports (e.g., that generates a high horizontal velocity. Notably, neuro-
soccer, rugby, American football, basketball, futsal, or field muscular function is vital to sprint performance, because
hockey). For example, straight sprinting is the most frequent the activity and the interaction of the central nervous system
action in goal situations in professional soccer [1]. In addi- with the muscles ultimately influence muscle RFD, given
tion, professional soccer players, for whom sprinting is key, that the stretch–shortening cycle (SSC) actions are par-
have become faster over time [2, 3], independent of sex and/ ticularly prevalent in sports involving sprinting [15]. Thus,
or age [4]. However, the maximum expression of a sprint is efficient usage of the SSC can recover approximately 60%
represented in the 100-m dash final in the Olympic games, of the total mechanical energy, with a higher contribution
a sporting event with some of the largest social media and of non-metabolic energy sources with increases in running
sporting repercussions worldwide. Therefore, sprinting has speed [18, 19].
been extensively studied from both a biomechanical and a For sprint training, given the above-mentioned factors,
physiologic point of view [5]. it is typical to recommend different classifications of train-
In a 100-m race, for instance, as in many other sports ing methods, such as that based on the specificity of the
in which speed is of crucial importance, there are several method and its similitude with the sprint’s technique [20].
clearly defined phases, but these can be summarized in two: MacDougall and Sale [21] suggested that training should
acceleration and maximum-velocity phases. The accelera- be specific with regards to movement pattern, contrac-
tion phase is characterized by the athlete starting the sprint tion velocity, muscle activation type, and applied force.
from a semi-static position and increasing their speed rapidly In fact, a recent study showed that chronic performance
in a short period of time; in the maximum-velocity phase, and biomechanical adaptations are associated with ver-
the athlete moves at high speed that is maintained without tically and horizontally oriented conditioning regimens
any apparent acceleration. It is of utmost importance to [22]. Plisk [20] established primary methods that simu-
emphasize that, in some cases, these phases are not related, late the sprint movement pattern (sprint-technique drills,
Effectiveness of RST for Sprint Performance

stride length and frequency exercises, and sprints of vary- RST, with some resulting controversy. Monte et al. [40] indi-
ing distances and intensities); secondary methods that cated that the maximum power produced when using RST
simulate the sprint action but with a slight overload or occurs with loads close to 20% BM, without inducing sig-
degree of assistance (resisted or assisted methods, respec- nificant changes in the sprint technique when this load was
tively); and tertiary methods, characterized by non-specific used, whereas Cross et al. [41] indicated that the maximum
sprint development (resistance training, plyometric train- power output is achieved with loads near 80% BM. However,
ing, complex training, stretching, etc.) [23–25]. It is also this latter study [41] has the limitation of not having meas-
typical to classify sprint training methods according to the ured the effect on kinematic parameters and measuring the
force–velocity (F–V) relationship of muscle shortening, horizontal power when the athletes achieved the maximum
described by Hill and colleagues [26, 27] many years ago. velocity of sprint, given that the maximum power output in
The F–V relationship describes a characteristic property sprinting is known to usually occur in the first steps. Moreo-
of the muscle that dictates its power-production capacities ver, the horizontal force and power decreased by 82.0% and
[28]. Because mechanical power is the expression of both 62.5%, respectively, from the first to the last step during
force and velocity, it is generally accepted that improving the 20-m sprint with sled towing [40]. Interestingly, both
force-production potential and/or velocity of movement is studies have used the method developed by Samozino et al.
effective when seeking to improve short-distance sprint- [42] to calculate the horizontal power production but in dif-
ing [29]. Based on contemporary scientific knowledge, it ferent disciplines (sprinters vs. soccer players) and in differ-
is evident that maximal (relative) strength, the RFD, and ent phases. These authors indirectly estimated the anterio-
peak power-generating capacity are all important attributes posterior (horizontal) force of a sprint from spatio-temporal
that need to be developed when implementing strength parameters and, consequently, calculated the horizontal
and conditioning programs [30] to increase sprint perfor- power production during the sprint.
mance during the entire season. Haff and Nimphius [30] From a kinematic point of view, cross-sectional studies
suggested a mixed-methods approach in which a variety have focused on the effects of RST on different variables
of loads and exercise types are used in a periodized fash- such as stride length (SL), CT, flight times (FT), and joint
ion to optimize power output. In this context, resisted and angles. For example, Alcaraz et al. [31] showed a decrease
assisted sprint training protocols play a key role [30]. in SL and running velocity with an RST load of 16% BM but
One of the most traditional secondary methods to improve observed no significant changes in running technique (i.e.,
sprint performance is resisted sled training (RST), mainly by analyzing the joint angles). In addition, various studies
due to its greater effect on horizontal forces [31, 32] when [37, 40, 43, 44] showed a decrement in FT and SL, and an
compared with tertiary methods, which have a greater ver- increase in CT as a function of sled load. From a kinetic
tical orientation of resistance forces [29]. RST has been point of view, Martínez-Valencia et al. [45] found an acute,
applied in sports for decades, evolving from the pull of a significant RFD increase when added loads were 15–20%
wheel to current electromechanical devices that regulate BM compared with an UR sprint. In addition, other authors
the load to produce the loss of speed desired by the athlete have concluded that RST, mainly with high loads (30% BM),
(i.e., 1080 Motion™). Traditionally, and mainly address- increases relative net horizontal and propulsive impulse
ing the principle of specificity, the most popular manuals production compared with UR sprinting by directing force
and training guides [33, 34] have recommended that normal production more horizontally and by allowing longer time
unresisted (UR) sprint biomechanics should be maintained to apply force against the ground [32], which could be trans-
when RST is used. Hence, loads should be chosen based ferred to a better application of horizontal GRF. However,
on the sport and the athlete’s physical status. For example, the same research team compared the short-term effect
track and field sprinters may use loads that do not decrease (8 weeks) of using RST with low loads (~ 13% BM) versus
running velocity by more than 10–12% of body mass (BM) higher loads (~ 43% BM) [39] and did not find significant
[35–37]. In contrast, field sport athletes who overcome differences between the groups, either for improvement of
external resistance while blocking and tackling can use loads performance or horizontal GRFs. Therefore, there is no
20–30% of BM to improve early acceleration [38, 39]. This agreement on whether low or high loads are more benefi-
hypothesis has been widely accepted, since it is believed that cial from both a kinetic and a kinematic point of view when
the production of mechanical power when the load increases increases in performance in the short and medium term are
is significantly reduced [40]. This reduction of mechanical sought.
power is associated with an improper configuration of the In sprinting with RST, the additional resistance expe-
athlete’s levers (involving different motor units or even mus- rienced by the athlete arises mainly from the frictional
cle groups), an increase in the CT, and no possibility of force between the base of the sled and the running surface.
adequate use of the SSC. Thus, recent studies have tried This frictional force is approximately proportional to the
to elucidate the load that maximizes mechanical power in total weight of the sled, and the coach may manipulate the
 P. E. Alcaraz et al.

resistance experienced by the athlete by changing the weight 2.1 Literature Research and Data Sources
placed on the sled. The greater the added weight, the greater
the friction, and hence the slower the athlete’s acceleration The search was performed by two independent reviewers
and the lower the maximum speed achieved by the athlete. (AM and BO) using the following databases to identify
The coefficient of friction between the sled and the running studies for this review: PubMed, SPORTDiscus, and Web
surface is determined by the type and surface roughness of of Science. The search results were limited to studies pub-
the materials used in the base of the sled and running sur- lished up to and including 9 January 2018. Reviewers used a
face. For this reason, the use of different surfaces (grass vs. computerized and manual library search with the following
athletics track) can produce different stimuli with the same Boolean search phrases in all of the mentioned databases:
load [46, 47]. Another factor that can alter friction is the (“Sprint” OR “Pulling” OR “Towing” OR “Training”) AND
inertia of the sled, which is higher when the sled moves at (“Sled” OR “Resisted”). Figure 1 shows the flow diagrams
low speed. for the entire search process for both variables.
Finally, a recent systematic review by Petrakos et  al.
[47] described in detail the training recommendations in 2.2 Inclusion Criteria
the existing literature. However, the authors concluded that
performance benefits of RST over UR training remain to be The following inclusion criteria had to be met for a study
conclusively demonstrated. Between-study comparisons are to be considered for this review: (1) type of publication:
limited, primarily due to discrepancies in participant training the current research only considered articles published in
status and level and sled load prescription, indicating that scientific journals; (2) type of study: original experimental
future work is required to define the optimal training load and longitudinal studies were considered; (3) participants:
for RST, depending on the specific components of sprint the study participants were at least recreationally active who
performance to be enhanced. towed or pulled the sled while running at maximum inten-
In an effort to clarify discrepancies in the literature, the sity; (4) intervention: sled towing and sprinting must have
following questions should be addressed: (1) does RST been one of the training methods used; (5) outcome meas-
improve sprint performance; (2) will the effect be greater ures: studies must have identified the load of the sled used,
when RST is used for the different phases of sprint; (3) is distance tested up to a maximum of 100 m in at least one
RST more effective than UR sprinting; (4) what is the opti- of sprint time and/or sprint velocity, pre-test and post-test
mal load when applying RST for obtaining higher sprint results, and test used; (6) tests must have been measured by
adaptations; (5) should the load be different for the different an automated electronic machine, such as timing gates or a
phases of sprint; (6) does RST have a different effect on radar gun, or by a manual method such as a stopwatch; (7)
athletes depending on age; (7) will the effect be greater if only studies in English were considered; (8) quality assess-
the athlete is highly trained; (8) how many days per week ment: a score > 6 points on the Physiotherapy Evidence
should RST be applied; (9) for how many weeks should RST Database scale (PEDro) for systematic review [49–51].
be applied; and (10) can the surface affect the performance All studies that included the time used to cover a maxi-
adaptations? mum distance of 10 m were utilized for the subgroup analy-
Therefore, and in response to the questions raised by the sis of the acceleration phase. All studies that measured the
current literature, the objective of this systematic review time used to cover a distance of at least 15 m at maximum
with meta-analysis is twofold: (1) to review the state of the intensity using a run-in distance of  ≥ 10 m before recording
current literature on intervention studies that have analyzed time were used for the subgroup analysis of the maximum-
the effects of RST on sprint performance both in the accel- velocity phase. All studies that included the time used to
eration and the maximum-velocity phase in healthy athletes cover a distance of at least 20 m were used for the subgroup
and (2) to establish which RST load characteristics are asso- analysis of the full sprint. For studies that did not include the
ciated with the greatest improvements in sprint performance. time but showed the average speed and the covered distance,
the time was calculated as time = distance/average speed.
Likewise, for the studies that included the sprint time and
2 Methods presented the acceleration phase and the full sprint data, or
the acceleration phase and the maximum-velocity phase, the
We followed the Preferred Reporting Items for Systematic time was calculated as the subtraction or summation (based
review and Meta-Analyses (PRISMA) guidelines [48] for on the conditions, respectively) of the known mean data.
search procedures, study selection, data collection, and
analysis.
Effectiveness of RST for Sprint Performance

Records identified from PubMed,

SPORTDiscus and Web of Science Additional records identified
(n = 2376) through reference search (n = 0)

Records after duplicates removed

(n = 659)
Excluded based on abstract/title screening (n = 633)

• Not resisted sled training (n = 546)

• Not athletes (n = 6)
• Not English language (n = 6)
• Acute studies (n = 37)
• Towing but not running (n = 1)
• Not towing on ground (n = 18)
• Theoretical review paper (n = 14)
• Mathematical model (n = 1)
• Not journal articles (n = 4)

Articles selected for full-text

retrieval (n = 26)

Excluded after full-text screening (n = 13)

• No outcome measure of sprint (n = 13)

Articles included in meta-analysis

(n = 13)

Fig. 1  Flow diagram of the process of study selection

2.3 Study Selection 2.4 Data Extraction and Analysis

Two reviewers (AM and BO) independently evaluated the One reviewer (AM) extracted the following information
titles and abstracts of the studies that resulted from the from each full-text article, and a second reviewer (BO) con-
search. Disagreements between the two reviewers were firmed the extraction. Disparities in data abstraction were
resolved by discussion; if necessary, a third reviewer (PA) resolved by a third reviewer (PA).
was consulted to reach a consensus. Review Manager Software (RevMan 5.2; Cochrane Col-
For this review, only studies that met the eligibility cri- laboration, Oxford, UK) and Comprehensive Meta-analysis
teria were selected. Reviewers independently assessed the software (Version 2; Biostat, Englewood, NJ, USA) were
methodological quality of the eligible studies using the used for meta-analysis. A randomized effect model was used.
PEDro scale [49–51]. Heterogeneity among studies was assessed using I2 statistics.
 P. E. Alcaraz et al.

Subgroup analyses were performed to evaluate the poten- sessions·week−1. The distance covered in sprint assessment
tial moderating factors or variables. For continuous variables ranged from 10 to 50 m. Regarding the sprint time assess-
comparison, the cut-off values based on medians from the ment, nine studies used photoelectric cells [39, 55, 57, 58,
full sprint analysis were used. However, in specific cases, the 61–65], three studies used a stopwatch [56, 59, 60], and only
cut-off was established in an arbitrary way (i.e., load). Pub- one study used an indirect method [66] that was recently val-
lication bias was evaluated using the estimating funnel plot idated by Samozino et al. [42] to record sprint performance.
asymmetry test. A p value of < 0.05 was considered statisti-
cally significant. The standard deviation (SD) was calculated 3.3 Acceleration Phase
as the square root of the summation of the squared SDs of
the mean time in the known conditions. Cohen’s d was used The effect of RST on sprint time was measured in 144 par-
to calculate the effect size (ES, 95% confidence limit) of ticipants. The results of the overall effects on sprint time
each study using the following equation [52]: showed a significant and moderate improvement between
pre- and post-test on the sprint performance (ES 0.61;
Mpre − Mpost ( 3
)
ES = 1− p = 0.0001; standardized mean difference [SMD] 0.57; 95%
Spre 4n − 5
confidence interval [CI] − 0.85 to − 0.28), with an average
where Mpre is the mean value before the CT intervention, heterogeneity of I2 = 28% (Fig. 2) [39, 55–57, 59–66]. Fur-
Mpost is the mean value after the intervention, n is the sam- thermore, in the studies that included a control group, a non-
ple size of the CT group, and Spre is the SD pre-intervention. significant improvement was found in participants belonging
Threshold values for Cohen’s ES statistics were > 0.2 to the RST group compared with the control group (ES 0.09;
(small), > 0.6 (moderate), and > 1.2 (large) [53]. p = 0.64; SMD 0.07; 95% CI − 0.37 to 0.23), with an average
heterogeneity of I2 = 0% (Fig. 3) [55–57, 59–62, 64, 66].
2.5 Risk of Bias Table 3 presents the subgroup analysis assessing poten-
tial moderating factors for sprint time on the acceleration
Methodological quality and risk of bias were independently phase of sprint. Regarding the population characteris-
assessed via visual interpretation of the funnel plot by two tics, significant (p ≤ 0.05) improvements were found for
authors (AM, BO), with disagreements resolved by a third age, sex, and level. Moderate ESs were obtained for age
party evaluator (PA), in accordance with Cochrane Collabo- (ES < 21 = 0.60; ≥ 21 = 0.62), male sex (ES 0.73), recrea-
ration guidelines [54]. tionally active (ES 0.75), and trained (ES 0.84). However,
non-significant improvements were found for female sex (ES
0.14) and highly trained (ES 0.30). Additionally, between-
3 Results subgroup analyses revealed significant (p ≤ 0.05) differences
for sex and level.
3.1 Characteristics of Included Studies Concerning the exercise characteristics, significant
(p ≤ 0.05) improvements were found for lower loads (< 20%
A total of 2376 studies were found following the study BM) (Fig. 4), training frequency, training period duration,
selection procedures, and 659 studies remained after dupli- session volume, total training volume, and rigid and grass
cates were removed. Finally, 13 studies [39, 55–66] were surfaces. A large ES was obtained for a frequency over twice
included in this meta-analysis (Table 1). All included stud- a week (ES 1.85) [55], and moderate ESs were found for
ies had an RST group that accounted for a total of 32 sub- lower loads (ES 0.61) [39, 55–57, 59–65], > 6-week train-
groups analyzed for sprint time. However, only ten studies ing periods (ES 0.63) [39, 55, 56, 60, 63–66], session vol-
[55–62, 64, 66] had control groups, which represented a ume > 160 m (ES 0.92) [55, 56, 59, 60, 64], total weekly
total of 15 subgroups analyzed. training volume > 2680 m (ES 0.83) [55, 59, 60, 64], and
The quality (internal validity) of the trials, according to rigid surface (ES 0.69) [39, 56, 61, 64]. Small ESs were
a PEDro scale [51], was high. The mean score was eight of also found for a training frequency equal to or fewer than
a possible ten points. two trainings per week (ES 0.52) [39, 56, 57, 59–66], a
total weekly training volume < 2680 m (ES 0.53) [39, 56,
3.2 Characteristics of the Interventions 57, 61–63, 65, 66], and grass surface (ES 0.47) [59, 60, 65,
66]. However, non-significant improvements were found for
Table 2 shows the characteristics of the different RST inter- higher loads (≥ 20% BM; ES 0.63) [39, 63, 66], a train-
vention groups. The sled-towing exercise load performed ing period ≤ 6 weeks (ES 0.55) [57, 59, 61, 62], and track
ranged from 5 to 80% of BM. The interventions ranged from surface (ES 0.64) [55, 57, 62, 63]. Additionally, significant
4 to 10 weeks in duration, with a frequency of one to three (p ≤ 0.05) differences were found for frequency of train-
ing between subgroups; however, it should be taken into
Effectiveness of RST for Sprint Performance

Table 1  Main ­characteristicsa of studies included in the meta-analysis

Study Type N Females (%) Age (years) Weight Height Level
CG RST

Alcaraz et al. [62] A: 0–15 m 11 11 45 21.5 ± 2.2 69.8 ± 14.7 173.0 ± 10.5 Highly trained

B: 15–50 m
C: 0–50 m
Bachero-Mena et al. [63] LL–A: 0–20 m – 7 0 21.9 ± 2.3 75.8 ± 10.7 180.9 ± 6.8 Recreationally active
LL–B: 20–40 m
LL–C: 0–40 m
ML–A: 0–20 m – 6 0 20.8 ± 2.1 66.8 ± 8.5 173.8 ± 4.6
ML–B: 20–40 m
ML–C: 0–40 m
HL–A: 0–20 m – 6 0 19.8 ± 1.6 70.2 ± 11.9 175.4 ± 6.8
HL–B: 20–40 m
HL–C: 0–40 m
Clark et al. [58] B: 18.3–54.9 m 7 7 0 19.7 ± 1.0 87.9 ± 17.3 181.15 ± 6.8 Trained
De Hoyo et al. [65] A: 0–20 m – 13 0 17.0 ± 1.0 73.1 ± 2.56 178.24 ± 1.3 Highly trained
B: 20–50 m
C: 0–50 m
Harrison and Bourke [57] A: 0–10 m 7 8 0 20.5 ± 2.8 87.0 ± 10.5 NA Highly trained
Kawamori et al. [39] LL: 0–10 m – 11 0 22.3 ± 5.2 82.5 ± 9.0 183.0 ± 0.07 Trained and recreationally
HL: 0–10 m 10 0 22.8 ± 3.3 77.5 ± 7.3 179.0 ± 0.08 active
Lockie et al. [59] A: 0–10 m 9 9 0 23.1 ± 4.2 83.1 ± 8.6 182.0 ± 0.1 Trained
Luteberget et al. [64] A: 0–10 m 8 10 100 20.4 ± 3.1 74.6 ± 5.9 170.3 ± 5.3 Highly trained
B: 10–30 m
C: 0–30 m
Makaruk et al. [60] A: 0–20 m 12 12 100 22.0 ± 0.9 61.5 ± 4.7 167.0 ± 0.1 Recreationally active
Morin et al. [66] A: 0–20 m 6 10 0 26.3 ± 4.0 74.5 ± 5.3 177.0 ± 0.1 Trained
Spinks et al. [56] A:0–15 m 10 10 0 21.8 ± 4.2 83.3 ± 8.7 181.9 ± 6.2 Trained
West et al. [61] A: 0–10 m 10 10 0 26.8 ± 3.0 90.2 ± 10.3 186.0 ± 8.0 Highly trained
B: 10–30 m
C: 0–30 m
Zafeiridis et al. [55] A: 0–20 m 11 11 0 20.1 ± 1.9 73.1 ± 2.4 178.0 ± 7.0 Recreationally active
B: 20–50 m
C: 0–50 m

Data are mean, mean ± standard deviation, or n

A acceleration phase distance, B maximum-velocity phase distance, C full sprint distance, CG control group, HL high load, LL low load, ML
medium load, NA not available, RST resisted sled training exercise-group
a
 All characteristics refer to the RST group

consideration that only one study evaluated a training fre- Furthermore, the studies with a control group found a non-
quency of more than twice a week. significant improvement in the RST group compared with
the control group (ES 0.29; p = 0.23; SMD 0.26; 95% CI
3.4 Maximum‑Velocity Phase − 0.16 to 0.68), with an average heterogeneity of I2 = 5%
(Fig. 6) [55, 58, 61, 62, 64].
The effect of RST on sprint time in the maximum-veloc- Subgroup analysis assessing potential moderating factors
ity phase was measured in 81 participants. The results of for sprint time on the maximum-velocity phase of sprint is
the overall effects on sprint time showed a non-significant presented in Table 4. Both for population and exercise char-
improvement between pre- and post-test on the sprint time acteristics, the ESs were small or trivial (ES 0.00–0.43) in
(ES 0.27; p = 0.25; SMD 0.18; 95% CI − 0.49 to 0.13), with the maximum-velocity phase (Fig. 7).
an average heterogeneity of I2 = 0% (Fig. 5) [55, 58, 61–65].
 P. E. Alcaraz et al.

Table 2  Characteristics of the resisted sled training interventions and sprint time assessment of the studies included in the meta-analysis
Study Fre- Session volume (m) Total train- Duration Surface Load (% BM) Sprint time assessment
quency ing volume (weeks)
­(week−1) (m) Instrument Total
distance
(m)

Alcaraz et al. [62] 2 90–180 1080 4 Track ~ 8 to 9 Photoelectric cells 50

Bachero-Mena et al. 2 100–210 2115 7 Track 5 Photoelectric cells 40
[63] 12.5
20
Clark et al. [58] 2 240–400 4060 7 Rigid 10.2 Photoelectric cells 36.6
De Hoyo et al. [65] 1–2 120–200 2680 8 Grass 12.6 Photoelectric cells 50
Harrison and Bourke 2 120 1440 6 Track ~ 13 Photoelectric cells 10
[57]
Kawamori et al. [39] 2 90–140 1740 8 Rigid ~ 13 Photoelectric cells 10
~ 43
Lockie et al. [59] 2 195–320 3100 6 Grass 12.6 Velocimeter with 10
stopwatch
Luteberget et al. [64] 2 240–280 5200 10 Rigid 12.4 Photoelectric cells 30
Makaruk et al. [60] 2 180–360 6210 9 Grass 7.5–10 Stopwatch 20
Morin et al. [66] 2 100 1600 8 Grass 80 Indirect method 20
Spinks et al. [56] 2 215–340 4090 8 Rigid 12.6 Stopwatch 15
West et al. [61] 2 60 720 6 Rigid 12.6 Photoelectric cells 30
Zafeiridis et al. [55] 3 280 6720 8 Track ~ 6.8 Photoelectric cells 50

Data are mean or range

BM body mass

Fig. 2  SMD between post and pre-intervention for sprint time in the CI confidence interval, HL high load, IV independent variable, LL
acceleration phase. Squares represent the SMD for each trial. Dia- low load, ML moderate load, SD standard deviation, SMD standard-
monds represent the pooled SMD across trials. A acceleration phase, ized mean difference

3.5 Full Sprint average heterogeneity of I2 = 0% (Fig. 8) [55, 60–66]. How-

ever, the studies with a control group found a non-significant
The effect of RST on sprint time in the full sprint was meas- improvement in the RST groups compared with the con-
ured in 96 participants. The results of the overall effects trol groups (ES 0.05; p = 0.89; SMD 0.03; 95% CI − 0.40
on sprint time showed a significant (p ≤ 0.05) improvement to 0.47), with an average heterogeneity of I2 = 0% (Fig. 9)
between pre- and post-test in sprint performance (ES 0.36; [55, 61, 62, 64].
p = 0.009; SMD 0.38; 95% CI − 0.67 to − 0.10), with an
Effectiveness of RST for Sprint Performance

Fig. 3  SMD in post-intervention sprint time between intervention and across trials. A acceleration phase, CI confidence interval, IV inde-
control athletes for sprint time in the acceleration phase. Squares rep- pendent variable, SD standard deviation, SMD standardized mean dif-
resent the SMD for each trial. Diamonds represent the pooled SMD ference

Table 5 presents the subgroup analysis assessing poten- whether RST effectively improves sprint performance, in
tial moderating factors for sprint time on the full sprint. both its acceleration and maximum-velocity phases. Second,
Small ESs (0.24–0.53) were found for both population and we intended to establish which variables are associated with
exercise characteristics. Furthermore, significant (p ≤ 0.05) the largest RST-induced improvements. The main finding
improvements with small ESs were found for younger ath- of the present analysis is that RST improves sprint perfor-
letes (aged < 21 years; ES 0.37) [55, 63–65], male sex (ES mance, mainly in its early acceleration phase. However, its
0.34) [55, 61, 63, 65, 66], team-sports athletes (ES 0.42) [60, effect is trivial or low in the maximum-velocity phase or in
61, 64–66], using a load < 20% BM (ES 0.35) (Fig. 10) [55, sprints of ≥ 20 m. Also, no additional benefit was observed
60–65], a training frequency equal to or less than twice a when RST was compared with UR conditions. Furthermore,
week (ES 0.35) [60–66], a period of training > 6 weeks (ES it was determined that the magnitude of its effect on sprint
0.39) [55, 60, 63–66], a session volume > 160 m (ES = 0.53) performance is related to the selected population and/or
[55, 60, 64], and total training volume > 2680 m (ES 0.53) training characteristics.
[55, 60, 64].
4.1 Acceleration Phase
3.6 Evaluation of Potential Bias
As mentioned in Sect. 1, training with sled towing is an
Visual interpretation of the funnel plot was performed to appropriate method for improving the early acceleration
evaluate potential bias. SMD between pre- and post-inter- phase of the sprint (here defined as 0–10 m), with a moderate
vention sprint time in RST participants was considered nota- and significant effect. However, it is no more effective than
bly symmetrical, suggesting the absence of a significant pub- performing the same sprint training without overload. The
lication bias. Similar results were obtained for the evaluation explanation for these results is that most studies use loads
of potential bias of the SMD in post-intervention sprint time close to a subject’s BM, making the differences in load very
between RST and control group athletes. small. This fact has led some researchers [41, 66] to think
that higher loads may have to be used for the improvements
to be significantly greater than when lower loads are used.
4 Discussion However, given the low number of studies using loads > 20%
BM, the effect is similar (0.61 vs. 0.63, respectively), and the
RST has been used extensively to improve sprint perfor- effect pre-post is not significant with high loads. This finding
mance. However, because there are many variables to may be explained by some of the main characteristics that
manipulate (load, distance, friction, etc.) with this type of make the human being run at great speed, such as muscular
device, substantial controversy has been generated, both in mechanical properties (i.e., the capacity for adequate use of
the scientific community and in the field of training, regard- the elastic elements, both in series and in parallel), consid-
ing the optimal training protocol. The main objective of ering here the SSC, RFD, as well as the H reflex. It is clear
this systematic review with meta-analysis was to determine that the SSC and H reflex can be developed in situations
 P. E. Alcaraz et al.

Table 3  Subgroup analysis assessing potential moderating factors for sprint time in the acceleration phase in the studies included in the meta-
analysis
Population characteristics Studies Resisted sled training
a
Number References SMD (95% CI) ES I2 (%) p pDifference

Age (~ 17 to 27 years)

  < 21 y.o. 7 Bachero-Mena et al. [63]: LL-A, ML-A, − 0.53 (− 1.01 to − 0.04) 0.60 40 0.03 0.81
HL-A
de Hoyo et al. [65]: A
Harrison and Bourke [57]
Luteberget et al. [64]: A
Zafeiridis et al. [55]: A
  ≥ 21 y.o. 8 Alcaraz et al. [62]: A − 0.60 (− 0.97 to − 0.23) 0.62 25 0.001
Kawamori et al. [39]: LL, HL
Lockie et al. [59]
Makaruk et al. [60]
Morin et al. [66]
Spinks et al. [56]
West et al. [61]: A
Sex
 Female ≥ 45% sample 3 Alcaraz et al. [62]: A − 0.12 (− 0.60 to 0.37) 0.14 0 0.64 0.05
Luteberget et al. [64]: A
Makaruk et al. [60]
 Male 12 Bachero-Mena et al. [63]: LL-A, ML-A, − 0.70 (− 1.01 to − 0.39) 0.73 20 < 0.001
HL-A
de Hoyo et al. [65]: A
Harrison and Bourke [57]
Kawamori et al. [39]: LL, HL
Lockie et al. [59]
Morin et al. [66]
Spinks et al. [56]
West et al. [61]: A
Zafeiridis et al. [55]: A
Level
 Recreationally active 7 Bachero-Mena et al. [63]: LL-A, ML-A, − 0.79 (− 1.24 to − 0.34) 0.75 29 < 0.001 0.04
HL-A
Kawamori et al. [39]: LL, HL
Makaruk et al. [60]
Zafeiridis et al. [55]: A
 Trained 5 Kawamori et al. [39]: LL, HL − 0.85 (− 1.27 to − 0.43) 0.84 0 < 0.001
Lockie et al. [59]
Morin et al. [66]
Spinks et al. [56]
 Highly trained 5 Alcaraz et al. [62]: A − 0.17 (− 0.56 to 0.22) 0.30 0 0.39
de Hoyo et al. [65]: A
Harrison and Bourke [57]
Luteberget et al. [64]: A
West et al. [61]: A
Sport characteristics
 Team sports 9 de Hoyo et al. [65]: A − 0.58 (− 0.91 to − 0.26) 0.66 11 < 0.001 0.07
Harrison and Bourke [57]
Kawamori et al. [39]: LL, HL
Lockie et al. [59]
Luteberget et al. [64]: A
Morin et al. [66]
Spinks et al. [56]
West et al. [61]: A
 Individual sports 1 Alcaraz et al. [62]: A 0.23 (− 0.61 to 1.07) 0.23 NA 0.59
Effectiveness of RST for Sprint Performance

Table 3  (continued)
Population characteristics Studies Resisted sled training
a
Number References SMD (95% CI) ES I2 (%) p pDifference

Exercise characteristics
 Load (5–80% BM)
   < 20% BM 12 Alcaraz et al. [62]: A − 0.61 (− 0.97 to − 0.25) 0.61 43 < 0.001 0.63
Bachero-Mena et al. [63]: LL-A, ML-A
de Hoyo et al. [65]: A
Harrison and Bourke [57]
Kawamori et al. [39]: LL
Lockie et al. [59]
Luteberget et al. [64]: A
Makaruk et al. [60]
Spinks et al. [56]
West et al. [61]: A
Zafeiridis et al. [55]: A
   ≥ 20% BM 3 Bachero-Mena et al. [63]: HL-A − 0.45 (− 0.99 to 0.10) 0.63 0 0.11
Kawamori et al. [39]: HL
Morin et al. [66]
 Frequency (1–3 week−1)
   ≤ 2 week−1 14 Alcaraz et al. [62]: A − 0.47 (− 0.72 to − 0.22) 0.52 0 < 0.001 0.008
Bachero-Mena et al. [63]: LL-A, ML-A,
HL-A
de Hoyo et al. [65]: A
Harrison and Bourke [57]
Kawamori et al. [39]: LL, HL
Lockie et al. [59]
Luteberget et al. [64]: A
Makaruk et al. [60]
Morin et al. [66]
Spinks et al. [56]
West et al. [61]: A
   > 2 week−1 1 Zafeiridis et al. [55]: A − 1.92 (− 2.97 to − 0.88) 1.85 NA < 0.001
 Duration (4–10 weeks)
   ≤ 6 weeks 4 Alcaraz et al. [62]: A − 0.38 (− 0.87 to 0.11) 0.55 12 0.13 0.41
Harrison and Bourke [57]
Lockie et al. [59]
West et al. [61]: A
   > 6 weeks 11 Bachero-Mena et al. [63]: LL-A, ML-A, − 0.64 (− 0.99 to − 0.28) 0.63 35 < 0.001
HL-A
de Hoyo et al. [65]: A
Kawamori et al. [39]: LL, HL
Luteberget et al. [64]: A
Makaruk et al. [60]
Morin et al. [66]
Spinks et al. [56]
Zafeiridis et al. [55]: A
 Session volume (60–360 m)
   ≤ 160 m 10 Alcaraz et al. [62]: A − 0.40 (− 0.70 to − 0.11) 0.46 0 0.008 0.18
Bachero-Mena et al. [63]: LL-A, ML-A,
HL-A
de Hoyo et al. [65]: A
Harrison and Bourke [57]
Kawamori et al. [39]: LL, HL
Morin et al. [66]
West et al. [61]: A
 P. E. Alcaraz et al.

Table 3  (continued)
Population characteristics Studies Resisted sled training
a
Number References SMD (95% CI) ES I2 (%) p pDifference

   > 160 m 5 Lockie et al. [59] − 0.89 (− 1.53 to − 0.25) 0.92 57 0.006

Luteberget et al. [64]: A
Makaruk et al. [60]
Spinks et al. [56]
Zafeiridis et al. [55]: A
 Total training volume (720–6720 m)
   ≤ 2680 m 11 Alcaraz et al. [62]: A − 0.49 (− 0.79 to − 0.19) 0.53 9 0.001 0.48
Bachero-Mena et al. [63]: LL-A, ML-A,
HL-A
de Hoyo et al. [65]: A
Harrison and Bourke [57]
Kawamori et al. [39]: LL, HL
Morin et al. [66]
Spinks et al. [56]
West et al. [61]: A
   > 2680 m 4 Lockie et al. [59] − 0.78 (− 1.52 to − 0.03) 0.83 62 0.04
Luteberget et al. [64]: A
Makaruk et al. [60]
Zafeiridis et al. [55]: A
 Surface
  Rigid 5 Kawamori et al. [39]: LL, HL − 0.71 (− 1.24 to − 0.18) 0.69 40 0.009 0.70
Luteberget et al. [64]: A
Spinks et al. [56]
West et al. [61]: A
  Track 6 Alcaraz et al. [62]: A − 0.57 (− 1.19 to 0.05) 0.64 54 0.07
Bachero-Mena et al. [63]: LL-A; ML-A;
HL-A
Harrison and Bourke [57]
Zafeiridis et al. [55]: A
  Grass 4 de Hoyo et al. [65]: A − 0.42 (− 0.84 to 0.01) 0.47 0 0.05
Lockie et al. [59]
Makaruk et al. [60]
Morin et al. [66]

Subgroup analyses were performed on SMD between post and pre-intervention sprint time. Median values of continuous variables were used as
cut-off values for grouping studies. Changes in moderating factors were calculated as post-intervention minus pre-intervention values
A acceleration phase, B maximum-velocity phase, BM body mass, C full sprint, CI confidence interval, ES Cohen´s d effect size, HL high load,
I2 heterogeneity, LL low load, ML medium load, NA not available, p test for overall effect, pDifference test for subgroup differences, SMD standard-
ized mean difference
a
 Number of intervention groups in the studies. Some enrolled studies were not included because the value used for subgroup analysis was not
reported

where specific patterns of movement are replicated and load phase than in the maximum-velocity phase, where the move-
specificity occurs [67]. Therefore, an excess load may limit ment velocity, the SSC, and H reflex are critical [67]. That
the development of the SSC, and—to a lesser degree—the is, in the acceleration phase, CT is greater than during the
coupling of the H reflex, despite the athlete’s attempts to maximum-velocity phase, as is the involvement of horizontal
maintain the movement pattern. On the other hand, the mus- force and the larger H:V force ratio. Moreover, it has been
cle has been observed to work quasi-isometrically when suggested that increases in maximal strength are more likely
sprinting [68, 69], allowing for greater tendon lengthening to increase short-distance (5-m) sprint performance [73].
as load intensity increases [70] and for the tendon to act as One of the most analyzed variables that has consequently
a power amplifier as it recoils at high velocities [71, 72]. generated the greatest discrepancy among the scientific com-
Therefore, if we take into account that the tendon demands munity in recent years, is the load that should be used when
are dependent on the movement velocity, the adaptations performing sled towing. Some authors [40] have shown that
induced will also be velocity dependent. This appears to also the maximum power production when using these methods
explain why RST is much more effective in the acceleration occurs with loads around 20% BM, whereas others have
Effectiveness of RST for Sprint Performance

Fig. 4  SMD between post and pre-intervention  for sprint time in moderate load, LL low load, BM body mass, CI confidence interval,
the acceleration phase based on different load (subgroup analysis). IV independent variable, SD standard deviation, SMD standardized
Squares represent the SMD for each trial. Diamonds represent the mean difference
pooled SMD across trials. A acceleration phase, HL high load, ML

Fig. 5  SMD between post and pre-intervention for sprint time in the velocity phase, CI confidence interval, HL high load, IV independ-
maximum-velocity phase. Squares represent the SMD for each trial. ent variable, LL low load, ML moderate load, SD standard deviation,
Diamonds represent the pooled SMD across trials. B maximum- SMD standardized mean difference

suggested that the maximum power output can be achieved for improving sprint performance. Hence, more research is
with 69–96% BM [41]. It should be noted that Monte et al. needed in this regard.
[40] calculated and examined the load that maximizes power As mentioned, the present meta-analysis observed that
output with sprinters and for the full sprint (20 m) and ana- the effect on performance improvement was moderate when
lyzed the effect on the kinematics of the joints and segments. comparing loads < 20% BM and loads ≥ 20% BM, with no
However, Cross et al. [41] calculated the maximum power significant improvements achieved with loads ≥ 20% BM.
output at the moment when participants (team-sports ath- It should be noted that only three groups [39, 63, 66] used
letes) reached maximum sprint speed and did not calculate loads ≥ 20% BM and that these three groups were composed
effect on sprint technique. In any case, it is not clear for this of team-sports athletes. Therefore, it cannot be inferred that
training method that training with the load that produces the use of high loads is more effective than the use of low
the maximum power output is the most effective method loads. In fact, the only study to use very high loads (80%
 P. E. Alcaraz et al.

Table 4  Subgroup analysis assessing potential moderating factors for sprint time in the maximum-velocity phase in the studies included in the
meta-analysis
Population characteristics Studies SMD (95% CI) Resisted sled training
a
Number References ES I2 (%) p pDifference

Age (~ 17 to 27 years)

  < 21 y.o. 7 Bachero-Mena et al. [63]: LL-B, ML-B, HL-B − 0.17 (− 0.53 to 0.19) 0.26 0 0.35 0.92
Clark et al. [58]
de Hoyo et al. [65]: B
Luteberget et al. [64]: B
Zafeiridis et al. [55]: B
  ≥ 21 y.o. 2 Alcaraz et al. [62]: B − 0.21 (− 0.81 to 0.40) 0.28 0 0.50
West et al. [61]: B
Sex
 Female ≥ 45% simple 2 Alcaraz et al. [62]: B − 0.38 (− 0.99 to 0.24) 0.18 0 0.23 0.47
Luteberget et al. [64]: B
 Male 7 Bachero-Mena et al. [63]: LL-B, ML-B, HL-B − 0.11 (− 0.47 to 0.25) 0.23 0 0.54
Clark et al. [58]
de Hoyo et al. [65]: B
West et al. [61]: B
Zafeiridis et al. [55]: B
Level
 Recreationally active 4 Bachero-Mena et al. [63]: LL-B, ML-B, HL-B − 0.04 (− 0.54 to 0.47) 0.21 0 0.89 0.67
Zafeiridis et al. [55]: B
 Trained 1 Clark et al. [58] 0.00 (− 1.05 to 1.05) 0.00 NA 1
 Highly trained 4 Alcaraz et al. [62]: B − 0.31 (− 0.73 to 0.11) 0.39 0 0.15
de Hoyo et al. [65]: B
Luteberget et al. [64]: B
West et al. [61]: B
Sport characteristics
 Team sports 4 Clark et al. [58] − 0.29 (− 0.73 to 0.15) 0.34 0 0.20 0.84
de Hoyo et al. [65]: B
Luteberget et al. [64]: B
West et al. [61]: B
 Individual sports 1 Alcaraz et al. [62]: B − 0.19 (− 1.03 to 0.65) 0.20 NA 0.65
Exercise characteristics
 Load (5–20% BM)
   < 20% BM 8 Alcaraz et al. [62]: B − 0.18 (− 0.50 to 0.14) 0.26 0 0.28 0.97
Bachero-Mena et al. [63]: LL-B, ML-B
Clark et al. [58]
de Hoyo et al. [65]: B
Luteberget et al. [64]: B
West et al. [61]: B
Zafeiridis et al. [55]: B
   ≥ 20% BM 1 Bachero-Mena et al. [63]: HL-B − 0.20 (− 1.34 to 0.93) 0.28 NA 0.73
 Frequency (1–3 week−1)
   ≤ 2 week−1 8 Alcaraz et al. [62]: B − 0.23 (− 0.57 to 0.10) 0.25 0 0.17 0.41
Bachero-Mena et al. [63]: LL-B, ML-B, HL-B
Clark et al. [58]
de Hoyo et al. [65]: B
Luteberget et al. [64]: B
West et al. [61]: B
   > 2 week−1 1 Zafeiridis et al. [55]: B 0.15 (− 0.69 to 0.98) 0.13 NA 0.73
 Duration (4–8 weeks)
   ≤ 6 weeks 2 Alcaraz et al. [62]: B − 0.21 (− 0.81 to 0.40) 0.28 0 0.50 0.92
West et al. [61]: B
Effectiveness of RST for Sprint Performance

Table 4  (continued)
Population characteristics Studies SMD (95% CI) Resisted sled training
a
Number References ES I2 (%) p pDifference

   > 6 weeks 7 Bachero-Mena et al. [63]: LL-B, ML-B, HL-B − 0.17 (− 0.53 to 0.19) 0.26 0 0.35
Clark et al. [58]
de Hoyo et al. [65]: B
Luteberget et al. [64]: B
Zafeiridis et al. [55]: B
 Session volume (60–400 m)
   ≤ 160 m 6 Alcaraz et al. [62]: B − 0.20 (− 0.58 to 0.18) 0.32 0 0.31 0.87
Bachero-Mena et al. [63]: LL-B, ML-B, HL-B
De Hoyo et al. [65]: B
West et al. [61]: B
   > 160 m 3 Clark et al. [58] − 0.15 (− 0.67 to 0.38) 0.15 0 0.59
Luteberget [64]: B
Zafeiridis et al. [55]: B
 Total training volume (720–6720 m)
   ≤ 2680 m 6 Alcaraz et al. [62]: B − 0.20 (− 0.58 to 0.18) 0.32 0 0.31 0.87
Bachero-Mena et al. [63]: LL-B, ML-B, HL-B
De Hoyo et al. [65]: B
West et al. [61]: B
   > 2680 m 3 Clark et al. [58] − 0.15 (− 0.67 to 0.38) 0.15 0 0.59
Luteberget et al. [64]: B
Zafeiridis et al. [55]: B
 Surface
  Rigid 3 Clark et al. [58] − 0.30 (− 0.84 to 0.24) 0.31 0 0.28 0.80
Luteberget et al. [64]: B
West et al. [61]: B
  Track 5 Alcaraz et al. [62]: B − 0.08 (− 0.51 to 0.36) 0.20 0 0.73
Bachero-Mena et al. [63]: LL-B, ML-B, HL-B
Zafeiridis et al. [55]: B
  Grass 1 de Hoyo et al. [65]: B − 0.27 (− 1.04 to 0.50) 0.43 NA 0.49

Subgroup analyses were performed on SMD between post and pre-intervention sprint time. Median values of continuous variables were used as
cut-off values for grouping studies. Changes in moderating factors were calculated as post-intervention minus pre-intervention values
A acceleration phase, B maximum-velocity phase, BM body mass, C full sprint, CI confidence interval, ES Cohen´s d effect size, HL high load,
I2 heterogeneity, LL low load, ML medium load, NA not available data, p test for overall effect, pDifference test for subgroup differences, SMD
standardized mean difference
a
 Number of intervention groups in the studies. Some enrolled studies were not included because the value used for subgroup analysis was not
reported

Fig. 6  SMD in post-intervention sprint time between intervention pooled SMD across trials. B maximum-velocity phase, CI confidence
and control athletes for sprint time in the maximum-velocity phase. interval, IV independent variable, SD standard deviation, SMD stand-
Squares represent the SMD for each trial. Diamonds represent the ardized mean difference
 P. E. Alcaraz et al.

Fig. 7  SMD between post and pre-intervention for sprint time in the mass, CI confidence interval, HL high load, LL low load, ML moder-
maximum-velocity phase based on different load (subgroup analy- ate load, IV independent variable, SD standard deviation, SMD stand-
sis). Squares represent the SMD for each trial. Diamonds represent ardized mean difference
the pooled SMD across trials. B maximum-velocity phase, BM body

BM) [66] found a small and trivial effect on performance human leg has a major positive influence on various athletic
of the 5- and 20-m sprint, respectively, which is similar variables, including RFD, elastic energy storage and utiliza-
to those found in the control group with the same training tion, and therefore the SSC, and sprint kinematics (i.e., CT
but without overload. Thus, Morin et al. [66] found only and FT, and SL and frequency) [79]. However, only Alcaraz
a moderate effect on the F0, (i.e., the force at velocity 0), et al. [62] analyzed the effects of sled towing on stiffness, and
or what would be equivalent to the maximum theoretical they found a slight non-significant reduction after a training
force of the participants, but not on V0 (i.e., which would period of 4 weeks (with low loads). However, this study
be the equivalent of the maximum theoretical velocity of observed that the group training without a sled experienced
the players), according to the model proposed by the same an improvement in stiffness that approached significance,
authors [42]. As a consequence, the proposal of using RST suggesting that training that respects the principle of speci-
with very heavy loads would only be justified as a tertiary ficity can have a positive effect on one of the most important
method, which would affect the improvement of the force variables in sprinting (i.e., stiffness), whereas increasing the
at low or null velocities “simulating” the movement pattern load can have the opposite effect. Given these findings and
(and only in athletes who can maintain it). However, it is the recommendations based on a mixed-methods approach
far from the load and velocity specificity, which are both in which a variety of loads and exercise types are used in a
decisive in the training of maximum power production, as periodized fashion to optimize power output, training with
explained by Cormie et al. [28] in their narrative review on high loads could be an alternative to traditional resistance
maximal power production training. Another problem that training. Thereby, using high-load RST to improve maxi-
can occur when using excessively high loads is that neither mum strength, athletes train—to some extent—replicating
the SSC nor the H reflex are properly activated. In fact, train- the movement patterns of the sprint. However, when the aim
ing activities aimed at improving SSC performance should is to work in the area of the force-velocity curve close to
fulfill two criteria [15, 74–77]: (1) they should involve skill- the demands of the competition, the RST must be carried
ful, multi-joint movements that transmit forces through the out with low loads to develop all the mechanisms involved
kinetic chain and exploit elastic–reflexive mechanisms; and in this situation, such as the RFD, SSC, H reflex, and leg
(2) they should be structured around brief work bouts or stiffness.
clusters separated by frequent rest periods to manage fatigue Regarding the characteristics of the population, statisti-
and emphasize work quality and technique. cally significant differences were observed only between lev-
Another point to consider is that chronic exposure to els of participants, with the effect being moderate for both
movements eliciting the SSC can increase muscle stiffness, recreationally active (ES 0.75) and trained participants (ES
which is a potential physiological advantage for sprint ability 0.84) but small for highly trained participants (ES 0.30).
[78]. An optimal development of mechanical stiffness in the The explanation for these results lies in the fact that highly
Effectiveness of RST for Sprint Performance

Table 5  Subgroup analysis assessing potential moderating factors for sprint time during full sprint in the studies included in the meta-analysis
Population characteristics Studies Resisted sled training
a
Number References SMD (95% CI) ES I2 (%) p pDifference

Age (~ 17 to 27 years)

  < 21 y.o. 6 Bachero-Mena et al. [63]: LL-C, ML-C, HL-C − 0.41 (− 0.80 to − 0.02) 0.37 0 0.04 0.83
De Hoyo et al. [65]: C
Luteberget et al. [64]: C
Zafeiridis et al. [55]: C
  ≥ 21 y.o. 4 Alcaraz et al. [62]: C − 0.35 (− 0.78 to 0.08) 0.35 0 0.11
Makaruk et al. [60]
Morin et al. [66]
West et al. [61]: C
Sex
 Female ≥ 45% sample 3 Alcaraz et al. [62]: C − 0.42 (− 0.91 to 0.07) 0.41 0 0.09 0.84
Luteberget et al. [64]: C
Makaruk et al. [60]
 Male 7 Bachero-Mena et al. [63]: LL-C, ML-C, HL-C − 0.36 (− 0.71 to − 0.01) 0.34 0 0.04
De Hoyo et al. [65]: C
Morin et al. [66]
West et al. [61]: C
Zafeiridis et al. [55]: C
Level
 Recreationally active 4 Bachero-Mena et al. [63]:LL-C, ML-C, HL-C − 0.38 (− 0.90 to 0.13) 0.35 0 0.14 0.91
Zafeiridis et al. [55]: C
 Trained 0 – – – – –
 Highly trained 4 Alcaraz et al. [62]: C − 0.35 (− 0.77 to 0.08) 0.32 0 0.11
De Hoyo et al. [65]: C
Luteberget et al. [64]: C
West et al. [61]: C
Sport characteristics
 Team sports 5 De Hoyo et al. [65]: C − 0.43 (− 0.81 to − 0.06) 0.42 0 0.02 0.51
Luteberget et al. [64]: C
Makaruk et al. [60]
Morin et al. [66]
West et al. [61]: C
 Individual sports 1 Alcaraz et al. [62]: C − 0.12 (− 0.96 to 0.71) 0.12 NA 0.77
Exercise characteristics
 Load (5–80% BM)
   < 20% BM 8 Alcaraz et al. [62]: C − 0.38 (− 0.70 to − 0.07) 0.35 0 0.02 0.99
Bachero-Mena et al. [63]: LL-C, ML-C
De Hoyo et al. [65]: C
Luteberget et al. [64]: C
Makaruk et al. [60]
West et al. [61]: C
Zafeiridis et al. [55]: C
   ≥ 20% BM 2 Bachero-Mena et al. [63]: HL-C − 0.39 (− 1.09 to 0.32) 0.42 0 0.28
Morin et al. [66]
 Frequency (1–3 week−1)
   ≤ 2 week−1 9 Alcaraz et al. [62]: C − 0.37 (− 0.67 to − 0.06) 0.35 0 0.02 0.76
Bachero-Mena et al. [63]: LL-C, ML-C, HL-C
de Hoyo et al. [65]: C
Luteberget et al. [64]: C
Makaruk et al. [60]
Morin et al. [66]
West et al. [61]: C
   > 2 week−1 1 Zafeiridis et al. [55]: C − 0.51 (− 1.36 to 0.35) 0.46 NA 0.24
 P. E. Alcaraz et al.

Table 5  (continued)
Population characteristics Studies Resisted sled training
a
Number References SMD (95% CI) ES I2 (%) p pDifference

 Duration (4–10 weeks)
   ≤ 6 weeks 2 Alcaraz et al. [62]: C − 0.24 (− 0.85 to 0.37) 0.24 0 0.44 0.61
West et al. [61]: C
   > 6 weeks 8 Bachero-Mena et al. [63]: LL-C, ML-C, HL-C − 0.42 (− 0.75 to − 0.10) 0.39 0 0.01
De Hoyo et al. [65]: C
Luteberget et al. [64]: C
Makaruk et al. [60]
Morin et al. [66]
Zafeiridis et al. [55]: C
 Session volume (60–360 m)
   ≤ 160 m 7 Alcaraz et al. [62]: C − 0.29 (− 0.65 to 0.06) 0.29 0 0.10 0.40
Bachero-Mena et al. [63]: LL-C, ML-C, HL-C
De Hoyo et al. [65]: C
Morin et al. [66]
West et al. [61]: C
   > 160 m 3 Luteberget et al. [64]: C − 0.56 (− 1.05 to − 0.06) 0.53 0 0.03
Makaruk et al. [60]
Zafeiridis et al. [55]: C
 Total training volume (720–6720 m)
   ≤ 2680 m 7 Alcaraz et al. [62]: C − 0.29 (− 0.65 to 0.06) 0.29 0 0.10 0.40
Bachero-Mena et al. [63]: LL-C, ML-C, HL-C
De Hoyo et al. [65]: C
Morin et al. [66]
West et al. [61]: C
   > 2680 m 3 Luteberget et al. [64]: C − 0.56 (− 1.05 to − 0.06) 0.53 0 0.03
Makaruk et al. [60]
Zafeiridis et al. [55]: C
 Surface
  Rigid 2 Luteberget et al. [64]: C − 0.49 (− 1.12 to 0.14) 0.42 0 0.13 0.90
West et al. [61]:C
  Track 5 Alcaraz et al. [62]: C − 0.31 (− 0.75 to 0.12) 0.31 0 0.16
Bachero-Mena et al. [63]: LL-C, ML-C, HL-C
Zafeiridis et al. [55]: C
  Grass 3 De Hoyo et al. [65]: C − 0.40 (− 0.88 to 0.07) 0.42 0 0.10
Makaruk et al. [60]
Morin et al. [66]

Subgroup analyses were performed on SMD between post and pre-intervention sprint time. Median values of continuous variables were used as
cut-off values for grouping studies. Changes in moderating factors were calculated as post-intervention minus pre-intervention values
A acceleration phase, B maximum-velocity phase, BM body mass, C full sprint, CI confidence interval, ES Cohen´s d effect size, HL high load,
I2 heterogeneity, LL low load, ML medium load, NA not available data, p test for overall effect, pDifference test for subgroup differences, SMD
standardized mean difference
a
 Number of intervention groups in the studies. Some enrolled studies were not included because the value used for subgroup analysis was not
reported

trained athletes have a lower margin of improvement than in this type of population, since RST is used regularly in the
recreationally active or trained athletes when any training design of training plans in athletics, specifically in force-
protocol is applied [80]. On the other hand, it has been velocity disciplines (such as sprints, jumps, and hurdles),
observed that, when an RST is applied to improve accel- and understanding is needed on the effects of RST on sprint
eration capacity, the effect is moderate in team sports (ES performance in both trained and highly trained athletes for
0.66) but small in individual sports (athletics) (ES 0.23). It proper programming during the season. The effect shown
is necessary to consider, in this case, that only one study was for team sports is also statistically significant. Therefore, the
included [60] that analyzed the effects of RST on national RST is clearly recommended for the improvement of early
level sprinters and jumpers. Thus, more studies are needed acceleration in sports such as soccer or rugby, particularly
Effectiveness of RST for Sprint Performance

Fig. 8  SMD between post and pre-intervention for sprint time during val, HL high load, IV independent variable, LL low load, ML moder-
full sprint. Squares represent the SMD for each trial. Diamonds repre- ate load, SD standard deviation, SMD standardized mean difference
sent the pooled SMD across trials. C full sprint, CI confidence inter-

Fig. 9  SMD in post-intervention sprint time between intervention and trials. C full sprint, CI confidence interval, IV independent variable,
control athletes for sprint time during full sprint. Squares represent SD standard deviation, SMD standardized mean difference
the SMD for each trial. Diamonds represent the pooled SMD across

since sprinting is the most frequent physical action for scor- program for the development of sprint performance through
ing and assisting players before goals [1], and total sprint RST, based on the results of the present meta-analysis, we
distance and number of sprints undertaken during games recommend a minimum duration of > 6 weeks.
have increased significantly in different European league Another variable that should be considered when design-
players in the last decade [81]. ing training programs with RST is volume, both for each
Regarding the characteristics of training, weekly train- session and for the microcycle. The volume per session var-
ing frequencies > 2 days were observed to produce a sig- ied widely, from 60 to 360 m. Although most (ten) groups
nificantly greater effect than frequencies ≤ 2 (ES 1.85 vs. included volumes < 160 m, volumes between 160 and 360 m
0.52). However, these data must be taken with caution since (five groups) produced greater effects than lower volumes
only one study was included [53] that had applied > 2 days (ES 0.92 vs. 0.46, respectively). Similarly, when weekly
(3 days·week−1) of training. In addition, this study was con- volumes were compared, higher volumes (> 2680 m) pro-
ducted with physical education students, who had a lower duced a moderate effect (ES 0.83) compared with volumes
level of sprint performance and, a priori, higher potential 720–2680 m (ES 0.53), although the effects were statistically
for improvement than those who are highly trained, as significant for both volume per session and weekly volume.
noted in Sect. 3.3. Therefore, training frequencies between Therefore, we can infer that slightly higher volumes, both
2–3 days·week−1 can be optimal for the development of per session and microcycle, have a greater effect; however,
acceleration capacity when using RST. With respect to the whether volume is as important as other variables that have
duration of the program, four groups [57, 59, 61, 62] had been previously analyzed is unclear.
a duration of 4–6 weeks, with a small and non-significant Finally, with respect to the early acceleration phase, it has
effect, whereas 11 groups [39, 55, 56, 60, 63–66] used dura- been observed that the effect differs according to the surface
tions of 6–10 weeks, with a significant (p < 0.001) moderate used. For example, rigid surfaces or athletic tracks have a
effect (ES 0.63). The explanation for this finding may be moderate effect (ES 0.69 and 0.64, respectively) and grass
that shorter durations do not produce sufficient neuromus- produces a small effect. These results can be explained by
cular and mechanical adaptations to have a positive effect the high variability in friction related to the different types
on sprint performance [82]. Consequently, when designing a of surfaces (natural vs. artificial turf). Since the coefficient of
 P. E. Alcaraz et al.

Fig. 10  SMD between post and pre-intervention  for sprint time dur- SMD across trials. BM body mass, C full sprint, CI confidence inter-
ing full sprint based on different load (subgroup analysis). Squares val, HL high load, IV independent variable, LL low load, ML moder-
represent the SMD for each trial. Diamonds represent the pooled ate load, SD standard deviation, SMD standardized mean difference

friction differs greatly between surfaces, this largely affects cases. However, sub-analysis for the full sprint (≥ 20 m)
resistance when using RST, as suggested by Linthorne and indicated that, although the effects were still trivial or small
Cooper [46]. These authors concluded that different sprint in some cases, some were significant. For example, regard-
surfaces would elicit varying degrees of coefficient of friction. ing the population characteristics, the effect was small but
significant for age < 21 years (ES 0.37; p = 0.04), men (ES
0.34; p = 0.04), and team sports (ES 0.42; p = 0.02). With
4.2 Maximum‑Velocity Phase and Full Sprint respect to the training characteristics, the effect was small
in all cases and significant with loads < 20% BM (ES 0.35;
Unlike the findings for the acceleration phase, RST has a p = 0.02), weekly training frequencies of  ≤  2 days (ES 0.35;
small effect on performance in the maximum-velocity phase p = 0.02), for duration > 6 weeks (ES 0.39; p = 0.01), vol-
(ES 0.27) and/or when performing sprints ≥ 20 m (ES 0.36); umes per session > 160 m (ES 0.53; p = 0.03), and weekly
however, in the latter case, the effect was significant. These values > 2680 m (ES 0.53; p = 0.03).
findings can be explained by the fact that resistance coming Although this meta-analysis answers many of the ques-
from the friction between the surface of the sled and the tions from the scientific literature, there is still a significant
contact surface (track, grass, etc.) when the sprint is per- lack of research that focuses on highly trained athletes in
formed with a sled will be different if the sled is stopped individual sports such as athletics, in both males and females
or in motion because of the inertia of the system. There- of different ages.
fore, when the maximum power output is calculated, the
maximum power output in sprinting usually occurs in the
first steps, and the horizontal force and power are decreased 5 Conclusions
by 82% and 63%, respectively, from the first to the last
step when using 20-m RST [40]. Therefore, if the aim is RST has been used extensively in both team and individual
to develop maximum power production in the maximum- sports to improve sprint performance. However, to date,
velocity phase, training strategies other than sled towing there has been no consensus on whether this training method
may be needed, since vertical forces are predominant in this actually improves sprint performance, in either the accelera-
phase [9, 16, 17]. Perhaps plyometric training or another tion or the maximum-velocity phases. Furthermore, whether
type of resisted training, such as the weighted vest, could the effects differ according to population characteristics is
produce a greater effect in these phases. unclear, as is the optimal training load (intensity, volume,
When analyzing both the characteristics of the population etc.) for adaptations to be optimised. Based on the present
and the training method for the maximum-velocity phase, systematic review with meta-analysis, it can be affirmed that
the effects were small or trivial and not significant in all RST is an effective method to improve sprint performance,
Effectiveness of RST for Sprint Performance

mainly via improvement of the early acceleration phase. 6. Does RST have a different effect on athletes depend-
However, it cannot be said that this method is more effec- ing on age? This is not clear. Therefore, more studies
tive than the same training without overload. Regarding the comparing the effects of RST on different age groups
population characteristics, the effect is greater in men, rec- are necessary to determine whether the effect will dif-
reationally active or trained, but is small in highly trained fer.
individuals who practice team sports such as football (soc- 7. Will the effect be greater if the athlete is highly
cer) or rugby. Finally, with regards to the training charac- trained? No, in fact, for the acceleration phase, the
teristics, the intensity (load) is not a determinant of sprint effect is significantly greater in recreationally active
performance improvement, but the recommended volume and trained athletes than in highly trained athletes.
is > 160 m per session, and approximately 2680 m per week, 8. How many days per week should RST be applied?
with a training frequency of 2–3 times per week, for at least These methods should be applied between 2 and 3 days
6 weeks. Finally, rigid surfaces appear to enhance the effect per week, depending on the sprint demands of the sport
of RST on sprint performance. in which the athlete is working.
Based on these findings and given the limitations of this 9. How many weeks should RST be applied? The effect
meta-analysis, we provide the following answers to the is significantly greater when the training is > 6 weeks.
questions posed by coaches and the scientific community 10. Can the surface affect the performance adaptations?
in Sect. 1: Yes, the effect is greater for rigid surfaces than for
grass, probably due to the lower friction that exists
1. Does RST improve sprint performance? Yes, but the with the same load on this surface.
improvements will depend on the training phase.
2. Will the effect be greater when RST is used for the dif-
ferent phases of sprint? Yes. For the early acceleration Compliance with Ethical Standards 
phase (≤ 10 m), there will be a reduction in the average
sprint time of 2.3%; for the maximum-velocity phase Funding  No sources of funding were used to assist in the preparation
of this article.
(≥ 15-m flying sprints at maximum intensity using a
run-in distance of ≥ 10 m), the time will be reduced
Conflict of interest  Pedro E. Alcaraz, Jorge Carlos-Vivas, Bruno O.
by 1.7%; and for the full sprint (≥ 20 m), the reduction Oponjuru, and Alejandro Martínez-Rodríguez have no conflicts of in-
will be 1.5%. The effect is moderate and significant terest relevant to the content of this review.
only for the early acceleration phase.
3. Is RST more effective than UR sprinting? No; no dif-
ference was observed between groups with sleds and
controls (same training without sled). Therefore, RST References
or UR will produce a similar level of adaptation.
4. What is the optimal load to use when applying RST for 1. Faude O, Koch T, Meyer T. Straight sprinting is the most fre-
obtaining higher sprint adaptations? There is no opti- quent action in goal situations in professional football. J Sports
Sci. 2012;30(7):625–31.
mal load for RST, since it will depend on the desired 2. Haugen TA, Tønnessen E, Seiler S. Speed and countermove-
objective. However, when training with sled towing as ment-jump characteristics of elite female soccer players, 1995–
a secondary method (i.e., replicating sprint demands in 2010. Int J Sports Physiol Perform. 2012;7(4):340–9.
terms of movement pattern, load, and movement veloc- 3. Haugen TA, Tønnessen E, Seiler S. Anaerobic performance
testing of professional soccer players 1995–2010. Int J Sports
ity) but with a slight overload, the load must never Physiol Perform. 2013;8(2):148–56.
be > 20% BM. Conversely, when the aim is to improve 4. Vescovi JD. Impact of maximum speed on sprint performance
maximum strength, “respecting” the movement pat- during high-level youth female field hockey matches: Female
tern, loads > 20% BM could be suitable, as long as the Athletes in Motion (FAiM) study. Int J Sports Physiol Perform.
2014;9(4):621–6.
athlete does not substantially modify his/her running 5. Mann RA, Hagy J. Biomechanics of walking, running, and
technique. Loads > 20% BM should not be used with sprinting. Am J Sports Med. 1980;8(5):345–50.
low-level or inexperienced athletes with sled towing. 6. Young W, Benton D, Duthie G, Pryor J. Resistance training
5. Should the load be different for the different phases for short sprints and maximum-speed sprints. Strength Cond J.
2001;23(2):7–13.
of sprint? Yes, keeping in mind that the main effect 7. Mero A, Komi P, Gregor R. Biomechanics of sprint running. A
of RST occurs in the early acceleration phase. For the review. Sports Med. 1992;13(6):376–92.
maximum-velocity phase, instead of using a different 8. Alexander RM. Mechanics of skeleton and tendons. Hand-
load, perhaps another sprint training method with a book of physiology—the nervous system. Am Physiol Soc.
1981;2:17–42.
more vertical resistance component, like weighted 9. Rabita G, Dorel S, Slawinski J, Saez-de-Villarreal E, Couturier
vests, could be used. A, Samozino P, et al. Sprint mechanics in world-class athletes: a
 P. E. Alcaraz et al.

new insight into the limits of human locomotion. Scand J Med Sci 29. Randell AD, Cronin JB, Keogh JWL, Gill ND. Transference of
Sports. 2015;25(5):583–94. strength and power adaptation to sports performance-horizontal
10. Mero A, Komi PV. Force-, EMG-, and elasticity-velocity rela- and vertical force production. Strength Cond J. 2010;32(4):100–6.
tionships at submaximal, maximal and supramaximal run- 30. Haff GG, Nimphius S. Training principles for power. Strength
ning speeds in sprinters. Eur J Appl Physiol Occup Physiol. Cond J. 2012;34(6):2–12.
1986;55(5):553–61. 31. Alcaraz PE, Palao JM, Elvira JLL, Linthorne NP. Effects of three
11. Delecluse C. Influence of strength training on sprint running per- types of resisted sprint training devices on the kinematics of sprint-
formance. Sports Med. 1997;24(3):147–56. ing at maximum velocity. J Strength Cond Res. 2008;22(3):890–7.
12. Weyand PG, Sandell RF, Prime DNL, Bundle MW. The biological 32. Kawamori N, Newton R, Nosaka K. Effects of weighted sled
limits to running speed are imposed from the ground up. J Appl towing on ground reaction force during the acceleration phase of
Physiol. 2010;108(4):950–61. sprint running. J Sports Sci. 2014;32(12):1139–45.
13. Weyand PG, Sternlight DB, Bellizzi MJ, Wright S. Faster top 33. Brown SC, Craik FI. Encoding and retrieval of information. In:
running speeds are achieved with greater ground forces not more Tulving E, Craik FIM, editors. The Oxford handbook of memory.
rapid leg movements. J Appl Physiol. 2000;89(5):1991–9. 2000. p. 93–107.
14. Clark KP, Weyand PG. Are running speeds maximized 34. Haff GG, Stone MH. Methods of developing power with special
with simple-spring stance mechanics? J Appl Physiol. reference to football players. Strength Cond J. 2015;37(6):2–16.
2014;117(6):604–15. 35. Alcaraz PE, Palao JM, Elvira JL. Determining the optimal load
15. DeWeese BH, Bellon C, Magrum E, Taber CB, Suchomel TJ. for resisted sprint training with sled towing. J Strength Cond Res.
Strengthening the springs: improving sprint performance via 2009;23(2):480–5.
strength training. Track Tech. 2016;9(3):8–20. 36. Jakalski K. The pros and cons of using resisted and assisted train-
16. Nagahara R, Mizutani M, Matsuo A, Kanehisa H, Fukunaga T. ing methods with high school sprinters: parachutes, tubing and
Step-to-step spatiotemporal variables and ground reaction forces towing. Track Coach. 1998;144:4585–9.
of intra-individual fastest sprinting in a single session. J Sports 37. Lockie RG, Murphy AJ, Spinks CD. Effects of resisted sled tow-
Sci. 2017;36(12):1392–401. ing on sprint kinematics in field-sport athletes. J Strength Cond
17. Nagahara R, Mizutani M, Matsuo A, Kanehisa H, Fukunaga T. Res. 2003;17(4):760–7.
Association of sprint performance with ground reaction forces 38. Cottle CA, Carlson LA, Lawrence MA. Effects of sled towing on
during acceleration and maximal speed phases in a single sprint. sprint starts. J Strength Cond Res. 2014;28(5):1241–5.
J Appl Biomech. 2017;34:104–10. 39. Kawamori N, Newton RU, Hori N, Nosaka K. Effects of weighted
18. Dalleau G, Belli A, Bourdin M, Lacour J-R. The spring-mass sled towing with heavy versus light load on sprint acceleration
model and the energy cost of treadmill running. Eur J Appl Phys- ability. J Strength Cond Res. 2014;28(10):2738–45.
iol Occup Physiol. 1998;77(3):257–63. 40. Monte A, Nardello F, Zamparo P. Sled towing: the optimal over-
19. Voigt M, Bojsen-Møller F, Simonsen EB, Dyhre-Poulsen P. The load for peak power production. Int J Sports Physiol Perform.
influence of tendon Youngs modulus, dimensions and instantane- 2016;12(8):1052–8.
ous moment arms on the efficiency of human movement. J Bio- 41. Cross MR, Brughelli M, Samozino P, Brown SR, Morin J-B. Opti-
mech. 1995;28(3):281–91. mal loading for maximising power during sled-resisted sprinting.
20. Plisk SS. Speed, agility, and speed-endurance development. In: Int J Sports Physiol Perform. 2017;12(8):1069–77.
Baechle TR, Earle RW, editors. National strength and conditioning 42. Samozino P, Rabita G, Dorel S, Slawinski J, Peyrot N, Saez de
association: essentials of strength training & conditioning. 2nd ed. Villarreal E, et al. A simple method for measuring power, force,
Champaign IL: Human Kinetics; 2000. p. 471–91. velocity properties, and mechanical effectiveness in sprint run-
21. MacDougall D, Sale D. Continuous vs. interval training: ning. Scand J Med Sci Sports. 2016;26(6):648–58.
a review for the athlete and the coach. Can J Appl Sport Sci. 43. Cronin JB, Hansen K, Kawamori N, McNair P. Effects of weighted
1981;6(2):93–7. vests and sled towing on sprint kinematics. Sports Biomech.
22. Iacono AD, Martone D, Milic M, Padulo J. Vertical-vs. hori- 2008;7(2):160–72.
zontal-oriented drop jump training: Chronic effects on explosive 44. Maulder PS, Bradshaw EJ, Keogh JWL. Kinematic altera-
performances of elite handball players. J Strength Cond Res. tions due to different loading schemes in early acceleration
2017;31(4):921–31. sprint performance from starting blocks. J Strength Cond Res.
23. Rumpf MC, Lockie RG, Cronin JB, Jalilvand F. Effect of different 2008;22(6):1992–2002.
sprint training methods on sprint performance over various dis- 45. Martinez-Valencia MA, Romero-Arenas S, Elvira JLL, Gon-
tances: a brief review. J Strength Cond Res. 2016;30(6):1767–85. zalez-Rave JM, Navarro-Valdivielso F, Alcaraz PE. Effects of
24. Freitas TT, Martinez-Rodriguez A, Calleja-González J, Alcaraz sled towing on peak force, the rate of force development and
PE. Short-term adaptations following complex training in team- sprint performance during the acceleration phase. J Hum Kinet.
sports: a meta-analysis. PLoS One. 2017;12(6):e0180223. 2015;46(1):139–48.
25. Seitz LB, Reyes A, Tran TT, de Villarreal ES, Haff GG. 46. Linthorne NP, Cooper JE. Effect of the coefficient of friction of
Increases in lower-body strength transfer positively to sprint a running surface on sprint time in a sled-towing exercise. Sports
performance: a systematic review with meta-analysis. Sports Biomech. 2013;12(2):175–85.
Med. 2014;44(12):1693–702. 47. Petrakos G, Morin JB, Egan B. Resisted sled sprint training to
26. Hill A. The heat of shortening and the dynamic constants of improve sprint performance: A systematic review. Sports Med.
muscle. Proc R Soc Lond B Biol Sci. 1938;126(843):136–95. 2016;46(3):381–400.
27. Hill A, Long C, Lupton H. The effect of fatigue on the relation 48. Moher D, Liberati A, Tetzlaff J, Altman DG, Prisma G. Preferred
between work and speed, in contraction of human arm muscles. reporting items for systematic reviews and meta-analyses: the
J Physiol. 1924;58(4–5):334–7. PRISMA statement. PLoS Med. 2009;6(7):e1000097.
28. Cormie P, McGuigan MR, Newton RU. Developing maxi- 49. Bhogal SK, Teasell RW, Foley NC, Speechley MR. The PEDro
mal neuromuscular power part 2-training considerations scale provides a more comprehensive measure of methodological
for improving maximal power production. Sports Med. quality than the Jadad scale in stroke rehabilitation literature. J
2011;41(2):125–46. Clin Epidemiol. 2005;58(7):668–73.
Effectiveness of RST for Sprint Performance

50. de Morton NA. The PEDro scale is a valid measure of the meth- 66. Morin JB, Petrakos G, Jimenez-Reyes P, Brown SR, Samozino
odological quality of clinical trials: a demographic study. Aust J P, Cross MR. Very-heavy sled training for improving horizon-
Physiother. 2009;55(2):129–33. tal force output in soccer players. Int J Sports Physiol Perform.
51. Maher CG, Sherrington C, Herbert RD, Moseley AM, Elkins M. 2016;12(6):840–4.
Reliability of the PEDro scale for rating quality of randomized 67. Ross A, Leveritt M, Riek S. Neural influences on sprint running.
controlled trials. Phys Ther. 2003;83(8):713–21. Sports Med. 2001;31(6):409–25.
52. Cohen J. Statistical power analysis for the behavioral sciences 68. Van Hooren B, Bosch F. Is there really an eccentric action of
(revised ed.). New York: Academic; 1977. the hamstrings during the swing phase of high-speed run-
53. Hopkins W, Marshall S, Batterham A, Hanin J. Progressive statis- ning? part I: a critical review of the literature. J Sports Sci.
tics for studies in sports medicine and exercise science. Med Sci 2017;35(23):2313–21.
Sports Exerc. 2009;41(1):3. 69. Van Hooren B, Bosch F. Is there really an eccentric action of the
54. Higgins JP, Green S. Cochrane handbook for systematic reviews hamstrings during the swing phase of high-speed running? Part
of interventions. New York: Wiley; 2011. II: implications for exercise. J Sports Sci. 2017;35(23):2322–33.
55. Zafeiridis A, Saraslanidis P, Manou V, Ioakimidis P, Dipla K, 70. Reeves ND, Narici MV. Behavior of human muscle fascicles
Kellis S. The effects of resisted sled-pulling sprint training on during shortening and lengthening contractions in vivo. J Appl
acceleration and maximum speed performance. J Sports Med Phys Physiol. 2003;95(3):1090–6.
Fit. 2005;45(3):284–90. 71. Ettema G. Mechanical efficiency and efficiency of storage and
56. Spinks CD, Murphy AJ, Spinks WL, Lockie RG. The effects release of series elastic energy in skeletal muscle during stretch-
of resisted sprint training on acceleration performance and kin- shorten cycles. J Exp Biol. 1996;199(9):1983–97.
ematics in soccer, rugby union, and Australian football players. J 72. Nagano A, Komura T, Fukashiro S. Effects of the length ratio
Strength Cond Res. 2007;21(1):77–85. between the contractile element and the series elastic element
57. Harrison AJ, Bourke G. The effect of resisted sprint training on on an explosive muscular performance. J Electromyogr Kinesiol.
speed and strength performance in male rugby players. J Strength 2004;14(2):197–203.
Cond Res. 2009;23(1):275–83. 73. Comfort P, Bullock N, Pearson SJ. A comparison of maximal
58. Clark KP, Stearne DJ, Walts CT, Miller AD. The longitudinal squat strength and 5-, 10-, and 20-meter sprint times, in ath-
effects of resisted sprint training using weighted sleds vs weighted letes and recreationally trained men. J Strength Cond Res.
vests. J Strength Cond Res. 2010;24(12):3287–95. 2012;26(4):937–40.
59. Lockie RG, Murphy AJ, Schultz AB, Knight TJ, de Jonge XAKJ. 74. McLean B. Biomechanics of running. In: Hawley JA, editor.
The effects of different speed training protocols on sprint accel- Handbook of sports medicine and science: running. John Wiley
eration kinematics and muscle strength and power in field sport & Sons; 2008. p. 28–43.
athletes. J Strength Cond Res. 2012;26(6):1539–50. 75. Schmidtbleicher D. Strength training: part 2: structural analysis of
60. Makaruk B, Sozański H, Makaruk H, Sacewicz T. The effects of motor strength qualities and its application to training. Sci Period
resisted sprint training on speed performance in women. Hum Res Tech Sport. 1985;4:1–10.
Mov Sci. 2013;14(2):116–22. 76. Schmidtbleicher D. Strength training: part 1: structural analysis of
61. West DJ, Cunningham DJ, Bracken RM, Bevan HR, Crewther motor strength qualities and its application to training. Sci Period
BT, Cook CJ, et  al. Effects of resisted training on accelera- Res Tech Sport. 1985;4:1–12.
tion in professional rugby union players. J Strength Cond Res. 77. Siff M, Verkhoshansky Y. Supertraining. Denver: Supertraining
2013;27(4):1014–8. Institute; 2003.
62. Alcaraz PE, Elvira JLL, Palao JM. Kinematic, strength, and stiff- 78. Farley CT, Gonzalez O. Leg stiffness and stride frequency in
ness adaptations after a short-term sled towing training in athletes. human running. J Biomech. 1996;29(2):181–6.
Scand J Med Sci Sports. 2014;24(2):279–90. 79. Brughelli M, Cronin J. A review of research on the mechanical
63. Bachero-Mena B, Gonzalez-Badillo JJ. Effects of resisted sprint stiffness in running and jumping: methodology and implications.
training on acceleration with three different loads account- Scand J Med Sci Sports. 2008;18(4):417–26.
ing for 5, 12.5, and 20% of body mass. J Strength Cond Res. 80. Kraemer WJ, Adams K, Cafarelli E, Dudley GA, Dooly C, Feigen-
2014;28(10):2954–60. baum MS, American College of Sports Medicine position stand,
64. Luteberget LS, Raastad T, Seynnes O, Spencer M. Effect of tra- et al. Progression models in resistance training for healthy adults.
ditional and resisted sprint training in highly trained female team Med Sci Sports Exerc. 2002;34(2):364–80.
handball players. Int J Sports Physiol Perform. 2015;10(5):642–7. 81. Barnes C, Archer D, Hogg B, Bush M, Bradley P. The evolution
65. de Hoyo M, Gonzalo-Skok O, Sañudo B, Carrascal C, Plaza- of physical and technical performance parameters in the English
Armas JR, Camacho-Candil F, et al. Comparative effects of in- Premier League. Int J Sports Med. 2014;35(13):1095–100.
season full-back squat, resisted sprint training, and plyometric 82. Bolger R, Lyons M, Harrison AJ, Kenny IC. Sprinting perfor-
training on explosive performance in U-19 elite soccer players. J mance and resistance-based training interventions: a systematic
Strength Cond Res. 2016;30(2):368–77. review. J Strength Cond Res. 2015;29(4):1146–56.

Affiliations

Pedro E. Alcaraz1,2 · Jorge Carlos‑Vivas1 · Bruno O. Oponjuru1 · Alejandro Martínez‑Rodríguez3

1 3
UCAM Research Center for High Performance Sport, Department of Analytical Chemistry, Nutrition and Food
Catholic University of Murcia, Murcia, Spain Sciences, Faculty of Sciences, University of Alicante,
2 Alicante, Spain
Faculty of Sport Sciences, UCAM, Catholic University
of Murcia, Murcia, Spain

View publication stats