Upper-Extremity Rehab

01What the measurements show

The Numbers Behind an Upper-Extremity Program

Eight findings that turn upper-extremity rehab from a sequence of exercises into a dose schedule.

02Four phases in order

Upper-Extremity Rehab Runs in Ordered Phases Because Each Stage Builds What the Next One Loads

Upper-extremity rehab works because it is sequential. The arm moves through four ordered phases rather than racing to load a joint that cannot yet protect itself. The first calms the tissue and restores pain-free range. The second rebuilds dynamic stability and strength. The third drives concentric, eccentric and closed-chain power. The fourth returns the athlete to sport-specific performance.

The published framework says the same thing with different labels. A review of rehabilitation in the overhead throwing athlete organizes the work around five emphases Wilk 2002. Those are controlling inflammation, restoring muscle balance, improving soft tissue flexibility, enhancing proprioception and neuromuscular control, and returning the athlete to competitive throwing.

That review also names the problem the whole program is solving, and calls it the thrower's paradox. The throwing shoulder has to stay lax enough to allow excessive external rotation and stable enough to prevent symptomatic subluxation of the humeral head. Every early exercise is chosen to guard the front of the joint while dynamic stability is rebuilt from the inside out.

The order is agreed, the settings are not

Phases overlap as an athlete progresses, and the order still holds. Each stage prepares the nervous system for the demand of the next one. Skipping ahead loads ligament before the muscles that defend that ligament can fire on time, which is the failure the sequence exists to prevent.

How much motion to allow, and when, is where published programs diverge. A systematic review of rehabilitation protocols after ulnar collateral ligament repair found 13 usable protocols, 12 of which braced the elbow at varying range-of-motion settings for the first four to six weeks Ose 2024.

Most of those protocols were built around a five-phase structure, and all 13 included a return to throwing. The staging is consensus. The dose inside each stage is a judgment, and that judgment is what this page is about.

03Scapular retraining

The Scapula Is the Base the Arm Fires From, and Rehab Returns to It in Every Phase

Upper-extremity rehab begins at the scapula and comes back to it at every phase. The shoulder blade is the platform the arm fires from. A base that does not set lets the rotator cuff and the glenohumeral joint leak force they have already produced, and serratus anterior is the muscle that holds the tracking while the arm elevates.

Electromyography turns that principle into a ladder with rungs. In 20 healthy subjects tested across eight scapulohumeral exercises performed below shoulder height, five drove serratus anterior activity above 20 percent of maximal voluntary contraction Decker 1999.

Those five were the serratus anterior punch, scaption with external rotation, the dynamic hug, the knee push-up plus and the push-up plus. The two at the top shared a mechanic. They held the scapula upwardly rotated while accentuating protraction, and they produced the greatest serratus activity of the eight exercises tested.

Read that as a dose ladder rather than a list of favorites. A quadruped position, a knee push-up plus and a full push-up plus are the same movement at three magnitudes, and a program picks the rung the arm can currently organize. A scapula that sets without a superior shrug is the signature of a base that will hold under speed.

What scapular work moves, and what it leaves alone

Scapular retraining changes symptoms and control rather than mobility. A meta-analysis of eight randomized trials and 387 participants with subacromial pain found scapular stabilization exercise better than conventional physical therapy for pain and for the Shoulder Pain and Disability Index Zhong 2024. Range of motion did not improve.

That split is informative rather than disappointing. Range is a property of the tissue. Control is a property of the nervous system driving it. An input aimed at control moved the control measures and left the mechanical one where it was, which is what a program should expect from base work.

The findings that identify which athlete needs this work, scapular dyskinesis and the rotation deficit that comes with it, belong to The Shoulder. What an upper-extremity program does about them is this page.

04Rotation strength ratios

Muscle Balance Is the Measurable Target, and the Throwing Shoulder Does Not Reach It on Its Own

Muscular balance between the rotators is what an upper-extremity program can actually measure, and the throwing shoulder arrives below it. Isokinetic testing of 150 healthy professional baseball pitchers put the external to internal rotation strength ratio at 65 percent at 180 degrees per second Wilk 1993. At 300 degrees per second the same group measured 61 percent.

Those pitchers were healthy and competing. Bilateral comparison in that group found no significant difference between the internal rotators of the throwing and non-throwing shoulders at either speed. The ratio a rehabilitation program sets as its goal is therefore a goal, not a description of what a thrower walks in with.

A functional version of the ratio sharpens the picture. Eccentric external rotation was measured against concentric internal rotation at 300 degrees per second in 59 men Noffal 2003. Throwers showed significantly lower ratios than non-throwers in the dominant limb, with no group difference in the non-dominant limb.

The cause was named in the same study. Throwers had significantly greater concentric internal rotation torque without matching gains in eccentric external rotation. The accelerating half of the throw grew and the decelerating half did not, which is why an upper-extremity program trains the cuff with a heavy emphasis on the eccentric phase. Eccentric control is what stops a throwing arm.

The eccentric case, and the trial that tested it

The rationale is sound and the isolated version of it did not outperform ordinary exercise. Thirty-six patients with rotator cuff tendinopathy were randomized to a 12 week isolated eccentric program or a conventional one Dejaco 2017. Both groups improved on the Constant Murley score and on pain at 26 weeks, with no difference between them.

The authors put the practical reading plainly. Two eccentric exercises performed twice a day were as effective as six mixed concentric and eccentric exercises performed once a day. That is a statement about total magnitude delivered rather than about contraction type, and it is the kind of result an upper-extremity program should be reading.

The tendon science underneath that trial, including the loading continuum and the heavy slow resistance evidence, is carried by Tendon Pain. What belongs here is the dosing conclusion: the cuff responds to how much load arrives across a week, and the label on the contraction is a smaller variable than the total.

05The medial elbow

The Medial Elbow Is Rebuilt by Loading Everything Around the Ligament Before the Ligament Itself

Elbow rehabilitation in a throwing athlete is a staged climb that protects the joint while it heals. The early weeks hold elbow motion inside a protected arc in a hinged brace, and the rest of the body is trained hard through that window. The lower extremity, the core and the periscapular muscles all carry load while the ligament is off limits.

Early shoulder work is chosen with the elbow in mind. Programs hold back internal rotation strengthening in the first phase, and the biomechanics explain why. On the throws that produced peak elbow varus torque of 100 newton meters, shoulder internal rotation torque peaked at 101 newton meters in the same subjects Fleisig 2011. The two loads arrive together.

Only when motion and the surrounding chain are restored does the flexor-pronator mass, the dynamic stabilizer of the medial elbow, get strengthened in earnest. Two-hand plyometrics precede one-hand drills for the same reason the push-up plus has a kneeling version. Each is the same movement at a smaller magnitude. The valgus load that made the injury belongs to The Elbow.

What the staged return actually produced

The largest published series puts numbers on the climb. Ulnar collateral ligament reconstruction was performed in 1281 athletes over 19 years, and 743 were followed for a minimum of two years Cain 2010. Of those, 617 returned to their previous level of competition or higher, which is 83 percent.

The timing in that series is the part an upper-extremity program should sit with. Average time from surgery to the first throw was 4.4 months, with a range of 2.8 to 12 months. Average time to full competition was 11.6 months, with a range of 3 to 72 months. Complications occurred in 20 percent of patients, 16 percent minor and 4 percent major.

Read the ranges rather than the averages. The same operation and the same class of protocol produced a spread of more than twenty to one in time to full competition. Those are findings from one series and not a schedule any athlete is owed. The spread is the argument for dosing against the arm in front of you.

Newer repair techniques move the whole curve. Across 13 published protocols for ligament repair with suture tape augmentation, all 13 began return to throwing between 10 and 12 weeks Ose 2024. Eleven named a return-to-sport point, as early as 16 weeks, with a mean recommendation of 21.5 weeks.

Those are recommendations attached to a different operation, and they show how far the calendar moves when the structure underneath it changes. The criteria do not move with it. What clears an athlete is covered by Return to Play.

06Relearning the arm

Rehab Is a Neurological Process, and Throwing Has Already Quieted the Reflex That Protects the Arm

The limit on a recovering arm is rarely raw strength and almost always control. Upper-extremity rehab is a neurological process before it is a muscular one, and the evidence for that is unusually blunt. The reflex that decelerates a throwing arm is slower in throwers than in people who have never thrown.

Fifteen intercollegiate baseball players and 15 untrained men were tested against a sudden internal rotation force Brindle 1999. Each sat with the dominant shoulder at 90 degrees of abduction and external rotation, elbow flexed to 90 degrees. The throwers reacted more slowly at infraspinatus and teres minor.

Supraspinatus and posterior deltoid also stayed active for shorter periods in the trained group. The investigators read the pattern as downregulation of the rotator cuff against sudden internal rotation, an adaptation that permits higher internal rotation velocity and may contribute to glenohumeral joint pathology. The arm is fast in part because its brake has been turned down.

That reframes what an upper-extremity program is restoring. A program that returns strength and leaves the reflex where throwing left it has restored the accelerating half of the arm. Proprioception sits alongside soft tissue work, articulation, flexibility and strength as the five things the program manages, and it is the one trained from the first phase rather than the last.

The drills that train the incoming signal

The 90 degree abduction and 90 degree elbow flexion position used in that experiment is the position rehabilitation trains reflexive control in. Rhythmic stabilization there asks the cuff to hold a joint against unpredictable manual perturbation. Ball drops train reaction time at the same angles, and unstable-surface work on a balance disc loads the shoulder through the hand.

Closed-chain loading is the other half. Driving co-contraction across the glenohumeral joint restores joint position sense, and that returning accuracy is the deeper goal of every drill in the program. A joint that knows where it is protects itself, fires on time, and tolerates load.

The maintenance series that carries this into training is built on the same principle. The Advanced Throwers Ten program is organized around coactivation, dynamic stabilization, high-level neuromuscular control, endurance and coordination, and it links the upper and lower extremities in one series Wilk 2011.

The neck belongs in this section too. Cervical mechanics feed the same maps that aim the arm, which is why an upper-extremity program that stops at the shoulder is working on part of the loop. That link is developed in The Cervical Spine in the Athlete and in Proprioception and Joint Position Sense.

When the arm is reorganized this way it expresses the readiness that underlies durable performance, the central integrative state in which the nervous system meets load without losing control.

07The interval throwing program

The Interval Throwing Program Is a Dose Schedule, and the Number It Is Graded On Is Not the One the Elbow Pays

Return to play is earned through criteria, not a calendar, and the gateway for the overhead athlete is the interval throwing program. Before it begins the athlete must show full pain-free range and strength through flexion, extension, pronation and supination. Clean sport-specific mechanics are part of the entry standard, not a later refinement.

The whole kinetic chain is protected through this climb, because a throw is a wave of force that starts in the legs and finishes at the fingertips. Lower-extremity strength, single-leg balance and trunk rotation are trained right alongside the cuff and the scapula. The sequencing that makes that wave add up rather than leak belongs to The Kinetic Chain.

The progression itself is graded. Tossing gives way to throwing with a wind-up, long toss extends the distance, mound work restores the slope, and competition comes last. Throwing stops the moment discomfort appears, because the program is a series of doses and pain is the readout that the last one was too large.

The modern version of the program was rebuilt from workload data rather than from experience. Elbow varus torque per throw was calculated from the relationship between torque and throwing distance, and the schedule was redrawn around it Reinold 2024. The original program ran 136 days to a final chronic workload of 15.0.

The rebuilt program runs 217 days to a final chronic workload of 10.8. Its peak acute to chronic workload ratio fell from 1.61 to 1.33, and the share of days spent outside the conventional safe band fell from 18 percent to 9 percent. That workload ratio and the methodological arguments against it are carried by Injury Prevention and Load Management.

Long toss is not one dose

The same drill at a different distance is a different input. Seventeen college pitchers threw fastballs from a mound at 18.4 meters and threw 37 meters, 55 meters and maximum distance from flat ground Fleisig 2011. The 37 and 55 meter throws were made hard and on a horizontal line by instruction.

Those constrained flat-ground throws produced biomechanical patterns similar to pitching, which makes them reasonable rehabilitation work. The unconstrained maximum-distance throws did not. They produced the greatest shoulder external rotation at 180 degrees, the greatest elbow flexion at 109 degrees, and elbow extension velocity of 2573 degrees per second.

They also produced the highest torques of the session, 101 newton meters of shoulder internal rotation and 100 newton meters of elbow varus. Forward trunk tilt at release fell as distance grew. The instruction attached to the throw changes the dose as much as the distance does, which is why a throwing program specifies both.

The instrument the program is graded on

The number a program watches and the number the elbow pays come apart. Sixty high school and collegiate pitchers wore a validated inertial sensor through a structured long toss program at 90, 120, 150 and 180 feet, then pitched from a mound Leafblad 2019. Ball velocity changed significantly at every step.

Elbow varus torque did not. Mound pitching placed no more torque on the elbow than long toss from 120 feet and beyond. That is an argument for bringing the mound in earlier rather than treating it as the final and largest dose of the program.

The reliability figures decide how the program should be run. Within a thrower, the metrics were excellent across the whole progression, with intraclass correlation coefficients above 0.75. Between throwers, 91 percent had acceptable agreement for ball velocity and only 79 percent for elbow torque. Each athlete is a precise instrument about himself and a rough one about the group.

That is the case for grading a throwing program against the athlete's own record rather than a population schedule. The investigators framed the study around that question and titled it accordingly, asking whether one size fits all or whether these programs should be individualized.

08Load kept deliberately small

A Deliberately Small Load Can Carry a Large Message to an Arm That Cannot Take a Heavy One

An arm early in rehabilitation often cannot tolerate the load that builds strength. Restricting blood flow to a limb while it trains with light weight is one way an upper-extremity program answers that, and the upper-limb trials are worth reading closely because they disagree in an informative way.

Twenty-four healthy subjects were randomized to a six week low weight program performed with or without restriction on one extremity Bowman 2020. Against the untreated limb and against controls, the restricted limb gained 30 percent in shoulder scaption strength, 23 percent in flexion, 22 percent in abduction and 13 percent in grip.

Arm and forearm circumference rose in the restricted limb, and no adverse events were reported. The result that matters most for a program is the one nobody trained. Grip strength in the opposite untreated extremity was 9 percent greater than in controls, which the investigators read as a possible systemic effect.

The same input, two different answers

A clinical trial in injured shoulders gave a narrower result. Twenty-eight patients with rotator cuff tendinopathy were randomized to eight weeks of shoulder rehabilitation with or without blood flow restriction Kara 2024. The restricted group gained more biceps thickness and more internal rotation strength at 60 degrees per second.

Nothing else separated the groups, deltoid thickness and external rotation strength included. Rotator cuff and scapular retractor thickness, rotation strength, pain and function all improved in both groups over time. That is a partial null reported in full, and it is the more useful of the two studies for planning a program.

Healthy limbs and tendinopathic ones answered the same input differently, which is what a dosing account expects. The arm that can already take heavy load has no need of the restriction. The arm that cannot can still be given something. Where the other modalities sit on that same axis is the subject of Adjunctive Therapies.

09What we corrected

Four Claims Removed From This Page

This page previously set the healthy external to internal rotation ratio at 68 to 75 percent and the abduction to adduction ratio at 75 to 85 percent. Neither figure could be traced to a source. The measured values in 150 healthy professional pitchers are 65 percent at 180 degrees per second and 61 percent at 300, and those replace them Wilk 1993.

The elbow schedule has been rebuilt for the same reason. The page previously specified a braced pain-free window of ten to one hundred degrees, motion opening five to ten degrees per week, and full range by week six. None of those figures could be sourced. What replaces them is the published record of 13 rehabilitation protocols, most bracing for four to six weeks Ose 2024.

A second maintenance program was previously named alongside the Thrower's Ten. It does not appear in the indexed literature, so it is no longer named here, and the Advanced Throwers Ten is cited in its place Wilk 2011.

The page also carried a gold pull-quote attributed to Dr. Jason Dulberg that was not drawn from anything he said or wrote. It has been removed. Every claim here is either sourced to the literature or named explicitly as the model's.

10The model's claim

Upper-Extremity Dosing Is One Question Asked of Every Input, at Every Stage

Two layers run through this page and they should stay separate. The established science is the measured rotation ratios, the serratus anterior activation ladder, and the reflex latencies recorded in throwers. It is also the torque figures across throwing distances, the workload rebuild of the interval program, and the restriction trials. Each belongs to the investigators who ran it.

The Unified Model of Tone reads the throwing arm as a system under repeated dosing. Every exercise and every throw is an input with a magnitude. The model sets them on one axis that runs from the lightest manual perturbation to the maximum-effort throw. It asks a single question of each: does the magnitude of this input match what this system, right now, can use?

The overhead athlete is where that question is hardest to answer, because tissue tolerance and performance demand sit close together. A maximum-distance flat-ground throw is a rehabilitation drill in most programs and it produced 100 newton meters of elbow varus torque, the highest of any throw tested. There is very little room between the dose that rebuilds an arm and the dose that reinjures it.

The prediction this page makes

The model treats the readouts a throwing program already collects as several views of one organization rather than as separate qualities to be trained one at a time. That is a claim about how the return is organized rather than a claim about what treatment does, and the instruments to record it already exist in this literature.

Take one squad of throwers through an interval program. At each stage record joint position sense error in degrees at 90 degrees of abduction, and rotator cuff reflex latency in milliseconds against a sudden internal rotation perturbation. Record the external to internal rotation strength ratio and RMSSD recovery time after a standardized throwing bout at the same visits.

The model makes a directional call inside that design. The thrower whose reflex latency has not returned toward baseline will show elbow varus torque rising at a fixed throwing distance before ball velocity shows anything. Torque and velocity have already been shown to come apart across a long toss progression Leafblad 2019.

The comparison has to run within athlete, because that is where the instruments are precise. Within-thrower agreement across a long toss program exceeded an intraclass correlation of 0.75, while between-thrower agreement for elbow torque reached acceptable levels in only 79 percent of subjects. The variability readout itself is carried by Heart Rate Variability.

If joint position sense error, rotator cuff reflex latency, the external to internal rotation strength ratio and RMSSD recovery time move together within the same throwers across one program, the unification claim is confirmed.

11The tone reading

The Rebuilt Arm as One Regulated System

Three signatures of tone show up in upper-extremity rehab, each inside a measurement the program already takes.

The other foundations show up in the same program. Input quality is why an instructed horizontal throw at 55 meters and an unconstrained maximum-distance throw are different inputs at the same drill. Constraint is the braced arc that holds elbow motion inside a protected window. Coupling is the relationship that puts 101 newton meters of shoulder internal rotation torque and 100 at the elbow on the same throw. Set-point is the rotation ratio a program defends, measured at 65 percent in healthy professional pitchers. Prediction is the feedforward model the cuff runs before a deceleration it cannot wait to feel. Oscillation is the weekly rhythm of throwing days and rest days that a chronic workload is calculated from. The full framework is set out in the Unified Model of Tone.

12Where this sits

How This Page Relates to the Rest of the Library

Seven places this argument continues, each with the claim that earns the link.

13Questions athletes ask

Questions Athletes Ask

14The sources

References

15 primary sources, each linked to its record. Figures quoted on this page were checked against the published abstract.