What is a SLAP lesion, and why the superior labrum?

The superior labrum is not a simple fibrocartilaginous seal: it is the place where the tendon of the long head of biceps attaches to the glenoid. Understanding the SLAP lesion means understanding why that precise zone gives way, and why it hurts when other portions of the labrum tear in silence.

The glenoid labrum is a fibrocartilaginous ring that borders the glenoid and deepens its concavity. Its superior portion has a decisive anatomical feature: the tendon of the long head of biceps takes its intra-articular insertion there, continuing directly into the labral fibres rather than attaching to bone. The biceps and superior labrum complex therefore forms a single mechanical unit, and it is from that continuity that the lesion follows.

In 1990, Snyder and his colleagues identified, in a retrospective review of more than 700 shoulder arthroscopies, a constant injury pattern in 27 patients: a detachment of the superior labrum beginning posteriorly and extending anteriorly, stopping before or at the mid-glenoid notch, and taking the anchor of the biceps tendon with it. They named it the SLAP lesion, for superior labrum anterior and posterior.¹ The series comprised 23 men and 4 women, of mean age 37.5, with a mean delay of 29.3 months between the injury and surgery. That delay is not incidental: it says that even then, the lesion could not be recognised quickly.

A rare lesion, a frequent operation

The real frequency of SLAP lesions is modest and remarkably stable from one series to another. Snyder found 27 in more than 700 arthroscopies. A Belgian multicentre study of 530 glenohumeral arthroscopies found 32, that is an incidence of 6 %, which the authors point out is identical to Snyder's figure.² In that series, 53 % of the lesions were type II, and the mechanisms divided between traction (22 %), compression (28 %) and overhead sporting movements (25 %).

That rarity contrasts violently with the surgical volume. In the United States, between 2004 and 2009, 25,574 arthroscopic SLAP repairs were recorded in an insurance database, with the incidence rising from 17.0 to 28.1 per 10,000 patients carrying an orthopaedic code: a 65 % increase in five years.³ Three quarters of the patients were men, and the two age peaks were at 20 to 29 and 40 to 49, that second peak raising a question on its own, since it corresponds to the age at which asymptomatic labral lesions become common.

The sternest finding comes from the certification registry of the American Board of Orthopedic Surgery : from 4,975 repairs reported by young surgeons between 2003 and 2008, the authors conclude that the percentage of SLAP repairs declared represents three times the incidence the literature supports.⁴ In that same series, pain was absent at follow-up in only 26.3 % of patients and function judged normal in 13.1 %.

The particular case of the thrower

In the overhead athlete, the mechanism is not a fall but the repetition of an extreme movement. Burkhart, Morgan and Kibler described the peel-back mechanism: in the position of maximum cocking, with abduction and external rotation pushed to their limit, the direction of the biceps fibres changes and applies a torsional load to the posterosuperior labrum that progressively peels it off the glenoid rim.⁵ They argue that these lesions occur in the acceleration phase, in late cocking, and not during the deceleration of the follow-through as had been supposed.

The same work brings two ideas that structure the examination of the thrower. First, type II lesions are not homogeneous: there are three distinct forms, anterior, posterior and combined, and it is the posterior and combined forms that disable the thrower, through posterosuperior instability and anteroinferior pseudolaxity. Second, a tight posteroinferior capsule predisposes to these lesions; the authors propose the 180-degree rule, according to which a shoulder at risk of a dead arm has an arc of rotation of less than 180 degrees at 90 degrees of abduction.⁵ This glenohumeral internal rotation deficit, or GIRD, will be at the heart of rehabilitation.

Why this zone hurts

A rarely asked question deserves asking: why should a superior labral detachment be painful when so many structural abnormalities of the shoulder are silent? Immunohistochemistry work provides an element. In ten biceps and superior labrum complexes taken from cadaveric specimens, the distribution of neurofilament proved uneven, with the highest density in the anterior superior labrum, in front of the origin of the biceps.⁶ A second study, on eleven specimens, confirmed a significantly higher number and density of axons in the anterosuperior portion compared with the posterosuperior one, and demonstrated nociceptive fibres, marked with CGRP and substance P, along those axons.⁷

The authors draw from this a hypothesis that directly concerns the physiotherapist faced with a painful postoperative patient: the biceps anchor is a richly innervated region, and the placement of suture anchors could contribute to the variability of pain after repair.⁷ That does not prove that every SLAP lesion hurts, but it explains that a lesion of this zone can be a genuine generator of pain, and that the surgery that repairs it can create a new one.

What does the Snyder classification really change in practice?

Four types in 1990, ten today. The classification is essential for practitioners to understand each other, but it does not on its own say what should be done, and its reliability depends entirely on the tool with which it is applied.

Snyder divided superior labral pathology into four distinct types, defined on the arthroscopic appearance and not on the mechanism or the symptoms.¹ That classification has been revised as new configurations were identified: Maffet and his colleagues showed, in 84 patients with significant labral abnormalities, that 32 of them, that is 38 %, had lesions that fitted none of the four boxes, and proposed three further categories.² There are ten today.³

The ten types, and which ones count

The table below reproduces the full classification as published in an open-access illustrated review of 2023, which sets it against the arthroscopic images.³ The right-hand column, by contrast, is a reading of our own, based on Handelberg's multicentre series: it concludes that only lesions of types II, IV and V should be considered unstable and warrant fixation.⁴

The subtyping of type II, the only subdivision that changes the clinical picture

The subdivision of type II into anterior, posterior and combined forms is not a nosological nicety: it is the only distinction in the classification that changes what is expected on examination. Burkhart and his colleagues established it by observing that Jobe's relocation test is positive, with posterosuperior pain, in patients with a posterior or combined type II, and negative in those with an anterior type II.⁵ It is also the posterior and combined forms that disable the thrower, and that are most often associated with lesions of the articular side of the cuff.

The reliability of that subtyping has been measured. Among five experienced shoulder surgeons judging 90 video recordings of arthroscopy on two occasions, the intra-observer agreement for the three subvariants of type II reached a mean kappa of 0.598, described as moderate, and the inter-observer agreement a mean correlation of 0.804, described as substantial.⁶

Reliability depends on the tool, not on the classification

This is where the subject becomes instructive. The same classification, applied by the same categories of practitioner, gives opposite results depending on whether an arthroscopy or an MR arthrogram is being looked at.

At arthroscopy, among surgeons with more than ten years of practice, agreement is good: mean intra-observer kappa of 0.670 and inter-observer correlation of 0.804 when the choice is between a normal labrum and the four types.⁶ The authors conclude that the Snyder classification is a reliable system for identifying SLAP lesions, and they note in passing that diagnostic confidence was strongly correlated with the perceived quality of the video (mean Pearson correlation 0.718).

On MR arthrography, the picture collapses. In 20 selected arthrograms, classified into seven categories by three orthopaedic surgeons and three musculoskeletal radiologists on two occasions, overall inter-observer agreement was poor: kappa of 0.177 then 0.124.⁷ Between surgeons alone, the values ran from minus 0.056 to 0.114, and the analysis established that none of the values obtained among the orthopaedic surgeons could be considered statistically different from zero. The radiologists did better, with fair to moderate agreement (kappa 0.479 then 0.340).

An agreement that is not statistically different from zero means that the classification obtained is compatible with a coin toss. That is not a methodological nuance: it is the limit of what can be asked of an MR arthrogram.

The consequence for physiotherapy practice is direct. When an MR arthrogram report announces a « type II », that label carries considerable uncertainty, and it must not decide the course of action on its own. When it comes from an operative report, it is far more solid, but it then arrives after the surgical decision, not before.

Why does no clinical test diagnose a SLAP lesion?

This chapter is the heart of the subject, and it is disagreeable. The literature has produced a dozen provocative manoeuvres, almost all published with excellent figures, and almost none has held up when other teams tried to reproduce them. Here are the values as they were measured, including where they contradict each other.

Let us start with the result that is shortest to state. A meta-analysis pooled all the physical shoulder tests intended for SLAP lesions and calculated their overall diagnostic odds ratio, that is by how much they shift the probability of the disease. The result is 1.38, with a 95 % confidence interval of 1.13 to 1.69.¹ An odds ratio of 1 means a test brings nothing. The only merit of this 1.38 is to be statistically greater than 1: clinically, it shifts almost nothing.

In the same analysis, the best single test for SLAP lesions is the compression-rotation test, with a diagnostic odds ratio of 6.36, a sensitivity of 0.43 and a specificity of 0.89.¹ That is the top of the category, and it misses nearly six lesions out of ten.

The table of values, sourced line by line

The table below gathers the published diagnostic values. Each line carries its source, because averaging these figures would make no sense: they are not repeated measurements of the same quantity, they are measurements made on different populations, with different protocols and thresholds.

Why the original figures do not reproduce

Reading this table raises a question that must be put plainly: how can a test show 91 % sensitivity in one author and 46 % in another? Three mechanisms, all documented, explain it.

The first is the selection of the population. Liu assessed the crank test in 62 patients with shoulder pain resistant to three months of conservative treatment, explicitly excluding anterior glenohumeral dislocations and cuff tears.⁵ In other words, he removed from the sample the diagnoses that produce false positives. Stetson, for his part, tested 65 all-comer patients from a shoulder clinic and concludes that the results were often falsely positive in patients with other conditions, impingement and cuff tears included.⁴ It is not the test that changed, it is the population.

The second is design bias. A systematic review of 15 studies showed that the probable overestimation of accuracy came from the use of case-control designs, verification bias and reference standards of lower quality.¹⁷ Its authors identify six accurate tests, with spectacular positive likelihood ratios (Biceps Load I at 29.09; Biceps Load II at 26.32; jerk test at 34.71), but they immediately point out that these values come from single high-quality studies conducted in selected populations, and that further evaluation is needed before they can be used with confidence.

The third is inter-examiner reliability. Walsworth and his colleagues measured the inter-tester reliability of the crank, the anterior slide and active compression: it sits between 0.20 and 0.24.¹⁸ They draw from it the most economical explanation of the whole mess: the variability in the diagnostic accuracy reported from one study to another is partly explained by the modest reliability of these tests. Two clinicians who do not agree on the result of a test cannot produce the same sensitivity twice.

The harshest verdict, and what it does not say

Calvert and his colleagues applied Sackett's critical appraisal criteria to 15 eligible studies out of 29 identified. Only one satisfied them all. The confidence intervals of the positive and negative likelihood ratios contained the value 1. Their conclusion fits in a sentence: the literature used as a teaching resource in medical schools and in continuing education lacks the validity needed to be useful, and there is no good physical test for effectively diagnosing a SLAP lesion.¹⁶

« There are no good physical examination tests that exist for effectively diagnosing a SLAP lesion. » That is the literal conclusion of Calvert 2009, and no subsequent publication has overturned it.

A careful reading of Hegedus's update tempers without contradicting. It does identify a few promising tests, among them the passive distraction test, highly specific and coming from a study at low risk of bias. But it warns in the same paragraph that more than one shoulder test, active compression and Biceps Load II by name, was introduced with remarkable diagnostic statistics before subsequent research failed to reproduce the results of the original authors.¹⁵ Its general conclusion is that the use of any single test to make a pathognomonic diagnosis cannot be recommended without reservation.

What that verdict does not say, on the other hand, is that examining should stop. It says that the examination does not settle it alone, which is different, and that is the subject of the next chapter.

How do you examine anyway, and what are combinations of tests worth?

The previous chapter established that no test settles it alone. It does not follow that examining should be given up: it follows that what is expected of the examination has to change. The aim is no longer to prove the lesion, it is to guide a course of action and to rule out what is not ours.

The first thing to know about combinations is a rule of logic, and it is solid because it depends on no particular study. A 2023 meta-analysis calculated the performance of the manoeuvres not only in isolation but combined, in series and in parallel.¹ In series, every test must be positive for the combination to be so; in parallel, one is enough. The result is the one theory predicts: combinations in series improve specificity, combinations in parallel improve sensitivity. In that analysis, the combination of the O'Brien test and the crank test was the most sensitive of the two-test combinations, in series as in parallel; the combination of the Yergason and the anterior slide was the most specific.¹

That principle gives a way of using mediocre tests without deceiving oneself. Early in the reasoning, they are strung together in parallel so as not to miss anything; if the aim is to strengthen a hypothesis before undertaking anything invasive, they are all required to be positive.

The history counts for more than the manoeuvres

The most useful result of all this literature is perhaps this one: it is the combinations including an element of history that shift the probability most, not the combinations of manoeuvres.

Michener and his colleagues showed, in 55 patients with blinded arthroscopic confirmation, that the anterior slide alone had only limited usefulness (area under the curve 0.70; LR+ 2.25; LR− 0.44), but that combining it with a history of catching, clicking or locking took the positive likelihood ratio to 6.00 for confirming a type II to IV lesion.² That is more than double the best likelihood ratio published for a single test, the 2.81 of the compression-rotation test.

Walsworth obtained the same kind of result in another cohort of 55 patients. Combining a history of catching or locking with a positive crank or anterior slide gave specificities of 0.91 and 1.00, and positive likelihood ratios of 3.0 and infinity respectively.³ Symmetrically, requiring a single positive result in the combination of catching or locking plus anterior slide or crank gave sensitivities of 0.82 and 0.89, with negative likelihood ratios of 0.31 and 0.33: that is the configuration that allows a labral lesion to be reasonably ruled out when everything is negative.

The best likelihood ratio in the whole SLAP literature does not come from a manoeuvre, but from a question put to the patient: « does your shoulder catch, click or lock? »

That hierarchy is confirmed by the declared practice of surgeons themselves. In a survey of 175 shoulder specialist surgeons, cross-referenced with a systematic review of articles reporting SLAP repairs, diagnostic arthroscopy and the history were ranked as the two most important elements for making the diagnosis.⁴ The authors conclude that the diagnosis of SLAP is a matter of clinical impression, and that the literature as much as practice is variable and inconsistent on the criteria used.

The counterpoint, which must be known

It would be dishonest to present combinations as the solution. Cook and his colleagues tested five manoeuvres, in isolation and then grouped, in 87 patients with arthroscopic confirmation, in a deliberately strict methodological design intended to reduce the bias that usually inflates accuracy.⁵ Their conclusion is blunt: no grouping demonstrated better diagnostic accuracy than the results taken in isolation.

The two findings are not contradictory, they concern different things. Cook grouped manoeuvres with each other. Michener and Walsworth combined a manoeuvre with an element of history. It is that second combination that pays, and it is one more reason to begin the examination with the interview rather than with the hands.

The manoeuvres, described as their authors described them

Since they continue to be done, they may as well be done as they were validated. The descriptions below are those of the original publications, and not the second-hand versions that circulate in courses.

Active compression, known as the O'Brien test. Patient standing, arm flexed to 90 degrees, elbow in full extension, then adducted 10 to 15 degrees medial to the sagittal plane of the body, and internally rotated with the thumb pointing down. The examiner, standing behind the patient, applies a uniform downward force. The movement is repeated, arm in the same position, palm fully supinated. The test is positive if pain appears during the first manoeuvre and decreases or disappears during the second. Pain localised to the acromioclavicular joint, described as « on top », points to that joint; pain or a painful click described as « inside » the shoulder points to the labrum.⁶ That distinction between pain « on top » and « inside » is constitutive of the test: without it, the test loses what made it specific.

Biceps load II. Patient supine, arm elevated to 120 degrees and taken into maximum external rotation, elbow flexed to 90 degrees, forearm supinated. Elbow flexion against resistance is requested. The test is positive if the patient describes pain during the resisted flexion. It is negative if no pain appears, or if the pain already present on elevation and external rotation stays unchanged or decreases during the resisted flexion.⁷ That last point is the subtlety of the test, and it is the one most often forgotten.

Crank test. Arm elevated to 160 degrees in the plane of the scapula, an axial load applied along the humerus, with maximum internal and then external rotation.⁸

Resisted supination and external rotation. The test deliberately reproduces the peel-back mechanism, which gives it a mechanical plausibility the others do not have.⁹

Passive distraction. A precaution from the original protocol is worth knowing: the authors excluded 53 cases out of 319 because of elevation limited to less than 150 degrees or pain in the starting position.¹⁰ In other words, this test does not apply to a stiff or very painful shoulder, and using it outside its conditions means leaving the domain in which it was measured.

The decision tree that follows from this

What follows is not a learned society recommendation, since none exists on this precise point. It is a reading of the literature cited in this chapter, offered as a framework for reasoning.

Why does a positive MRI not make a diagnosis?

This is the most important point of this article for the course of action, and the one most often glossed over. Superior labral lesions exist in people who hurt nowhere, in considerable proportions, and that proportion increases with age. A positive image is therefore information, not proof.

Snyder already wrote it in the founding article: no imaging examination correctly defined superior labral pathology before the operation.¹ Thirty-six years on, imaging has improved considerably, but the problem has shifted rather than been solved. It is no longer a matter of whether MRI sees the lesion, but of what a lesion seen means.

What is found in shoulders that do not hurt

The most disturbing study is that of Schwartzberg and his colleagues.² They recruited 53 adults aged 45 to 60, with no history of surgery or trauma to either shoulder, all asymptomatic, and performed an MRI without injection on a randomly chosen shoulder. Two musculoskeletal radiologists, blinded to the aim of the study and to the age of the participants, read the examinations. They concluded that there was a superior labral tear in 55 % and 72 % of the cohort respectively.

Two details of that study deserve emphasis. First, the agreement between the two radiologists was only moderate (kappa 0.410): not only are many lesions found in healthy people, but there is no agreement on which. Second, the prevalence did not differ significantly by age within that band, nor by sex, nor by dominant side, nor by physical work, nor by playing an overhead sport. The authors explicitly conclude that these images may be normal age-related findings, and that physicians should remember this in order to avoid over-treatment.²

In athletes, the finding is older and just as clear. Miniaci performed bilateral MRI in 14 asymptomatic professional baseball pitchers, with no significant history of trauma, ten of them having a stable shoulder and full painless mobility. The labrum was abnormal in 79 % of the 28 shoulders examined.³ The authors conclude that non-injected MRI of the asymptomatic shoulder of a high-level pitcher reveals abnormalities that may constitute a spectrum of non-clinical findings, and that these data are useful precisely in order to separate symptomatic abnormalities from these variants.

The counterpoint, because it exists

It would be dishonest to present only the studies that point one way. An Australian community population study compared thirty people in three groups of ten: current shoulder pain, a history of pain, no history. Tendinosis and cuff tears were present in the majority of participants in each group, but labral abnormalities were rare in all the groups.⁴

How is that result to be reconciled with Schwartzberg's 55 to 72 %? Two explanations, not mutually exclusive. Size: ten asymptomatic participants do not allow a prevalence to be estimated precisely. And protocol: the definition of what counts as a superior labral tear, the sequence used, the field strength and the reader's readiness to conclude vary considerably. The disagreement between two radiologists on the same examinations, measured at a kappa of 0.410 in the American study, shows well enough that the boundary between normal and pathological is poorly defined here.²

What that divergence does not allow is the conclusion that a positive image amounts to a diagnosis. It demonstrates on the contrary that the prevalence of labral findings depends on who looks and how, which is exactly the argument against using the image as arbiter.

What MRI and MR arthrography are really worth

The performance of imaging has been measured, with arthroscopy as the reference. It is good on specificity, moderate on sensitivity, which defines an examination that confirms but does not rule out.

Ajuied's result is worth pausing on, because it contradicts a widespread practice. At 3 tesla, arthrography does not improve sensitivity for SLAP lesions compared with plain MRI, and it significantly degrades its specificity, from 0.99 to 0.92.⁷ Translated into clinical terms: injecting gadolinium to look for a SLAP produces more false positives, in a condition where the false positive is precisely the problem. The injection keeps its value for the anterior and posterior labrum, where it genuinely improves detection.

The anatomical variants that imitate the lesion

One last source of false positives is purely anatomical. The junction between the superior labrum and the glenoid normally shows variations that look like a detachment. A meta-analysis of 20 studies and 7,601 upper limbs gives the pooled prevalences: sublabral recess 57.2 %, sublabral foramen 13.5 %, Buford complex 3.0 %.⁹

A sublabral recess is therefore present in more than one person in two. Mistaking that recess for a labral detachment is the classic pitfall in reading an MR arthrogram, and a multicentre study had already warned as early as 1998 of the possible diagnostic and therapeutic errors in the face of anatomical variations creating a large sublabral hole.¹⁰

The same meta-analysis brings an interesting nuance: of those three variants, only the Buford complex is statistically associated with an increased risk of a SLAP lesion, with a relative risk of 2.4 (CI 1.3 to 4.7), and above all of type II in 95.5 % of cases.⁹ The sublabral recess and foramen are not associated with such a risk: they are merely confounding.

What place for conservative treatment as a first line?

The two previous chapters leave the clinician without diagnostic certainty. That uncertainty is not paralysing: on the contrary, it designates the trial of treatment as the most rational course, and the available data support it better than is generally believed.

The position of the authors who write the algorithms is explicit. Non-operative treatment is the first line for most young, active patients with no history of trauma, no mechanical symptoms and no overhead activity demands; it combines rest, avoidance of aggravating factors, injections and physiotherapy centred on correcting scapular dyskinesis, restoring range and strength, and assessing the throwing movement.¹

What rehabilitation actually produces

The most complete systematic review on the subject retained five articles totalling 244 athletes, 162 of them elite level or above, of mean age between 20.3 and 38.0.² Type II lesions dominated; baseball, softball and weightlifting were the sports most represented. Its results deserve reading in detail, because they contain vital information about how to conduct the treatment.

The overall return-to-sport rate is 53.7 % across all athletes, and 52.5 % in elite athletes. That is the figure usually quoted, and it is mediocre. But in athletes who were able to see their rehabilitation programme through to the end, that rate rises to 78 % overall and 76.6 % for the elite. Return to the previous level follows the same logic: 42.6 % overall, 72 % in those who completed the programme.²

The data that explain that gap are simple and telling: patients who abandoned the protocol in favour of surgery had received on average 8 sessions of physiotherapy, against 20 sessions in those whose conservative treatment succeeded.² That comparison is observational, and the naive causal reading must therefore be resisted: the patients who drop out are perhaps also the ones who were doing worst. But it at least puts a figure on what « a trial of rehabilitation » means, and eight sessions are not one.

Eight sessions in those who switch to surgery, twenty in those whose rehabilitation succeeds. A conservative treatment stopped halfway is not a conservative treatment that failed: it is a conservative treatment that was not delivered.

These results converge with an earlier case series. Edwards and his colleagues followed 39 patients with a clinically documented superior labral lesion, 19 of them treated without surgery, with a mean follow-up of 3.1 years.³ In those whose conservative treatment succeeded, function improved significantly (total ASES score from 58.5 to 84.7; p = 0.001), pain fell (visual analogue scale from 4.5 to 2.1; p = 0.043) and quality of life improved (EuroQol from 0.76 to 0.89; p = 0.009). All the patients with conservative success returned to sport; 71 % of all the athletes regained their previous level of play, but only 66 % of the overhead athletes.

The conservative protocol used in that study is described precisely, and it is the reference one: non-steroidal anti-inflammatory drugs combined with a physiotherapy programme centred on scapular stabilisation exercises and stretching of the posterior capsule.³ The authors also report honestly that 20 patients out of the 39 in the overall group, that is 51 %, were finally operated on and can be considered conservative failures.

Who responds poorly, and why knowing it changes the conversation

The same systematic review identifies the factors associated with failure of non-surgical treatment: older age, playing an overhead sport and particularly the position of baseball pitcher, a traumatic lesion, a positive compression-rotation test, associated rotator cuff involvement, a longer baseball career, a longer symptomatic period, and the presence of a Bennett spur.²

That list has a concrete use: it makes it possible to calibrate what is announced to the patient from the first session. A 45-year-old office worker with non-traumatic shoulder pain and an MRI showing a superior labral lesion has good reason to hope, and should be told that their image is probably an age-related finding. A professional pitcher with a traumatic lesion, a long career and associated cuff involvement has an unfavourable conservative prognosis, and should be told that too, otherwise a season is lost.

Modalities and level of evidence

The evidence table is thin, and it must be presented as such. A publication-type search in Europe PMC, run on 15 August 2026, returns thirteen indexed randomised trials mentioning « superior labral ». None of them compares a rehabilitation programme with no treatment or with another conservative modality : they concern surgical techniques, imaging protocols, the validation of a clinical test or postoperative analgesic modalities. The gap thus measured is a gap in MEDLINE indexing, and it must be put that way, but it is clear.

The internal rotation deficit, a measurable and treatable target

If one is looking in this file for an intervention supported by a randomised trial, it is towards the glenohumeral internal rotation deficit that one should look. A double-blind randomised trial of 2026 compared the sleeper stretch and the cross-body stretch in 35 overhead athletes aged 18 to 40, in competitive practice, with shoulder pain and a GIRD, at three sessions a week for four weeks.⁴

Both techniques reduced pain and the deficit, with no significant difference between them: the sleeper stretch group lost 2.17 points of pain (95 % CI −3.13 to −1.21) and the cross-body group 1.54 points (CI −2.53 to −0.55); the GIRD reduced by 14.69 degrees (CI −19.48 to −9.89) and 14.77 degrees (CI −19.70 to −9.84) respectively.⁴ Adherence, tolerance and affective responses were comparable.

That trial does not concern confirmed SLAP lesions, and that must be said: it concerns painful overhead athletes with GIRD, which is the population in which a SLAP lesion is suspected. It nonetheless establishes two useful things. The GIRD reduces by about fifteen degrees in four weeks of diligent work, which is clinically substantial when one recalls that Burkhart made the internal rotation deficit a predisposing factor. And the choice between the two techniques can be made on the patient's tolerance rather than on any supposed superiority.

The scapula, a constant component but never isolated

Every effective conservative protocol includes scapular work, and no trial has isolated it for a SLAP lesion. The best support available is a clinical reasoning algorithm proposed by Cools and his colleagues for the rehabilitation of scapular dyskinesis, from the office worker to the high-level overhead athlete, which stresses that general guidelines are lacking despite the well-established association between abnormal scapular positions and movements and glenohumeral pathology.⁵

In practice, this means that the content of scapular work is a matter of clinical reasoning and not of a validated protocol. It can be done seriously without pretending that it is demonstrated, and that is the honest formulation to give the patient.

Should the labrum be repaired or the long head of biceps tenodesed?

The surgical question has changed in nature over fifteen years. It is no longer only how to repair, but whether to repair, and a controlled trial asked the most embarrassing question of all: does surgery do better than sham surgery?

The trial against sham surgery

In 2017, Schrøder and his colleagues published in the British Journal of Sports Medicine a double-blind, three-armed randomised trial controlled by sham surgery.¹ The protocol deserves describing precisely, because its rigour is what gives the result its weight. One hundred and eighteen candidates for surgery, of mean age 40, with a history, clinical symptoms and an MR arthrogram all pointing to an isolated type II lesion, were randomised intraoperatively, once arthroscopy had confirmed the isolated lesion: labral repair (n = 40), biceps tenodesis (n = 39) or sham surgery (n = 39).

The primary outcomes at 6 and 24 months were the Rowe clinical score, from 0 to 100, and the Western Ontario Shoulder Instability Index, from 0 to 2100. The result is unambiguous: no significant difference between the groups, at any follow-up point, on any outcome. The between-group differences on the Rowe score at two years were 1.0 point between tenodesis and repair (95 % CI −5.4 to 7.4; p = 0.76), 1.6 points between tenodesis and sham surgery (CI −5.0 to 8.1; p = 0.64), and 0.6 points between repair and sham surgery (CI −5.9 to 7.0; p = 0.86).¹ Five patients developed postoperative stiffness after repair, four after tenodesis.

« Neither labral repair nor biceps tenodesis had any significant clinical benefit over sham surgery for patients with SLAP II lesions in the population studied. » The literal conclusion of Schrøder 2017.

The final restriction of that sentence must be read, and it is in good faith rather than an escape clause: in the population studied. That population had a mean age of 40, which is neither that of the college pitcher nor that of the 55-year-old patient. This trial does not say that SLAP surgery is useless in everybody. It says that at that age, in that indication, it did not prove a benefit beyond the act itself. That is already considerable.

The shift from repair to tenodesis

Independently of that trial, practice has moved. An analysis of a large private insurance database in the United States, covering 46,650 diagnoses of SLAP lesion between 2007 and 2016, of which 3,347 patients were operated on for an isolated lesion, documents the movement.² Isolated SLAP repair fell by 69.3 % over the period, while biceps tenodesis for that same diagnosis rose by 370 %. In patients over 40, the increase in tenodesis reaches 1,500 %.

That shift rests on revision data rather than on functional superiority. A retrospective propensity-score-matched cohort study, covering 11,081 arthroscopic repairs, 9,960 arthroscopic tenodeses and 9,420 open tenodeses, shows that patients who had a SLAP repair were 1.8 times more likely to undergo revision surgery than those who had an arthroscopic tenodesis (2.9 % against 1.6 %; p < 0.0001), and 1.6 times more likely to receive a subsequent diagnosis of a cuff lesion than those who had an open tenodesis (5.1 % against 3.2 %; p = 0.0002).³

But the same study documents the price of tenodesis, and it is rarely quoted: arthroscopic tenodesis is accompanied by 1.3 times more episodes of acute postoperative pain than repair (5.2 % against 4.0 %; p = 0.011), and open tenodesis by 1.8 times more acute pain (6.9 % against 3.8 %) and 1.3 times more shoulder stiffness (11.8 % against 9.0 %).³ For the physiotherapist, that is information from the field: the tenodesis patient more often arrives in pain and stiff, the repair patient more often goes back to theatre.

Shin's meta-analysis, the most specific one on the overhead athlete, deserves reading for exactly what it says.⁴ It included 13 studies after appraisal with the MINORS tool, requiring a minimum follow-up of two years and patients with SLAP lesions and no other major pathology. Postoperative ASES scores were slightly higher in the tenodesis group, the rates of return to sport and of return to the previous level slightly higher as well, the complication rate slightly lower, and none of those differences reached statistical significance. Its cautious conclusion is that tenodesis is not inferior to repair, and that high-level randomised trials remain necessary.

The thrower, the population in which the least is promised

All the data converge on this point, and it is the most important message to pass on to an overhead athlete before they decide.

A systematic review of 22 articles totalling 944 patients operated on with an arthroscopic repair establishes an overall return-to-sport rate of 69.6 %, a return to the previous level of 69.0 %, and a mean interval of 8.9 months plus or minus 2.4 (range 6.0 to 11.7 months).⁷ But the return rate in pitchers falls to 57.5 % against 87.1 % in non-pitchers.

A series of 30 overhead athletes followed for a mean of 3.5 years after a type II repair gives a still finer picture.⁸ The mean ASES score reached 87.9, which looks excellent, but the Kerlan-Jobe score, designed specifically for sporting performance, rose only to 73.6. In baseball and softball players, the drop between the two scores was significant (87.9 against 72; p = 0.006). The athletes estimated that they had returned to 84.1 % of their previous level, with a mean interval to resumption of 11.7 months, and 93.3 % declared themselves satisfied. In other words: everyday life is repaired, performance is not entirely, and the patient can be satisfied with both at once.

A systematic review devoted to the chances of returning to the previous level confirms this in terms of odds ratios: athletes who are not baseball players, not overhead athletes and not pitchers are 2.3 to 5.8 times more likely to return fully than overhead athletes and pitchers after isolated labral repair.⁹

Finally, a 2025 systematic review of 547 patients from 16 studies sums up the real state of knowledge in one word: variability. Return-to-play rates range from 37.5 % to 94.7 % after repair and from 35.3 % to 93.1 % after tenodesis; in pitchers specifically, from 40 to 80 % after repair, against 16.7 % after tenodesis, but on 34 pitchers only.⁵ The authors conclude that the results of the two techniques show massive variability in the pitcher.

Tenodesis in the overhead athlete has been the subject of a dedicated systematic review, covering 8 articles and 99 athletes, mostly baseball and softball players, of whom 0 to 44 % had failed a previous SLAP repair.¹⁰ The overall return-to-sport rate there reaches 70 %, with 69 % in recreational athletes, 80 % in competitive and college athletes, and 60 % in professionals. The authors conclude that it can be performed as a first line rather than a repair, including in younger patients, but that particular attention is needed for elite overhead athletes, and pitchers in particular, whose results tend to be poorer.

Which rehabilitation, in phases, and with what criteria for progression?

The status of this chapter must be stated at the outset: there is no validated rehabilitation protocol after SLAP repair, and the extent of the disagreement between the published protocols has been measured. What follows is a framework for reasoning, anchored to the few markers that are documented.

The most recent review on rehabilitation after arthroscopic shoulder surgery is explicit on the point: further research is needed to compare rehabilitation protocols after SLAP repair, and there is a shortage of high-quality evidence concerning the arthroscopic management of SLAP lesions.¹ The authors add that this knowledge gap probably underlies the variability observed in clinical practice.

The variability, measured

That variability has been quantified. A team collected 60 rehabilitation protocols after SLAP repair published by university orthopaedic programmes and compared them.² The results give the measure of the disagreement, but also of the implicit consensus.

Where there is agreement: 61.7 % of the protocols (37 of 60) recommend a sling for four to six weeks, and 65 % (39 of 60) include scapular strengthening. The authors note that immobilisation time and scapular strengthening are the least variable elements.² Where there is no agreement: 90 % of the protocols provide for full range, but the interval for reaching it varies considerably; only 33 % (20 of 60) recommend return to sport at 24 weeks, and 38.3 % (23 of 60) allow throwing at 16 weeks. The exercises also differ: pendulum exercises in 53 %, submaximal isometrics in 55 %.

The authors finally note a difficulty that should give pause: the absence of a precise definition of what « return to throwing » means functionally made comparing the protocols difficult.² That imprecision is also ours in everyday practice.

The four-phase framework

The framework below brings together the consensus elements of the protocols surveyed and the few documented biomechanical constraints. It does not constitute a validated recommendation.

The only genuinely argued biomechanical constraint

Of all the restrictions imposed after repair, only one rests on explicit, published mechanical reasoning. Burkhart and his colleagues explain that, for a reattachment of the posterosuperior labrum to resist peel-back, the suture anchors must be placed behind the biceps, at the angle of the glenoid, and that the repair must be protected against external rotation beyond neutral for three weeks, in order to avoid premature and excessive torsional loads from the peel-back mechanism.³

That constraint has a logical elegance: the position that created the lesion is the one that threatens the repair. It remains an expert recommendation founded on mechanics, not a trial result, and it must be presented as such to the patient who is surprised by the slowness.

Rehabilitation without surgery does not follow the same calendar

The postoperative calendar must not be applied to the non-operated patient, who has no reattachment to protect. In them, the only constraint is tissue tolerance, and the return-to-sport intervals reported are generally under six months.⁸

The content of the non-operative programme is described in the reference study: non-steroidal anti-inflammatory drugs and physiotherapy centred on scapular stabilisation exercises and stretching of the posterior capsule.⁹ What is known about the dose matters more than the detail of the exercises: twenty sessions in patients whose treatment succeeded, eight in those who switched to surgery.⁸

What do published clinical cases teach us?

Three indexed publications, four patients, chosen because each illustrates a point that the aggregated series make disappear: the success of conservative treatment in a pitcher who has already been operated on, the SLAP lesion that shows itself through a neurological deficit, and the rare form that requires surgery.

Case 1: the college pitcher, operated on once, treated without surgery the second time

The situation. A right-handed senior NCAA division 1 pitcher, 21, 186 cm, 84 kg, throwing three-quarters, consults his team physician before the season for pain in the front of the right shoulder that had set in throughout the autumn pre-season. He had had labral surgery on that same shoulder during his final year at school. The pain, of mild to moderate intensity, occurred mainly at the start of the acceleration phase of the throw, and affected both his velocity and his accuracy. A cortisone injection into the subacromial space relieved it only for a few days.¹

The assessment. No instability test was positive: neither sulcus, nor global instability tests, nor load and shift. MRI showed a posterior type II SLAP lesion. Passive glenohumeral internal rotation measured supine was 32 degrees on the right against 51 on the left, that is a deficit of 19 degrees, with a total arc of rotation 15 degrees lower on the dominant side: a characterised GIRD. The scapular dyskinesis test found moderate to severe prominence of the medial border of the scapula in flexion and extension.¹

What surface EMG added. This is what makes the observation original. Surface electromyographic analysis showed that lower trapezius was abruptly inhibited on the dominant side during the descending phase of flexion, compared with the non-dominant side, with a markedly higher upper-to-lower trapezius activity ratio on the dominant side during that same phase.¹

The management and the result. The off-season programme, run by the team's strength and conditioning coach, combined scapular stability exercises with light dumbbells and kinetic chain work: wall scapular slide with a band, abduction of the throwing arm with a lateral lunge, a free-movement kinetic chain exercise known as the « lawnmower », prone extension-abduction-external rotation with a 0.9 to 1.4 kg dumbbell, prone scapular row with 1.4 to 1.8 kg, supine « scapular punch » with 6 to 9 kg, side-lying external rotation with 0.9 to 1.4 kg.¹ The player pitched his final college season: 22 relief appearances in 57 games, four wins, an earned run average of 3.70 over 41.1 innings.

Cases 2 and 3: when the SLAP lesion speaks through a nerve

A physical and rehabilitation medicine team reported two observations of paralabral cysts linked to a type 2 SLAP lesion, sitting at the spinoglenoid notch, responsible for isolated entrapment of the infraspinatus branch of the suprascapular nerve.²

First patient, 62. Pain in the back of the right shoulder for three months, coming on abruptly while carrying a heavy load. Treated in the community with non-steroidal anti-inflammatory drugs, physiotherapy and rest, with no benefit at all. Examination found a painful arc limited to 160 degrees of active abduction and flexion, posterior pain worse on external rotation, with no passive limitation. Manual muscle testing was normal. The Neer and Hawkins tests were positive, as was the empty can test, and the radiographs unremarkable.²

It was electromyography, requested to rule out a cervical radiculopathy or a plexus lesion, that settled it: only infraspinatus showed fibrillations and positive sharp waves at rest, with polyphasic motor unit potentials and slightly reduced recruitment. MRI then revealed a type 2 SLAP lesion with a large paralabral cyst extending into the suprascapular and spinoglenoid notches.²

Second patient, 56. Pain in the right shoulder for a month, treated with anti-inflammatory drugs, physiotherapy, rest and corticosteroid injections, with minimal benefit. Examination this time found wasting of infraspinatus, normal strength in all the muscles tested except the external rotators, and severe pain rated 8 out of 10 while holding a steering wheel. The Neer, Hawkins, O'Brien and Speed tests were positive. Motor conduction studies showed low-amplitude responses on stimulation of the suprascapular nerve with recording over infraspinatus, and needle examination confirmed the denervation.²

Case 4: the rare form that requires surgery

The last case shows the other end of the spectrum. An 18-year-old professional footballer had a type IX SLAP lesion, that is to say a circumferential labral tear, in an athlete who is not a thrower.³ The decisive point is the chronology: it is the recurrence of the instability after previous non-operative treatment that led to the surgical indication. An arthroscopic pan-labral repair with suture anchors was performed. Three months after a personalised postoperative rehabilitation programme, the player had returned to competition at his previous level, with no recurrence of instability and no other symptom over the 18 months of follow-up.³

This case illustrates two things. First, that extensive labral lesions with true instability are not a matter for the reasoning developed in this article: when the shoulder subluxes, the dominant problem is the instability, and the course of action joins the one described in shoulder instability and recurrent dislocation. Second, that the classification keeps a descriptive usefulness: recognising a circumferential lesion rather than a type II genuinely changes the procedure.

What these cases do not prove

Four observations, published because they were remarkable. They tell us nothing about the frequency of the situations described, nor about the probability that a given patient will follow the same course. A published case is a case judged worth publishing, which is precisely the opposite of a representative sample.

Their usefulness lies elsewhere: they set markers for reasoning. The first says that a pitcher with a documented lesion can succeed without further surgery if what is modifiable is treated. The two that follow say that a shoulder that does not respond must reopen the diagnosis, and where to look. The last says that true instability changes the subject.

How do you apply this concretely in the clinic?

A patient arrives with an MRI announcing a SLAP lesion, or with shoulder pain and an overhead movement. Here is what the data in this article change at the first session, in what is measured, and above all in what is said.

What is done at the first session

Start with the history, and take it seriously. It is the only part of the examination whose diagnostic yield is established: combined with a manoeuvre, the question about catching, clicking or locking takes the positive likelihood ratio to 6.00, against 2.81 at best for a single test. Three elements to obtain precisely: the initial mechanism (a fall on the outstretched arm, traction, or gradual onset with a repeated movement), the presence of a mechanical phenomenon, and the exact phase of the sporting movement in which the pain appears.

Test, without over-interpreting. A reasonable sequence in parallel: O'Brien active compression with the « on top » versus « inside » distinction, Biceps Load II if the position is tolerated, and resisted supination external rotation in the thrower, because it reproduces the mechanism. All negative in a patient with no mechanical phenomenon makes the lesion less likely; all positive does not prove it.

Measure what is modifiable. This is the point that changes the session. Glenohumeral internal rotation compared with the sound side, the total arc of rotation, scapular control on the way up and on the way down, the strength of the external rotators. Those four measurements guide the treatment, whereas the debate about the presence of a labral detachment does not.

Look for what is not ours. Wasting of the infraspinous fossa, an external rotation deficit, frank apprehension, non-mechanical night pain. Both cases reported in the previous chapter had been treated for several weeks before electromyography corrected the diagnosis.

The differential diagnosis, and where to read on

What is said to the patient, and why the words count here

This condition is a textbook case of the weight of words, for a simple reason: the patient often arrives with a report asserting a lesion, in a context where that lesion exists in a majority of people who hurt nowhere. Three formulations are worth preparing.

On the image. « Your MRI shows an abnormality of the labrum. That is not neutral information, but nor is it proof that it is what is hurting you: in people of your age with no pain at all, the same thing is found in more than one case in two. So we are going to treat your shoulder, not your image. » That is not a consolation, it is a fact measured in 53 asymptomatic people.

On the diagnosis. « No examination test allows this lesion to be confirmed or excluded with certainty, and that is not a limitation of this practice: it is the conclusion of the reviews that have assessed all these tests. So we are going to reason on what we can measure and modify. » Saying the uncertainty does not weaken the care: it protects it from escalation.

On the duration. « Patients in whom rehabilitation succeeds have done about twenty sessions; those who switch to surgery stop on average at eight. You will judge at the end of the programme, not in the middle. » That is probably the most useful sentence of the first session, because it forestalls the abandonment that then produces a conservative « failure ».

Frequently asked questions about SLAP lesions

The questions patients ask, and two that colleagues ask.

Can a SLAP lesion heal without an operation?

The question deserves rephrasing, because it contains an assumption. A labral detachment does not reattach spontaneously, but that is not what determines the result. What the data show is that symptoms can disappear and function be restored without the structure changing: 78 % return to sport in athletes who see their rehabilitation through to the end, with significant improvements in pain, function and quality of life.¹ The right formulation is therefore: rehabilitation does not repair the labrum, it makes the shoulder functional and painless, which is the real objective.

How long before you know whether rehabilitation is enough?

No trial defines that interval. The available markers are indirect but consistent: patients whose conservative treatment succeeded had received about twenty sessions, against eight in those who switched to surgery², and the return-to-sport intervals reported after non-operative treatment are generally under six months.¹ A serious conservative trial is therefore counted in months and in tens of sessions, not in weeks.

Can you keep throwing during treatment?

The published case of the college pitcher is instructive: the programme was carried out in the off-season, and that is what made the following season possible.³ Continuing the painful movement while trying to correct scapular control amounts to sustaining what is being treated. Stopping altogether, on the other hand, is not necessary for painless movements below shoulder level.

Does age change anything?

A great deal, in both directions. After 45, a superior labral lesion seen on MRI has a substantial probability of being an age-related finding rather than the cause of the symptoms.⁴ Older age is also among the factors associated with failure of conservative treatment.² And the surgical strategy shifts: biceps tenodesis rose by 1,500 % in people over 40 between 2007 and 2016, while repair declined.⁵

Should an MR arthrogram be requested rather than a plain MRI?

Not to look for a SLAP. At 3 tesla, arthrography does not improve sensitivity for this lesion and significantly degrades its specificity, from 0.99 to 0.92.⁶ It keeps its value for the anterior labrum, where it genuinely improves detection. This request is the physician's, but the physiotherapist who receives the report gains from knowing what the examination could and could not show.

What is the difference from a long head of biceps tendinopathy?

The tendon and its anchor are continuous, which explains the confusion. Tendinopathy concerns the tendon itself, mainly in its groove, and is associated with cuff pathology in more than nine cases out of ten; its management is developed in the article devoted to it. The SLAP lesion concerns the insertion of that tendon on the glenoid labrum: it is a structural detachment whose crux is diagnostic. The two meet on tenodesis, which treats both problems by taking the tendon out of the joint.

If surgery has not beaten placebo, why do people still operate?

Three reasons, two of them legitimate. First, the trial against sham surgery covered patients of 40 on average with an isolated type II lesion, and its authors explicitly restrict their conclusion to that population⁷: it says nothing about the young patient with instability or about extensive lesions. Second, some forms, notably type V or IX with true instability, are a different problem. Finally, the third reason is a poorer one: the volume of repairs declared represents three times the incidence the literature supports⁸, and the registry in which that figure was measured reports pain absent in only 26.3 % of operated patients at follow-up.