Review

Br J Sports Med: first published as 10.1136/bjsports-2020-103683 on 17 May 2021. Downloaded from http://bjsm.bmj.com/ on December 9, 2021 by guest. Protected by copyright.
Do exercise-­based prevention programmes reduce
non-­contact musculoskeletal injuries in football
(soccer)? A systematic review and meta-­analysis with
13 355 athletes and more than 1 million
exposure hours
Italo Ribeiro Lemes  ‍ ‍,1 Rafael Zambelli Pinto  ‍ ‍,1,2 Vitor N Lage,2
Bárbara A B Roch,2 Evert Verhagen  ‍ ‍,3 Caroline Bolling,3 Cecilia Ferreira Aquino,4,5
Sérgio T Fonseca  ‍ ‍,1,2 Thales R Souza  ‍ ‍1,2

►► Additional supplemental ABSTRACT rate among professional and amateur players may
material is published online Objective  The aim of this systematic review was to vary. Previous research has shown that injury rates
only. To view, please visit the
journal online (http://d​ x.​doi.​investigate the effect of exercise-­based programmes in among amateurs and professionals are 9.6 and 8.1
org/1​ 0.​1136/​bjsports-​2020-​ the prevention of non-­contact musculoskeletal injuries per 1000 hours of exposure, respectively,7 8 and up
103683). among football players in comparison to a control group. to half of these injuries are muscle injuries.9
1
Design  Systematic review and meta-­analysis of Although football is a sport with frequent physical
Graduate Program in
randomised controlled trials. contact, occurrence of non-­ contact musculoskel-
Rehabilitation Sciences,
Universidade Federal de Minas Data sources  MEDLINE, EMBASE, CENTRAL, CINAHL, etal injuries, such as hamstring strain and anterior
Gerais (UFMG), Belo Horizonte, PEDro and SPORTDiscus databases were searched from cruciate ligament (ACL) rupture, is common.10
2
MG, Brazil the earliest record to January 2021. There is evidence showing that more than 90% of
Department of Physical Eligibility criteria  Studies were eligible if they (1)
Therapy, Universidade Federal
all muscle injuries and 51%–64% of joint/ligament
de Minas Gerais (UFMG), Belo included football players aged 13 years or older, (2) injuries (ie, ACL) in football occur in non-­contact
Horizonte, MG, Brazil used exercise-­based programmes as intervention, (3) situations.11–13 Lower limb injuries are the most
3
Amsterdam Collaboration on presented the number of non-­contact musculoskeletal common type of injury in football and most of
Health and Safety in Sports injuries (ie, defined as any acute sudden onset these (66%) are non-­contact injuries and, therefore,
& Department of Public and
musculoskeletal injury that occurred without physical preventable.14 Recovery from non-­ contact injury
Occupational Health, VU
University Medical Center, contact) and exposure hours for each group, and (4) had depends on the type and severity of injury but often
Amsterdam, The Netherlands a control group (eg, usual training, minimal intervention, requires athletes to take time off from sports. For
4
Department of Physical education). All types of exercise-­based prevention instance, an acute hamstring strain (grade I or II)
Therapy, Universidade José do programmes were eligible for inclusion. Risk of bias for
Rosário Vellano, Divinópolis, may take up to 8 weeks of rehabilitation,15 while
MG, Brazil each included study and overall quality of evidence for time loss following an ACL injury is ~7.5 months.12
5
Department of Physical the meta-­analysis were assessed. Sport injuries can affect the athlete’s physical
Therapy, Universidade do Estado Results  Ten original randomised controlled trials
and mental health and, consequently, impact team’s
de Minas Gerais (UEMG), with 13 355 football players and 1 062 711 hours
Divinópolis, MG, Brazil performance.16–19 Professional teams lose approx-
of exposure were selected. Pooled injury risk ratio imately £45 million (~US$55 million) per season
showed very low-­quality evidence that exercise-­based due to injuries from 2012 to 2017.20 To reduce
Correspondence to
Thales R Souza, Department of prevention programmes reduced the risk of non-­contact absence from training and competition, interven-
Physical Therapy, Universidade musculoskeletal injuries by 23% (0.77 (95% CI 0.61 to
tions to reduce injuries among football players have
Federal de Minas Gerais 0.97)) compared with a control group.
(UFMG), Belo Horizonte 31270- Conclusion  Exercise-­based prevention programmes been developed.21 22 One of the available interven-
901, MG, Brazil; tion programmes for football players is the general
may reduce the risk of non-­contact musculoskeletal
t​ halesrs@​ufmg.​br (ie, targeting multiple body parts) exercise-­ based
injuries by 23% among football players. Future high-­
programme FIFA 11+.23 This general programme
Accepted 4 May 2021 quality trials are still needed to clarify the role of
has been shown to reduce overall injury rate by
Published Online First exercise-­based programmes in preventing non-­contact
17 May 2021 39%.22 Focused programmes, comprised of exer-
musculoskeletal injuries among football players.
cises targeting a specific muscle group (eg, the
PROSPERO registration number  CRD42020173017.
Nordic Hamstring Exercise (NHE)), may provide
extra protection for the targeted tissue21 as it is
shown to reduce the risk of hamstring injury by
© Author(s) (or their
INTRODUCTION 46% among amateur football players.24 Despite the
employer(s)) 2021. No
commercial re-­use. See rights Football (soccer) is one of the most popular sports burden of non-­contact musculoskeletal injuries to
and permissions. Published worldwide. It is estimated that more than 250 million athletes and their teams, the efficacy of focused and
by BMJ. men and women, from children to elderly, with general exercise-­based programmes to prevent non-­
To cite: Lemes IR, different socioeconomic backgrounds and levels of contact musculoskeletal injuries and the overall
Pinto RZ, Lage VN, expertise, play football.1 2 Besides the well-­known incidence rate of such injuries remains unclear.
et al. Br J Sports Med health benefits of sports participation,3–6 there is Previous systematic reviews investigating exercise-­
2021;55:1170–1178. an increased risk of musculoskeletal injuries. Injury based programmes to prevent injuries in football

Lemes IR, et al. Br J Sports Med 2021;55:1170–1178. doi:10.1136/bjsports-2020-103683    1 of 10

 Review

Br J Sports Med: first published as 10.1136/bjsports-2020-103683 on 17 May 2021. Downloaded from http://bjsm.bmj.com/ on December 9, 2021 by guest. Protected by copyright.
players combined data from all types of injuries and/or inter- confirm or clarify the number of non-­contact injuries. A total of
ventions.22 25–28 Given that the injury profile and incidence vary three attempts to contact the authors were made. If missing or
across teams and leagues,29 30 it would be informative to estimate unclear data could not be provided by the authors of included
the efficacy of exercise-­based programmes, including focused studies, they were excluded.
and general exercise programmes, specifically for preventing
non-­contact musculoskeletal injuries. In addition, informing
Study selection and data extraction
injury incidence of non-­contact injuries would contribute to the
Two independent reviewers (VNL and BABR) applied the inclu-
development of future interventions. This information may be
sion criteria and screened all titles and abstracts. Full texts
incorporated in pre-­season and during-­season training plans.
were evaluated for potential inclusion and disagreements were
To the best of our knowledge, there is no systematic review
resolved by consensus. If consensus was not reached, a third
assessing the efficacy of exercise-­ based programmes in the
reviewer (RZP) was consulted.
prevention of non-­contact musculoskeletal injuries among foot-
Two independent reviewers (VNL and IRL) performed the
ball players. The primary aim of this systematic review and meta-­
data extraction of included studies using a standardised data
analysis was to evaluate the effects of exercise-­based programmes
extraction form. In case of disagreement, a third reviewer (RZP)
in the prevention of non-­contact musculoskeletal injuries among
arbitrated the decision. We extracted the following information
football players in comparison with a control group. In addi-
from each eligible study: country, participants’ characteristics
tion, we aimed to (1) investigate whether there are differences
(ie, age, sex and skill level), sample size (total and per group),
in the estimates between focused and general exercises on the
characteristics of the intervention (focused or general exer-
prevention of specific non-­contact injuries, and (2) report the
cises), number of non-­contact injuries and exposure hours for
injury incidence of non-­contact injuries using data from control
each group, and study length. Interventions were categorised as
groups.
general when the exercises targeted many body segments and
joints, with no priority to train or protect a specific muscle group
METHODS or joint. Focused interventions were defined as a set of exer-
The Preferred Reporting Items for Systematic Reviews and cises chosen to train and protect a specific muscle or joint. When
Meta-­Analysis (PRISMA) guidelines were followed for reporting exposure information was reported as ‘athletic exposure’ (ie,
of this review.31 The protocol for this review was prospectively one athlete participating in one training or game), we converted
registered in the PROSPERO database (CRD42020173017).32 the data using the assumption that one athletic exposure was
equivalent to 2 hours of exposure.28
Search strategy and inclusion criteria
The search for relevant studies was performed in six data- Quality assessment
bases (MEDLINE, EMBASE, CENTRAL, CINAHL, PEDro The PEDro scale was used to assess the risk of bias of the
and SPORTDiscus) from the earliest record to 10 March 2020 included clinical trials.34 35 All studies included in this review
and updated on 14 January 2021. Clinical trial registries, such were listed in the PEDro database36; therefore, these assessments
as ​clinicaltrials.​
gov, the International Standard Randomised were adopted.35 Given the nature of the intervention assessed in
Controlled Trial Number Register (IRSCTN) and the Austra- this review, we adapted the scale and did not consider the items
lian–New Zealand Clinical Trials Registry (ANZCTR), were also participant and therapist blinding, as it is not possible to blind
searched for potential ongoing or unpublished trials. The search participant and therapists in trials testing efficacy of exercise
strategy used a combination of terms related to football, soccer, prevention programmes. Risk of bias of trials were confirmed by
prevention and randomised controlled trial (online supplemental a reviewer (RZP) with 5 years of experience in rating trials using
table 1). The reference list of previous systematic reviews in the the PEDro scale.
topic was checked to find potential studies that could also be The Grading of Recommendations Assessment, Develop-
used in this review. There was no restriction regarding language ment, and Evaluation (GRADE) approach was used to assess the
of publication. overall quality of the evidence.37 38 The GRADE tool is a system-
Only randomised controlled trials investigating the efficacy atic and explicit approach that allows judgements to be made
of exercise-­based programmes compared with a control group about strength of evidence resulting from systematic reviews.
for preventing non-­contact musculoskeletal injuries in football Briefly, the GRADE classification was initially regarded as ‘high’
players were included. An exercise-­based prevention programme but downgraded by one level for each of the following domains
was defined as any exercise therapy that was performed in order we considered: (1) risk of bias (downgraded by one level when
to develop or improve function, skills or physical fitness. Control more than 25% of participants included in the meta-­analysis
group was defined as usual training/warm-­up, minimal interven- were from studies with ‘high risk of bias’ (ie, when the study did
tion, education or not exposed to the intervention. Non-­contact not meet one or more of the following items: random allocation,
musculoskeletal injuries were the primary outcome in this review allocation concealment, assessor blinding, complete outcome
and was defined as any acute sudden onset musculoskeletal follow-­up data and intention-­ to-­
treat analysis)); (2) inconsis-
injury that occurred without physical contact by another player tency (downgraded by one level considering: the proportion
or object on the field.33 Studies were eligible if they (1) included of the observed variance may be substantial (I2>50%), visual
football players aged 13 years or older, (2) used exercise-­ inspection for minimal or no overlap of CIs, and χ2 test (p value
based programmes as intervention, (3) presented number of <0.05); (3) indirectness (downgraded by one level if meta-­
non-­contact injuries and exposure hours for each group, and analysis included participants with heterogeneous characteris-
(4) had a control group. All types of exercise-­based preven- tics with regard to sex, age and level of sport (eg, men, women,
tion programmes were eligible for inclusion. Studies reporting youth, adults, amateur, professionals)); (4) imprecision (down-
overall injuries were included if it was possible to extract the graded by one level when the clinical course of action differed
data specific to non-­contact injuries. When information about considering the upper and lower CI as the true estimate, or
the injuries was unclear, we contacted the authors via email to the difference between upper and lower CI around the pooled

2 of 10 Lemes IR, et al. Br J Sports Med 2021;55:1170–1178. doi:10.1136/bjsports-2020-103683

 Review

Br J Sports Med: first published as 10.1136/bjsports-2020-103683 on 17 May 2021. Downloaded from http://bjsm.bmj.com/ on December 9, 2021 by guest. Protected by copyright.
estimate of the injury risk ratio (IRR) was >0.5)28; and (5) publi- Deviation from protocol
cation bias (assessed funnel plot asymmetry by visual inspection Post hoc sensitivity analyses were performed to investigate the
and quantified using the Egger test (p value <0.1), if there were effect of exercise-­based programmes in the prevention of non-­
at least 10 studies in the meta-­analysis). contact musculoskeletal injuries when (1) including only male
The following categories were used to define the quality of or female participants, and (2) including only youth or adult
evidence: high quality (ie, further research is unlikely to change participants. In addition, we decided to replace the terms
our confidence in the estimate); moderate quality (ie, further unimodal and multimodal exercise programmes described in
research is likely to have an important impact on our confidence the registered protocol with the terms focused and general exer-
in the estimate and might change the estimate); low quality (ie, cise programmes. Although the most appropriate terms to refer
further research is likely to have an important impact on our to these types of exercises may be open to question, the terms
confidence in the estimate and is likely to change the estimate); focused and general exercise programmes were considered more
and very low quality (ie, we are uncertain about the estimate).37 38 suitable for this review.

Statistical analysis
Number of non-­contact injuries and exposure hours were used RESULTS
to calculate the injury incidence rate per 1000 hours and the Included studies
IRR. The inverse variance random-­effect model was used to The search strategy identified 7512 studies. After removal of
compute the pooled IRR and 95% CIs. As previously planned duplicates, 4366 studies remained. Title and abstract screening
in our protocol, the available evidence allowed us to perform a identified 53 potentially eligible studies, and 10 original
subgroup analysis stratifying by intervention specificity (general randomised controlled trials met the criteria to be included in
vs focused exercises). Given that focused interventions targeted this review (figure 1). In this review, a total of three studies were
the hamstring muscles, subgroup analysis was restricted to non-­ excluded due to lack of clear information.
contact hamstring injury. Heterogeneity between studies (ie, The included studies were conducted in the USA,33 40
how much the effect size varies across studies) was estimated Norway,23 41 the Netherlands,24 42 Germany,43 Japan,44 Nigeria45
based on the proportion of the variation in point estimates and Sweden.46 Six studies included only male football
due to between-­study differences (I2). Number needed to treat players24 40 42–45 and four included only female football
(NNT) analysis was performed by a mathematical formula.39 players,23 33 41 46 and the total sample sizes for intervention and
Meta-­analyses were calculated using RevMan software, V.5.3. control groups were 6900 and 6455, respectively. One study

Figure 1  Flow chart of the included studies.

Lemes IR, et al. Br J Sports Med 2021;55:1170–1178. doi:10.1136/bjsports-2020-103683 3 of 10

 Review

Br J Sports Med: first published as 10.1136/bjsports-2020-103683 on 17 May 2021. Downloaded from http://bjsm.bmj.com/ on December 9, 2021 by guest. Protected by copyright.
Table 1  Characteristics of the included studies
Participants
Study, year Country (all football players) Sample size Intervention Study length Outcome
33
Gilchrist, 2008 USA Female collegiate IG: 583 General 12 weeks Non-­contact ACL
Age CG: 852 Prevent Injury and injuries
IG: 19.9 Enhance Performance
CG: 19.9 (PEP) Programme
3 times/week
Hammes et al, 201543 Germany Male veteran (≥32 IG: 146 General 9 months Overall injuries
years) CG: 119 FIFA 11+
Age Every training session
IG: 45.2±7.7
CG: 43.1±6.5
Hasebe et al, 202044 Japan Male high school IG: 156 Focused 27 weeks Overall hamstring
Age CG: 103 Nordic Hamstring injuries
IG: 16.7±0.5 Exercise
CG: 16.3±0.6 After training session
2 times/week
Owoeye et al, 201445 Nigeria Male youth IG: 212 General 6 months Overall injuries
Age CG: 204 FIFA 11+
IG: 17.8±0.9 2 times/week
CG: 17.5±1.1
Silvers-G
­ ranelli et al, USA Male collegiate IG: 675 General 5 months Overall ACL injuries
201740 Age CG: 850 FIFA 11+
IG: 20.0±2.0 2–3 times/week
CG: 21.0±1.0
Soligard et al, 200823 Norway Female youth IG: 1055 General 8 months Overall lower limb
Age CG: 837 FIFA 11+ injuries
IG: 15.4±0.7 2 times/week
CG: 15.4±0.7
Steffen et al, 200841 Norway Female youth IG: 1073 General 8 months (including pre-­season Overall injuries
Age CG: 947 FIFA 11 and summer break)
IG: 15.4±0.8 Every training session for
CG: 15.4±0.8 15 consecutive sessions,
then 1 time/week for the
rest of the season
van de Hoef et al, 201842 The Netherlands Male amateur IG: 229 Focused 39 weeks Overall hamstring
Age CG: 171 Bounding Exercise injuries
IG: 23.8±6.4 Programme (BEP)
CG: 22.2±3.1 Every training session
van der Horst et al, The Netherlands Male amateur IG: 292 Focused 13 weeks Overall hamstring
201560 Age CG: 287 Nordic Hamstring injuries
IG: 24.5±3.6 Exercise
CG: 24.6±4.1 After training session
2 times/week
Waldén et al, 201246 Sweden Female adolescents IG: 2479 General 7 months Overall knee injuries
Age CG: 2085 Neuromuscular training
IG: 14.0±1.2 (Knäkontroll)
CG: 14.1±1.2 2 times/week
CG, control group; IG, intervention group.

used veteran players,43 two used adult players24 42 and seven used Programme42 and one study used a neuromuscular training
youth players (eg, high school or collegiate).23 33 40 41 44–46 All programme (Knäkontroll).46 Focused prevention programmes
studies included amateur players. Three studies used a focused consisted of exercises for the quadriceps or hamstring muscles,
exercise programme,24 42 44 while seven used a general interven- while general programmes involved agility, balance, mobility,
tion.23 33 40 41 43 45 46 The intervention period ranged from 12 plyometrics, running and strength exercises for the lower
weeks to 9 months (table 1). limb. All interventions were applied at least twice a week to
Three ongoing trials were identified in clinical registries: one
every training session. A total of 545 non-­contact injuries and
from Sweden and two from Saudi Arabia. The status for all regis-
1 062 711 hours of exposure were computed.
tered trials is ‘Not yet recruiting’ (online supplemental table 2).
Outcomes of the included studies were overall injuries,41 43 45
overall lower limb injuries,23 overall hamstring injuries,24 42 44
Study characteristics
Four studies used the FIFA 11+ warm up and strengthening overall knee injuries,46 overall ACL injuries40 and non-­contact
programme,23 40 43 45 two studies used the Nordic Hamstring ACL injuries.33 Regarding comparators, control groups were
Exercise,24 44 one study used the FIFA 11 programme,41 instructed to perform their usual warm-­up exercises23 33 41 and/or
one study used the Prevention Injury and Enhance Perfor- training routines.24 42 43 45 46 Two studies did not provide infor-
mance programme,33 one study used the Bounding Exercise mation on control groups.40 44

4 of 10 Lemes IR, et al. Br J Sports Med 2021;55:1170–1178. doi:10.1136/bjsports-2020-103683

 Review

Br J Sports Med: first published as 10.1136/bjsports-2020-103683 on 17 May 2021. Downloaded from http://bjsm.bmj.com/ on December 9, 2021 by guest. Protected by copyright.
Figure 2  Meta-­analysis investigating the effect of exercise-­based prevention programmes compared with a control group in reducing (A) overall
non-­contact injuries and (B) non-­contact hamstring injuries in football players. The size of the blue boxes is proportional to the weight of each study in
the analysis.

Quality assessment by 35% (IRR 0.65; 95% CI 0.44 to 0.97; n=1238; I2=0%)
Overall risk of bias of individual included studies is shown in compared with the control group. There is very low-­ quality
online supplemental table 3. All studies had random allocation, evidence (ie, downgraded one level due risk of bias, one level
30% had concealed allocation, 40% blinded the assessors, 50% due to inconsistency, one level due to indirectness and one level
had complete outcome follow-­up data and 70% had intention-­ due to imprecision) that general programmes were not more
to-­treat analysis. Due to the nature of the interventions, none of effective than control group (IRR 0.63; 95% CI 0.19 to 2.12;
the included studies were blinded for participants and therapists. n=2573; I2=51%). Focused programmes were not different
from general interventions for the prevention of non-­contact
Effect of exercise-based prevention programmes on non- hamstring injuries (p value=0.95) (figure 2B). Detailed infor-
contact musculoskeletal injuries mation regarding the quality of evidence (GRADE) is shown
The results of meta-­analysis pooling data from 10 trials showed in table 2. The NNT value was 148 for overall intervention
very low-­quality evidence (ie, downgraded one level due to risk programmes and 31 for focused programmes. The NNT values
of bias, one level due to indirectness and one level due to publi- for men and women were 118 and 181, respectively, for overall
cation bias (online supplemental figure 1) that exercise-­based intervention programmes.
prevention programmes reduce, on average, the risk of non-­
contact musculoskeletal injuries by 23% (IRR 0.77; 95% CI 0.61 Sensitivity analysis
to 0.97; n=13 355; I2=30%) compared with the control group The pooled effect restricted to studies with male athletes showed
(figure 2A). an IRR of 0.68 ((95% CI 0.48 to 0.96); n=3444; I2=20%)
There is low-­quality evidence (ie, downgraded one level due favouring exercise-­based prevention programmes over control,
to risk of bias and one level due to imprecision) that focused while for studies with female athletes the IRR was 0.85 ((95%
programmes reduce the risk of non-­contact hamstring injuries CI 0.63 to 1.16); n=9911; I2=38%) (online supplemental figure

Lemes IR, et al. Br J Sports Med 2021;55:1170–1178. doi:10.1136/bjsports-2020-103683 5 of 10

 Review

Br J Sports Med: first published as 10.1136/bjsports-2020-103683 on 17 May 2021. Downloaded from http://bjsm.bmj.com/ on December 9, 2021 by guest. Protected by copyright.
Table 2  Summary of findings and quality of evidence (GRADE)
Quality assessment Participants, n Effect
Risk of Publication GRADE
Meta-­analysis bias* Inconsistency† Indirectness‡ Imprecision§ bias¶ IG CG IRR (95% CI) Quality
Exercise-­based prevention ◯ ⨁ ◯ ⨁ ◯ 6900 6455 0.77 (0.61 to 0.97) Very low
programmes 10 studies ⨁⨁◯◯◯
Focused programmes for non-­ ◯ ⨁ ⨁ ◯ – 677 561 0.65 (0.44 to 0.97) Low
contact hamstring injuries ⨁⨁◯◯
three studies
General programmes for non-­ ◯ ◯ ◯ ◯ – 1413 1160 0.63 (0.19 to 2.12) Very low
contact hamstring injuries ◯◯◯◯
three studies
*More than 25% of participants from studies with ‘high risk of bias’.
†Downgraded by one level considering: the proportion of the observed variance may be substantial (I2 >50%), visual inspection for minimal or no overlap of CIs, and χ2 test.
‡Based on the characteristics of participants included in the meta-­analysis.
§Downgraded if the upper and lower CIs had >0.5 difference; or if the clinical course of action differed considering the upper and lower CI as the true estimate.
¶Assessed with visual inspection of the funnel plot and two-­tailed Egger test (if >10 studies were included in the meta-­analysis).
CG, control group; IG, intervention group; IRR, injury risk ratio.

2). The IRR for studies with youth and adult participants were considered. Although most of the included studies (80%, n=8)
0.75 ((95% CI 0.54 to 1.02); n=12 111; I2=36%) and 0.77 used the consensus statement proposed by Fuller et al47 to
((95% CI 0.51 to 1.16); n=1244; I2=39%), respectively (online define injuries, 60% (n=6) focused on specific type of injuries
supplemental figure 3). (hamstring, knee or ACL) and did not report other non-­contact
injuries. This may have influenced the pooled estimated found
Incidence of non-contact injuries in this review. Future studies should investigate and report data
The incidence rate of non-­contact injuries in this review (for for all non-­contact injuries to allow estimates to be calculated
control groups only) was 0.54 (95% CI 0.47 to 0.60) per for overall and specific types of non-­contact injuries. We only
1000 hours of exposure (table 3). When considering studies with included lower limb non-­contact injuries, and most of partic-
overall non-­contact injuries as outcome,41 43 45 the injury inci- ipants were women and young amateur athletes, which limits
dence rate was 0.96 (95% CI 0.79 to 1.13) per 1000 hours of the generalisability of our findings. Nevertheless, the majority
exposure. The average incidence rate of non-­contact injuries for of injuries in football affect the lower extremities, and there are
male (age range 16.3–43.1 years)24 40 42–45 and female (age range more amateur than professional players worldwide (265 million
14.1–19.9)23 33 41 46 participants were 0.57 (95% CI 0.46 to 0.68) vs 200 000).48 Finally, we used visual inspection and statistical
and 0.52 (95% CI 0.44 to 0.59) per 1000 hours of exposure, test for asymmetry of funnel plots to assess publication bias.
respectively. Among youth and adult participants, the incidence These approaches are prone to error and their results should be
rates were 0.43 (95% CI 0.37 to 0.49) and 1.49 (95% CI 1.16 interpreted with caution.
to 1.82) per 1000 hours of exposure, respectively. The average A limitation of the available evidence is the high risk of bias
incidence rates of non-­contact injuries to the hamstring23 24 42–45 (4 of the 10 studies) and small number of studies included in
and ACL40 41 were 0.41 (95% CI 0.32 to 0.50) and 0.87 (95% CI the meta-­analysis. Despite the advance in football prevention
0.70 to 1.05) per 1000 hours of exposure, respectively. Detailed strategies, there is still a lack of randomised trials investigating
information is shown in table 3. non-­contact musculoskeletal injuries, especially among high-­risk
populations (eg, elite youth female players).49 Ninety per cent
DISCUSSION (n=9) of the included studies did not report non-­contact injuries
Statement of principal findings in training and matches separately; therefore, we were not able
This systematic review shows that exercise-­ based prevention to report injury incidence separately for training and matches.
programmes may be effective in reducing the risk of non-­ Finally, our search strategy identified three ongoing trials that
contact musculoskeletal injuries by 23% (95% CI 3% to 39%) might change our current estimates and should be considered
among football players when compared with a control group. for future reviews.
This result is based on very low-­ quality evidence and more
high-­quality studies in this area are needed to clarify the role of Comparison with previous reviews on injury risk reduction
exercise-­based programmes in preventing non-­contact musculo- Our results showed an estimated risk reduction of 23% in the
skeletal injuries. Hamstring-­focused programmes did not reduce prevention of non-­ contact musculoskeletal injuries favouring
hamstring injury any more than general programmes. The inci- exercise-­based programmes over control, but the true effect
dence rate of overall non-­contact injuries, for control groups could vary largely from 3% to 39%. This estimate is lower than
only, was 0.96 per 1000 hours of exposure. a previous systematic review showing a 37% risk reduction for
exercise-­based programmes on prevention of sport injuries.50
Strengths and limitations of the review and the available The diversity of interventions, sports and participation level may
evidence help to explain the higher heterogeneity (I2=70%; p<0.001)
The strengths of this systematic review include the use of a of this previous review in comparison to our results. In foot-
prespecified protocol with no language and date restriction ball, specifically, there is evidence that exercise-­based prevention
criteria, the inclusion of only randomised controlled trials, the programmes reduced overall injuries by 22% and 25% among
assessment of risk of bias and overall quality of the evidence. women and recreational/subelite athletes, respectively.22 28 These
On the other hand, this review has limitations that should be results combined evidence from overall injuries and different

6 of 10 Lemes IR, et al. Br J Sports Med 2021;55:1170–1178. doi:10.1136/bjsports-2020-103683

 Review

Br J Sports Med: first published as 10.1136/bjsports-2020-103683 on 17 May 2021. Downloaded from http://bjsm.bmj.com/ on December 9, 2021 by guest. Protected by copyright.
interventions, which might be related to the significant hetero-

Injuries per 1000 hours

geneity (I2=60.3% and 68.9%).22 28 Our review presents low
heterogeneity probably due to inclusion of specific football inju-
ries (ie, non-­contact). Our results could be interpreted as prom-
ising, but future high-­quality studies are needed to determine
whether the results are clinically relevant to be applied in the

0.537
0.094

0.104
0.262
0.226

1.277
1.308
1.381
0.902
0.194
7.831
sport settings. For instance, the NNT analysis indicated that at
2937
least 148 football athletes are needed to be exposed to interven-
105 838

28 910
61 045
44 212

45 428
65 725
21 717
27 724

532 620
129 647
tion to prevent one non-­contact musculoskeletal injury, which is
Total

considerably higher than previously pooled NNT estimates on

non-­contact injury (NNT=89).51
1073

3597

6276
23 686

13 624

14 342
19 856

10 074
34 402
126 930
Interestingly, our sensitivity results for gender showed that
Match
Exposure hours

exercise-­ based prevention programmes were not more effec-

NR

tive to prevent non-­contact injuries compared with control in

Training

1864
60 210

57 448
30 588

31 086
45 869
15 441
17 650
94 659
354 815
female athletes. The four trials included in the sensitivity analysis
were conducted with female amateur athletes and three of these
NR

trials were considered to be of high risk of bias (ie, no concealed

allocation, no assessor blinding, no adequate follow-­up, or no
Non-­contact injuries

intention-­to-­treat analysis). Our results contrast with a previous

review showing that exercise-­ based prevention programmes
reduce overall injuries by 22% in female football players.28 In
addition, there is evidence that a subgroup of female athletes,
†Total exposure hour includes activities performed outside training/matches; therefore, exposure hours in training and matches will not sum to total exposure hours.

that is, elite youth female athletes, are at a higher risk of injury.49
286
10
23
3
16
10

58
86
30
25
25

More high-­quality studies are needed, especially with high-­risk

subgroups (eg, youth elite female athletes), to strengthen our
Participants

findings and to provide information on how health and football

Control

policy makers can adopt and optimise exercise-­based preven-

6455
852
119
103
204
850

837
947
171
287
2085

tion programmes across all ages and skill levels (ie, amateurs vs
professionals and youth vs adult players).
Injuries per 1000 hours

The pooled effect calculated in the sensitivity analyses showed

considerable overlap of 95% CI, which means that the available
studies do not clearly show differences according to sex and age.
0.488
0.057
8.150
0.044
0.137
0.057

1.102
1.400
1.097
0.416
0.121

General versus focused exercise programmes

Our subgroup analysis did not reveal significant difference
between focused and general interventions (substantial overlap
4172
70 440

45 374
51 017
35 226

49 899
66 423
31 900
26 426

530 091
149 214
Total

between both estimates). Our analysis for the subgroup of focused

intervention included three studies focusing on hamstring exer-
‡Data reported on training and matches hours of exposure do not sum to total exposure hours.

cises (ie, Nordic Hamstring Exercise and Bounding Exercise

1238

3383

9664
9928
16 336

10 935

16 057
20 731

40 160
128 432
Match

Programme)24 42 44 and showed low-­quality evidence that the

Exposure hours

NR

focused interventions may reduce the risk of non-­contact inju-

ries, in this case hamstring injuries, by 35% (95% CI 3% to 56%)
Training

2934
36 740

47 634

33 842
45 692
22 236
16 498

338 943
109 076
24 91

when compared with the control group. A previous systematic

*Calculated from ‘athletic exposure’ (one athletic exposure=2 exposure hours).
NR

review has shown that injury prevention programmes that include

Nordic Hamstring Exercise decreased the risk of hamstring inju-
Table 3  Injury and exposure data of the included studies

Non-­contact injuries

ries by 51% when compared with the control group.21 This esti-
mate is higher than the estimate found in our review and might
be due to differences in study designs, mechanism of injury and
participants. We included only randomised controlled trials, non-­
contact injuries and both sexes, while previous reviews included
259
2
34
2
7
2

55
93
35
11
18

both randomised and prospective cohort studies,21 overall inju-

ries (ie, contact and non-­contact)21 28 and only women.28 The
Intervention

Participants

pooled NNT estimates for focused programmes showed that 31

athletes would need to participate in the intervention to prevent
6900
583
146
156
212
675

1055
1073
229
292
2479

one non-­contact hamstring injury, which is slightly higher than

previous individual evidence of specific exercise (ie, eccentric
van der Horst et al, 201560

training) on hamstring injuries (NNT=13).52

van de Hoef et al, 201942
Gilchrist et al, 200833*†

Waldén et al, 201246‡

Hammes et al, 201543

Soligard et al, 200823

Owoeye et al, 201445

Given that NNT is highly dependent on the incidence of an

Silvers-­Granelli et al,
Hasebe et al, 202044

Steffen et al, 200841

NR, not reported.

outcome, and not only on the efficacy of an intervention, the NNT

values in this review should be interpreted with caution. Considering
that incidence may vary according to type of injury, characteristics
201740
Study

of participants and sport, comparisons of NNT should be done

Total

between similar outcomes, participants and sports.

Lemes IR, et al. Br J Sports Med 2021;55:1170–1178. doi:10.1136/bjsports-2020-103683 7 of 10

 Review

Br J Sports Med: first published as 10.1136/bjsports-2020-103683 on 17 May 2021. Downloaded from http://bjsm.bmj.com/ on December 9, 2021 by guest. Protected by copyright.
Incidence of non-contact injuries in football
What is already known?
The incidence of non-­ contact injuries in this review (0.54 per
1000 hours) is higher than those observed for overall injuries in
►► More than 90% of all muscle injuries and 51%–64% of joint/
amateur players (0.49 per 1000 hours),53 and lower than observed for
ligament injuries (ie, ACL) in football (soccer) occur in non-­
non-­contact (7.4 per 1000 hours)54 and overall (8.1 per 1000 hours)7
contact situations.
injuries in professional football players. None of the studies in this
►► Exercise-­based prevention programmes reduce the overall
review included professional players, which may explain these
injury rate.
differences. We would argue that professionals play at the highest
►► There is no systematic review assessing the efficacy of
level and, therefore, are more susceptible to injuries than amateur
exercise-­based programmes in the prevention of non-­contact
players. The highest incidence in our study was for non-­contact
musculoskeletal injuries among football players.
ACL injuries (0.87 per 1000 hours), which is lower than previous
findings showing an incidence rate of ~1.5 per 1000 hours across
a seven-­season study with professional players.9 Among a total of
4483 injuries observed, 18% (n=828) were ligament-­related inju-
ries (eg, sprain/ligament injury).9 Since the majority of ligament inju- What are the new findings?
ries (ie, 64%) occurs in non-­contact situations,11 more high-­quality
randomised trials should be conducted to investigate the efficacy of ►► Exercise-­based prevention programmes reduce, on average,
exercise-­based programmes in preventing non-­contact ACL injuries the risk of non-­contact musculoskeletal injuries by 23%.
among amateur and professional football players. It is important to ►► Focused exercise-­programmes reduce the risk of non-­contact
mention that only 55% of athletes who undergo ACL reconstruc- hamstring injuries by 35%.
tion return to competitive pre-­injury level,55 which may cause some ►► The injury incidence rate of overall non-­contact injuries (for
athletes to not progress on to professional level when suffering an control group only) was 0.96 per 1000 hours of exposure.
ACL injury at lower levels of competition.
Although this review reported injury incidence for overall non-­
contact injuries from control group data, many of the included trials
reported only a subset of non-­ contact injuries (eg, hamstring or to 85%,59 while adherence to focused exercise (ie, Nordic Hamstring
ACL); thus, the injury incidence rates do not reflect the true rate of Exercise) is 69% among amateur football players.60 Clinicians should
overall non-­contact injuries. It is possible that non-­contact injuries consider using, in addition to athletes’ preferences for exercises, the
which were actually sustained during the exposure time included in top strategies to maximise player adherence to injury prevention
this review were not recorded. Even for overall non-­contact injuries, programmes: education, trust and communication, and coach atten-
there is an inherent limitation of incidence rates due to differences in dance at sessions.61
reporting systems, injury definition and athletes’ perception.
CONCLUSION
Exercise-­based prevention programmes reduce non-­ contact
Implications for clinicians musculoskeletal injuries by 23% (95% CI 3% to 39%) in
Given that football is the most popular recreational and amateur amateur football players. Despite this evidence being rated as
sport worldwide and provides important health benefits,56 57 well-­ very low quality and considering the burden that an injury
documented strategies to reduce injuries among football players, may have on athletes’ health, exercise-­ based prevention
especially amateurs, is of utmost importance to sports and public programmes should be implemented in football settings. Given
health. Based on the findings of our review and the available the number of ongoing trials, the estimate of effect found in
evidence,13 21 54 we recommend that exercise-­ based programmes this review is likely to change once the results of these trials
should be implemented in amateur football to reduce non-­contact become available. At present, it is still unclear whether focused
musculoskeletal injuries. Although this recommendation is based or general programmes provide greater reduction in overall
on very low-­quality evidence, we should not diminish the merits of non-­contact musculoskeletal injuries. Further research inves-
exercise-­based programmes in preventing sports injuries. The very tigating the efficacy of these programmes in amateur and
low-­quality evidence strengthens the need for more high-­ quality professional settings are warranted to provide a more directive
research on this topic. Regarding specificity of exercise-­ based clinical recommendation.
programmes (ie, general or focused), our findings suggest that there
might be a role for focused interventions, such as hamstring-­specific Twitter Italo Ribeiro Lemes @itolemes, Rafael Zambelli Pinto @Rafael_Z_Pinto,
Evert Verhagen @Evertverhagen and Caroline Bolling @cs_bolling
exercises (ie, eccentric), when the aim is to prevent non-­contact
hamstring injuries, which is the most prevalent non-­contact injury Acknowledgements  The authors acknowledge the Coordination for the
Improvement of Higher Education Personnel (CAPES) – Finance code 001, and
in football.10 11 Since athletes with history of hamstring injury are the corresponding authors of included studies for their help in data acquisition.
2.7 times more likely to re-­injury,58 we would argue that it might be STF and RZP are fellowship recipients from the National Council for Scientific and
feasible to incorporate hamstring-­focused exercises to exercise-­based Technological Development (CNPq).
prevention programmes for preventing occurrence and recurrence Contributors  IRL, RZP, VNL, BABR and TRS conceived and designed the study. IRL
of this type of injury. Based on the available evidence on the effi- and VNL conducted the search. IRL, VNL and BABR performed the screening, study
cacy of general exercises for reducing the risk of overall non-­contact selection and data extraction. IRL, RZP, STF and TRS analysed and interpreted the
data. IRL, RZP and TRS drafted the manuscript with input from STF, EV, CB and CFA.
musculoskeletal injuries,22 28 our view is that exercise-­based preven-
All authors have read and approved the final version.
tion programmes with general exercises (eg, eccentric strengthening,
Funding  The authors have not declared a specific grant for this research from any
plyometrics, running, agility and neuromuscular training) would be
funding agency in the public, commercial or not-­for-­profit sectors.
more feasible to implement for amateur football players because
Competing interests  None declared.
exercises are often easy to perform and may require less supervision.
Interestingly, adherence to general and focused exercises may vary. Patient consent for publication  Not required.
Previous research has shown that adherence to general exercises is up Provenance and peer review  Not commissioned; externally peer reviewed.

8 of 10 Lemes IR, et al. Br J Sports Med 2021;55:1170–1178. doi:10.1136/bjsports-2020-103683

 Review

Br J Sports Med: first published as 10.1136/bjsports-2020-103683 on 17 May 2021. Downloaded from http://bjsm.bmj.com/ on December 9, 2021 by guest. Protected by copyright.
Supplemental material  This content has been supplied by the author(s). 23 Soligard T, Myklebust G, Steffen K, et al. Comprehensive warm-­up programme to
It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not prevent injuries in young female footballers: cluster randomised controlled trial. BMJ
have been peer-­reviewed. Any opinions or recommendations discussed are 2008;337:a2469.
solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all 24 van der Horst N, Smits D-­W, Petersen J, et al. The preventive effect of the Nordic
liability and responsibility arising from any reliance placed on the content. hamstring exercise on hamstring injuries in amateur soccer players: a randomized
Where the content includes any translated material, BMJ does not warrant the controlled trial. Am J Sports Med 2015;43:1316–23.
accuracy and reliability of the translations (including but not limited to local 25 Al Attar WSA, Alshehri MA. A meta-­analysis of meta-­analyses of the effectiveness of
regulations, clinical guidelines, terminology, drug names and drug dosages), and FIFA injury prevention programs in soccer. Scand J Med Sci Sports 2019;29:1846–55.
is not responsible for any error and/or omissions arising from translation and 26 van Beijsterveldt AMC, van der Horst N, van de Port IGL, et al. How effective are
adaptation or otherwise. exercise-­based injury prevention programmes for soccer players? : A systematic
review. Sports Med 2013;43:257–65.
ORCID iDs 27 Gomes Neto M, Conceição CS, de Lima Brasileiro AJA, et al. Effects of the FIFA
Italo Ribeiro Lemes http://​orcid.​org/​0000-​0001-​9245-​287X 11 training program on injury prevention and performance in football players: a
Rafael Zambelli Pinto http://​orcid.​org/​0000-​0002-​2775-​860X systematic review and meta-­analysis. Clin Rehabil 2017;31:651–9.
Evert Verhagen http://o​ rcid.​org/​0000-​0001-9​ 227-​8234 28 Crossley KM, Patterson BE, Culvenor AG, et al. Making football safer for women:
Sérgio T Fonseca http://​orcid.​org/0​ 000-​0002-​2979-​8744 a systematic review and meta-­analysis of injury prevention programmes in 11 773
Thales R Souza http://o​ rcid.​org/0​ 000-​0001-8​ 081-​1687 female football (soccer) players. Br J Sports Med 2020;54:1089–98.
29 Eirale C, Gillogly S, Singh G, et al. Injury and illness epidemiology in soccer - effects
of global geographical differences - a call for standardized and consistent research
studies. Biol Sport 2017;34:249–54.
REFERENCES 30 Reis GF, Santos TRT, Lasmar RCP, et al. Sports injuries profile of a first division Brazilian
1 Stølen T, Chamari K, Castagna C. Physiology of soccer. Sports Medicine soccer team: a descriptive cohort study. Braz J Phys Ther 2015;19:390–7.
2005;35:501–36. 31 Moher D, Liberati A, Tetzlaff J, et al. Preferred reporting items for systematic
2 Dvorak J, Junge A, Graf-­Baumann T, et al. Editorial. Am J Sports Med 2004;32:3–4. reviews and meta-­analyses: the PRISMA statement. Ann Intern Med
3 Oja P, Titze S, Kokko S, et al. Health benefits of different sport disciplines for adults: 2009;151:264.
systematic review of observational and intervention studies with meta-­analysis. Br J 32 Oliveira CB, Elkins MR, Lemes Ítalo Ribeiro, et al. A low proportion of systematic
Sports Med 2015;49:434–40.
reviews in physical therapy are registered: a survey of 150 published systematic
4 Pedersen MT, Vorup J, Nistrup A, et al. Effect of team sports and resistance training
reviews. Braz J Phys Ther 2018;22:177–83.
on physical function, quality of life, and motivation in older adults. Scand J Med Sci
33 Gilchrist J, Mandelbaum BR, Melancon H, et al. A randomized controlled trial to
Sports 2017;27:852–64.
prevent noncontact anterior cruciate ligament injury in female collegiate soccer
5 Eime RM, Young JA, Harvey JT, et al. A systematic review of the psychological and
players. Am J Sports Med 2008;36:1476–83.
social benefits of participation in sport for children and adolescents: informing
34 Macedo LG, Elkins MR, Maher CG, et al. There was evidence of convergent and
development of a conceptual model of health through sport. Int J Behav Nutr Phys Act
construct validity of physiotherapy evidence database quality scale for physiotherapy
2013;10:98.
trials. J Clin Epidemiol 2010;63:920–5.
6 Turi-­Lynch BC, Monteiro HL, Fernandes RA, et al. Impact of sports participation on
35 Maher CG, Sherrington C, Herbert RD, et al. Reliability of the PEDro scale for rating
mortality rates among Brazilian adults. J Sports Sci 2019;37:1443–8.
quality of randomized controlled trials. Phys Ther 2003;83:713–21.
7 López-­Valenciano A, Ruiz-­Pérez I, Garcia-­Gómez A, et al. Epidemiology of injuries
36 Moseley AM, Elkins MR, Van der Wees PJ. Using research to guide practice: the
in professional football: a systematic review and meta-­analysis. Br J Sports Med
physiotherapy evidence database (PEDro). Brazilian J Phys Ther 2019.
2020;54:711–8.
37 Atkins D, Best D, Briss PA, et al. Grading quality of evidence and strength of
8 van Beijsterveldt AMCA-­M, Stubbe JH, Schmikli SL, et al. Differences in injury risk and
recommendations. BMJ 2004;328:1490.
characteristics between Dutch amateur and professional soccer players. J Sci Med
38 Guyatt GH, Oxman AD, Kunz R, et al. What is "quality of evidence" and why is it
Sport 2015;18:145–9.
important to clinicians? BMJ 2008;336:995–8.
9 Ekstrand J, Hägglund M, Waldén M. Injury incidence and injury patterns in
professional football: the UEFA injury study. Br J Sports Med 2011;45:553–8. 39 Chatellier G, Zapletal E, Lemaitre D, et al. The number needed to treat: a clinically
10 Stubbe JH, van Beijsterveldt A-­MMC, van der Knaap S, et al. Injuries in professional useful nomogram in its proper context. BMJ 1996;312:426–9.
male soccer players in the Netherlands: a prospective cohort study. J Athl Train 40 Silvers-­Granelli HJ, Bizzini M, Arundale A, et al. Does the FIFA 11+ Injury Prevention
2015;50:211–6. Program Reduce the Incidence of ACL Injury in Male Soccer Players? Clin Orthop Relat
11 Ekstrand J, Hägglund M, Waldén M. Epidemiology of muscle injuries in professional Res 2017;475:2447–55.
football (soccer). Am J Sports Med 2011;39:1226–32. 41 Steffen K, Myklebust G, Olsen OE, et al. Preventing injuries in female youth football--a
12 Rekik RN, Tabben M, Eirale C, et al. ACL injury incidence, severity and patterns in cluster-­randomized controlled trial. Scand J Med Sci Sports 2008;18:605–14.
professional male soccer players in a middle Eastern League. BMJ Open Sport Exerc 42 van de Hoef PA, Brink MS, Huisstede BMA, et al. Does a bounding exercise program
Med 2018;4:e000461. prevent hamstring injuries in adult male soccer players? - A cluster-­RCT. Scand J Med
13 Waldén M, Krosshaug T, Bjørneboe J, et al. Three distinct mechanisms predominate Sci Sports 2019;29:515–23.
in non-­contact anterior cruciate ligament injuries in male professional 43 Hammes D, Aus der Fünten K, Kaiser S, et al. Injury prevention in male veteran
football players: a systematic video analysis of 39 cases. Br J Sports Med football players - a randomised controlled trial using "FIFA 11+". J Sports Sci
2015;49:1452–60. 2015;33:873–81.
14 Jones S, Almousa S, Gibb A, et al. Injury incidence, prevalence and severity in high-­ 44 Hasebe Y, Akasaka K, Otsudo T, et al. Effects of Nordic hamstring exercise on
level male youth football: a systematic review. Sports Med 2019;49:1879–99. hamstring injuries in high school soccer players: a randomized controlled trial. Int J
15 Sherry MA, Johnston TS, Heiderscheit BC. Rehabilitation of acute hamstring strain Sports Med 2020;41:154–60.
injuries. Clin Sports Med 2015;34:263–84. 45 Owoeye OBA, Akinbo SRA, Tella BA, et al. Efficacy of the FIFA 11+ warm-­up
16 Psychological issues related to illness and injury in athletes and the team physician: a programme in male youth football: a cluster randomised controlled trial. J Sports Sci
consensus Statement-2016 update. Med Sci Sport Exerc 2017;49:1043–54. Med 2014;13:321–8.
17 Hägglund M, Waldén M, Magnusson H, et al. Injuries affect team performance 46 Waldén M, Atroshi I, Magnusson H, et al. Prevention of acute knee injuries in adolescent
negatively in professional football: an 11-­year follow-­up of the UEFA champions female football players: cluster randomised controlled trial. BMJ 2012;344:e3042.
League injury study. Br J Sports Med 2013;47:738–42. 47 Fuller CW, Ekstrand J, Junge A, et al. Consensus statement on injury definitions and
18 Sepúlveda F, Sánchez L, Amy E, et al. Anterior cruciate ligament injury: return to play, data collection procedures in studies of football (soccer) injuries. Br J Sports Med
function and long-­term considerations. Curr Sports Med Rep 2017;16:172–8. 2006;40:193–201.
19 Drawer S, Fuller CW. Propensity for osteoarthritis and lower limb joint pain in retired 48 Fédération Internationale de Football Association (FIFA). FIFA big count 2006: 270
professional soccer players. Br J Sports Med 2001;35:402–8. million people active in football. FIFA Commun Div Inf Serv 2007.
20 Eliakim E, Morgulev E, Lidor R, et al. Estimation of injury costs: financial damage of 49 Bricca A, Juhl CB, Bizzini M, et al. There are more football injury prevention reviews
English Premier League teams’ underachievement due to injuries. BMJ Open Sport than randomised controlled trials. Time for more RCT action! Br J Sports Med
Exerc Med 2020;6:e000675. 2018;52:1477–8.
21 Al Attar WSA, Soomro N, Sinclair PJ, et al. Effect of injury prevention programs that 50 Lauersen JB, Bertelsen DM, Andersen LB. The effectiveness of exercise interventions
include the Nordic hamstring exercise on hamstring injury rates in soccer players: a to prevent sports injuries: a systematic review and meta-­analysis of randomised
systematic review and meta-­analysis. Sports Med 2017;47:907–16. controlled trials. Br J Sports Med 2014;48:871–7.
22 Thorborg K, Krommes KK, Esteve E, et al. Effect of specific exercise-­based football injury 51 Grindstaff TL, Hammill RR, Tuzson AE, et al. Neuromuscular control training programs
prevention programmes on the overall injury rate in football: a systematic review and meta-­ and noncontact anterior cruciate ligament injury rates in female athletes: a numbers-­
analysis of the FIFA 11 and 11+ programmes. Br J Sports Med 2017;51:562–71. needed-t­ o-­treat analysis. J Athl Train 2006;41:450–6.

Lemes IR, et al. Br J Sports Med 2021;55:1170–1178. doi:10.1136/bjsports-2020-103683 9 of 10

 Review

Br J Sports Med: first published as 10.1136/bjsports-2020-103683 on 17 May 2021. Downloaded from http://bjsm.bmj.com/ on December 9, 2021 by guest. Protected by copyright.
52 Petersen J, Thorborg K, Nielsen MB, et al. Preventive effect of eccentric training on lipid profile, body composition, muscle strength and functional capacity. Br J Sports
acute hamstring injuries in men’s soccer: a cluster-­randomized controlled trial. Am J Med 2015;49:568–76.
Sports Med 2011;39:2296–303. 58 Green B, Bourne MN, van Dyk N, et al. Recalibrating the risk of hamstring strain
53 Herrero H, Salinero JJ, Del Coso J. Injuries among Spanish male amateur soccer injury (HSI): a 2020 systematic review and meta-­analysis of risk factors for index and
players: a retrospective population study. Am J Sports Med 2014;42:78–85. recurrent hamstring strain injury in sport. Br J Sports Med 2020;54:1081–8.
54 Delecroix B, McCall A, Dawson B, et al. Workload and non-­contact injury incidence in 59 Steffen K, Meeuwisse WH, Romiti M, et al. Evaluation of how different
elite football players competing in European leagues. Eur J Sport Sci 2018;18:1280–7.
implementation strategies of an injury prevention programme (FIFA 11+) impact
55 Ardern CL, Taylor NF, Feller JA, et al. Fifty-­five per cent return to competitive sport
team adherence and injury risk in Canadian female youth football players: a cluster-­
following anterior cruciate ligament reconstruction surgery: an updated systematic
randomised trial. Br J Sports Med 2013;47:480–7.
review and meta-­analysis including aspects of physical functioning and contextual
factors. Br J Sports Med 2014;48:1543–52. 60 van der Horst N, Hoef Svande, Otterloo Pvan, et al. Effective but not adhered to: how
56 Bangsbo J, Junge A, Dvorak J, et al. Executive summary: Football for health - can we improve adherence to evidence-­based hamstring injury prevention in amateur
prevention and treatment of non-­communicable diseases across the lifespan through football? Clin J Sport Med 2021;31:42–8.
football. Scand J Med Sci Sports 2014;24(Suppl 1):147–50. 61 McCall A, Dupont G, Ekstrand J. Injury prevention strategies, coach compliance and
57 Bangsbo J, Hansen PR, Dvorak J, et al. Recreational football for disease prevention player adherence of 33 of the UEFA elite Club injury study teams: a survey of teams’
and treatment in untrained men: a narrative review examining cardiovascular health, head medical officers. Br J Sports Med 2016;50:725–30.

10 of 10 Lemes IR, et al. Br J Sports Med 2021;55:1170–1178. doi:10.1136/bjsports-2020-103683