The baseline measurements showed that 12 of the 28 athletes had hip abduction strength of MRC grade 4. In total, at baseline, 14 out of 28 athletes had MRC grade 4 strength or greater than a 25% side-to-side difference in performance. These profound lopsided performance and/or pure hip abduction deficiencies have never been previously reported in asymptomatic athletes.
After two months of the exercise all had improved to grade 5. In fact some athletes doubled or even tripled their baseline scores. These observed changes were large compared to the threshold of detection set with the small pilot study. These improvements should reduce the chance that athletes still had Trendelenburg gait
 and increased valgus knee moments with landing manoeuvres
 but this was not tested in this study.
An unsupervised exercise program in an adolescent population can have significant compliance problems. This exercise without knowing exact compliance details resulted in no athletes continuing to have fundamental hip abduction deficiencies.
Fredericson in his review of iliotibial band syndrome in athletes
 makes reference to the long held belief that athletes with a greater than 10% disparity between muscle groups are more prone to injury. In our cohort, 11 players originally had greater than 25%. In fact the side-to-side differences in this subgroup were huge with a mean difference between the sides of 48%. While it was noted at retesting that 6 players still recorded difference of greater than 25% between sides, the mean difference had dropped to 34%. With close supervision it would be reasonable to anticipate achieving no athletes with a greater than 10% side to side difference although measuring this accurately in the clinical setting may prove difficult.
These results would suggest that if the 10% disparity is valid, and Trendelenburg positive gait is shown to be a risk factor for lower limb injury, the group’s injury profile should be diminished following their relatively short period of focused rehabilitation and into the future with continued hip abduction training.
The exercise used in this intervention has not been previously studied. There are no normative values for hip abduction performance in this position. Pure hip abduction performance has been measured in a smaller group of slightly older elite association footballers
. A mean side-to-side difference of 14% in these athletes was noted for hip torque but not noted in controls. No fundamental strength deficiencies were reported. Hip abduction deficits have been shown in a group of injured runners but only on their injured side
. Had these athletes been tested in the position used in our study it’s possible that more common and profound deficits would have been noted in injured and non-injured limbs and in the control group.
This exercise is used clinically by the teams sports physician (HRO) to treat a number of lower limb biomechanically related injuries as anecdotally similar deficiencies and subsequent gains are seen across the spectrum of injured athletes and non athletes alike. This exercise is used by the clinician for several reasons and because the literature fails to support any one exercise over another a process of reasoned assumption has been made. If injuries such as ilio tibial band syndrome and greater trochanteric pain syndrome are related to GMed weakness, how is it that when patients are performance tested in the clinical setting in pure abduction, strength is almost always grade 5 yet in the position of the exercise used in this study performance deficiencies noted? It is therefore logical to strengthen in the position of weakness. In the clinical scenario it has been observed using manual palpation of the muscles that the position used in this study minimizes tensor fascia lata and gluteus maximus contraction, making it likely that gluteus medius is the dominant muscle however this needs further laboratory based research.
Anatomically GMed is the largest abductor of the femur from the pelvis, tensor fascia lata and gluteus maximus causing leg abduction via the fascia lata to the tibia. It has been shown that in males peak contraction of GMed compared with gluteus maximus prior to heel strike during single leg landings expressed as percentage of maximum voluntary isometric contraction is the same (48% compared to 47% respectively)
 but the mean contraction of GMed compared with gluteus maximus similarly expressed is much greater in GMed (26%) compared to gluteus maximus (16%)
. This information would suggest that not only is GMed most ideally positioned anatomically but it is the key controller of lateral pelvic control at heel strike If GMed is fundamentally weak then coronal plane control of the pelvis at heel strike must be compromised i.e. Trendelenburg gait type variants
. Given that this proximal control is required for distal strength and co-ordination it follows that heel strike with poor GMed strength is the beginning of the abnormal movement patterns that lead to injury related to landing and impact whether acute or repeated.
If GMed is the key controller of pelvic tilt at heel strike then this would go some way to explaining the very large increases in performance seen in this study that would not normally be expected from a low load, high repetition activity. The weaknesses observed in this study at baseline are commonly seen clinically, in fact it’s not that uncommon to observe a grade 3 deficiency i.e. a patient unable to lift their knee from the floor. GMed must have a threshold of performance below which it cannot participate in a useful way in the normal gait cycle, otherwise it would not remain fundamentally weak. This is observed clinically as many patients with overload injuries related to GMed weakness (e.g. iliotibial band syndrome, greater trochanteric pain syndrome) have significant reduction in their pain by approximately four weeks of rehabilitation using this exercise. Once this currently unknown threshold (estimated to occur clinically after 4 weeks of exercises) is reached the GMed is then likely starting to actively participate in normal gait cycle and therefore the number of repetitions performed over the next month would increase exponentially to perhaps 5000 repetitions daily (its common for fit health people to take 10000 steps per day). Moreover GMed would no longer just be lifting the weight of the leg performing the exercise but actually supporting most of the body weight at heel strike. This would account for the very large increases in performance observed. It would be prudent with further research to do serial testing to observe if there is a threshold of performance after which there is a significant increase in performance increments compared to below this threshold.
Risk of injury is associated with overuse, a phenomenon which is relatively high in AFL
. Intuitively, it would be assumed that weaker muscle groups are prone to overuse injuries earlier. Overuse is associated with sports related chronic groin pain
 which is the third commonest reason for missed club games in the AFL
. In particular, players in this code are associated with a relatively high incidence of osteitis pubis which has been associated with a weakness in GMed
Other codes of football show similar injury patterns. Bathgate’s review of Australian rugby union injuries
 showed that injuries to the knee accounted for the largest proportion of severe injuries (25%). Of these 34% occurred in non-tackling situations. Although this is somewhat lower than in AFL
 where 76% of anterior cruciate ligament injures occur in non tackling situations it still represents an area where GMed weakness could potentially be contributing to a very significant problem. The relatively high level of injuries seen in association football especially to the lower limb again raises the issue of overall proximal control of pelvis and centre of mass over the planted leg
Current research trends into anterior cruciate ligament ruptures include the difference in landing techniques between males and females
. There is focus on the technical problems of landing with interest in quadriceps dominant landings, ankle dorsiflexion angles and also in prevention strategies including GMed strengthening and landing practice. However it has yet to be published that GMed weakness exists in those who rupture their anterior cruciate. This study moves towards finding that missing link, that GMed weakness is common and would help to account for the high level of injuries linked to its weakness despite the weakness being assumed but not previously shown.
This would especially be true for females where the broader female pelvis makes pelvi femoral control more difficult and leads to significantly more problems with injuries related to GMed weakness such as patello femoral syndrome and greater trochanteric pain syndrome (where increased knee valgus angles and GMed weakness have clear associations). It would be expected to find GMed deficiencies similar to those found in this study in elite female athletes. This is an area where future research should be focused.
This study has some obvious limitations. There was no control group. The exercise technique has not been shown to be an exercise that focuses on GMed over other hip abductors in a laboratory based study. Further anatomical and electromyographic research should be done on this particular exercise to determine that in this position maximum recruitment of GMed occurs with minimal recruitment of tensor fascia lata and gluteus maximus, in comparison with other hip abduction exercises. Other hip abduction exercises are described often in the literature but generally the authors are focused on hip abduction strength rather than individual muscle performance deficiencies
. There is a chance that the exercise does not primarily use GMed, but the results of the study remain significant and the exercise remains anecdotally a powerful clinical tool.
The lack of any weight lifted other than that of the athletes own leg can be criticized but a lower limb is of sizable weight. Hand held dynamometry is not the gold standard for strength testing and certainly isometric testing isn’t as good as laboratory based strength testing. However hand held dynamometry quite closely correlates with laboratory based dynamometry and is a valid clinical tool
The athletes were given 24 hours to practice the exercise between it being taught and the hand held dynamometry being performed. We believe this makes it unlikely that the large gains observed in the following two months are likely to be related to a learning phenomenon given the simplicity of the exercise. However the large threshold of detection set by the small pilot study could be improved with tighter testing protocols and strategies to minimise performance variance in the athletes.
However many of these criticisms can be seen as a positive as this study was undertaken at the coalface of elite sport on the sideline of a sporting field using readily available equipment. That such huge deficiencies were found and then corrected in a traditionally difficult age group to obtain compliance and that it was done essentially unsupervised would suggest that these are real results that can be achieved by a clinician in their clinic without the need for expensive equipment and lots of supervision time.