You Can't Prevent Injury; You Can Only Reduce the Risk
"Injury prevention" is the wrong word
No programme, screen or exercise can prevent injury. The best we can honestly offer an athlete is a lower probability of getting hurt, and mostly for non-contact injuries.
Our industry sells certainty. "Injury prevention programme", "bulletproof your knees", "fix your movement before it breaks you". Coaches, parents and athletes hear those phrases and it helps perpetuate the myth of moving ‘wrong’. When a player then tears a hamstring in week three, the programme looks like it failed, or worse, the player gets blamed for "not doing their correctives" or their leg length/rotated pelvis is to blame.
The language matters because it shapes what we assess, what we prescribe and what we tell people. "Risk reduction" is less catchy, but it is more accurate. This post explains why, using Willem Meeuwisse's injury causation model, and why the older idea that faulty movement patterns cause injury, best known through Shirley Sahrmann's kinesiopathological model, no longer holds up as an explanation we should build programmes on.
Meeuwisse: injury is a chain, not a cause
Meeuwisse's model shows that an injury needs a stack of circumstances to line up, and the last link in that chain is usually outside the athlete's control.
In 1994 Willem Meeuwisse proposed a multifactorial model of injury causation. In 2007 he and colleagues updated it into the dynamic, recursive model that most of us now teach. It has four layers:
1. Intrinsic risk factors: what the athlete brings: age, previous injury, strength, tissue capacity, fitness. This makes them predisposed.
2. Extrinsic risk factors: what the environment adds: opponents, equipment, playing surface, weather, rules, fixture density. Exposed to these, the athlete becomes susceptible.
3. The inciting event: the mechanism, whether the tackle, the mistimed landing, the slip on a wet pitch or the sprint at minute 85.
4. Recursion: every session changes the athlete. Exposure can produce adaptation, maladaptation, injury, or full or partial recovery, and the athlete then returns to the cycle with a new risk profile.
The key point is the 2007 paper's own framing: injury is usually preceded by a chain of shifting circumstances that only together constitute a sufficient cause. No single factor, whether a weak glute, a valgus knee or a tight hip flexor, is the cause.
Two things follow for practice:
• Risk is never zero. Even an athlete with every intrinsic factor optimised still steps onto a pitch with opponents, a surface, weather and a task that demands extreme positions at speed.
• Risk is never fixed. A player screened in preseason is not the same player in March. Load, fatigue, sleep, an argument with the other half, a previous niggle and the fixture list have all affected capacity on the day.
Later work, notably Bittencourt and colleagues' 2016 complex systems approach, argued that even the recursive model underplays how factors interact. Injury emerges from a web of determinants rather than a list of risk factors. That makes prediction at the individual level harder still, and it reinforces the conclusion: we can shift probabilities, not guarantee outcomes.
Contact and noncontact: what training can actually touch
Training changes the athlete, so it mainly moves the risk of noncontact injury. It does very little about the person running into them.
Think about what sits in each layer of Meeuwisse's model and who controls it:
Intrinsic factors
Examples: strength, tissue capacity, sprint exposure, fitness, previous injury.
Can we train it? Largely yes (except age and injury history).
Extrinsic factors
Examples: opponents, referee, surface, weather, boots, fixture congestion.
Can we train it? Mostly no; partly through rules, equipment and scheduling.
Inciting event
Examples: tackle, collision, slip, being landed on, a forced change of direction.
Can we train it? Rarely; technique helps at the margins.
In collision sports you can’t control the majority of extrinsic factors. In South African Super Rugby teams over five seasons, 79% of match injuries happened in contact situations and 54% in the tackle. Strengthening a winger's hamstrings does nothing about a late tackle that catches you off guard. Research on contact events in rugby points to tackle technique and strict law enforcement, not gym work, as the main controls for those injuries (English community rugby study).
Even "non-contact" injuries are not fully in our hands. A sprinter's hamstring at the end of a congested week, an ACL in a player who has to cut on a wet, uneven pitch to avoid an opponent, a calf in the cold: the task, the surface, fatigue, preparedness and the weather are all part of the puzzle.
So the honest framing for athletes is this. We can make you more robust, meaning better able to tolerate the forces your sport throws at you. We cannot make you invulnerable, and we cannot control the opponent, the ball, the pitch or the weather.
Why the kinesiopathological model is out of date
Sahrmann's model was a genuine step forward in its time, but its core claim, that "faulty" movement causes pain and injury and that correcting it fixes or prevents them, has not survived testing.
Shirley Sahrmann's Movement System Impairment (MSI) approach rests on the kinesiopathological model. Its basic premise is that repeated movements and sustained alignments in daily life induce tissue pathology. Muscles become relatively stiff or flexible, a joint takes the path of least resistance, and that "directional susceptibility" eventually becomes painful. The treatment logic follows: classify the faulty pattern, then retrain it.
It deserves credit. It moved physiotherapy away from purely structural diagnoses and put movement at the centre of assessment. But it is a linear model, built on a single cause, of a problem we now know is multifactorial. Four problems stand out.
1. Matched treatment does no better than generic exercise. In the Washington University team's own trial, with a year of follow up, people with chronic low back pain who received treatment matched to their classification did no better on function than those given general treatment; both groups improved meaningfully. A separate trial found that matching treatment to MSI or Treatment Based Classification did not improve outcomes. An independent RCT in Brazil found no difference between groups in pain or disability between MSI treatment and stretching and strengthening guided by symptoms.
2. It changes movement without changing outcomes. Treatment specific to MSI can reduce lumbopelvic motion during hip rotation in the lab. That is a measurable change, but if outcomes match general exercise, the "impairment" was probably not the driver.
3. It is a model of pain, being used as a model of injury. The trials are in chronic low back pain. Extending the same logic to sport, "fix the valgus and you prevent the ACL", stretches it further than its evidence ever reached. Sahrmann's own team acknowledges that efficacy has only been tested in RCTs for chronic low back pain.
4. It cannot predict who gets hurt. If faulty patterns caused injury, screening for them should identify athletes at risk. Roald Bahr's critical review concluded that no screening test for sports injuries has adequate test properties, because players at high and low risk overlap too much on any continuous measure. Statistical association with injury at group level is not prediction for the individual.
The running research points the same way. Two well known studies of hip strengthening found that getting stronger and changing movement are separate things:
Ferber, Kendall & Farr 2011
Who: 15 runners with patellofemoral pain, 10 controls.
What they did: 3 weeks of hip abductor strengthening. (duration of intervention a factor?)
Result: strength increased and pain reduced, but peak knee valgus was unchanged; there was less variability from stride to stride.
Willy & Davis 2011
Who: 20 women with ‘excessive’ hip adduction when running.
What they did: 6 weeks of hip strengthening plus single leg squat movement training.
Result: strength increased (pain not measured), but there was no change in hip or knee mechanics when running; the single leg squat did improve.
In Ferber's runners, pain fell even though the "faulty" knee valgus angle stayed the same. In Willy and Davis's study, stronger hips and coached squats did not carry over into how people ran. The strength gains seem to matter more than the frontal plane angle. These are small, short studies on pain rather than injury rates, so they support the argument rather than solve it. But the message is consistent: build capacity, and don't make a perfect movement pattern the target.
More recent studies point the same way. Greaves and colleagues (2021) put people with patellofemoral pain through a six week programme based on the international expert guidelines. All were free of pain at the end, with better function and less fear of movement, yet their running mechanics during stance did not change. And a 2025 systematic review by Hall, Barry and Hayes pooled every study up to October 2024 that measured running gait before and after hip abductor strengthening: five studies, 144 people. Strength went up quickly, but there was little evidence that it changed the running mechanics linked to overuse injury. The authors concluded that if you want to change how someone runs, strengthening alone will not do it. For our purposes the more interesting finding is the reverse: people get better without the movement changing at all.
The model also sits uneasily with Meeuwisse. It places the cause inside the athlete's movement pattern and largely ignores the extrinsic layer and the inciting event. In a tackle, a collision or a slip, the athlete's habitual lumbar flexion or hip rotation pattern is close to irrelevant.
None of this means movement quality never matters. Technique matters for performance, and load through a sore structure matters in rehab. The problem is treating movement "faults" as the cause that must be corrected before someone is safe to train.
The corrective exercise trap, nearly ten years on
The 2017 article got the core argument right. What it lacked was an injury causation model to explain why corrective screening fails, and the positive evidence for what to do instead. Both now exist.
What the article argued
"The Corrective Exercise Trap" by Nick Tumminello, Jason Silvernail and Ben Cormack was published in the NSCA's Personal Training Quarterly (4.1) in March 2017. Its key moves were:
• The difference is the standard, not the exercises. Good training corrects technique against general biomechanics and physiology, tailored to the client. Corrective exercise corrects against a proprietary evaluation model's idea of "normal", and uses that model to predict injury or poor performance.
• Injury prediction fails Bahr's three steps. Using Bahr's framework (find the risk factor, validate it in other groups, prove screening plus intervention works in an RCT), the authors noted that no approach combining screening and correction had completed all three. A systematic review of the Functional Movement Screen did not support its predictive validity, and FMS scores improve simply when people know the grading criteria.
• Better test scores do not mean better athletes. Training to a screen improves performance on that screen, but generally does not transfer to athletic performance.
• "Dysfunctions" are mostly normal variation. Case control studies found no posture differences between people with and without shoulder impingement or neck pain. Most asymptomatic people have an anterior pelvic tilt. Elite Australian footballers have marked psoas and quadratus lumborum asymmetry unrelated to injury, and in cricket fast bowlers it was symmetry that went with back pain. Hip and femoral anatomy varies so much that there is no single correct squat.
• Core stability is not special. Cochrane reviews, and an RCT comparing tailored with general exercise, found motor control exercise no better than general exercise for low back pain.
• Assessment can harm. Telling clients their bodies are fragile or must move a certain way can make them less resilient, a nocebo effect. Nocebo is the flip side of placebo: negative expectations, often planted by what a clinician says, that make pain worse or recovery slower.
• The alternative is good training. Work around pain rather than through it; pick exercises the person can do with comfort and control; fit the exercise to the person; build broad capacity in mobility, strength and work tolerance; and progress steadily.
The authors were careful to say they were not calling corrective approaches "bad", only that the priorities were reversed. Their conclusion was that "the trap is thinking small": time spent chasing questionable dysfunctions is time not spent on real strength and conditioning for the client.
What I would add in 2026
1. A causation model that explains the failure. The article shows screens fail but does not say why they always will. Meeuwisse does. If injury arises from a chain of intrinsic factors, extrinsic factors and an inciting event, and the athlete's risk changes every session, then a single movement score taken in preseason cannot carry much predictive weight. Bittencourt's complex systems model makes that even clearer. The article also never separates contact from noncontact injury, which is where the limits of any training intervention become obvious.
2. Sahrmann's model has now been tested directly. The article dealt with Janda's crossed syndromes and core stability. Since then, an independent RCT published in 2018 found treatment based on MSI classification no better than stretching and strengthening guided by symptoms for chronic low back pain. That adds the kinesiopathological model to the list of systems where matching treatment to a movement fault adds nothing.
3. The positive case is far stronger. In 2017 the alternative was "general exercise works too". We can now say more: strength training reduced sports injuries to under a third in Lauersen's review, and Nordic hamstring programmes roughly halve hamstring injuries (van Dyk, 2019). Both build capacity; neither depends on diagnosing a fault. That is exactly what the article's section on functional capacity was pointing towards.
4. The same trap caught load monitoring. Soon after the article, the acute:chronic workload ratio (ACWR) became the new single number for injury risk. Impellizzeri and colleagues have since shown there is no evidence supporting ACWR for reducing injury risk and recommend going back to progressive overload adjusted to the athlete's response (Impellizzeri et al., 2020). That is close to the article's own "be progressive" guideline. (Keep it simple - add ~10%)
5. Nocebo has grown into a bigger conversation. The article's nocebo section was based on low back pain beliefs research. Cormack's later work argues that the biopsychosocial model itself is often lost in translation, applied as three separate boxes, and proposes an enactive update. For injury risk, the message is to avoid swapping one reductive explanation ("fix the glute") for another ("fix the stress").
6. The language. The article itself already spoke of reducing injury risk, not preventing injury. The research literature has largely followed. Course titles, marketing and conversations with athletes still lag behind. If you see a course that purports to help you prevent injury - challenge it, ask for the evidence…
What actually reduces risk
The programmes with the best evidence cut injury rates substantially, roughly by a third to a half, but none gets close to zero. That gap is the whole argument of this post.
• Strength training: injuries reduced to less than a third (RR 0.32). Lauersen et al. 2014, 25 trials, 26,610 participants.
• Proprioception training: roughly 45% reduction (RR 0.55). Lauersen et al. 2014.
• Programmes including the Nordic hamstring exercise: hamstring injuries reduced by up to 51%. van Dyk et al. 2019, 15 studies, 8,459 athletes.
• FIFA 11+ warm up: football injuries reduced by 39%. Thorborg et al. 2017.
• Stretching: no meaningful effect (RR 0.96). Lauersen et al. 2014.
Even the strongest result, strength training, still leaves around a third of the injuries happening. The Nordic exercise halves hamstring injuries; it does not abolish them. And these figures come from controlled trials with good adherence, which real squads/clients rarely match.
Notice too what these interventions have in common. None of them depends on first diagnosing a movement fault. They build general capacity, expose tissues to the forces sport demands, and get the athlete used to high speed, high load tasks. That is robustness, not correction.
In practice, risk reduction looks like this:
• Build capacity: progressive strength work, eccentric hamstring loading, plyometrics and landing practice relevant to the sport.
• Prepare for the sport's demands: regular high speed running for field athletes so a match sprint is not a novel event.
• Manage load sensibly: progressive overload, adjusted to how the athlete responds, without chasing a single magic ratio. Use the ACWR as a guide, not a definitive answer - and use internal loading as a key measurement.
• Respect the big intrinsic factor: previous injury. Complete rehab and a graded return matter.
• Influence the extrinsic layer where you can: tackle technique coaching, law enforcement, appropriate footwear, surface checks, sensible scheduling.
• Use screening to profile, not predict: baseline data is useful for return to play decisions and individualising training, not for telling an athlete they are "at risk".
Tests worth doing
Test capacity, not movement quality: endurance, aerobic fitness and strength can all be measured reliably, and they tell you what to train. They still will not predict who gets hurt.
Two words matter when choosing a test. A test is reliable if it gives the same result when nothing has changed, so a real change stands out from the noise. It is valid if it measures what it claims to measure. Reliability is often reported as an intraclass correlation coefficient (ICC), where 1 means perfect agreement between sessions and anything above about 0.90 is excellent.
No test reduces injury risk on its own. Its value lies in what you do with the result: each one below points training at a capacity that does reduce risk, or gives an objective marker for a safer return to play.
Aerobic capacity (lab)
Test: VO2max on a treadmill or bike with gas analysis.
Evidence: The criterion measure that field tests are validated against.
What it tells you: Maximal aerobic capacity; best for endurance athletes or research, but costly.
How it helps reduce risk: A bigger aerobic engine means fatigue arrives later in training and matches, and a fitter athlete tolerates the training load they need. The result sets training intensities and shows whether the base is improving.
Aerobic capacity (field)
Test: Yo Yo Intermittent Recovery Test Level 1.
Evidence: Validated against physiological measures by Krustrup and colleagues; repeat test ICC 0.86 to 0.95, and scores correlated with VO2max at r = 0.87 in recreational team sport players.
What it tells you: Ability to repeat high intensity running with short recoveries; suits team sports and tests a whole squad at once.
How it helps reduce risk: Shows who is not yet ready for the repeated sprints and short recoveries of a match, so conditioning can be built up before they meet those demands; retesting after time out guides a gradual return rather than a sudden jump in load.
Maximal strength
Test: Isometric mid thigh pull on a force plate.
Evidence: Across 16 studies, median ICC 0.96 for peak force.
What it tells you: Whole body force production, with little fatigue or injury risk compared with a 1RM.
How it helps reduce risk: Strength training has the strongest evidence for reducing injury (injuries cut to under a third in Lauersen's review); this test finds who has the most to gain and shows whether the programme is actually making them stronger.
Eccentric hamstring strength
Test: Nordic hamstring test on an instrumented device (such as NordBord).
Evidence: Bilateral ICC 0.83 to 0.90; less reliable when one leg is tested at a time.
What it tells you: Hamstring capacity and side to side difference, useful in return to play after a hamstring injury.
How it helps reduce risk: Eccentric weakness and previous injury are both linked to hamstring strains; the test guides Nordic training, which roughly halves hamstring injuries, and gives an objective target before returning to sprinting, the riskiest point for reinjury.
Eccentric adductor strength
Test: Copenhagen adduction exercise, with force measured during the lowering phase; a timed Copenhagen hold or repetition count is the simple field version.
Evidence: Force during the lowering phase: ICC 0.84 to 0.87 in 20 male athletes. A smartphone field version showed moderate to good reliability (ICC 0.62 to 0.83), with repetition counts less reliable than strength measures. A plain timed hold has not yet been validated.
What it tells you: Adductor strength in the same position used to train it, and side to side difference.
How it helps reduce risk: Low adductor strength is a key risk factor for groin injury, and the Copenhagen exercise itself cut groin problems by 41% in a trial of 652 semiprofessional footballers. Testing sets the starting level on the exercise's progressions and shows whether strength is building.
Posterior chain endurance
Test: Single leg bridge test: heel on a 60 cm box, bridging repetitions to failure.
Evidence: Intratester ICC 0.77 to 0.89 and intertester ICC 0.89 to 0.91; but it does not isolate the hamstrings, as the glutes and back also contribute.
What it tells you: Combined hamstring and glute endurance; cheap, quick and needs only a box.
How it helps reduce risk: In amateur and subelite Australian footballers, those who went on to injure a hamstring had lower preseason scores. Low scores flag who needs more posterior chain endurance work, and retesting tracks it. Treat it as an endurance marker, not a hamstring strength test.
Calf endurance
Test: Single leg heel rise endurance test (repetitions to fatigue).
Evidence: Reliable, with norms from 500 adults: median about 25 repetitions.
What it tells you: Calf strength endurance, valuable for runners and in Achilles rehab; a phone app can also record height and work.
How it helps reduce risk: The calf takes a large share of the load in running; low endurance limits how much running a runner can tolerate. The test sets a calf loading target, tracks progress and gives a clear marker for returning to running after Achilles or calf injury.
Use these to profile and monitor, not to predict. They show where an athlete's capacity sits against teammates, norms or their own baseline. They show whether training is working. And they give objective markers for return to play. Even a reliable, valid test inherits Bahr's problem: high and low risk athletes overlap too much for any single score to tell you who will be injured.
Two practical points. Test under the same conditions each time (time of day, warm up, footwear, the same tester), because that is what makes the reliability figures hold in your setting. And present results as capacity to build, not faults to fear, so the testing does not become a nocebo of its own.
Changing the conversation
Stop promising prevention. Promise a better prepared athlete and a lower chance of injury, and be honest about what nobody controls.
That honesty protects everyone. Athletes are not told they are broken before they have been hurt. Coaches are not sold a programme as a guarantee. Practitioners are not blamed when an opponent's knee lands on a well conditioned ankle.
Some phrases worth swapping:
• Instead of "injury prevention programme", try "injury risk reduction programme" or "robustness programme".
• Instead of "This will stop you getting injured", try "This lowers your chances and helps you cope with what the sport throws at you".
• Instead of "Your knee caves in, that's why you'll do your ACL", try "Let's get you stronger and more practised at landing and cutting".
• Instead of "You need correctives before you can train", try "Training is the corrective".
Meeuwisse gave us the framework to say this with confidence. Injury emerges from a chain that includes the athlete, the environment and a moment of chance. We can strengthen the athlete's links in that chain. We cannot remove the chain.
This is what exercise therapists do
Everything in this article is how we work. We don't promise to prevent injury, and we don't chase perfect movement patterns. We test what matters with reliable measures, build capacity with progressive loading, and use the results to guide training and return to play, explained to athletes in language that builds confidence rather than fear.
If you want an athlete or patient treated this way, find out more at www.exercisetherapist.com. If you want to work this way yourself, the Exercise Therapist Pathway at MTE Academy is our qualification route into it.
Exercise Therapist Pathway at MTE Academy
References
1. Meeuwisse WH, Tyreman H, Hagel B, Emery C. A dynamic model of etiology in sport injury: the recursive nature of risk and causation. Clin J Sport Med. 2007;17(3):215–219.
2. Bittencourt NFN, Meeuwisse WH, Mendonça LD, et al. Complex systems approach for sports injuries: moving from risk factor identification to injury pattern recognition. Br J Sports Med. 2016;50(21):1309–1314.
3. Bahr R. Why screening tests to predict injury do not work—and probably never will…: a critical review. Br J Sports Med. 2016;50(13):776–780.
4. Sahrmann S, Azevedo DC, Van Dillen L. Diagnosis and treatment of movement system impairment syndromes. Braz J Phys Ther. 2017.
5. Van Dillen LR, Norton BJ, Sahrmann SA, et al. Efficacy of classification-specific treatment and adherence on outcomes in people with chronic low back pain. Man Ther. 2016;24:52–64.
6. Henry SM, Van Dillen LR, Ouellette-Morton RH, et al. Outcomes are not different for patient-matched versus nonmatched treatment in subjects with chronic recurrent low back pain. Spine J. 2014.
7. Azevedo DC, Ferreira PH, Santos HO, et al. Movement system impairment-based classification treatment versus general exercises for chronic low back pain. Phys Ther. 2018;98(1):28–39.
8. Tumminello N, Silvernail J, Cormack B. The corrective exercise trap. NSCA Personal Training Quarterly. 2017;4(1).
9. Cormack B, Stilwell P, Coninx S, Gibson J. The biopsychosocial model is lost in translation: from misrepresentation to an enactive modernization. Physiother Theory Pract. 2022. doi:10.1080/09593985.2022.2080130
10. Impellizzeri FM, Tenan MS, Kempton T, et al. Acute:chronic workload ratio: conceptual issues and fundamental pitfalls. Int J Sports Physiol Perform. 2020;15(6):907–913.
11. Impellizzeri FM, McCall A, Ward P, et al. Training load and its role in injury prevention, part 2: conceptual and methodologic pitfalls. J Athl Train. 2020;55(9):893–901.
12. Lauersen JB, Bertelsen DM, Andersen LB. The effectiveness of exercise interventions to prevent sports injuries. Br J Sports Med. 2014;48(11):871–877.
13. van Dyk N, Behan FP, Whiteley R. Including the Nordic hamstring exercise in injury prevention programmes halves the rate of hamstring injuries. Br J Sports Med. 2019;53(21):1362–1370.
14. Thorborg K, Krommes KK, Esteve E, et al. Effect of specific exercise-based football injury prevention programmes on the overall injury rate in football. Br J Sports Med. 2017;51(7):562–571.
15. Match injury incidence during the Super Rugby tournament is high: a prospective cohort study over five seasons involving 93,641 player-hours. Br J Sports Med. 2018.
16. Rugby Football Union community study. Collapsed scrums and collision tackles: what is the injury risk? Br J Sports Med. 2015;49(8):536–540.
17. Ferber R, Kendall KD, Farr L. Changes in knee biomechanics after a hip-abductor strengthening protocol for runners with patellofemoral pain syndrome. J Athl Train. 2011;46(2):142–149.
18. Willy RW, Davis IS. The effect of a hip-strengthening program on mechanics during running and during a single-leg squat. J Orthop Sports Phys Ther. 2011;41(9):625–632.
19. Greaves H, Comfort P, Liu A, Herrington L, Jones R. How effective is an evidence-based exercise intervention in individuals with patellofemoral pain? Phys Ther Sport. 2021;51:92–101.
20. Hall G, Barry G, Hayes P. The role of hip abductor strengthening and the effects on running gait: a systematic review. Adv Exerc Health Sci. 2025:16–23. doi:10.1016/j.aehs.2025.02.004
21. Krustrup P, Mohr M, Amstrup T, et al. The yo-yo intermittent recovery test: physiological response, reliability, and validity. Med Sci Sports Exerc. 2003;35(4):697–705.
22. The Yo-Yo test: reliability and association with a 20-m shuttle run and VO2max. Int J Sports Physiol Perform. 2006.
23. Grgic J, Scapec B, Mikulic P, Pedisic Z. Test-retest reliability of isometric mid-thigh pull maximum strength assessment: a systematic review. Biol Sport. 2022;39(2):407–414.
24. Opar DA, Piatkowski T, Williams MD, Shield AJ. A novel device using the Nordic hamstring exercise to assess eccentric knee flexor strength: a reliability and retrospective injury study. J Orthop Sports Phys Ther. 2013;43(9):636–640.
25. Sleeswijk Visser TSO, O'Neill S, Hébert-Losier K, Eygendaal D, de Vos RJ. Normative values for calf muscle strength-endurance in the general population assessed with the Calf Raise Application. Braz J Phys Ther. 2025.
26. Harøy J, Clarsen B, Wiger EG, et al. The Adductor Strengthening Programme prevents groin problems among male football players: a cluster-randomised controlled trial. Br J Sports Med. 2019;53(3):150–157.
27. Hickey and colleagues. Measuring eccentric hip adductor strength during the Copenhagen adduction exercise: a proof-of-concept and test re-test reliability study. Phys Ther Sport. 2025;73:34–38.
28. Dynamic field assessment of hip adductor function using a smartphone-based Copenhagen test: reliability and concurrent associations with isometric strength in amateur football players. Sports. 2026;14(4):125.
29. Freckleton G, Cook J, Pizzari T. The predictive validity of a single leg bridge test for hamstring injuries in Australian Rules Football Players. Br J Sports Med. 2014;48(8):713–717.
30. The single leg bridge test does not measure isolated hamstring endurance in healthy men. Int J Sports Phys Ther.
