Back in the height of World War II, the United States faced a big problem. Bomber flight pilots were dropping like fleas. For every 100 sent to fight the war in Europe, less than 45 were returning. They were, in the words of military history Kevin Wilson, “dead men walking.” 

In an effort to find solutions, the Air Force started collecting returning planes. They needed a more resilient aircraft if they wanted any pilots alive by the end of the war. The problem was that it couldn’t be “too” resilient. Reinforcing the entire entire plane with armor would make it too heavy and unable to maneuver. They could only reinforce a limited amount of surface area.

As a result, military experts decided it would be best to reinforce the areas which came back with the most damage. They plotted and mapped bullet holes on the planes that returned from battle. They used this information to try and draw patterns. If they could figure out what parts of the plane took the most shots, they had the ability to best allocate armor for maximal protection and maneuverability.

That is, until Abraham Wald proposed an alternate solution.

Wald, born in Hungary in 1902, had been studying America’s lack of success through the air as part of a hand-picked, elite group of mathematicians. They gathered on the fourth floor of a communal building in Harlem, NY. Their task was to design a war-tested aircraft that had strength, speed, and agility. When Wald examined the returning planes, he brought up a glaring issue. The planes returning from battle were not the problem. The planes that were not returning were the problem. 

Original thinking suggested the areas which took on the most shots were most vulnerable. Wald disagreed. These areas were of least concern. They could withstand the most amount of fire and still return safely back home. The areas that returned with the least amount of damage were actually most vulnerable. Those two areas were the cockpit and tail. Wald’s simple, but critical shift in thinking ended up turning into a major breakthrough for military aviation.

The initial plan to collect and examine returning planes wasn’t a bad idea. It was a great idea. They were just weren’t looking at a critical part of the problem: The planes that did not return. The issue wasn’t the data they were seeing. It was the data they were not seeing. If you only focus on what you can see, you become blind to what you cannot see. Baseball had to learn this the hard way when healthy populations of pitchers started getting treated like injured ones.The problem didn’t stem from a bad set of data.

It originated because a critical set of data had been left out.




In the early 2000s, medical professionals faced a problem. An increasing number of baseball pitchers were reporting symptoms of pain in the throwing shoulder while throwing. In an effort to find solutions, they started testing athletes so they could collect data and draw patterns about symptomatic populations. One of the tests they ran measured glenohumeral internal rotation (GHIR) and external rotation (GHER) of the throwing shoulder.

The glenohumeral (GH) joint refers to the site of articulation between the humeral head and glenoid fossa of the scapula. The joint is one of the two ball and socket joints in our body – the other being the hip. Think of the humeral head as a golf ball and the glenoid fossa as a golf tee. You can move and maneuver the golf ball omnidirectionally without losing congruency to the tee. The GH joint works the same way. Our humeral head can move and traverse multiple planes of motion while still maintaining joint integrity with the glenoid fossa. One of these is GHIR and GHER. 

Aroldis Chapman with his throwing shoulder at max external rotation (GHER)

Glenohumeral external rotation, as it relates to the pitching delivery, occurs when the humerus retroverts and tilts posteriorly in the joint capsule. Glenohumeral internal rotation is where it shifts in the opposite direction and tilts anteriorly in the capsule. Testing for GHER and GHIR is popular in baseball populations because of the role both play in throwing a baseball. Between foot plant and ball release, GHER is in high demand. In order to create enough space for the arm to unwind and capture energy around the body, external rotation needs to occur at the shoulder as the torso begins ballistic rotation. If the thrower does not have access to enough GHER, the throwing arc becomes compromised. The humerus and forearm don’t have enough room to unwind as the torso rotates to create for a clean transfer of energy.

Justin Verlander internally rotating his throwing shoulder after ball release (see throwing thumb pointed down)

After the arm reaches its point of maximum external rotation, it transitions into internal rotation and forearm pronation. This part of the throwing arc delivers the baseball and initiates deceleration of the arm post ball release. You can often point it out by looking at the hand post release. If the hand is turned over and the thumb is pointed down towards the ground (pronation), the thrower has internally rotated the throwing shoulder.

Testing for both internal and external rotation at the shoulder gives us information on passive range of motion. This can give us insight into potential hardware limitations, soft tissue restrictions, or abnormal ranges that can negatively influence shoulder health. If a hardware restriction is impacting a software solution, you can’t just download new software. You need to address the faulty hardware first. 

That is, if the hardware was the problem in the first place.

When medical professionals started testing injured athletes, they found a common pattern: Most, if not all, lacked some degree of internal rotation in their throwing arm. This was abnormal compared to what they typically saw from asymptomatic patients. As a result, they developed a theory. This theory is better known as Glenohumeral Internal Rotation Deficit, or GIRD for short. They concluded throwers who lacked an adequate amount of shoulder internal rotation were at risk for injury at the shoulder. As a result, they designed stretches and modalities – such as the sleeper stretch – to improve GHIR and reduce pain when throwing. 

They just had one big problem: The testing they ran was only on injured throwers. They had no information on ranges of motion from healthy throwers. 

If we go back to Abraham Wald’s discovery, the problem wasn’t the information they were looking at. The planes that returned from battle weren’t the issue. It was the information they were not looking at: The planes that had never returned. They were blind to the information they couldn’t see. Medical professionals were no different. All they could see was information they had collected on injured populations of throwers. If they had tested healthy throwers, they would have found the same exact “limitation.” Injured throwers weren’t the only ones who presented with a GHIR deficit. Healthy throwers did too.

Lacking internal rotation isn’t a problem in throwing populations. It’s a completely normal adaptation due to the high demands for GHER. Over time, the humeral head in most throwers is going to retrovert and shift into a position more favorable for GHER. Think of it like a strategic head start: The humerus is starting in more of an externally rotated position, making it easier to access GHER at the beginning of the throwing arc. When this happens, throwers are naturally going to lose internal rotation.

Our shoulder doesn’t have an endless amount of degrees of freedom. Between GHER and GHIR, it only has about 180 to work with. If you gain some in one direction, you’re going to lose some in the other direction. This is why the pitchers originally tested had a GHIR deficit. Because they threw for a living, their throwing shoulder naturally gained GHER. As a result, they lost GHIR.

This was a completely normal adaptation, but it appeared abnormal because the control group wasn’t a population of throwing athletes. It was the general population. This was arguably the biggest mistake medical practitioners made. Comparing average Joes to big league pros is as apples to oranges as it gets. Elite athletes shouldn’t present like normal people. What they do for a living is anything but normal. Their body should reflect that.

They’re not elite in spite of their hardware. They’re elite because of it. When we compare desk job workers to professional pitchers, we lose sight of this.

GIRD was never a problem in throwing populations. You need zero internal rotation to throw a baseball. The only internal rotation that ever occurs is created from the natural unwinding of the hand and forearm around the humerus into release. If the hand and forearm unwind perpendicular to the humerus – which is how most practitioners test for GHIR – you’re adding sheer force to the elbow joint. The arm can’t naturally unwind and dissipate energy. It vaults and pulls your joints into positions they cannot safely handle. 

Just imagine prescribing stretches that do the same exact thing…

(If you’re being prescribed the sleeper stretch to remedy your cranky shoulder, do yourself a favor and look for a new practitioner.)




Before you try to infer a relationship, consider that your decision making process is flawed. Are you looking at everything you need to see, or are you too consumed over what you can currently see? Do you have the right control group? If there was one thing you could look at that would help you make a better decision, what would that be?

We all have blind spots. Once you get married to what you see, you become blind to what you don’t see. For the United States, it was the planes that never made it back home. For medical professionals, it was the throwers that never needed an appointment. These kinds of populations are often the key to the solutions we seek. 

That is, if you’re looking for them in the first place. 

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes:

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>