Sports Medicine (2023) 53:837–848

https://doi.org/10.1007/s40279-022-01797-7

SYSTEMATIC REVIEW

Efficacy of Multi‑Component Exercise‑Based Injury Prevention

Programs on Injury Risk Among Footballers of All Age Groups:
A Systematic Review and Meta‑analysis
Rilind Obërtinca1,3 · Ilir Hoxha2,4,5 · Rina Meha1,4 · Arber Lama4,5,6 · Altina Bimbashi4 · Dorentina Kuqi4 ·
Bujar Shabani4,6 · Tim Meyer1 · Karen aus der Fünten1

Accepted: 21 November 2022 / Published online: 8 February 2023

© The Author(s) 2023

Abstract
Background Playing football is associated with a high risk of injury. Injury prevention is a priority as injuries not only nega-
tively impact health but also potentially performance. Various multi-component exercise-based injury prevention programs
for football players have been examined in studies.
Objective We aimed to investigate the efficacy of multi-component exercise-based injury prevention programs among foot-
ballers of all age groups in comparison to a control group.
Methods We conducted a systematic review and meta-analysis of randomized and cluster-randomized controlled trials.
CINAHL, Cochrane, PubMed, Scopus, and Web of Science databases were searched from inception to June 2022. The
following inclusion criteria were used for studies to determine their eligibility: they (1) include football (soccer) players;
(2) investigate the preventive effect of multi-component exercise-based injury prevention programs in football; (3) contain
original data from a randomized or cluster-randomized trial; and (4) investigate football injuries as the outcome. The risk
of bias and quality of evidence were assessed using the Cochrane Risk of Bias Tool and the Grading of Recommendations
Assessment, Development, and Evaluation (GRADE), respectively. The outcome measures were the risk ratio (RR) between
the intervention and the control group for the overall number of injuries and body region-specific, contact, and non-contact
injuries sustained during the study period in training and match play.
Results Fifteen randomized and cluster-randomized controlled trials with 22,177 players, 5080 injuries, and 1,587,327
exposure hours fulfilled the inclusion criteria and reported the required outcome measures. The point estimate (RR) for the
overall number of injuries was 0.71 (95% confidence interval [CI] 0.59–0.85; 95% prediction interval [PI] 0.38–1.32) with
very low-quality evidence. The point estimate (RR) for lower limb injuries was 0.82 (95% CI 0.71–0.94; 95% PI 0.58–1.15)
with moderate-quality evidence; for hip/groin injuries, the RR was 0.56 (95% CI 0.30–1.05; 95% PI 0.00–102.92) with low-
quality evidence; for knee injuries, the RR was 0.69 (95% CI 0.52–0.90; 95% PI 0.31–1.50) with low-quality evidence; for
ankle injuries, the RR was 0.73 (95% CI 0.55–0.96; 95% PI 0.36–1.46) with moderate-quality evidence; and for hamstring
injuries, the RR was 0.83 (95% CI 0.50–1.37) with low-quality evidence. The point estimate (RR) for contact injuries was
0.70 (95% CI 0.56–0.88; 95% PI 0.40–1.24) with moderate-quality evidence, while for non-contact injuries, the RR was
0.78 (95% CI 0.55–1.10; 95% PI 0.25–2.47) with low-quality evidence.
Conclusions This systematic review and meta-analysis indicated that the treatment effect associated with the use of multi-
component exercise-based injury prevention programs in football is uncertain and inconclusive. In addition, the majority
of the results are based on low-quality evidence. Therefore, future high-quality trials are needed to provide more reliable
evidence.
Clinical Trial Registration PROSPERO CRD42020221772.

Extended author information available on the last page of the article

Vol.:(0123456789)
838 R. Obërtinca et al.

in reducing the overall number of injuries as well as body

Key Points region-specific injuries, and considering footballers of all
age groups (children, youth, senior, and veteran). Addi-
The present meta-analysis is the first to use prediction tionally, contact-related injuries represent 50% of overall
intervals in the interpretation of results derived from injuries in professional football [17]. Previous research has
trials assessing the efficacy of multi-component exercise- not investigated the impact of the programs on preventing
based injury prevention programs among footballers of these injuries. Providing information about the age-specific
all age groups. efficacy and estimating the potential of these programs on
This study revealed that the evidence for meaningful contact-related injuries may guide future evidence-based
effects of exercise-based injury prevention programs directions regarding the implementation and development
remains inconclusive at best. of new interventions. Finally, providing only confidence
intervals (CIs) might not be the best way forward. A recent
The quality of evidence is a major issue in existing stud- meta-analysis examined the effect of the Nordic hamstring
ies; therefore, these findings call for future high-quality exercise [18]. The authors strongly recommended providing
trials to provide more reliable evidence. the prediction intervals (PIs) in addition to CIs. This is in
line with authors promoting the use of PIs in the interpreta-
tion of results from a random-effects meta-analysis of trials
assessing treatment effects [19]. Therefore, and for the first
time, this meta-analysis reports the PIs in addition to the
CIs. The aim of this meta-analysis was to investigate the
1 Background efficacy of multi-component exercise-based injury preven-
tion programs in reducing injuries of different types among
The overall injury incidence in professional male football footballers of all age groups.
players is between 5.9 [1] and 9.6 [2] injuries/1000 football
hours. In amateur and veteran football, reported incidences
are even higher and reach 9.6 [2] to 12.5 [3] and 12.4 [4]
injuries/1000 football hours, respectively. There are hardly 2 Methods
any data regarding players under the age of 11 years [5].
A professional football team with 25 players has approx- 2.1 Protocol and Registration
imately 50 injuries per season [6], and youth elite teams
about 30 [7]. Many efforts have been made in recent years to We report this systematic review in accordance with the
reduce these numbers. Various injury prevention programs guidelines of the Preferred Reporting Items for Systematic
for football players of both sexes and various age groups reviews and Meta-Analyses (PRISMA) [20]. The study was
have been established. Some of them target specific inju- registered at PROSPERO (ID: CRD42020221772).
ries, for example, Prevent injury and Enhance Performance
[8] and HarmoKnee [9], target knee injuries. Others take a 2.2 Study Eligibility Criteria
more general approach, trying to prevent non-contact lower
extremity injuries in general for example, ­FIFA® 11 [10], In the present study, we included all controlled, multi-com-
­FIFA® 11 + [11], and the Neuromuscular training program ponent exercise-based injury prevention programs contain-
[12]. 11 + Kids [13] aims to prevent football injuries by ing at least two or more exercises. Players of the intervention
increasing children’s fundamental and sport-specific motor group performed these programs during their training ses-
skills. sions in addition to their usual training and were compared
Previous systematic reviews and meta-analyses have eval- to a control group. Criteria for study inclusion were: (1)
uated the efficacy of either specific programs (e.g., FIFA include football (soccer) players; (2) investigate the preven-
11 and 11 +) [14, 15] or the effect of various programs on tive effect of multi-component exercise-based injury preven-
specific injuries (e.g., non-contact injuries) [16]. However, tion programs in football; (3) contain original data from a
recognizing the differences between programs regarding the randomized or cluster-randomized trial; and (4) investigate
content, the different age groups targeted, and the differ- football injuries as the outcome. Studies were excluded from
ent results reported compared to each other, a comprehen- the meta-analysis if they were: (1) studies with a single exer-
sive meta-analysis of pooled results across the studies will cise intervention; (2) studies with a primary target on per-
produce a more comprehensive result. To date, no meta- formance or other physical measurements than injuries; (3)
analysis is available that has evaluated the efficacy of all studies using protective equipment (e.g., bracing) as part
multi-component exercise-based injury prevention programs
Injury Prevention in Football 839

of the intervention; and (4) studies published in a language by two reviewers (RO, BSH). Discrepancies were resolved
other than English. by consensus. The overall quality of evidence was assessed
using the Grading of Recommendations Assessment, Devel-
2.3 Sources and Study Selection opment, and Evaluation (GRADE). This method assesses the
strength of evidence derived from systematic reviews [22].
Possible studies were identified using a systematic search In the GRADE system, randomized controlled trials (RCTs)
process. First, we searched the following databases begin as high-quality evidence [23]. Subsequently, the evi-
CINAHL, Cochrane, PubMed, Scopus, and Web of Science dence is downgraded by one level for each of the following
from the earliest record to June 2022, with the following domains considered: (1) risk of bias (downgraded by one
search strategy: (injury prevention OR warm-up program OR level if the trials scored an overall high risk of bias on the
neuromuscular program OR f-marc OR 11 +) AND (football Cochrane Collaboration Risk of Bias Tool); (2) inconsist-
OR soccer). The reference lists of the studies recovered were ency (downgraded by one level if statistical heterogeneity
hand searched to identify potentially eligible studies missed between studies was I2 > 50%); (3) indirectness (downgraded
by electronic searches. Two reviewers independently (AB, by one level if the meta-analysis included participants with
DK) performed the selection of studies based on the title and heterogeneous characteristics with regard to sex, age, and
abstract provided by the bibliographic databases. The full- level of sport); (4) imprecision (downgraded by one level
text evaluation followed on those selected studies from the if the upper and lower CIs had a > 0.5 difference); and (5)
first selection step. A third reviewer (RO) was responsible publication bias (assessed with a visual inspection of a fun-
for resolving any discrepancies in the selection process. nel plot and two-tailed Egger’s test if more than ten studies
were included in the meta-analysis). Evidence obtained was
2.4 Data Extraction and Administration categorized into four levels of evidence quality: high, moder-
ate, low, and very low [24] (Table 1).
For each eligible study, four reviewers (RM, AB, DK, AL)
extracted data independently using a standardized data 2.6 Outcome Measures
extraction form [14]. One section was added (type of inju-
ries: contact or non-contact) to the extraction form for an The primary outcome was the risk ratio (RR) for the over-
additional analysis that we performed regarding the effect all number of injuries. Body region-specific injury RRs for
on contact versus non-contact injuries. We extracted data the lower limb, hamstring, hip/groin, knee, and ankle were
on the studies’ basic information, design, participants, inter- secondary outcomes. Additionally, the overall number and
vention characteristics, and outcome measures. Thereafter, the region-specific injury RRs were assessed for a non-con-
the reviewers compared the extracted data for consistency. tact versus contact induced cause. All injuries occurring in
Reviewers resolved discrepancies by discussion and, when official training and match play during the respective study
necessary, a fifth party (RO) was involved. Final decisions period were included.
were made based on a majority vote. Primary outcome
results from individual studies were extracted and collated 2.7 Synthesis of Results
in Excel 365 (Microsoft Corporation, Redmond, WA, USA).
If studies did not report RR estimates, we converted them to
2.5 Quality Assessment RRs as far as possible [25, 26]. Out of the 15 included stud-
ies, six studies did not perform cluster adjustments. They
The risk of bias was assessed for each included trial accord- also did not provide information on the intra-cluster correla-
ing to the recommendations outlined in the Cochrane Hand- tion coefficient or other data that would allow for calculating
book for Systematic Reviews of Interventions [21]. The the design effect or inflation factor (as recommended by the
following items were considered: allocation sequence gener- Cochrane Handbook for Systematic Review of Interven-
ation, concealment of allocation, blinding of outcome assess- tions) [27]. Hence, we performed a cluster adjustment by
ment, incomplete outcome data, selective outcome report- increasing variance by 30% for effect estimates of studies
ing, and other sources of bias. As it is impossible to blind the with no adjustment for the cluster effect [28]. We performed
participants to the intervention, we removed the item “blind- a meta-analysis of RRs and their 95% CIs using the DerSi-
ing of participants and investigators”. Each bias domain was monian and Laird random-effects method [29]. A random-
judged as at low or high risk of bias according to its possible effects meta-analysis assumes that the true treatment effect
effect on the results of the study. When the possible effect varies among studies. The DerSimonian and Laird method
was unknown or insufficient detail was reported, we judged does not make any assumptions about the distribution of the
it as unclear. The risk of bias was examined independently random effects [30]. In addition to the presentation of overall
effect estimates and 95% CIs, we also calculated 95% PIs.
840 R. Obërtinca et al.

Table 1  Grades of recommendation, assessment, development and evaluation (GRADE) quality of evidence

CI confidence interval, RCT​randomized controlled trial, RR risk ratio

They enable the examination of treatment effects within an conducted in one each of the following countries: Canada
individual study setting, as this can differ from the average [12], Australia [36], Rwanda [37], Nigeria [38], and Iran
effect [19]. Heterogeneity was assessed using I2, τ2, and Q [39]. The overall number of participants was 22,177 includ-
value (χ2 test for heterogeneity). We interpreted I2 values ing both sexes. Participants were registered football players in
according to guidelines by Higgins and Green, a low het- one of the following age groups: children (7–14 years), youth
erogeneity for I2 values between 25 and 50%, a moderate f (12–19 years), senior, and veteran (> 32 years). The number
heterogeneity or 50–75%, and a high heterogeneity for ≥ 75% of participants ranged from 265 [4] to 4564 participants [9].
[27]. A small study effect was investigated using Egger’s test A total of 5080 injuries and 1,587,327 h of exposure were
for a meta-analysis with ten or more studies [31]. Statistical included. The study period lasted between 12 weeks [8] and
analysis was carried out using STATA 17 BE (Stata Corpo- 9 months [4, 13, 39, 33]. All interventions were applied at
ration, College Station, TX, USA). least twice a week in the training sessions. The control groups
performed their usual warm-up exercises and/or training rou-
tines. One study required an additional home-based stretch-
3 Results ing program [12]. Nine studies used a F ­ IFA® warm-up pro-
® ®
gram of the ­FIFA 11, the ­FIFA 11 + , or the 11 + Kids [4,
3.1 Literature Identification 10, 11, 13, 33, 35, 37–39]. Two studies used Neuromuscular
Training programs [12, 32], and one study each used the Neu-
The initial database search identified 7954 studies. Fol- romuscular Control Program [36], the Knäkontroll program
lowing the removal of duplicates (n = 4986), 2968 studies [9], the Prevention Injury and Enhance Performance program
remained. After screening the titles and abstracts, 69 full-text [8], and the Bounding Exercise Program [34] (Table 2).
articles were left. A further 54 studies had to be excluded as
they did not present data on injuries, included non-football 3.3 Risk of Bias
players, or were neither cluster RCTs nor RCTs. Finally, 15
articles were included in the meta-analysis (Fig. 1). Seven (46%) studies had a high risk of bias in two or more
domains. The domain “other bias” was the most frequent
3.2 Demographic and Study Characteristics cause for a high risk of bias within the studies (46%), with
seven studies neither reporting an intention-to-treat analysis
Eight trials stemmed from Europe [4, 9–11, 13, 32–34]. nor an adjustment for clustering (Fig. 1 and Table 1 of the
Two trials were conducted in the USA [8, 35]. One trial was Electronic Supplementary Material [ESM]).
Injury Prevention in Football 841

Fig. 1  Flow chart of the included studies. RCTs randomized controlled trials

3.4 Meta‑Analysis Results (downgraded one level because of a risk of bias and one
level because of inconsistency). For hip/groin injuries, the
3.4.1 Overall, Body Region, Contact, RR was 0.56 (95% CI 0.30–1.05; 95% PI 0.00–102.92) with
and Non‑Contact‑Related Injuries low-level evidence (downgraded one level because of a risk
of bias and one level because of imprecision). For hamstring
For the primary outcome analysis, i.e., the overall injury injuries, the RR was 0.83 (95% CI 0.50–1.37) with low-
risk, the pooled results showed a point estimate (RR) of level evidence (downgraded one level because of a risk of
0.71 (95% CI 0.59–0.85; 95% PI 0.38–1.32; I2 = 80.5%; bias and one level because of imprecision). With regard to
τ2 = 0.067; p < 0.001). The width of the 95% PI suggests that ankle injuries, the RR was 0.73 (95% CI 0.55–0.96; 95% PI
the effect in future similar studies lies between 0.38 and 0.36–1.46) with moderate-level evidence (downgraded one
1.32 (Fig. 2). In practical terms, the effect may vary from level because of a risk of bias). For each calculation, the 95%
being very protective to an increased risk of injury. The level PI was wider in comparison to the 95% CI.
of evidence was rated as very low (downgraded one level The pooled results for non-contact injuries showed a
because of a risk of bias, one level because of inconsistency, point estimate (RR) of 0.78 (95% CI 0.55–1.10; 95% PI
and one level because of publication bias) (Table 1). 0.25–2.47; I2 = 67.3%; τ2 = 0.100; p = 0.016), with evidence
Regarding the secondary outcome analyses, i.e., the body rated as low level (downgraded one level because of a risk
region-specific injury risk (Fig. 2), the point estimate (RR) of bias and one level because of inconsistency). Addition-
for the lower limb injuries was 0.82 (95% CI 0.71–0.94; ally, the point estimate (RR) for contact injuries was 0.70
95% PI 0.58–1.15; I2 = 45.3%; τ2 = 0.016; p = 0.067) with (95% CI 0.56–0.88; 95% PI 0.40–1.24 I2 = 29.2%; τ2 = 0.018;
moderate-level evidence (downgraded one level because of p = 0.227), with moderate-level evidence (downgraded one
a risk of bias). For knee injuries, the RR was 0.69 (95% level because of a risk of bias). The width of the 95% PI sug-
CI 0.52–0.90; 95% PI 0.31–1.50) with low-level evidence gested that the effect may vary from being very protective
842 R. Obërtinca et al.

Table 2  Summary of included multi-component randomized controlled trials investigating the effect of injury prevention programs
Study Intervention Population (age) Follow-up Outcome Number of ana- Exposure time Number of
program lysed (h) injuries
(players)

Emery et al. 2010 Neuromuscular Male and female 20 weeks Overall injuries IG: 380 IG: 24 051 IG: 50
[12] training program youth (13– CG: 364 CG: 24 597 CG: 79
18 years)
Finch et al. 2016 Neuromuscular Male senior 28 weeks Overall injuries IG: 679 IG: 12 ­790a IG: 335
[36] control program (18–30 years) CG: 885 CG: 15 ­537a CG: 438
Gilchrist et al. PEP Female senior, 12 weeks Knee injuries IG: 583 IG: 35 220 IG: ­40c
2008 [8] (19.88 years)b CG: 852 CG: 52 919 CG: ­58c
®
Hammes et al. FIFA 11 + Male veteran 9 months Overall injuries IG: 146 IG: 4 172 IG: 51
2015 [4] (≥ 32 years) CG:119 CG: 2 937 CG: 37
Hilska et al. 2021 Neuromuscular Male and female 20 weeks Lower limb inju- IG: 673 IG: 71 109 IG: ­310d
[32] training children ries CG: 730 CG: 63 404 CG: ­346d
(9–14 years)
®
Nuhu et al. 2021 FIFA 11 + Male senior (IG: 7 months Overall injuries IG: 309 IG: 65 333 IG: 168
[37] 19.9 years) (CG: CG: 317 CG: 63 389 CG: 252
19.7 years))
Owoeye et al. FIFA® 11 + Male youth 6 months Overall injuries IG: 212 IG: 51 017 IG: 36
2014 [38] (14–19 years) CG: 204 CG: 61 045 CG: 94
Rossler et al. 2018 11 + Kids Male and female 9 months Overall injuries IG: 2066 IG: 140 716 IG: 139
[13] children CG: 1829 CG: 152 033 CG: 235
(7–13 years)
Silvers-Granell FIFA® 11 + Male senior 5 months Overall injuries IG: 675 IG: 35 226 IG: 285
et al. 2017 [35] (18–25 years) CG: 850 CG: 44 212 CG: 665
Soligard et al. FIFA® 11 + Female youth 8 months Overall injuries IG: 1055 IG: 49 899 IG: 161
2008 [11] (13–17 years) CG: 837 CG: 45 428 CG: 215
Steffen et al. 2008 FIFA® program Female youth 8 months Overall injuries IG: 1073 IG: 66 423 IG: 242
[10] 11 (13–17 years) CG: 947 CG: 65 725 CG: 241
Walden et al. Knakontrol Female youth 7 months ACL injuries IG: 2479 IG: 149 214 IG: ­7e
2012 [9] (12–17 years) CG: 2085 CG: 129 084 CG: ­14e
Zarei et al. 2020 11 + kids Male children 9 months Overall injuries IG: 443 IG: 31 934 IG: 30
[39] (7–14 years) CG: 519 CG: 32 113 CG: 60
®
Van de Beijs- FIFA program Male senior 9 months Overall injuries IG: 233 IG: 21 605 IG: 207
terveldt et al. 11 (18–40 years) CG: 233 CG: 22 647 CG: 220
2012 [33]
Van de Hoef et al. BEP Male senior 39 weeks Hamstring inju- IG: 229 IG: 31 831 IG: ­35f
2019 [34] (18–45 years) ries CG: 171 CG: 21 717 CG: ­30f

ACL anterior cruciate ligament, BEP bounding exercise program, CG control group, IG intervention group, N/A Not applicable, PEP Prevent
injury and Enhance Performance
a
Match exposure only was reported
b
Average age only was reported
c
Knee injuries
d
Lower limb injuries
e
ACL injuries
f
Hamstring injuries

to an increased risk of injury for both outcomes, i.e., non- 3.4.2 Subgroup Analysis According to Sex
contact injuries (95% PI 0.55–1.10) and contact injuries
(95% PI 0.40–1.24) (Fig. 3). Regarding a distinction between male and female individu-
als, the point estimate (RR) for the overall number of injuries
Injury Prevention in Football 843

Fig. 2  Analysis of multi-component exercise-based injury prevention programs’ effect on the overall and region-specific injury risk compared
with control groups. I2 I square, p p value, RR risk ratio, τ2 tau square

in male football players was 0.70 (95% CI 0.55–0.90; and, in veterans, the RR was 0.91 (95% CI 0.53–1.57) (Fig. 4
I2 = 83.5%; τ2 = 0.082; p < 001). In female football play- of the ESM).
ers, the point estimate (RR) was 0.82 (95% CI 0.57–1.20;
I2 = 68.9%; τ2 = 0.064; p = 0.008) (Fig. 4 of the ESM).
4 Discussion
3.4.3 Subgroup Analysis According to Age Group
4.1 Principal Findings
The point estimate (RR) for the overall number of injuries in
children was 0.52 (95% CI 0.36–0.76; I2 = 0.0%; τ2 < 0.001; This systematic review and meta-analysis included 15 RCTs
p = 0.841), in youth, the RR was 0.74 (95% CI 0.56–0.97; that assessed the effect of injury prevention programs on the
I2 = 68.9%; τ2 = 0.048; p = 0.022), in seniors, the RR was overall and body region-specific injury risk in football play-
0.73 (95% CI 0.53–1.01; I2 = 91.1%; τ2 = 0.098; p < 0.001), ers. Based on calculated PIs, their efficacy remains uncertain
844 R. Obërtinca et al.

Fig. 3  Analysis of multi-component exercise-based injury prevention programs’ effect on the overall non-contact (a) and contact (b) injury risk
compared with control groups. I2 I square, p p value, RR risk ratio, τ2 tau square

and inconclusive regarding all primary and secondary out- participants). A previous meta-analysis [14] investigated the
comes. In addition, the majority of the results are based on effect of the ­FIFA® exercise-based injury prevention programs
low-quality evidence. on specific body regions. The observed efficacy on hamstring
(RR 0.83 vs IRR 0.40), knee (RR 0.69 vs IRR 0.52), and ankle
4.1.1 Comparison with Existing Literature on Injury Risk injuries (RR 0.73 vs IRR 0.68) was lower in our study, but
Reduction comparable for hip/groin injuries (RR 0.56 vs IRR 0.59). A
likely explanation for the differing results between the reviews
Riley et al. [40] suggested that if a random-effects approach is is that we included a higher number of studies that examined
used, the pooled result must be interpreted as the average inter- different types of programs in the analysis. An additional
vention effect across studies, rather than the common effect. explanation could be the inclusion of studies with children
Previous meta-analyses have not reported PIs, which means, because injury patterns vary with age [42]. The most obvious
an appropriate comparison is not possible. Therefore, we can difference from other studies was regarding hamstring injuries.
only compare our point estimates with those reported in the The results may be expected as we did not include trials inves-
literature. In contrast with the currently available evidence tigating the Nordic Hamstring as a single component exercise,
[14–16, 41], our study included footballers of all age groups which has been shown to be very effective for preventing ham-
and skill levels (amateur and professional). The point estimate string injuries [43]. Moreover, in comparison to Thorborg et al.
(RR) of 0.71 (95% CI 0.59–0.85) in the current analysis is at [14], we included the Bounding Exercise Program [34], which
the lower end of those reported in previous systematic reviews, showed very little effect in reducing these injuries.
which reported an incidence rate ratio (IRR) of 0.73 (95% CI
0.59–0.91) [41], IRR of 0.75 (95% CI 0.57–0.98) [14], IRR 4.1.2 Effectiveness of Injury Prevention Programs
of 0.77 (95% CI 0.64–0.91) [15], and IRR of 0.77 (95% CI on Contact Versus Non‑contact Injuries
0.61–0.97) [16]. This was to be expected as we also included
interventions in children, which showed a substantially higher For the first time, this study investigated the effect of multi-
injury reduction of 48% [13] and 50% [39] compared with component exercise-based injury prevention programs not
older players. This effect was somewhat counterbalanced by only on non-contact injuries but also on contact-related
the reduced effect of the programs among veterans, which injuries. The point estimate (RR) for contact injuries was
was only 9%. However, the relative weight of the studies 0.70 (95% CI 0.56–0.88). Surprisingly, the estimated risk
with children was higher (higher in the number of studies and reduction was higher than for non-contact injuries for which
Injury Prevention in Football 845

the vast majority of programs are designed. Most programs comparison to confidence intervals. Based on this evidence,
include strength exercises that mostly focus on core stability. there is a lack of compelling data to affirm the certainty
Furthermore, plyometrics (hopping, jumping, and landing) of preventive effects from multi-component exercise-based
are often part of the programs. They have the potential to injury prevention programs. However, for our meta-anal-
improve lower leg strength, functional leg stability, and bal- ysis, we have to take into account that the use of PIs has
ance, thus improving the ability to absorb external forces, for its shortcomings. IntHout et al. [19] mentioned that they
example, induced by contact. The 11 + Kids [13] program show a wider range compared with CIs when there is any
also includes one exercise specifically on correct falling heterogeneity. Our main outcome provided an I2 = 80.5%,
techniques. The point estimate (RR) for non-contact injuries which should be interpreted as high heterogeneity accord-
in the current study was 0.78, in line with a previous study ing to the Cochrane Handbook for Systematic Reviews of
that reported a RR of 0.77 [16]. Interventions [27]. In addition, Riley et al. [40] stated that a
PI will be most appropriate when the studies included in the
4.1.3 Effectiveness of Injury Prevention Programs Across meta-analysis have a low risk of bias. However, the majority
Sexes and Age Groups of studies in our analysis had a high risk of bias. Therefore,
these shortcomings would have affected the use of PIs in
The subgroup analysis showed a point estimate (RR) of 0.70 our meta-analysis.
in male football players. These results mimic the data of the
Al Attar et al. study [15]. However, the estimated effect is 4.3 Strengths and Limitations
slightly lower than data reported by Lemes et al. [16] show-
ing a point estimate (RR) of 0.68. To the best of our knowledge, this review is the first to ana-
Regarding female individuals, the pooled results showed lyze the efficacy of multi-component exercise-based injury
a point estimate (RR) of 0.82. This result falls within the prevention programs among footballers of all age groups.
range of results reported by studies with similar inclusion One strength of this systematic review is that it included
criteria [15, 16]. However, the meta-analysis with the largest multiple analyses. It investigated the risk reduction for the
estimated effect [41] included RCTs that used various injury overall number of injuries as well as of body region-specific,
prevention strategies. In addition to physical exercises, they contact, and non-contact injuries. Subgroup analyses for age
included studies that used braces and education as a method and sex were also performed. Additionally, the PIs for the
for prevention. Furthermore, they included studies with par- main outcomes were calculated. A further strength is the
ticipants of varying backgrounds and sports (i.e., middle and large number of participants (22,177), injuries (5080), and
high school non-footballer athletes). These dissimilarities exposure hours (1,587,327 h) included in comparison with
might have caused these considerable differences. In con- other reviews [14–16]. Furthermore, we followed best prac-
trast, small differences compared with other reviews [15, tice by including only randomized trials and cluster-RCTs,
16] may reflect the diversity of interventions, i.e., the inclu- using a risk of bias assessment and grading the quality of
sion of single-component exercise-based injury prevention evidence.
programs. However, this review also has some limitations, mainly
The subgroup analysis for age groups showed a point esti- that > 50% of the reported effects were based on studies
mate (RR) of 0.52 in children, a RR of 0.74 in youth, 0.73 in with a very low or low level of evidence. The main outcome
seniors, and 0.91 in veteran football players. The point esti- variable provided high heterogeneity among the studies
mate in youth and seniors is homogeneous with the current (I2 = 80.5%). The lack of information about compliance with
available evidence [14, 41]. The low point estimate found in the prevention program in many studies is another limitation
children may be expected by the fact that there is rarely any of this review. Furthermore, there was missing information
prior use of preventative measures at all; therefore, using on content and compliance with the usual warm-ups/training
the program is likely to evoke the biggest benefit. Only one routines of the control groups. Another limitation is the high
trial [4] assessed the effects of injury prevention programs risk of bias, especially from the “other bias” domain, with
in veteran football players. The comparably small effect in seven studies failing to report the use of an intention-to-treat
this population is likely owing to the infrequent application analysis and of an adjustment for clustering. Finally, two
of the program (only once a week) as well as relatively low deviations (lack of a compliance analysis and the modifica-
compliance. tion of literature databases) from the original study protocol
have to be mentioned as limitations of this review.
4.2 Factors to Take into Account When Assessing PIs

In the current analysis, we calculated the PIs for the main

investigated outcomes. Prediction intervals were wider in
846 R. Obërtinca et al.

4.4 Differences Between the Protocol and Review Supplementary Information The online version contains supplemen-
tary material available at https://d​ oi.o​ rg/1​ 0.1​ 007/s​ 40279-0​ 22-0​ 1797-7.
Owing to the lack of respective information provided in
the studies, a compliance analysis was impossible. We
Declarations
contacted the corresponding authors to provide us with
these data, but within the set time of 2 weeks, we only Funding Open Access funding enabled and organized by Projekt
received information on one of the studies. Our planned DEAL.
bibliographic databases for literature identification were
modified during the study implementation. Because of the Conflicts of interest/competing interests Rilind Obërtinca, Ilir Hoxha,
lack of access, we did not search in EMBASE and SPORT- Rina Meha, Arber Lama, Altina Bimbashi, Dorentina Kuqi, Bujar Sha-
bani, Tim Meyer, and Karen aus der Fünten have no conflicts of inter-
Discus. However, we additionally searched in the origi- est that are directly relevant to the content of this article.
nally unplanned database Scopus. In addition, to empower
the review, although it was not registered in the protocol, Ethics approval Not applicable.
we assessed the quality of evidence using the GRADE
Consent to participate Not applicable.
approach and calculated the PIs for the main outcomes.
Consent for publication Not applicable.
4.5 Recommendations for Future Studies
Availability of data and material The datasets generated and/or ana-
lyzed during the study implementation are available from the corre-
Based on the data obtained, we recommend future high- sponding author upon request.
quality trials to investigate the efficacy of multi-component
exercise-based injury prevention programs. In upcoming Code availability Not applicable.
studies, data on compliance and the content of the train-
Authors’ contributions RO, IH, and KadF conceived and designed the
ing of the control groups should be included. Adjustment study. RO, IH, RM, AL, AB, DK, and BSH conducted the search,
for clustering and more extensive reporting of outcomes study selection, data extraction, and quality appraisal. IH analyzed the
should be emphasized. In addition, it appears important data. RO, IH, and KadF contributed to the interpretation of the data.
RO drafted the manuscript with input from KadF and TM. All authors
to create new injury prevention programs that reflect the
have read and approved the final manuscript.
development and changes in football training. This should
include increasing their attractiveness to promote compli-
Open Access This article is licensed under a Creative Commons Attri-
ance (also outside of study settings), which appears crucial bution 4.0 International License, which permits use, sharing, adapta-
to reduce injury risk. Currently, a large number of different tion, distribution and reproduction in any medium or format, as long
exercises are included because it is unknown which exer- as you give appropriate credit to the original author(s) and the source,
provide a link to the Creative Commons licence, and indicate if changes
cises (or which combination of them) are most effective
were made. The images or other third party material in this article are
in general or in relation to specific injuries. Tailoring the included in the article's Creative Commons licence, unless indicated
exercises would potentially mean fewer injuries and more otherwise in a credit line to the material. If material is not included in
efficiency. the article's Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will
need to obtain permission directly from the copyright holder. To view a
copy of this licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/.
5 Conclusions

This meta-analysis indicated that evidence for the meaning-

ful effects of multi-component exercise-based injury preven- References
tion programs in football remains inconclusive at best. This
statement is based on PIs that were wider than the frequently 1. Aus der Fünten K, Faude O, Lensch J, Meyer T. Injury charac-
teristics in the German professional male soccer leagues after a
employed CIs, with a range from very protective effects to an
shortened winter break. J Athl Train. 2014;49(6):786–93. https://​
increased injury risk. In addition, the quality of evidence is a doi.​org/​10.​4085/​1062-​6050-​49.3.​51.
major issue in existing studies. These findings call for future 2. van Beijsterveldt AM, Stubbe JH, Schmikli SL, van de Port IG,
high-quality trials to provide more reliable evidence regard- Backx FJ. Differences in injury risk and characteristics between
Dutch amateur and professional soccer players. J Sci Med Sport.
ing the efficacy of injury prevention programs in football.
2015;18(2):145–9. https://​doi.​org/​10.​1016/j.​jsams.​2014.​02.​004.
3. Szymski D, Krutsch V, Achenbach L, Gerling S, Pfeifer C, Alt
V, et al. Epidemiological analysis of injury occurrence and
current prevention strategies on international amateur foot-
ball level during the UEFA Regions Cup 2019. Arch Orthop
Injury Prevention in Football 847

Trauma Surg. 2022;142(2):271–80. https://​d oi.​o rg/​1 0.​1 007/​ 18. Impellizzeri FM, McCall A, van Smeden M. Why methods mat-
s00402-​021-​03861-9. ter in a meta-analysis: a reappraisal showed inconclusive injury
4. Hammes D, Aus der Fünten K, Kaiser S, Frisen E, Bizzini M, preventive effect of Nordic hamstring exercise. J Clin Epidemiol.
Meyer T. Injury prevention in male veteran football players: a 2021;140:111–24. https://​doi.​org/​10.​1016/j.​jclin​epi.​2021.​09.​007.
randomised controlled trial using “FIFA 11+”. J Sports Sci. 19. IntHout J, Ioannidis JP, Rovers MM, Goeman JJ. Plea for rou-
2015;33(9):873–81. https://​doi.​org/​10.​1080/​02640​414.​2014.​ tinely presenting prediction intervals in meta-analysis. BMJ
975736. Open. 2016;6(7): e010247. https:// ​ d oi. ​ o rg/ ​ 1 0. ​ 1 136/ ​ b mjop​
5. Faude O, Rößler R, Junge A. Football injuries in children and en-​2015-​010247.
adolescent players: are there clues for prevention? Sports Med. 20. Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC,
2013;43(9):819–37. https://​doi.​org/​10.​1007/​s40279-​013-​0061-x. Mulrow CD, et al. The PRISMA 2020 statement: an updated
6. Ekstrand J, Hägglund M, Waldén M. Injury incidence and injury guideline for reporting systematic reviews. BMJ. 2021;372: n71.
patterns in professional football: the UEFA injury study. Br J https://​doi.​org/​10.​1136/​bmj.​n71.
Sports Med. 2011;45(7):553–8. https://​doi.​org/​10.​1136/​bjsm.​ 21. Higgins JP, Altman DG, Gøtzsche PC, Jüni P, Moher D, Oxman
2009.​060582. AD, Cochrane Bias Methods Group; Cochrane Statistical Methods
7. Materne O, Chamari K, Farooq A, Weir A, Hölmich P, Bahr R, Group, et al. The Cochrane Collaboration’s tool for assessing risk
et al. Injury incidence and burden in a youth elite football acad- of bias in randomised trials. BMJ. 2011;343:d5928. https://​doi.​
emy: a four-season prospective study of 551 players aged from org/​10.​1136/​bmj.​d5928.
under 9 to under 19 years. Br J Sports Med. 2021;55(9):493–500. 22. Atkins D, Best D, Briss PA, Eccles M, Falck-Ytter Y, Flottorp S,
https://​doi.​org/​10.​1136/​bjspo​rts-​2020-​102859. GRADE Working Group, et al. Grading quality of evidence and
8. Gilchrist J, Mandelbaum BR, Melancon H, Ryan GW, Silvers HJ, strength of recommendations. BMJ. 2004;328(7454):1490. https://​
Griffin LY, et al. A randomized controlled trial to prevent noncon- doi.​org/​10.​1136/​bmj.​328.​7454.​1490.
tact anterior cruciate ligament injury in female collegiate soccer 23. Meader N, King K, Llewellyn A, Norman G, Brown J, Rodgers
players. Am J Sports Med. 2008;36(8):1476–83. https://​doi.​org/​ M, et al. A checklist designed to aid consistency and repro-
10.​1177/​03635​46508​318188 ducibility of GRADE assessments: development and pilot
9. Waldén M, Atroshi I, Magnusson H, Wagner P, Hägglund M. Pre- validation. Syst Rev. 2014;3(1):1–9. https://​doi.​org/​10.​1186/​
vention of acute knee injuries in adolescent female football play- 2046-​4053-3-​82.
ers: cluster randomised controlled trial. BMJ. 2012;344: e3042. 24. Guyatt GH, Oxman AD, Kunz R, Vist GE, Falck-Ytter Y,
https://​doi.​org/​10.​1136/​bmj.​e3042. Schünemann HJ, GRADE Working Group. What is “qual-
10. Steffen K, Myklebust G, Olsen OE, Holme I, Bahr R. Preventing ity of evidence” and why is it important to clinicians? BMJ.
injuries in female youth football—a cluster-randomized controlled 2008;336(7651):995–8. https://​d oi.​o rg/​1 0.​1 136/​b mj.​3 9490.​
trial. Scand J Med Sci Sports. 2008;18(5):605–14. https://d​ oi.o​ rg/​ 551019.​BE.
10.​1111/j.​1600-​0838.​2007.​00703.x. 25. 25. Grant RL. Converting an odds ratio to a range of plausible
11. Soligard T, Myklebust G, Steffen K, Holme I, Silvers H, Bizzini relative risks for better communication of research findings.
M, et al. Comprehensive warm-up programme to prevent injuries BMJ. 2014;348: f7450. https://​doi.​org/​10.​1136/​bmj.​f7450.
in young female footballers: cluster randomised controlled trial. 26. Shor E, Roelfs D, Vang ZM. The “Hispanic mortality paradox”
BMJ. 2008;9(337): a2469. https://​doi.​org/​10.​1136/​bmj.​a2469. revisited: meta-analysis and meta-regression of life-course dif-
12. Emery CA, Meeuwisse WH. The effectiveness of a neuromuscular ferentials in Latin American and Caribbean immigrants’ mor-
prevention strategy to reduce injuries in youth soccer: a cluster- tality. Soc Sci Med. 2017;186:20–33. https://​doi.​org/​10.​1016/j.​
randomised controlled trial. Br J Sports Med. 2010;44(8):555–62. socsc​imed.​2017.​05.​049.
https://​doi.​org/​10.​1136/​bjsm.​2010.​074377. 27. Higgins JPT, Green S. Cochrane handbook for systematic
13. Rössler R, Junge A, Bizzini M, Verhagen E, Chomiak J, Aus der reviews of interventions. Version 5.1.0 [updated March 2011].
Fünten K, et al. A multinational cluster randomised controlled trial The Cochrane Collaboration; 2011.
to assess the efficacy of “11+ Kids”: a warm-up programme to pre- 28. Fawzi WW, Chalmers TC, Herrera MG, Mosteller F. Vitamin A
vent injuries in children’s football. Sports Med. 2018;48(6):1493– supplementation and child mortality: a meta-analysis. JAMA.
504. https://​doi.​org/​10.​1007/​s40279-​017-​0834-8. 1993;269(7):898–903. https://​doi.​org/​10.​1001/​jama.​1993.​03500​
14. Thorborg K, Krommes KK, Esteve E, Clausen MB, Bartels EM, 07007​8033.
Rathleff MS. Effect of specific exercise-based football injury 29. Egger M, Davey-Smith G, Altman D, editors. Systematic reviews
prevention programmes on the overall injury rate in football: a in health care: meta-analysis in context. New York: Wiley; 2008.
systematic review and meta-analysis of the FIFA 11 and 11 + 30. Borenstein M, Hedges LV, Higgins JP, Rothstein HR. Introduction
programmes. Br J Sports Med. 2017;51(7):562–71. https://​doi.​ to meta-analysis. New York: Wiley; 2011.
org/​10.​1136/​bjspo​rts-​2016-​097066. 31. Egger M, Davey Smith G, Schneider M, Minder C. Bias in
15. Al Attar WS, Soomro N, Pappas E, Sinclair PJ, Sanders RH. meta-analysis detected by a simple, graphical test. BMJ.
How effective are F-Marc Injury Prevention programs for soc- 1997;315(7109):629–34. https://​doi.​org/​10.​1136/​bmj.​315.​7109.​
cer players? A systematic review and meta-analysis. Sports Med. 629.
2015;46(2):205–17. https://​doi.​org/​10.​1007/​s40279-​015-​0404-x. 32. Hilska M, Leppänen M, Vasankari T, Aaltonen S, Kannus P,
16. Lemes IR, Pinto RZ, Lage VN, Roch BAB, Verhagen E, Bolling Parkkari J, et al. Neuromuscular training warm-up prevents
C, et al. Do exercise-based prevention programmes reduce non- acute noncontact lower extremity injuries in children’s soccer:
contact musculoskeletal injuries in football (soccer)? A systematic a cluster randomized controlled trial. Orthop J Sports Med.
review and meta-analysis with 13 355 athletes and more than 1 2021;9(4):23259671211005770. https://​doi.​org/​10.​1177/​23259​
million exposure hours. Br J Sports Med. 2021;55(20):1170–8. 67121​10057​69.
https://​doi.​org/​10.​1136/​bjspo​rts-​2020-​103683. 33. van Beijsterveldt AM, van de Port IG, Krist MR, Schmikli SL,
17. López-Valenciano A, Ruiz-Pérez I, Garcia-Gómez A, Vera-Garcia Stubbe JH, Frederiks JE, et al. Effectiveness of an injury preven-
FJ, De Ste CM, Myer GD, et al. Epidemiology of injuries in pro- tion programme for adult male amateur soccer players: a cluster-
fessional football: a systematic review and meta-analysis. Br J randomised controlled trial. Br J Sports Med. 2012;46(16):1114–
Sports Med. 2020;54(12):711–8. https://​doi.​org/​10.​1136/​bjspo​ 8. https://​doi.​org/​10.​1136/​bjspo​rts-​2012-​091277.
rts-​2018-​099577.
848 R. Obërtinca et al.

34. van de Hoef PA, Brink MS, Huisstede BMA, van Smeden M, de Iranian male high-level football (soccer) players: a cluster-ran-
Vries N, Goedhart EA, et al. Does a bounding exercise program domised controlled trial. J Sci Med Sport. 2020;23(5):469–74.
prevent hamstring injuries in adult male soccer players? A cluster- https://​doi.​org/​10.​1016/j.​jsams.​2019.​12.​001.
RCT. Scand J Med Sci Sports. 2019;29(4):515–23. https://d​ oi.o​ rg/​ 40. 40. Riley RD, Higgins JP, Deeks JJ. Interpretation of random
10.​1111/​sms.​13353. effects meta-analyses. BMJ. 2011;342: d549. https://​doi.​org/​10.​
35. Silvers-Granelli HJ, Bizzini M, Arundale A, Mandelbaum BR, 1136/​bmj.​d549.
Snyder-Mackler L. Does the FIFA 11 + injury prevention pro- 41. Crossley KM, Patterson BE, Culvenor AG, Bruder AM, Mos-
gram reduce the incidence of ACL injury in male soccer players? ler AB, Mentiplay BF. Making football safer for women: a
Clin Orthop Relat Res. 2017;475(10):2447–55. https://d​ oi.o​ rg/1​ 0.​ systematic review and meta-analysis of injury prevention pro-
1007/​s11999-​017-​5342-5. grammes in 11 773 female football (soccer) players. Br J Sports
36. Finch CF, Twomey DM, Fortington LV, Doyle TL, Elliott BC, Med. 2020;54(18):1089–98. https:// ​ d oi. ​ o rg/ ​ 1 0. ​ 1 136/ ​ b jspo​
Akram M, et al. Preventing Australian football injuries with a rts-​2019-​101587.
targeted neuromuscular control exercise programme: comparative 42. Wik EH, Lolli L, Chamari K, Materne O, Di Salvo V, Gregson
injury rates from a training intervention delivered in a clustered W, et al. Injury patterns differ with age in male youth football: a
randomised controlled trial. Inj Prev. 2016;22(2):123–8. https://​ four-season prospective study of 1111 time-loss injuries in an elite
doi.​org/​10.​1136/​injur​yprev-​2015-​041667. national academy. Br J Sports Med. 2021;55(14):794–800. https://​
37. Nuhu A, Jelsma J, Dunleavy K, Burgess T. Effect of the FIFA 11 + doi.​org/​10.​1136/​bjspo​rts-​2020-​103430.
soccer specific warm up programme on the incidence of injuries: 43. van der Horst N, Smits DW, Petersen J, Goedhart EA, Backx FJ.
a cluster-randomised controlled trial. PLoS ONE. 2021;16(5): The preventive effect of the Nordic hamstring exercise on ham-
e0251839. https://​doi.​org/​10.​1371/​journ​al.​pone.​02518​39. string injuries in amateur soccer players: a randomized controlled
38. Owoeye OB, Akinbo SR, Tella BA, Olawale OA. Efficacy of the trial. Am J Sports Med. 2015;43(6):1316–23. https://​doi.​org/​10.​
FIFA 11 + warm-up programme in male youth football: a cluster 1177/​03635​46515​574057.
randomised controlled trial. J Sports Sci Med. 2014;13(2):321–8.
39. Zarei M, Abbasi H, Namazi P, Asgari M, Rommers N, Rössler R.
The 11+ Kids warm-up programme to prevent injuries in young

Authors and Affiliations

Rilind Obërtinca1,3 · Ilir Hoxha2,4,5 · Rina Meha1,4 · Arber Lama4,5,6 · Altina Bimbashi4 · Dorentina Kuqi4 ·
Bujar Shabani4,6 · Tim Meyer1 · Karen aus der Fünten1

3
* Rilind Obërtinca Department of Physiotherapy, University of Gjakova “Fehmi
[email protected] Agani”, Gjakova, Kosovo
4
1 Research Unit, Heimerer College, Pristina, Kosovo
Institute of Sports and Preventive Medicine, Saarland
5
University, Saarbrücken Campus, Building B8 2, Evidence Synthesis Group, Pristina, Kosovo
66123 Saarbrücken, Germany 6
University Clinical Center of Kosovo, Pristina, Kosovo
2
Dartmouth Institute for Health Policy and Clinical Practice,
Lebanon, NH, USA