Recently I’ve undertaken the task of reviewing some of the research on the very popular Functional Movement Screen.  Previously I’ve reviewed the Interrater Reliability of the Functional Movement Screen and Core strength: A New Model for Injury Prediction and Prevention.

.

Today I’ll be taking on the 3rd of 4 studies I hope to review.  After the final review I’ll talk a little bit more about my overall impression of the FMS and how I believe it should be used.

.

Can Serious Injury in Professional Football be Predicted by a Preseason Functional Movement Screen?

Kiesel K, et al.  North American Journal of Sports Physical Therapy Aug 27, 2:3

.

Background

.

Risk factors for injuries in high school and collegiate football include previous injury, body mass index, body fat percentage, playing experience, femoral intercondylar notch width, cleat design, playing surface, muscle flexibility, ligamentous laxity, and foot biomechanics.  However, injury risk is likely a combination of many of the above.

.

Further, evaluation of isolated risk factors does not take into consideration how the athlete performs functional movement patterns required for sport.  The goal of this study was to examine functional movement scores (assessed by the FMS) and to determine the relationship between professional football players’ score on the FMS and the likelihood of serious injury.

.

Methods

.

FMS scores were obtained prior to the start of the season for 46 professional football players.  A receiveroperator characteristic curve the FMS score was used to predict injury during one complete football season.  For the sake of this study, injury was defined as membership on the injured reserve for at least 3 weeks.

.

A dependent t-test was used to determine if a difference existed between of the FMS scores of those who were injured versus those who were not.

.

Sidebar – Definitions

.

To have a clear understanding of the methods and the results of this study a brief discussion is needed to definte sensitivity, specificity, and how these are used to create a ROC curve (receiveroperator characteristic curve).

.

Sensitivity is basically the power to detect a true positive.  For example, if you were to go through a scanner at the airport to detect for metal it would be very sensitive to decrease the likelihood that someone were able to slip onto an airplane with a weapon.  On the other hand, the scanner doesn’t have very high specificity in that it will sound for almost any piece of metal not just weapons.  In this case, a high sensitivity is most important because it is important that weapons do not sneak aboard the aircraft.

.

A dog trained to sniff for narcotics would have a high specificity since only those carrying drugs would need to be stopped.  If the dog didn’t have a high specificity for a specific substance, but was highly sensitive it would possibly alert people needlessly to any scent and make the purpose of having the dog useless (since every bag would have to be checked anyway).

.

In a perfect world every test would have 100% sensitivity and 100% specificity (i.e., identifying every weapon at the airport without going off for every other piece of metal), but this is rarely the case.  There is usually a tradeoff between one and the other and the ROC curve plots sensitivity against specificity to determine the ideal cutoff number to use to maximize both.

.

With the FMS the the cutoff was chosen using the ROC curve such that the test correctly identifies the greatest number of athletes at risk of injury (true positives) while minimizing incorrectly identifying athletes not at risk of injury (false positives).

.

Results

.

A score of 14 or less on the FMS was able to predict injury with specificity of 0.91 and sensitivity of 0.54. 

.

The test had a very high specificity indicating that the majority of people with a score below 14 had a greater chance of injury.

.
Those with a score under 14 that got an injury = 7

Those with a score under 14 that didn’t get an injury = 3

.
Unfortunately, the test had a only a moderate sensitivity so it did not detect those with a score over 14 who did experience an injury.

.
Those with a score over 14 that got an injury = 6

Those with a score over 14 that didn’t get an injury = 30

.
In other words, the bulk of the people with an FMS score over 14 did not get an injury and the bulk of those with a score under 14 did.  Using something called an odds ratio the authors determined that the likelihood of injury was 11 times more likely if the player had a score below 14 on the FMS.

.

However, 6 people that did have a score higher than 14 did end up getting injured.  These ones were missed by the screen.  In fact, it failed to identify almost as many people as it did identify as being at risk.

.

Funding

.

None declared, but I believe at least two of the authors have affiliation with the FMS.  (Not that there is anything wrong with that, but I would declare this as a possible conflict of interest.)

.

My Thoughts

.

All in all I think that the FMS did a great job of determining that those with a score less than 14 were at risk for injury.  In terms of practical application, these players could have been flagged for specific work with a fitness/rehab professional.  On the other hand, the test wasn’t sensitive enough to detect risk of injury such that 6 athletes slipped through the cracks and ended up being injured without this being detected by the screen.

.

Unfortunately, there was no differentiation between the types of injuries that landed people on the reserve list so it is possible that some of the injured athletes suffered from contact injuries that could not have been predicted by any test or screen.  Perhaps if contact injuries were ruled out (since you can’t really test for these) the FMS would have proven to be more sensitive.  Then again, maybe some of those with a score below 14 suffered contact injuries as well.  It would have been interesting to see if the results were different if these types of injuries were excluded.

.

It is also possible that one reason the FMS predicted injuries so well below the score cutoff of 14 with this group is because it is the same group whose results were used to create the cutoff in the first place.  Only future research will tell if this pass/fail cutoff is equally as effective for other groups.

.
Summing Up

.

The FMS indicated correctly that those with a score less than 14 were more prone to injury.  However, the results of the present study indicate that the FMS may also miss equally as many people as it detected (which may be the reason why the pass/fail score for the FMS when it is typically used is actually lower than this).

.

It would also have been interesting to see which of the individual scores within the FMS were most related in injury.  Since the FMS typically suggests that side to side imbalances are most important to address, it would have been nice to see this data to see if this hypothesis holds true.

.

In the end though, the FMS does appear to effectively predict injury in this below a value of 14 in the group studied.  It does not catch all injuries and as such is not a perfect screen, but effective at picking out some who are at risk.  And since the results are those of professional football players, we should be careful when generalizing them to other populations.

.
What do you think?

.