Scand J Med Sci Sports 2010: 20: 814–821 & 2009 John Wiley & Sons A/S

doi: 10.1111/j.1600-0838.2009.00998.x

Exercise program for prevention of groin pain in football players: a

cluster-randomized trial
P. Hölmich1,3, K. Larsen2, K. Krogsgaard2, C. Gluud3
1
Department of Orthopaedic Surgery, Orthopaedic Research Centre, Amager Hospital, Copenhagen University Hospital,
Copenhagen, Denmark, 2Clinical Research Centre, Hvidovre Hospital, Copenhagen, Denmark, 3Copenhagen Trial Unit, Centre for
Clinical Intervention Research, Rigshospitalet, Copenhagen, Denmark
Corresponding author: Per Hölmich, Department of Orthopaedic Surgery, Orthopaedic Research Centre, Amager Hospital,
Copenhagen University Hospital, Italiensvej 1, DK-2300 Copenhagen S, Denmark. Tel: 10045 32 34 32 63, Fax: 10045 32 34
39 95, E-mail: [Link]@[Link]
Accepted for publication 8 June 2009

Groin injuries cause major problems in sports and particularly tion, and core stability exercises for the muscles related to
in football. Exercise is effective in treating adductor-related the pelvis. Physiotherapists assigned to each club registered
groin pain, but no trials have been published regarding the all groin injuries. Twenty-two clubs in each group comple-
specific prevention of groin pain or prevention specifically ted the study, represented by 977 players. There was no
targeting overuse injuries in sport using exercise programs. significant effect of the intervention (HR 5 0.69, 95% CI
We performed a cluster-randomized trial including 55 foot- 0.40–1.19). The risk of a groin injury was reduced by 31%,
ball clubs representing 1211 players. The clubs were rando- but this reduction was not significant. A univariate analysis
mized to an exercise program aimed at preventing groin showed that having had a previous groin injury almost doubles
injuries (n 5 27) or to a control group training as usual the risk of developing a new groin injury and playing at a
(n 5 28). The intervention program consisted of six exercises higher level almost triples the risk of developing a groin
including strengthening (concentric and eccentric), coordina- injury.

Groin pain in athletes is a major problem in sports tor-related groin injuries in athletes than traditional
medicine, and among football players it is a serious ‘‘passive’’ physiotherapy (Hölmich et al., 1999). The
and very common injury, which may end the sports training program resulted in return of 80% of the
career. In a study of long-standing adductor-related players to the previous level of sport without any
groin pain, 72% of the athletes had ceased to groin pain (Hölmich et al., 1999). Based on our
participate in sport because of the groin pain (Höl- experiences with this training program, we have
mich et al., 1999). developed a new training program with the aim of
The annual frequency of groin injuries is 8–18% in preventing groin injuries in connection with football.
football (Hölmich, 1998; Ekstrand & Hilding, 1999) The program comprises strength and coordination
and 20% in ice hockey (Emery et al., 1999). Treat- exercises focusing on the muscles related to the
ment of these injuries may take months and is costly pelvis. It also includes core stability as well as
both for the player and for the health service system. eccentric exercises. The purpose of this cluster-ran-
Prevention would be beneficial for the athletes as well domized clinical trial was to investigate the effect of a
as for the society. specific training program aimed at preventing groin
Training as rehabilitation has been widely used for injuries in male football players. The effect of the
many years. In the past 10 years, using training with program was evaluated by comparing an interven-
specific exercises as a treatment has been a rising tion group with a control group (CG) that received
issue in sports medicine as well as in the general warmup and training as usual.
treatment of musculo-skeletal overuse pain (Alfred-
son et al., 1998; Young et al., 2005; Slade et al.,
2006). In a randomized trial, we have shown that a Material and methods
training program including strength training of the Study population
adductor, abdominal and low-back muscles, com-
In cooperation with the Danish Football Federation (DBU),
bined with coordination and balance exercises, was 120 football clubs from the Denmark, Copenhagen, and
significantly better in treating long-standing adduc- Zealand series and series 1–3 were invited to participate. The

814
 Exercise program for prevention of groin pain in football players
clubs represented both urban as well as non-urban clubs. The both eccentrically and concentrically. The stretching exercise
football players were amateurs training and playing at a for the iliopsoas (no. 6) was included, as pain and tightness of
competitive level. The number of training sessions each week this muscle in our clinical experience (Hölmich, 2007) is often
varied between 2 and 4; generally the higher the level, the more a significant problem. For such a program to be implemented,
the training sessions. The coaches and managers of the eligible it is our belief that it must be easy to perform without the need
clubs were invited to a number of meetings, where detailed for additional equipment, fast to carry out, and adoptable
information was given regarding the trial. The Ethics Com- within the regular football warmup.
mittee of Copenhagen and Frederiksberg Municipalities and The exercises were
The Danish Data Protection Agency approved the study (KF
01-171-97 and 1997-1200-271). All players and trainers gave 1. Isometric adduction against a football placed between
written informed consent before entering the study. The trial is the feet when lying supine, the first toe pointed straight
registered at [Link] as NCT00226603. forward and pressure against the ball as hard as
possible; each adduction 10 s, five repetitions (stimulat-
ing the adductor muscle group and the torso-to-pelvis
Design stability).
2. Isometric adduction against a football placed between
Randomization at the individual level could not be carried the knees when lying supine with flexed hips and knees
out, as we feared spillover to the controls. We therefore chose and feet flat on the surface, the first toe pointed straight
a cluster randomization design, where the football club was forward and pressure against the ball as hard as
the unit of randomization. A physiotherapist was allocated to possible; each adduction 10 s, five repetitions (stimulat-
each club before randomization. The physiotherapist was in ing the adductor muscle group and the torso-to-pelvis
cooperation with the coach, responsible for collecting data stability).
and to report to the data manager. The randomization was 3. Combined abdominal sit-up and hip flexion, starting
carried out as cluster randomization, with each club being a from the supine position and with a football placed
cluster. Before randomization, the clubs were stratified ac- between the knees (‘‘folding knife exercise’’); two series
cording to level: (i) Denmark, Copenhagen, and Zealand series of 20 repetitions performed rhythmically in a slow pace
or (ii) series 1–3 and according to geography: (i) urban or (ii) (stimulating the coordination of both eccentric as well
non-urban. Because of the nature of the intervention, blinding as concentric work of the torso muscles and the hip- and
of the participants and observers (physiotherapist and coach) pelvis-related muscles).
was not possible. 4. One-leg coordination exercise flexing and extending the
Before the start of the 1997/1998 season, 78 clubs had knee and swinging the arms in the same rhythm (‘‘cross-
accepted to participate in the trial, and they were randomized country skiing on one leg’’); 1 min two times for each leg
to the prevention group (PG) or the CG by block randomiza- (stimulating the neuromuscular coordination utilizing
tion (block size two). The randomization was computer the principles of core stability including the muscles of
generated. The individual physiotherapists and coaches were the torso as well as of the lower extremities).
informed about the allocation of their club by a letter in a 5. Hip adduction against a partner’s hip abduction. In the
sealed and opaque envelope mailed by a secretary not involved sitting position, supported by the hands placed on the
in the analysis of the data. The data manager, the statistician, ground behind the trunk, one player places his straight
and the authors were all blinded to the result of the randomi- leg with the feet and lower shin on the outside of his
zation. partner’s feet and lower shin. He adducts as the partner
The clubs and the players were invited to participate in the abducts eccentrically and slowly gets his feet pressed
trial before the allocation, and both the players as well as the together. He then abducts concentrically as the partner
coaches gave written informed consent. Before initiating the adducts eccentrically and is slowly brought into abduc-
trial, the physiotherapists were trained in the use of the tion. They work like this for 1 min and then shift
questionnaire and in how to perform the intervention program positions; two series of 1 min in both positions (stimu-
correctly. The physiotherapists collected a self-administered lating concentric and eccentric strength of both the
questionnaire form from each included player providing adductor and the abductor muscles).
information about the demographic data, type of work, foot- 6. Stretching of the iliopsoas muscle with a standardized
ball experience, playing position, participation in other sports, technique; 20 s stretch for each leg repeated twice
and previous groin injuries. The coaches and the players of the (securing the length of the iliopsoas muscle, given the
clubs allocated to the PG were instructed on how to perform common secondary involvement of the muscle).
the exercises by the physiotherapist. The questionnaires were
given to the players before information of allocation. At the
next visit, the physiotherapist collected the questionnaires and The duration of the exercise program was approximately
the instructions regarding the allocation were given. The 13 min and the exercises were used as an integrated part of the
season (from September 13, 1997 to July 5, 1998) included warmup exercises before every regular football practice.
the 33 weeks of active playing and 9 weeks around Christmas The experimental and CGs were warming up and training
with no training or matches. A number of circumstances, not in the traditional way according to the standards of the coach
related to the study, resulted in an unfortunate delay in education program of the DBU ([Link] The
reporting of the trial. coaches in all clubs registered the number of hours and
sessions of training (including any intervention) and the
number of matches. The registration was carried out at the
team level. The physiotherapist also checked at random
Intervention whether the intervention program was performed correctly.
The intervention program consisted of six exercises; four of All injuries were reported to the physiotherapist, who attended
those were also part of the previously described treatment the club at least every second week and at shorter intervals if
program (Hölmich et al., 1999). Exercise no. 5 was specifically needed because of injuries. The injury was assessed by the
developed as an in-field partner exercise that should physiotherapist to determine whether it was a groin injury,
strengthen the adductor and abductor muscles of the femur and if so, it was recorded. When a player was recorded with a

815
Hölmich et al.
groin injury, he was censored, i.e., no player could be recorded clubs could not cope with the task and withdrew
with more than one injury. immediately without including any players: 12 clubs
from the experimental and 11 clubs from the CG.
During the trial, another 11 clubs withdrew: five
Injury definition
from the experimental and six from the CG (Fig. 1).
An injury was defined as any physical complaint that is the
The reasons were too much work registering the
result of participating in football training or a football
match incapacitating the player when playing football and/ injuries, bad results in the tournament, the coach
or demanding special medical attention for the player to be did not want anything ‘‘irrelevant’’ to disturb the
able to participate or preventing him from participating in players, or change of coach during the trial period
training or a match. We later found this definition to be in and the new coach did not want to participate. Forty-
concordance with the consensus statement by the Injury
four clubs, 22 clubs in each group, completed the
Consensus Group under the auspices of Federation Interna-
tionale de Football Association published in 2006 (Fuller trial, represented by 977 players, of whom 907 pre-
et al., 2006). sented with complete data. The baseline characteris-
tics are shown in Table 1.
The analysis of the number of groin injuries at the
Sample size club level reveals no within-club correlation
With an intracluster coefficient estimated to be 0.08 (Corn- (w2 5 35.46, df 5 44 4 5 40, P 5 0.67, conditional
field, 1978; Donner et al., 1981) and a cluster size of 20 players on the three cluster-level covariates, and w2 5 45.83,
and a power of 80% and a significance level of 5%, we
calculated that 80 clubs were needed to detect a reduction of
df 5 44 1 5 43, and P 5 0.36 for the unconditional
50% from 10% to 5% in the occurrence of a groin injury. analysis), i.e., the intraclass correlation equals zero.
Therefore, the cluster sampling may be disregarded
in the analysis of the primary response variable: time
Outcome measure and statistical analysis until first groin injury.
Double data entry was performed and the data management Results from the univariate analysis of each of the
were unaware of the allocation. covariates on the 907 complete cases are shown in
The primary outcome measure was the time until the first Table 2. There was no significant effect of the
groin injury. This is a rather complex response variable to
analyze in the presence of clustering, because it is not normally intervention (HR 5 0.69, P 5 0.18), although the
distributed, and it may be censored. Therefore, the approach estimate suggests a 31% smaller hazard in the inter-
adopted here was to first investigate whether clustering of vention group compared with the CG. Kaplan–
groin injuries at the club level was at all present, and then Meier estimates are shown for the two groups in
subsequently choose a proper analysis for the primary re- Fig. 2.
sponse variable, depending on whether the groin injuries
cluster. An investigation of clustering was carried out at the Having had a previous groin injury almost doubles
aggregated club level in a Poisson model with the number of the hazard of developing a new groin injury
groin injuries within the club as the response variable, and the (HR 5 1.97, P 5 0.015), and playing at a higher level
total number of weeks under risk, summed over players within almost triples the hazard of developing a groin injury
club, as the offset variable. Thus, the expected number of groin (HR 5 2.58, Po0.001). There was no effect of the
injuries is assumed to be proportional to the total time under
risk. The hypothesis of ‘‘no within club correlation’’ was then type of work (P 5 0.91), age (P 5 0.50), or the posi-
tested using the chi-square distributed deviance (McCullagh & tion played on the field (P 5 0.65) of the player on
Nelder, 1989). the time until groin injury, and it did not matter
If there was no within-club correlation at the club level, whether the club was urban or non-urban (P 5 0.09).
then the clustering may be disregarded in the analysis of the The results from the same analysis carried out on the
primary endpoint: time until first groin injury. Tests for
marginal associations between each of the covariates and the 977 available cases did show any relevant differences.
time until groin injury were carried out using log-rank tests. The results from the multiple regression analysis
Cox’s proportional hazards model (Cox, 1972) was used for on the covariates including the 907 players with no
multiple regression analysis of time until groin injury, and missing information are shown in Table 3. Exactly
likelihood ratio tests were used for hypothesis testing, includ- the same variables were significant as in the univari-
ing tests for interaction between covariates. The assumption of
proportional hazards was investigated using the chi-square ate analyses: having had an injury previously and
distributed test suggested by Grambsch and Therneau (1994). playing at a high level increase the hazard
The significance level was 5% in all hypothesis testing. (HR 5 1.95, P 5 0.017 and HR 5 2.56, Po0.001),
All analyses were carried out in the statistical package R while neither the intervention nor the other covari-
(R Development Core Team, 2006). ates had a significant effect on the time until groin
injury.
Results To investigate the fit of the model, tests for
interaction between variables with a significant
Of the 120 eligible football clubs, 78 clubs accepted main effect on the outcome, time until groin injury,
to participate, but when the inclusion of players and were carried out, and the assumption of proportional
the implementation of exercises had to begin, 23 hazards was investigated using the test suggested by

816
 Exercise program for prevention of groin pain in football players

120 eligible football

clubs invited to
participate

42 football clubs
declined to participate

78 football clubs
randomised

39 clubs allocated to 39 clubs allocated to

prevention group (PG) control group (CG)

12 clubs withdrawn 11 clubs withdrawn

without without
including players including players

27 clubs with 647 players 28 clubs with 562 players

active in the PG active in the CG

5 clubs with 123 players 6 clubs with 109 players

withdrawn during study withdrawn during study
period period

22 clubs with 524 22 clubs with 453 players

players completed the completed the trial

Fig. 1. Trial profile.

Grambsch and Therneau (1994) for the variables that sprains (Wedderkopp et al., 2003; Verhagen et al.,
showed a significant main effect on the outcome. 2004; Olsen et al., 2005).
None of the tests were significant. The 31% reduction in risk found in this study is
not significant (P 5 0.18). We based our sample size
estimation on a relative risk reduction of 50% based
on our experience with the treatment effect of a
Discussion similar training program used for treating groin
injuries (Hölmich et al., 1999). This may have been
Prevention specifically targeting overuse injuries in too optimistic. Second, we had anticipated that 10%
sport using exercise programs has, to our knowl- of the football players in the CG would develop a
edge, not yet been the subject of any pub- groin injury. In fact, only 8% did so. Third, we
lished randomized clinical trial to date, nor have experienced even larger difficulties with getting clubs
we been able to identify any randomized clinical to participate and remain in the trial than antici-
trial aimed specifically at the prevention of groin pated. These factors all increase our risk of type II
injuries. errors, i.e., risk of overlooking a real intervention
A number of studies have been published evaluat- effect. The compliance of the players to perform the
ing prevention programs for acute injuries as anterior exercises with both the intended frequency and the
cruciate ligament ruptures (Caraffa et al., 1996; intended intensity could also be a problem. We chose
Myklebust et al., 2003; Olsen et al., 2005) and ankle the best possible solution and had the coaches super-

817
Hölmich et al.
Table 1. Baseline characteristics of 907 players presenting with a Table 2. Log rank tests and maximum likelihood estimates from the
complete dataset univariate Cox regression analysis of the 907 complete cases

Prevention group Control group Variable HR 95% confidence P value

(N 5 477) (N 5 430) interval

Median age (years) 24.49 24.62 Intervention 0.18

Previous groin injury No 1
Yes 183 203 Yes 0.69 [0.40–1.19]
No 294 207 Age 0.50
Level of play Per additional year 0.98 [0.92–1.04]
High 108 130 Previous groin injury 0.015
Low 369 300 No 1
Location of club Yes 1.97 [1.13–3.42]
Urban 431 357 Level of play o0.001
Non-urban 46 73 Low 1
Physical work High 2.58 [1.49–4.46]
Not working 117 75 Location of club 0.09
Not physically 129 118 Urban 1
demanding Non-urban 1.80 [0.90–3.60]
Slightly demanding 83 61 Physical work 0.91
Demanding 99 100 Not working 1
Very demanding 49 76 Not physically demanding 0.89 [0.38–2.05]
Position on field Slightly demanding 1.28 [0.53–3.07]
Mixed 90 86 Demanding 1.21 [0.53–2.78]
Goalkeeper 42 32 Very demanding 1.07 [0.41–2.80]
Defence 120 102 Position on field 0.65
Midfield 157 152 Mixed 1
Striker 68 58 Goalkeeper 0.68 [0.22–2.07]
Defence 0.85 [0.41–1.79]
Midfield 0.63 [0.30–1.33]
Striker 0.52 [0.19–1.47]
vise and register the participation in the prevention
program. A trial including a much lager sample size,
as we originally intended, is needed to definitely test reporting time, which was due to a high number of
the beneficial and harmful effects of our preventive competing tasks.
training program. With the number of players who Twenty-three clubs withdrew before including
completed this study, no significant effect of the players. This was a matter of clubs agreeing to
prevention program could be shown. The 31% participate, but when faced with the start of the
reduction of the risk of developing a groin injury tournament they could not cope with the task.
would, however, if significant, be a considerable Because this group was evenly distributed bet-
advantage that would make it worthwhile for the ween the two allocations, we do not find this alarm-
football players to complete the program. ing from a trial quality point of view but very
The strength of this study is that the randomiza- unfortunate from a sample size point of view. The
tion was computer generated and that the allocation dropout after entering the study with players was
was unknown to the data manager, the statistician, 11/55 clubs (20%) and 232/1209 players (19%)
and the authors, thus preventing allocation bias. The without difference between the two groups.
registration and examination techniques used were The risk of sustaining a groin injury is almost
systematic and uniform in order to prevent registra- doubled if the player had a previous groin injury.
tion bias. This is the first study to evaluate prevention This is in concordance with the results of previous
strategies aimed at groin injuries in athletes. The studies on other injuries in sports (Arnason et al., 2004;
limitations of this study were that a large number of Petersen & Holmich, 2005; Hagglund et al., 2006).
eligible clubs did not participate and that a number Playing football at the highest level (Denmark,
of clubs that were randomized exited the study but Zealand, and Copenhagen series) compared with the
without including players. However, no sign of other levels included in this study almost tripled the
allocation bias was found when comparing the base- risk of a groin injury. This is the first time it is shown
line characteristics. A risk of assessment bias because that the risk of sustaining a groin injury increases with
the registration of the injuries was not blinded due to the competition level. This is probably not the result
the nature of the intervention and the practical of more training hours but may be the result of a
circumstances. Missing data in some cases could higher intensity in the training and match play. In
lead to attrition bias but no sign of this was found studies examining hamstring injuries, the risk of
when comparing the full data with the complete data. sustaining an injury increases with the age of the
A drawback of this trial is the extraordinarily long- athlete (Petersen & Holmich, 2005). This is not the

818
 Exercise program for prevention of groin pain in football players
Time to groin injury Table 3. Likelihood ratio tests and maximum likelihood estimates from
the multiple Cox regression analysis on the 907 complete cases
experienced a groin injury 0.08 No intervention
P value
Probability of having

Variable HR 95% confidence

interval
0.06 Intervention
Intervention 0.44
0.04 No 1
Yes 0.80 [0.46–1.40]
0.02 Age 0.59
Per additional year 0.98 [0.92–1.05]
Previous groin injury 0.017
0.00 No 1
0 5 10 15 20 25 30 Yes 1.95 [1.12–3.40]
Number of weeks since start of tournament Level of play o0.001
Low 1
Fig. 2. Time to groin injury. High 2.56 [1.48–4.42]
Location of club 0.50
Urban 1
Non-urban 0.75 [0.33–1.72]
case in the present study. There is no evident explana- Physical work 0.85
tion to this, but it could be speculated that the injuries Not working 1
Not physically demanding 0.80 [0.35–1.85]
are of a different nature (hamstring injuries are Slightly demanding 1.16 [0.53–2.78]
primarily acute muscle strains whereas groin injuries Demanding 1.24 [0.54–2.82]
are primarily overuse injuries) and as such caused by Very demanding 0.99 [0.38–2.61]
different etiologies not necessarily influenced by age. Position on field 0.56
Mixed 1
The present study is the result of an analysis of risk Goalkeeper 0.71 [0.23–2.19]
factors and incidence. According to previously de- Defence 0.90 [0.43–1.89]
scribed guidelines for the prevention of injuries (van Midfield 0.60 [0.29–1.26]
Striker 0.52 [0.19–1.46]
Mechelen et al., 1992), we have identified the inci-
dence and risk factors of groin injuries from the The tests and parameter estimates for the statistically significant variables
available literature, we have aimed at developing a are provided in a model with those variables only. The test and parameter
program that is likely to reduce the risk of groin estimates for statistically insignificant variables are corrected for the
injury, and we have evaluated this in a randomized effect of the significant variables.
clinical trial. Likewise, a study on acute knee and
ankle injuries in youth team handball (Olsen et al.,
2005) has shown a preventive effect of a general et al., 2004). The preventive program tested is derived
program including warm-up, technique, balance, from that treatment program and includes concentric
and strengthening exercises. The program was the as well as eccentric strengthening and coordination
result of an analysis of the known risk factors exercises for the muscles related to the pelvis, with
focusing on control of the knees and ankles during special emphasis on the adductor and abductor
pivoting and landing movements. A preventive pro- muscles.
gram utilizing exercises would probably not have Groin pain may be the result of a number of
effect from the first day, as the idea is to strengthen causes, the majority of them being related to muscles
certain physical abilities and thereby prevent injuries. and tendons, and often more than one cause can be
This will take some time, but it is not known how identified. A large and important list of differential
much. We started data collection at the same time as diagnosis should be kept in mind by anyone involved
the program was initialized and as can be seen in in the medical handling of these athletes. Groin pain
Fig. 2, it took a couple of weeks before a difference primarily related to hip joint disorders such as
between the two curves became apparent. femoro-acetabular impingement (FAI), arthritis,
The adductor-, iliopsoas-, and inguinal-related labral lesions, and others might not benefit directly
groin injuries are the most common causes of groin from a program like the one tested in our study.
pain in athletes (Hölmich et al., 2003; Hölmich, Another possible risk factor is the flexibility of the
2007), and strengthening the related muscles could adductor muscles, but neither a study of football
potentially be beneficial to avoid injury. Strength has (Witvrouw et al., 2003) nor a study of ice hockey
been indicated as a possible risk factor for adductor- (Emery & Meeuwisse, 2001) could show any influ-
related injuries in a study of ice hockey players (Tyler ence of adductor flexibility on the incidence of
et al., 2001). The present study utilizes the idea that adductor-related groin injuries.
the exercises of a treatment program (Hölmich et al., Eccentric exercises have been emphasized as being
1999) can potentially serve as the basis for preventing of major value in the treatment of tendon-related
that same injury (Wedderkopp et al., 2003; Verhagen overuse injuries (Fyfe & Stanish, 1992; Alfredson

819
Hölmich et al.
et al., 1998; Young et al., 2005). The exercises used in ness of physical activity used as both treatment and
the present study include eccentric strengthening in prevention of a large number of medical diseases
exercises number 3, 4, and especially 5,and the such as hypertension, diabetes, overweight, and
eccentric exercises were also utilized in the original others (Andersen, 2007; Church et al., 2007) has
treatment protocol for adductor-related groin pain substantially increased the problem of overuse in-
(Hölmich et al., 1999). The explanation for the juries, as a large number of new ‘‘athletes’’ with little
beneficial effect of eccentric exercises on tendon- or no experience are taking up sport. This underlines
related problems is not clear (Fyfe & Stanish, 1992; the need for prevention studies in order to avoid an
Rees et al., 2006). epidemic of athletic overuse injuries interfering with
Core stability (core strengthening) is another mod- the aim to prevent and treat lifestyle diseases. Groin
ality that has developed in the last 10 years into an injuries continuously constitute a major overuse
important element in the exercise treatment of mus- problem. Further studies into prevention of groin
culo-skeletal problems. Core stability is defined as the injuries are warranted. However, as this field is still
ability to control the position and motion of the trunk developing and new diagnose are emerging (FAI as
over the pelvis to allow optimum production, trans- an example), the design of the ‘‘next’’ study should be
fer, and control of force and motion to the terminal adequately adapted to this. The size of the study is
segment in integrated athletic activities (Akuthota & also an important and very demanding subject.
Nadler, 2004; Kibler et al., 2006). The exercises Perhaps more specific subgroups of groin injuries
utilized in the original treatment study as well as in should be targeted.
the prevention program of the present study includes
this principle, and especially exercises number 3 and 4 Key words: cluster-randomized trial, football, groin
stimulate the core stability both regarding the lumbar injury, overuse injury, prevention.
spine and the pelvis, as the trunk position must be
stabilized to perform the exercises correctly.
The present cluster-randomized clinical trial ad-
dressed a very common overuse injury using football Acknowledgements
players as the target group. A simple program based
on experience and the available literature including All the members of The Groin Pain Trial Study Group are
acknowledged for their important work of registering and
strength (concentric and eccentric), balance, and reporting from the participating football clubs: P. Akselbo,
core-stability exercises was used. The risk of sustain- J. Andersen, A. G. Andreasen, J. A. Pedersen, A. Bach,
ing a groin injury was decreased by 31%, but this M. Dall-Jepsen, B. Engdahl, N. Erichsen, J. Fredslund,
reduction was not significant. The risk of sustaining a A. G. Friis-Hansen, F. Hansen, G. Jensen, P. Johansen,
groin injury was significantly increased when the S. Jørgensen, M. Klausen, S. Knudsen, J. Kock,
L. D. Kristensen, S. Kristensen, R. Kromose, M. L. Larsen,
athlete had been groin injured previously. Playing T. Lund, C. Neergaard, B. Petersen, A-K. Pradel, M. Quvang,
at a higher level also increased the risk significantly. M. Reimer, A. Rosengren, N. B. Schmidt, L. Schrøder,
B. Skånstrøm, I. Sonne, D. Thorsen, L. Ulnits, G. Vestergaard,
H. Westen, M. Østergaard, and A-M. Aage.
Perspectives We thank the coordinating secretary Dorte Fischer for her
effort to keep track of all communications with the study
Overuse injuries are a major problem in society at group. We thank Kate Whitfield for her help in editing the
large and in sports specifically. The increasing aware- language.

Reference
Akuthota V, Nadler SF. Core Caraffa A, Cerulli G, Projetti M, Aisa G, Cornfield J. Randomization by group: a
strengthening. Arch Phys Med Rehab Rizzo A. Prevention of anterior formal analysis. Am J Epidemiol 1978:
2004: 85: S86–S92. cruciate ligament injuries in soccer. A 108: 100–102.
Alfredson H, Pietila T, Jonsson P, prospective controlled study of Cox D. Regression models and life tables
Lorentzon R. Heavy-load eccentric calf proprioceptive training. Knee Surg (with discussion). J R Stat Soc B 1972:
muscle training for the treatment of Sports Traumatol Arthrosc 1996: 4: 74: 187–220.
chronic Achilles tendinosis. 19–21. Donner A, Birkett N, Buck C.
Am J Sports Med 1998: 26: 360–366. Church TS, Earnest CP, Skinner JS, Blair Randomization by cluster. Sample size
Andersen LB. Physical activity and SN. Effects of different doses of requirements and analysis. Am J
health. BMJ 2007: 334: 1173. physical activity on cardiorespiratory Epidemiol 1981: 114: 906–914.
Arnason A, Sigurdsson SB, fitness among sedentary, overweight or Ekstrand J, Hilding J. The incidence and
Gudmundsson A, Holme I, obese postmenopausal women with differential diagnosis of acute groin
Engebretsen L, Bahr R. Risk factors elevated blood pressure: a randomized injuries in male soccer players.
for injuries in football. Am J Sports controlled trial. JAMA 2007: 297: Scand J Med Sci Sports 1999: 9:
Med 2004: 32: 5S–16S. 2081–2091. 98–103.

820
 Exercise program for prevention of groin pain in football players
Emery CA, Meeuwisse WH. Risk factors medicine. Blackwell Science Ltd, back pain: a systematic review. J
for groin injuries in hockey. Med Massachusetts, USA, 2003: 616–637. Manipulative Physiol Ther 2006: 29:
Sci Sports Exerc 2001: 33: 1423– Hölmich P, Uhrskou P, Ulnits L, 163–173.
1433. Kanstrup I-L, Nielsen MB, Bjerg AM, Tyler TF, Nicholas SJ, Campbell RJ,
Emery CA, Meeuwisse WH, Powell JW. Krogsgaard K. Effectiveness of active McHugh MP. The association of hip
Groin and abdominal strain injuries in physical training as treatment for long- strength and flexibility with the
the National Hockey League. Clin J standing adductor-related groin pain in incidence of adductor muscle strains in
Sport Med 1999: 9: 151–156. athletes: randomised trial. Lancet 1999: professional ice hockey players.
Fuller CW, Ekstrand J, Junge A, 353: 439–443. Am J Sports Med 2001: 29:
Andersen TE, Bahr R, Dvorak J, Kibler WB, Press J, Sciascia A. The role 124–128.
Hagglund M, McCrory P, Meeuwisse of core stability in athletic function. van Mechelen W, Hlobil H, Kemper HC.
WH. Consensus statement on injury Sports Med 2006: 36: 189–198. Incidence, severity, aetiology and
definitions and data collection McCullagh P, Nelder JA. Generalized prevention of sports injuries. A review
procedures in studies of football linear models. London, UK: of concepts. Sports Med 1992: 14: 82–
(soccer) injuries. Scand J Med Sci Chapmann & Hall, 1989. 99.
Sports 2006: 16: 83–92. Myklebust G, Engebretsen L, Braekken Verhagen E, van der BA, Twisk J, Bouter
Fyfe I, Stanish WD. The use of eccentric IH, Skjolberg A, Olsen OE, Bahr R. L, Bahr R, van Mechelen W. The effect
training and stretching in the treatment Prevention of anterior cruciate of a proprioceptive balance board
and prevention of tendon injuries. Clin ligament injuries in female team training program for the prevention of
Sports Med 1992: 11: 601–624. handball players: a prospective ankle sprains: a prospective controlled
Grambsch P, Therneau T. Proportional intervention study over three seasons. trial. Am J Sports Med 2004: 32: 1385–
hazards tests and diagnostics based on Clin J Sport Med 2003: 13: 71–78. 1393.
weighted residuals. Biometrika 1994: Olsen OE, Myklebust G, Engebretsen L, Wedderkopp N, Kaltoft M, Holm R,
81: 515–526. Holme I, Bahr R. Exercises to prevent Froberg K. Comparison of two
Hagglund M, Walden M, Ekstrand J. lower limb injuries in youth sports: intervention programmes in young
Previous injury as a risk factor for cluster randomised controlled trial 11. female players in European handball –
injury in elite football: a prospective BMJ 2005: 330: 449–456. with and without ankle disc.
study over two consecutive seasons. Br Petersen J, Holmich P. Evidence based Scand J Med Sci Sports 2003: 13:
J Sports Med 2006: 40: 767–772. prevention of hamstring injuries in 371–375.
Hölmich P. Adductor related groin pain sport. Br J Sports Med 2005: 39: 319– Witvrouw E, Danneels L, Asselman P,
in athletes. Sports Med Arthrosc Rev 323. D’Have T, Cambier D. Muscle
1998: 5: 285–291. R Development Core Team. R: a flexibility as a risk factor for developing
Hölmich P. Long-standing groin pain in language and environment for muscle injuries in male professional
sportspeople falls into three primary statistical computing. Vienna, Austria: soccer players. A prospective study.
patterns, a ‘‘clinical entity’’ approach: a R Foundation for Statistical Am J Sports Med 2003: 31: 41–46.
prospective study of 207 patients. Br J Computing, 2006. Young MA, Cook JL, Purdam CR, Kiss
Sports Med 2007: 41: 247–252. Rees JD, Wilson AM, Wolman RL. ZS, Alfredson H. Eccentric decline
Hölmich P, Renström P, Saartok T. Hip, Current concepts in the management of squat protocol offers superior results at
groin and pelvis. In: Kjaer M, tendon disorders. Rheumatology 12 months compared with traditional
Krogsgaard M, Magnusson P, (Oxford) 2006: 45: 508–521. eccentric protocol for patellar
Engebretsen L, Roos H, Takala T, Slade SC, Ther MM, Keating JL. Trunk- tendinopathy in volleyball players. Br J
Woo S L-Y, eds. Textbook of sports strengthening exercises for chronic low Sports Med 2005: 39: 102–105.

821