


Upper limb
Elbow rehabilitation: assess and empower your patients
Tessadit Aissaoui · Marion Burnier
Updated · published
Ulnohumeral osteoarthritis: 25 % of radiographic forms after 50, 0.9 % symptomatic. Diagnosis, the ulnar nerve, physiotherapy and surgical options.

After 50, one adult in four has the radiographic signs of elbow osteoarthritis, but fewer than one in a hundred suffers from it: an image does not make the diagnosis.
54.5%of operated ulnar neuropathies at the elbow that are due to primary osteoarthritis
What follows covers each of these points in detail, with sources. It is there if you need it.
This topic is taught in a course: Elbow rehabilitation: assess and empower your patients, online with Tessadit Aissaoui.
1 / 10
9 article chapters · 41 min in total
Vocabulary◔ 2 min
“Omarthrosis” comes from the Greek for the shoulder; the word never designates the elbow.
Defining◔ 6 min
This osteoarthritis makes bone rather than destroying cartilage, and stays silent in most people.
Examining◔ 7 min
No clinical test is validated for this osteoarthritis: no published sensitivity or specificity.
Imaging◔ 3 min
Two views are enough for the diagnosis; the rest serves to prepare a procedure, not to confirm.
Ulnar nerve◔ 4 min
Elbow osteoarthritis is the leading cause of operated ulnar neuropathy, and decides the surgical timetable.
Rehabilitating◔ 5 min
No controlled trial of physiotherapy exists in osteoarthritis of the elbow: everything is transposed.
Surgery◔ 5 min
Debridement and osteocapsular arthroplasty are sufficient in the majority of symptomatic elbow osteoarthritis.
Clinical cases◔ 3 min
No case is reconstructed: two come from series, one is an identified published case.
In practice◔ 6 min
Rare, with no trial or recommendation, this osteoarthritis calls for a course of action written in advance.
Train on this



Upper limb
Tessadit Aissaoui · Marion Burnier
Next comes practice: the course that teaches this topic, with Tessadit Aissaoui.
Our courses on this topic may be covered by DPC or FIFPL, the French continuing-education funding schemes, depending on your profession and situation.
An elbow that hurts at the end of extension, that locks from time to time, in a fifty-year-old man whose work has long put strain on his arms: the picture is stereotyped, the condition is little known, and its name is almost always mangled. This article goes through primary osteoarthritis of the elbow, from its vocabulary to its rehabilitation, stating at each step what the literature supports and what it does not.
A synthesis written from primary sources, each checked one by one on PubMed. Every figure carries its reference where it is written; the full bibliography, with PMID identifiers, appears at the end of the article.
Elbow osteoarthritis in three figures
What three independent studies measured, and why these three numbers tell the whole story of the disease
Sources: Nakayama K et al. J Shoulder Elbow Surg 2022;31(1):123-132 (PMID 34454037) for the first two figures; Sato N et al. J Orthop Sci 2020;25(2):235-240 (PMID 31005383) for the third.
What to take away before reading the rest
A vocabulary error is only a detail as long as it does not change what is done. This one changes the word the patient types into their search engine, the page they open, and sometimes the joint they are told about.
“Omarthrosis” comes from the Greek ômos, the shoulder. The term means glenohumeral osteoarthritis and nothing else. The check is easy to make yourself: two German-language articles titled literally Omarthrose and Omarthrose - Pathogenese, Klassifikation, Diagnostik und Therapie are indexed in PubMed under the English headings “Omarthrosis” and “Shoulder Osteoarthritis”12. Both deal with the glenohumeral joint; neither deals with the elbow. Using this word for an elbow is naming one joint while designating another.
The correct term is humeroulnar osteoarthritis, or ulnohumeral, named after the compartment that the disease affects first. It is, moreover, the term used in the English-language literature, where the reference operation is called ulnohumeral arthroplasty3 and where the series speak of humeroulnar4 involvement. One also meets “cubital osteoarthritis”, vocabulary from before the revision of the anatomical nomenclature, when the ulna was called the cubitus: it is the same bone and the same joint.
Three compartments coexist at the elbow, and the disease affects them neither at the same time nor in the same way:
In the series of Doherty and Preston, the first to have properly described the primary form, radiographic involvement was present in the humeroradial compartment in 100% of elbows, humeroulnar in 96% and radioulnar in 85%4. The disease is therefore pan-articular on imaging, whereas the complaint is almost always that of a hinge that hits a stop.
Naming it correctly is not fussiness: a patient who comes in for an “omarthrosis of the elbow” has already read, before seeing you, pages that were about their shoulder.
The glossary to give the patient
Elbow osteoarthritis = humeroulnar osteoarthritis = ulnohumeral osteoarthritis = cubital osteoarthritis (old term). Omarthrosis = osteoarthritis of the shoulder, another joint, another management. If the radiology report speaks of “humeroradial osteoarthritis” or “radiocapitellar osteoarthritis”, it designates a compartment of the same disease.
An osteoarthritis unlike any other: it makes bone rather than destroying cartilage, it stays silent in most of those who carry it, and it strikes an identifiable occupational profile.
The defining feature of primary elbow osteoarthritis is that it is hypertrophic: the bony response largely dominates cartilage loss. Osteophytes develop at the points of impingement (the tip of the olecranon, the coronoid process, the radial head), and the olecranon and coronoid fossae fill in, sometimes through a genuine thickening of the membrane lining them. Meanwhile, the joint space remains largely preserved5.
This particularity is not anecdotal; it is what drives the whole strategy. Where an advanced knee osteoarthritis requires the surfaces to be replaced, elbow osteoarthritis can often be treated by removing what impinges: osteophytes, loose bodies, retracted capsule. The relatively spared cartilage explains the good results of conservative procedures and also explains why prosthetic arthroplasty remains, in this indication, a last-resort solution5.
The case-control radiographic series by Dalal quantified exactly what the image shows, comparing 50 osteoarthritic elbows with 50 elbows matched for age and sex. Olecranon osteophytes were present in 96% of patients, coronoid in 90%, radial head in 86%, and of the coronoid and radial fossae in 64%. Thickening of the olecranon fossa membrane concerned 68% of cases. Narrowing, for its part, predominated in the radiocapitellar compartment (58%) and remained rare in the humeroulnar compartment (16%). In controls, only the coronoid osteophyte was frequent (36%); the other signs were rare6.
What the radiograph shows, and what normal ageing shows too
Frequency of each radiographic sign: 50 osteoarthritic elbows against 50 controls matched for age and sex
Source: Dalal S, Bull M, Stanley D. J Shoulder Elbow Surg 2007;16(3):358-361 (PMID 17188910). Thin bars represent the controls; only controls with a coronoid osteophyte reach a notable frequency.
The epidemiology of this disease remained unknown for a long time, for lack of a study in the general population. It was established in 2022 by a Japanese survey conducted on a random draw from the municipal register of a town: 415 residents over 50 sampled, 318 retained after exclusion of any cause of secondary osteoarthritis, all radiographed in two views7.
Radiographic prevalence was 25.2% (27.3% in men, 23.2% in women) for a Kellgren-Lawrence grade of 2 or higher. The prevalence of symptomatic forms, for its part, fell to 0.9%. In other words, one adult in four over 50 carries radiographic signs of elbow osteoarthritis, and barely one in a hundred suffers from it. Progression with age is steady and rapid7.
Radiographic prevalence by age band
318 adults aged 50 to 89, drawn at random from a municipal register, radiographed in two views: Kellgren-Lawrence grade 2 or higher
Source: Nakayama K, Kato H, Ikegami S et al. J Shoulder Elbow Surg 2022;31(1):123-132 (PMID 34454037). Japanese cohort, mean age 69.2 years.
Two practical lessons come out of these figures. The first: finding osteoarthritis on an elbow radiograph does not prove that it explains the pain. At 75, the probability that any given elbow carries signs of osteoarthritis approaches one in three, and nine times out of ten these signs are silent. The second: this osteoarthritis is, in the authors' own words, a disease that most people “accommodate” by themselves7. It is the rarity of the symptom, and not the rarity of the lesion, that explains why it is so seldom seen.
The surgical series paint a very consistent portrait. In Antuña, 46 elbows operated on: 44 men to one woman, mean age 483. In Doherty, 16 patients: 14 men, 2 women, mean age 61, with involvement of the dominant elbow in 14 of them4. The review by Gramstad and Galatz summarises the classic background: a middle-aged man engaged in intense manual activity5.
The Japanese survey adds factors that were measured rather than assumed. Age and body mass index were independently associated with the presence of osteoarthritis of grade 2 or higher. The use of vibrating tools emerged as a specific associated factor for the most advanced forms, those of grade 47. This last point deserves to be kept in mind in the clinic: jackhammer, grinder, chipping hammer, hammer drill, but also chainsaw and brush cutter in grounds maintenance workers.
A detail of the Doherty series often goes unnoticed and yet is useful: these patients frequently had osteoarthritis elsewhere: second and third metacarpophalangeal joints in 62% of them, knees in 38%, hips in 31%4. Faced with an isolated osteoarthritic elbow in a young subject, it is therefore worth looking at the hands: associated metacarpophalangeal involvement points towards a more generalised arthropathy, and notably towards a search for haemochromatosis or chondrocalcinosis, two diagnoses that this picture can mimic.
The same authors also established a point that tradition contradicts: the role of trauma is difficult to support in the primary form4. Post-traumatic osteoarthritis exists, it is frequent and well documented, but it is another entity: it occurs after a fracture of the radial head, a complex dislocation or a fracture of the distal end of the humerus, and it has neither the same profile nor the same prognosis.
Four practical benchmarks
Figures from the two reference series on the primary form
Sources: Antuña SA, Morrey BF, Adams RA, O'Driscoll SW. J Bone Joint Surg Am 2002;84(12):2168-2173 (PMID 12473704); Doherty M, Preston B. Ann Rheum Dis 1989;48(9):743-747 (PMID 2802795).
Key points
There is no validated clinical test for this condition: no published sensitivity, specificity or predictive value. The diagnosis therefore rests entirely on a body of arguments, and it is better to know which ones carry weight.
Pain in terminal extension. This is the cardinal symptom, and the word “terminal” matters: the pain does not appear in the mid-range but in the very last degrees, when the tip of the olecranon hits a filled-in fossa. In the Antuña series, all 46 elbows operated on complained of terminal extension pain3. The patient rarely describes it in these terms; they say they can no longer “straighten the arm fully”, or that it catches when pushing a heavy door or carrying a shopping bag with the arm held straight.
Locking. It reflects the presence of intra-articular loose bodies, found on 69% of the radiographs in the Doherty series4 and present in 36 of the 46 elbows operated on by Antuña3. Fourteen of these 46 patients reported frank locking. It is a useful sign because it is specific and because it points straight to a simple surgical procedure, removal of the loose body, often very rewarding.
Signs of the ulnar nerve. Twelve of Antuña's 46 patients complained of them before the operation3. This chapter is important enough to have its own section further down; for now, remember that not looking for them means missing the most frequent and most disabling complication of this osteoarthritis.
What, on the other hand, is not the picture: constant pain at rest, prolonged morning stiffness, a hot swelling, bilateral involvement from the outset in a young subject. These elements should prompt a search for something else.
Loss of range begins with extension and then progresses towards flexion. The flexion contracture sets in insidiously, and the patient compensates for it so well with the shoulder that they do not notice it. Pronation-supination remains normal for a long time, which distinguishes this osteoarthritis from post-traumatic radioulnar involvement.
The classic benchmark of the “functional arc” comes from an experimental study of 1995: one hundred healthy upper limbs, an orthosis progressively limiting range, twelve activities of daily living to perform at each step. Forty-nine subjects out of fifty completed all the tasks with a range limited to 75-120° of flexion8. This is the figure the surgical literature has used ever since to judge the need for a procedure.
This benchmark has nonetheless aged, and the patient should be told so. A three-dimensional kinematic study of 2011 measured the ranges actually required by contemporary activities: the flexion arc required rose to 130° ± 7 for telephoning, with a maximum of 142° ± 3; pronation reached 65° ± 8 for typing on a keyboard and supination 77° ± 13 for opening a door9. The authors explicitly conclude that the functional range of the elbow is probably wider than was thought.
The functional arc of the elbow: the 1995 benchmark and what today's activities demand
Flexion-extension range required by the tasks, measured by three-dimensional optical capture
Sources: Vasen AP, Lacey SH, Keith MW, Shaffer JW. J Hand Surg Am 1995;20(2):288-292 (PMID 7775772) for the 75-120° arc; Sardelli M, Tashjian RZ, MacWilliams BA. J Bone Joint Surg Am 2011;93(5):471-477 (PMID 21368079) for the contemporary tasks. The “table activities” band represents the order of magnitude common to both studies and is not a measured value.
| Diagnosis | What suggests it | What rules out osteoarthritis |
|---|---|---|
| Lateral epicondylalgia | Pain at the lateral epicondyle, provoked by resisted wrist extension and by gripping | Range preserved and painless at end range; no locking |
| Medial epicondylitis | Medial pain on resisted flexion-pronation | Painful tendon point rather than articular; no sign of impingement |
| Isolated cubital tunnel syndrome | Paraesthesiae of the 4th and 5th fingers, positive Tinel, worse with prolonged flexion | Normal radiograph; but the association is the rule rather than the exception |
| Radiocapitellar plica | Lateral pain and snapping; in the Lee series, radiocapitellar tenderness and pain in terminal extension in 65% of patients10 | Younger patient, normal radiograph, plica thickened on MRI (3.7 ± 1.0 mm)10 |
| Osteochondritis dissecans of the capitulum | Adolescent or young adult, throwing sport or gymnastics, locking | Age, sporting context, focal lesion of the capitulum on imaging |
| Inflammatory arthritis (rheumatoid arthritis, spondyloarthritis) | Prolonged morning stiffness, synovitis, polyarticular involvement, inflammatory markers | Osteoarthritis is mechanical; the joint fluids in the Doherty series were non-inflammatory4 |
| Microcrystalline arthropathy (gout, chondrocalcinosis) | Acute flares, effusion, calcific line, metabolic background | Acute and recurrent character, crystals on fluid analysis |
| Olecranon bursitis | Fluctuant posterior swelling, well circumscribed, superficial | Normal joint range, extra-articular swelling |
| Post-traumatic osteoarthritis | History of fracture of the radial head, dislocation or distal fracture of the humerus | It is another entity, with a different prognosis and different indications4 |
Red flags: stop rehabilitation and refer
Decision tree for a chronic painful elbow in adults
A pathway built from the signs reported in the cited series; it is not a prospectively validated algorithm
Editorial construction based on: Antuña 2002 (PMID 12473704), Doherty 1989 (PMID 2802795), Nakayama 2022 (PMID 34454037), Lee 2018 (PMID 30016689). No diagnostic algorithm has been prospectively validated in this condition.
Key points

X-rayLateral elbow radiograph: osteophytes on the radial head and on the anterior and posterior aspects of the ulnohumeral joint, typical signs of elbow osteoarthritis.
Source : Biswas et al., Arthritis, 2013, figure 1 · CC BY
Two views are enough for the diagnosis. The rest of the imaging does not serve to confirm; it serves to prepare a procedure or to rule out something else. And the classifications that the report uses are not all equally reliable.
Two views, AP and strict lateral, are enough to establish the diagnosis. What we look for is exactly what the Dalal series quantified: osteophytes of the olecranon, the coronoid process and the radial head, filling of the olecranon and coronoid fossae, loose bodies, and a joint space whose relative preservation is a positive argument and not an argument against6.
The strict lateral is the one that counts: it shows the filling of the fossae and the impingement osteophytes, that is, precisely what explains the extension pain. An isolated AP view can massively underestimate the disease.
The reading trap comes down to one sentence: the coronoid osteophyte is present in 36% of controls of the same age6. It is the only sign that normal ageing produces often enough to be misleading. The others (olecranon osteophyte, filling of the fossae, thickening of the membrane) are, for their part, discriminating.
Three systems circulate in reports and publications. A reproducibility study published in Osteoarthritis and Cartilage compared them on 81 cases, with each file scored by four orthopaedic surgeons, twice11. The results deserve to be known before basing a decision on a stage.
| Classification | Imaging | Inter-observer agreement (ICC) | Interpretation |
|---|---|---|---|
| Hastings and Rettig | Radiograph | 0.544 then 0.582 | Moderate agreement: two readers may classify the same elbow differently |
| Broberg and Morrey | Radiograph | 0.620 then 0.656 | Moderate to substantial agreement, the better of the two radiographic systems |
| CT-based classification | CT scan | 0.867 then 0.909 | Almost perfect agreement; correlated with the visual analogue scale (r = 0.754) and with the Mayo score (r = −0.614) |
The practical consequence is direct: a radiographic stage is not a solid datum, and the difference in severity between two successive reports may be only a difference between two readers. If a surgical decision hinges on the stage, it is the CT scan that decides, and it is also the one that, in this study, correlates best with what the patient feels.
Key points
This is the subpopulation that justifies a chapter on its own: elbow osteoarthritis is the leading identified cause of operated ulnar neuropathy, and it is also the variable that decides the surgical timetable.
A Japanese multicentre series reviewed the records of 413 elbows operated on for ulnar neuropathy at the elbow, diagnosed on the clinical and electrophysiological criteria of the American Association of Electrodiagnostic Medicine. A lesion of the elbow was present in 75.1% of cases, and the most frequent of all was primary elbow osteoarthritis, found in 54.5% of elbows. Next came the medial ganglion (8.5%) and cubitus valgus (6.5%), themselves most often associated with osteoarthritis14.
More worrying: the severity. McGowan grade III involvement, that is, with motor deficit and muscle wasting, concerned 50.8% of osteoarthritic elbows against 35.0% of elbows without an associated lesion14. In other words, when osteoarthritis is the cause, the nerve is on average more damaged at the time it is operated on. The slow onset of bony compression lets the patient get used to it, and the diagnosis comes late.
Faced with an osteoarthritic elbow, the question is not “should the ulnar nerve be examined?” but “how fast is it deteriorating?”
Compression was long explained by narrowing of the cubital tunnel in flexion. A dynamic computed tomography study measured this in vivo, scanning 13 osteoarthritic elbows with cubital tunnel syndrome and 25 osteoarthritic elbows without, in three positions: full extension, 90° of flexion, full flexion12.
The result reverses the classic explanation. The surface area of the humeral osteophytes and their proportion inside the tunnel increased markedly with flexion: 24.7 mm² and 49.9% in full flexion in symptomatic patients, against 9.0 mm² and 31.3% in extension. And these values were significantly higher in symptomatic patients than in the others. The authors conclude that it is the effect of the medial osteophytes (mainly humeral) on the nerve, and not narrowing of the tunnel, that explains the compression12.
The practical consequence is immediate and worth explaining to the patient: it is prolonged flexion that makes it worse, because it brings the osteophyte up against the nerve. Sleeping with the arm bent, long phone calls, driving with the elbow bent resting on the door, working at a keyboard with the elbows tightly flexed: these are the positions to correct, and this correction is the first therapeutic step for the associated neuropathy.
There is one randomised trial, on the cubital tunnel in general, not specifically on osteoarthritic forms. Seventy patients with mild or moderate symptoms were divided into three groups: night splint, nerve sliders, or control group. All received information on the cause of the symptoms and on the positions to avoid. At six months, 51 of the 57 patients followed (89.5%) were improved, and there was no difference between the three groups on any of the variables measured: neither the Canadian Occupational Performance Measure, nor the pain scales, nor strength, nor electrophysiology13.
In this trial, where all patients received the information, neither the night splint nor nerve sliders brought any additional measurable gain; with 57 patients followed, this does not prove the absence of an effect. A 2025 trial comparing neurodynamic techniques (sliders and tensioners) with a sham, single-blind (136 patients at baseline, 91 having completed the protocol), reports an improvement in all the parameters measured in the treated group and in only one, the shear modulus, in the sham group; it does not compare against information alone, and the two trials do not answer the same question28. Information and avoidance of provoking positions remain the starting point: this is an argument for investing the first session in explanation.
What this trial does not say, on the other hand: nothing on severe forms, nothing on osteoarthritic forms specifically, and nothing on McGowan grade III, which, as we have seen, represents half of the osteoarthritic elbows operated on. With an established or progressing motor deficit, the timetable is no longer that of rehabilitation.
The 2025 update of the Cochrane review on ulnar neuropathy at the elbow (15 trials, 970 participants) remains based, for conservative treatment, on isolated trials; it concerns idiopathic neuropathy, not elbow osteoarthritis27.
What should trigger a surgical opinion without waiting
Key points
This chapter begins with a confession, because omitting it would amount to manufacturing a certainty: to date there is no controlled trial of physiotherapy in primary elbow osteoarthritis. What follows is therefore reasoned transposition, and each proposition carries the level of evidence that really belongs to it.
The search for this article was conducted on PubMed by crossing elbow osteoarthritis with exercise, rehabilitation, corticosteroid injections, hyaluronic acid and platelet-rich plasma: no relevant randomised controlled trial exists for this location. The reference reviews confirm this by implication: the one by Del Core and Koehler recommends starting with “activity modification, anti-inflammatories, hand therapy and corticosteroids”15, without being able to back this sequence with anything other than a consensus of practice.
The patient must therefore be told where what is offered to them comes from. Two indirect sources are legitimate and a third is not:
Objective 1: preserve the useful arc. Loss of extension is the first to set in and the hardest to recover. Daily maintenance work, through range, without forcing against a hard stop, aims to delay the onset of the flexion contracture. The benchmark to give the patient is not “you must straighten the arm completely” but the functional arc: 75 to 120° of flexion allows almost all everyday activities8, while knowing that contemporary activities, telephoning first among them, require more9.
Objective 2: protect the ulnar nerve. Correct positions of prolonged flexion, at night as at work, on the basis of the mechanism demonstrated by dynamic computed tomography12. This is probably the intervention with the best benefit-to-effort ratio of the whole armamentarium.
Objective 3: reduce the impingement load. Analyse the occupational and sporting movements that bring the elbow into terminal extension under load, and modify them: carry loads against the body rather than with the arm outstretched, bring loads closer to the body, split up the use of vibrating tools, adapt the height of the work surface.
Objective 4: maintain strength without provoking. Strengthening is done in the pain-free range, avoiding end ranges. The triceps in terminal extension against resistance is precisely what should be avoided at the outset, since it compresses the posterior stop.
A safety principle, stated as reasoning and not as proof
Faced with a hard, bony stop at end range, forced passive mobilisations have no mechanical reason to work: an osteophyte cannot be deformed by pulling on it. They do, however, expose the patient to post-mobilisation inflammation and worsening of pain. This position is not supported by a trial (there is none) but by the nature of the obstacle. Faced with a soft stop, on the other hand, the capsular component justifies patient range work, over months rather than weeks: in the Lindenhovius trial, the arc was still gaining between the sixth and the twelfth month17.
The ranking below applies the GRADE principles (nature of the study design, directness of the evidence, precision, risk of bias) to the specific question of primary osteoarthritis of the elbow. This is an editorial judgement: no formal GRADE assessment has been published for this location, and no learned society has issued a recommendation concerning it.
Information and avoidance of positions of prolonged flexion, when mild to moderate ulnar neuropathy is associated. One randomised trial, where all patients received the information: 89.5% improvement at six months, with no measurable gain from the night splint or from nerve sliders13; a 2025 trial finds neurodynamic techniques (sliders and tensioners) better than a sham28. Rank to be read with caution: no arm was deprived of information, and the population is not exclusively osteoarthritic.
Prolonged range-of-motion work on stiffness with a capsular component, with or without a splint. Randomised trial on post-traumatic stiffness: gain of 47° to 49° of arc at twelve months, with no difference between progressive static and dynamic splinting17. The evidence is indirect for osteoarthritis, where the obstacle is partly bony.
Disease education and structured land-based exercise programme. Core treatments of osteoarthritis of the knee, hip and polyarticular forms according to OARSI 201916, but the elbow is not covered by these recommendations. An acknowledged extrapolation.
Activity modification and occupational load management. Recommended by the reference reviews as first-line treatment155; consistent with the measured risk factor that vibrating tools represent7. No efficacy data of its own.
Muscle strengthening outside end range. No trial for this location. Justified by functional reasoning and by the preservation of cartilage, not by demonstration.
Intra-articular corticosteroid injection. Cited as a non-surgical option by the reviews15, without any trial in elbow osteoarthritis. Note: perioperative corticosteroids are among the risk factors for complications of elbow arthroscopy18.
Analgesics and anti-inflammatories. Common use, mentioned by the reviews15. OARSI does not recommend oral anti-inflammatories in the presence of cardiovascular comorbidity or frailty, and strongly advises against opioids16.
Viscosupplementation, platelet-rich plasma, shock waves in elbow osteoarthritis: systematic search conducted for this article, no trial identified. Offering one of these treatments here means offering an extrapolation from another joint, and the patient is entitled to hear it.
Forced passive mobilisations against a hard stop. No data, and no plausible mechanism: an osteophytic block does not yield to stretching. The benefit/risk ratio is unfavourable by construction.
Key points

X-rayElbow radiograph at an advanced stage of osteoarthritis: destruction of the articular surface and intra-articular loose bodies, arrows, corresponding to the most severe radiographic grade.
Source : Casimiro et al., Acta ortopedica brasileira, 2016, figure 4 · CC BY
The physiotherapist does not indicate surgery, but is almost always the one who sees the patient for long enough to spot the moment when it becomes reasonable, and the one to whom the follow-up is entrusted.
The relative preservation of the cartilage changes everything. The surgical reviews agree on a gradation that has no equivalent at the knee or the hip: debridement and osteocapsular arthroplasty are sufficient in the majority of symptomatic elbow osteoarthritis, distraction arthroplasty has its place in the young patient at an advanced stage, and total replacement is reserved for end-stage forms in elderly, low-demand patients19.
Concretely, what the surgeon removes is what impinges: osteophytes of the olecranon and coronoid, loose bodies, retracted capsule, with, in the technique known as Outerbridge-Kashiwagi, a fenestration of the olecranon fossa that restores the travel of the olecranon. Open or arthroscopic, the principle is the same.
The reference series is that of Antuña: 46 elbows operated with an ulnohumeral arthroplasty between 1986 and 1996, reviewed at 80 months on average. The flexion-extension arc went from 79° to 101°; at last follow-up, 35 of 46 elbows (76%) were painless or only slightly painful. The Mayo score was excellent for 26 elbows, good for 8, fair for 4 and poor for 83.
The British series by Phillips, with a mean follow-up of 75 months on 20 ulnohumeral arthroplasties, gives convergent results: 85% good or excellent results on the DASH, 65% on the Mayo score, a flexion contracture reduced by 10° on average and 20° of flexion gained. Of the 16 patients who were working at the time of the operation, 12 (75%) returned to the same job20.
These gains in range, about twenty degrees, may seem modest. Set against the functional arc, they are not: going from 79° to 101° of arc means crossing the threshold that separates a hand that no longer reaches the face from a hand that gets back to it.
What surgical debridement gives back, and what it does not
Two independent series, mean follow-ups of 80 and 75 months
Sources: Antuña 2002, J Bone Joint Surg Am (PMID 12473704): 13 of 45 patients reported postoperative ulnar symptoms, that is 29%, of whom 6 were reoperated; Phillips 2003, J Bone Joint Surg Br (PMID 12729106); Tat 2022, J Shoulder Elbow Surg (PMID 35189372), survival without conversion to total replacement at ten years, 98.5% after open debridement and 98.0% after the Outerbridge-Kashiwagi procedure.
A systematic review of 18 articles totalling 625 patients and 634 elbows compared the two approaches. Both improve pain, the flexion-extension arc and functional scores. The open approach gains more in flexion; arthroscopy gives more homogeneous results. The price differs: complications 5.1% open against 2.0% arthroscopic, revisions 9.5% against 5.6%21.
An earlier controlled, non-randomised trial went in the same direction with a useful nuance: arthroscopic debridement relieved pain better, the Outerbridge-Kashiwagi technique improved flexion better22. The choice is therefore made on the patient's dominant complaint as much as on the surgeon's habit.
The specific risk of elbow arthroscopy deserves to be known and passed on, because it is specific to this joint: a systematic review of 52 articles reports 1.5% to 11% total complications and 1.26% to 7.5% nerve injuries, with obesity, age over 65, previous surgery and perioperative corticosteroid injections as risk factors18. No level I or II study exists on the subject.
A systematic review of 49 studies and 1,995 total elbow replacements measured what this procedure gives depending on the indication. For rheumatoid arthritis and fractures, long-term survival is satisfactory. For osteoarthritis, the reported rates are 50% complications and 11% revisions23. Conversely, survival without conversion to a prosthesis after simple debridement reaches 98% at ten years24.
The calculation is quickly made for a fifty-year-old man who works with his arms: a total elbow replacement is not fitted in a patient who will ask it to carry loads. The review by Gramstad and Galatz puts it bluntly: prosthetic arthroplasty is rarely indicated in primary osteoarthritis and should be reserved for elderly, low-demand patients in whom the other options have failed5.
No trial has compared rehabilitation protocols after debridement of the osteoarthritic elbow. What can be backed by evidence concerns stiffness: in the randomised trial by Lindenhovius on 66 patients, the flexion arc improved by 29° at three months, 40° at six months and 47° at twelve months, with no difference between dynamic splinting and progressive static splinting17. The message the authors draw from it is explicitly this: patience is justified.
Two points of vigilance specific to this surgery should guide the first weeks:
Key points
Three documented situations, taken as they were published. None is reconstructed for the example: the first two are series data, the third a reported case with its identifier.
Case 1: Elbow osteoarthritis revealed by a claw hand
Spalkit S, Sinha A, Prakash M, Tripathy S. BMJ Case Reports 2021;14(5):e24277326 (free full text: PMC8103949).
A 59-year-old woman, obese (body mass index 35.6), consults for weakness of the right hand evolving over two years, with paraesthesiae and numbness of the fourth and fifth fingers. She describes a weak grip and difficulties cooking and cutting.
Examination finds a moderate ulnar claw of the last two fingers, weakened grip and hypoaesthesia in the ulnar territory. The electromyogram shows a right ulnar axonal neuropathy, with sensory and motor conduction velocities not recordable. Ultrasound shows bowing of the nerve in the cubital tunnel with focal thickening, synovial thickening, a joint effusion and medial osteophytes arising from the olecranon. Radiographs show extensive degenerative changes with narrowing and large marginal humeroulnar osteophytes. MRI confirms ulnotrochlear osteophytes protruding into the cubital tunnel and displacing the nerve.
Treatment consisted of neurolysis with anterior transposition of the ulnar nerve, followed by regression of the paraesthesiae and numbness at three months.
What this case teaches. The reason for consultation was not the elbow: it was the hand. Two years passed before the diagnosis, and the involvement had become axonal. In any patient who describes a weak ulnar hand, examining the elbow and radiographing it costs little, and it is sometimes the only path to the cause.
Case 2: The collective picture of 46 operated elbows
Antuña SA, Morrey BF, Adams RA, O'Driscoll SW. J Bone Joint Surg Am 2002;84(12):2168-2173 (PMID 12473704). This is a series, presented here as a group portrait and not as a single patient.
Forty-six elbows, 45 patients, 44 men to a single woman, mean age 48. All described pain in terminal extension. Fourteen reported locking, twelve ulnar symptoms. At surgery, loose bodies were removed in 36 elbows; a capsular release was necessary in 19; a procedure on the ulnar nerve in 8.
At a mean follow-up of 80 months, the arc had gone from 79° to 101°, and 76% of elbows were painless or only slightly painful. But thirteen of 45 patients reported postoperative ulnar symptoms, six of whom had to be reoperated.
What this series teaches. The identikit is reliable: man, in his fifties, dominant elbow, impingement pain. And the ulnar nerve is the common thread: present before the operation in a quarter of patients, appearing or worsening afterwards in nearly a third. The authors draw an operational instruction from it: in a patient whose extension is limited by more than 60° or whose flexion does not reach 100°, consider preventive decompression.
Case 3: Sixteen patients followed without surgery, and a favourable course
Doherty M, Preston B. Ann Rheum Dis 1989;48(9):743-747 (PMID 2802795), free full text (PMC1003867).
Sixteen patients, 14 men and 2 women, mean age 61, with no obvious trauma. Symptom onset dated on average from the age of 53 and had lasted seven years on average, with extremes ranging from one to twenty years. The joint fluids taken from six of them were non-inflammatory; two biopsies showed non-specific synovitis.
Of the 26 affected elbows, 22 (85%) had a favourable clinical course. Thirteen of the sixteen patients had osteoarthritis elsewhere, notably in the second and third metacarpophalangeal joints (10 of 16).
What this series teaches. It is the indispensable counterweight to the two previous cases. The disease most often does well without surgery, over years, with symptoms that fluctuate. To announce surgery straight away to a patient whose elbow is bothersome but does not lock, and whose nerve is fine, is to ignore what this series documented more than thirty years ago.
Key points
A condition that is rare in the clinic, with no therapeutic trial and no recommendation, calls for a course of action written in advance; otherwise each practitioner reinvents their own.
| What is measured | Why | Pace |
|---|---|---|
| Flexion, extension, pronation-supination in degrees | The only objective marker of the disease; the arc conditions the surgical indication | Every 4 to 6 weeks |
| End-feel | A soft stop that hardens signals progression of the bony block | At every assessment |
| Ulnar examination (sensation, pinch strength, muscle bulk) | The most frequent and most disabling complication; its progression tips the balance towards surgery | At every assessment, and at every session if signs are present |
| Locking (number and circumstances) | Sign of loose bodies, whose removal is the most rewarding procedure | History-taking at every assessment |
| Mayo score (MEPS) or QuickDASH | Instruments used by the surgical series on the elbow, which makes the values comparable with the literature | Initially, then every 3 months |
| Work capacity | Return to the job is the most concrete outcome criterion: 75% after surgery20 | Initially, then at each change |
Recognition as an occupational disease: an often overlooked avenue
The schedule No. 69 of the general scheme, entitled “Conditions caused by vibration and shock transmitted by certain machine tools, tools and objects and by repeated impacts of the heel of the hand on fixed elements”, explicitly mentions “osteoarthritis of the elbow with radiological signs of osteophytosis”, with a 5-year time limit for coverage25. A minimum duration of exposure is required depending on the heading: it is for the occupational physician and the attending physician to build the file. This avenue directly concerns the profile identified by the epidemiology, since the use of vibrating tools is associated with grade 4 forms7. The site devotes a detailed article to the procedure, using the shoulder as the example: acromioclavicular osteoarthritis and occupational disease.
No: bony lesions do not regress. But the patient's question is in reality about symptoms, and there the answer is more encouraging. In the Doherty series, 22 of the 26 affected elbows had a favourable clinical course without surgery4. And at population level, the disease is radiographically present in one adult in four over 50 whereas fewer than one in a hundred suffers from it7.
Rarely stop it, often adapt it. What provokes it is end range under load and prolonged flexion. Split up the use of vibrating tools, bring loads closer to the body, avoid pushing with the arm outstretched and sustained bent-elbow positions. The use of vibrating tools is the only occupational factor formally associated with advanced forms in the general-population study7.
Because sleeping with the arm bent brings the medial osteophytes against the ulnar nerve. Dynamic computed tomography showed it: their surface area inside the cubital tunnel goes from 9.0 mm² in extension to 24.7 mm² in full flexion12. Correcting the night-time position is the first treatment; in mild to moderate forms, information alone did not give a result significantly different from that of a night splint in a randomised trial, which does not prove that the splint is useless13.
It is cited as an option by the reference reviews15, but no trial has evaluated it in elbow osteoarthritis. It must therefore be presented for what it is: an individual therapeutic trial, with no efficacy data specific to this location. Note also that perioperative corticosteroid injections are among the risk factors for complications of elbow arthroscopy18.
Months, not weeks. The most informative randomised trial on elbow stiffness (post-traumatic, and therefore with a capsular component) shows an arc gain of 29° at three months, 40° at six months and 47° at twelve months17. The authors explicitly conclude that patience is justified. In osteoarthritis, where part of the obstacle is bony, one must be even more cautious in what one promises.
That is very unlikely. At the elbow, the surgical strategy consists first of removing osteophytes and loose bodies, and survival without conversion to a prosthesis reaches 98% at ten years after simple debridement24. Total replacement remains reserved for end-stage forms in elderly, low-demand patients: fitted for osteoarthritis, it shows 50% complications and 11% revisions in the reference systematic review23.
No. It is in fact the most frequent situation: 25.2% radiographic prevalence after 50 against 0.9% symptomatic forms7. And an isolated coronoid process osteophyte is found in 36% of control subjects of the same age6. An image is not a disease.
Yes, provided they are clear about what is demonstrated and what is not. Two things rest on trials: for the ulnar nerve, information and correction of provoking positions, which all the patients in a randomised trial received13, and prolonged range-of-motion work on stiffness with a capsular component17. The rest (maintaining the arc, strengthening outside end range, load management) comes under coherent clinical reasoning, which the patient is entitled to know as such.
To go further on this site
Each reference was individually checked in PubMed: existence of the identifier, accuracy of the journal and year, and reading of the abstract to check that the source does establish what is attributed to it.
Method and limitations
The sources were searched in PubMed via the NCBI E-utilities services, crossing elbow osteoarthritis with epidemiology, diagnosis, imaging, ulnar neuropathy, exercise, injections and surgery. Each identifier was resolved and each abstract read before citation. Main limitation: the literature on this location is almost exclusively surgical and retrospective (only one randomised trial cited here concerns the elbow itself, and it concerns post-traumatic stiffness, not osteoarthritis). The evidence levels in the table of modalities are an editorial judgement applying GRADE principles, and not a published GRADE assessment. Article written on 14 August 2026.