Primary idiopathic shoulder stiffness: Frozen shoulder

Introduction and terminology

Shoulder stiffness was first described as a clinical entity in the late 19th century by both Duplay in France and Putnam in the United States. It was originally called scapulohumeral periarthritis, a term that included several painful afflictions of the shoulder resulting in stiffness. As mentioned in the previous section on general principles of shoulder stiffness, there is some confusion regarding the terminology used to describe the various presentations of shoulder stiffness. In 2016, Itoi et al. presented the consensus from the International Society of Arthroscopy, Knee Surgery and Orthopaedic Sports Medicine upper extremity committee on the definition, classification, and treatment of the stiff shoulder. Following the guidelines established by this group, we believe the term frozen shoulder should be used exclusively to describe the primary idiopathic stiff shoulder. Therefore frozen shoulder and primary idiopathic stiff shoulder are equivalent and refer to a condition that develops without any trauma or a specific shoulder disease. If a patient has a condition, such as diabetes, that may be linked to a stiff shoulder but not known to cause the stiffness specifically, it should still be considered idiopathic and not categorized as secondary frozen shoulder. The term adhesive capsulitis does not fully reflect the pathology present in this condition and has been frequently used to describe both the idiopathic frozen shoulder and shoulder stiffness related to other causes.


Frozen shoulder affects 2% to 5% of the general population. Several studies have shown that it presents most commonly in women between the ages of 40 and 60, affecting 3.38 women and 2.36 men per 1000 person-years. Recently, Kingston et al. compared 2190 patients with frozen shoulder with a sex-matched control group. Women compromised 58.4% of patients with an average age at presentation of 56.4 years (standard deviation, 13.1). Although most patients were 40 to 70 years old, 25% of frozen shoulder patients were younger than 40 or older than 70 years.

Some sort of comorbidity is present in more than 80% of the patients with a frozen shoulder. In their multivariate analysis, Cohen and Ejnisman identified hypothyroidism, diabetes, nephrolithiasis, and cancer as the only risk factors statistically related to frozen shoulder. Frozen shoulder is associated with hypoparathyroidism with an ongoing discussion if it is just a manifestation of hypoparathyroidism or if primary hypoparathyroidism and frozen shoulder have a common genetic or immunologic basis.

There is a clear relationship of frozen shoulder and diabetes in the literature. Tighe and Oakley reported a prevalence of diabetes (38%) and prediabetes (33%), measured with blood testing, much higher than in the normal population in a study of 88 patients with frozen shoulder. Austin et al. similarly reported the prevalence of diabetes as 10 times higher in patients with idiopathic primary shoulder stiffness. Other possible clinical scenarios related to a stiff shoulder include Parkinson disease, Dupuytren contracture, immobility, and neck and cardiac surgery. Although it is not common, frozen shoulder can be present bilaterally, more commonly in people with diabetes. Very rarely, the same shoulder may be affected twice with idiopathic primary stiffness. One scenario that has been rarely reflected in the literature, but is seen in clinical practice, is the development of frozen shoulder in patients with elbow or wrist pathology, treated surgically or conservatively. In fact, any patient with a long-standing elbow or wrist problem should be carefully examined for shoulder stiffness because some pain and reduced motion can be detected and may require treatment.

Patients with frozen shoulder may have severe pain, which leads to depression and anxiety. It is interesting to recognize that psychological symptoms are more related to the degree of pain than to the degree of motion loss.

Despite this condition being first described almost one century ago, the ultimate cause of frozen shoulder is still unknown. As described in the first section of this chapter, there is some evidence that cytokines are involved in the inflammatory and fibrotic process by a persistent stimulus of the matrix-bound transforming growth factor-β. Capsular tissue has a high density of cells, and there is a significant transformation from fibroblasts to myofibroblasts. Contracture of the rotator interval in frozen shoulder results clinically in loss of flexion and external rotation in adduction. When the anterior capsule is examined arthroscopically in patients with frozen shoulder, it is frequently seen adhered to the middle glenohumeral ligament and the subscapularis, resulting in further limitations of external rotation at the side. , The involvement of the anterior and posterior bands of the inferior glenohumeral ligaments further reduces shoulder elevation. The posterior capsule may be involved later during the development of a frozen shoulder, limiting internal rotation. ,

Clinical history

No specific criteria allow precise diagnosis of frozen shoulder. This explains the confusion on terminology and multiple classification systems that have been proposed. The diagnosis is usually made based on the presence of insidious shoulder pain in the absence of any anatomic or radiographic abnormality or history of trauma. Shoulder range of motion (ROM) is progressively reduced both actively and passively compared with the unaffected side. The diagnosis of primary idiopathic stiff shoulder requires a high index of suspicion, as pain and some degree of stiffness may be present in other pathologic entities. A careful history and clinical exam are of paramount importance to exclude other sources of pain because there are no fully reliable laboratory or radiographic tests. Frozen shoulder should be considered whenever a patient presents with some degree of shoulder pain and evidence of a limited ROM both actively, but more importantly, passively ( Fig. 18.1 ). Night pain and pain with sudden movements of the affected arm are typical of this condition. The only other condition that may present with progressive pain and stiffness is glenohumeral arthritis. However, the arthritic shoulder has a long-standing history of pain, whereas frozen shoulder patients complain of pain for the past few months or weeks. Shoulder arthritis can easily be excluded with a single radiograph of the shoulder. Additional patient characteristics and risk factors may help in confirming the diagnosis of this condition: in a middle-aged female patient with diabetes, significant shoulder pain, and limited passive motion, the diagnosis of frozen shoulder should be strongly considered.

Fig. 18.1

A 51-year-old woman with significant stiffness in all planes of motion without history of trauma. Patient refers to progressive loss of motion associated initially with night pain that progressed to pain with activities.

Physical examination

The most consistent and relevant finding when examining a frozen shoulder is the corresponding loss of both active and passive ROM. It is critical that before any provocative maneuver is performed, the patient is asked to demonstrate painless active ROM that is compared with the opposite unaffected arm. Passive ROM is then assessed and objectively recorded. The ROM exam for external rotation and forward elevation is better performed in the supine position because this position stabilizes the scapula against the chest wall.

Depending on the evolutive phase in which the patient is seen, the degree of stiffness may vary, but if the examination is meticulously performed, some degree of limited motion can always be detected. It is important to avoid any pain while motion is examined and to record the fixed reduction in ROM not influenced by pain. Typically a global reduction of active and passive ROM in all planes is found on examination, but external rotation loss is commonly an initial symptom. ROM, including elevation, external rotation in adduction, and internal rotation, should be concisely recorded in consecutive visits to monitor the clinical improvement.

After painless ROM is accurately recorded, a complete examination of the shoulder and cervical spine should be done. Neurologic compromise, signs of previous surgery, or history of a recent trauma should be ruled out because they could lead toward the diagnosis of secondary stiffness. Acromioclavicular joint and biceps groove palpation should be undertaken to detect additional symptoms from those areas.

The strength of the rotator cuff is usually normal if the standard manual motor tests are performed within the pain-free ROM. Causing pain while testing the rotator cuff by provocative maneuvers is the main confounding factor leading to a wrong diagnosis. In our experience, the vast majority of patients with frozen shoulder come with a previous diagnosis of rotator cuff pathology and with multiple imaging tests because the inexperienced examiner went directly to provocative tests performed within the painful arc of motion.

Considering that primary idiopathic shoulder stiffness is caused by capsular inflammation and fibrosis and that the capsule is extensively innervated, it seems logical to consider the possibility of eliciting shoulder pain by either palpating or stretching the capsule. In 2010, Carbone et al. described the clinical “coracoid pain test” to detect frozen shoulder. The test is positive when the digital pressure around the lateral aspect of the coracoid area causes a more intense pain compared with other areas such as the acromioclavicular joint or anterior acromion. This test elicits pain by direct palpation of the rotator interval area, just lateral to the coracoid. Although in the authors’ experience the test showed high sensitivity and specificity, it was also positive in 11% of patients with rotator cuff symptoms and it was not possible to use it in obese patients. In our experience, this test has not been really useful to make the diagnosis of frozen shoulder. However, eliciting pain by palpation of the rotator interval area may be helpful in considering the diagnosis of frozen shoulder during the early stages of the disease, when ROM has not been yet significantly reduced.

Wolf and Cox described the “external rotation test,” in which the patient’s arm is brought into external rotation with the elbow flexed to 90 degrees and the arm adducted. This maneuver was positive in patients with both frozen shoulder and glenohumeral arthritis. Noboa et al. described a variation of the external rotation test: the passive external rotation stretch test. This provocative test is performed with the patient standing up, the arm adducted, and the elbow bent at 90 degrees. From this position, the patient’s arm is placed in maximal external rotation while the examiner holds the wrist with one hand and, with the other, keeps the elbow adducted until the maximum painless point of the rotation is reached. From this point of maximum painless external rotation with the arm in adduction, a short abrupt stretch movement is made in external rotation. The test is considered positive if it causes significant pain in the shoulder ( Fig. 18.2 , ). The passive external rotation stretch test showed a sensitivity of 100% (95% confidence interval [CI], 91.8% to 100%) and a specificity of 90% (95% CI, 82.4% to 94.8%). The positive predictive value was 0.62, and the likelihood ratio was 10.22 (95% CI, 5.5% to 19.01%). False-positives were found only in patients with subscapularis tendinopathies or glenohumeral arthrosis. This clinical test seems to be useful not only for initial diagnosis but also to monitor the clinical progression of patients with frozen shoulder because it tends to become negative as the capsular irritation subsides.

Fig. 18.2

Passive external rotation stress test. This clinical test is performed by progressively rotating the adducted arm externally, without provoking any pain, until the point of maximal passive rotation is gently achieved. In this point, a sudden increase in external rotation causes significant pain, which is described as sharp within the joint.

Laboratory studies

Patients with shoulder stiffness do not need routine laboratory studies for diagnosis or management. On rare occasions when undiagnosed diabetes may be a concern, the physician may order a fasting glucose level. , However, a recent study by Safran et al. has shown that patients diagnosed with idiopathic frozen shoulder who are 60 years or younger have a similar probability of having diabetes or prediabetes than an age-matched asymptomatic population. Therefore no routine diabetic workup is warranted for these patients. Other biochemical mediators associated with frozen shoulder, such as matrix metalloproteinases, tissue inhibitors of metalloproteases, transforming growth factor-β, and interleukin-6, may be ordered in experimental studies, but they are not currently indicated in clinical practice.

Imaging studies


Simple radiographs are routinely ordered in patients with shoulder pain and stiffness. In idiopathic primary shoulder stiffness, they typically do not show any relevant findings. However, they are useful to rule out posttraumatic sequelae, glenohumeral arthritis, or osteonecrosis. Calcium deposits in the rotator cuff may be incidentally noted on the radiograph. It is essential to understand that this finding should not lead surgeons into the wrong diagnosis of calcifying tendinitis. Humeral head osteopenia may be present in half of the patients with frozen shoulder and has been attributed to the inflammatory process. A recent study discussing the indication of performing simple radiographs in patients with the clinical diagnosis of frozen shoulder has shown that simple radiographs are unnecessary unless the patient has a clinical history suggestive of other pathologies.


Ultrasound (US) is being more frequently used in the diagnosis of shoulder stiffness. When compared with magnetic resonance imaging (MRI), it offers the benefits of a shorter examination, lower cost, and the ability to use it in patients with metallic implants. However, it should be noted that in their consensus statement, the European Society of Musculoskeletal Radiology does not recommend US to evaluate patients with suspected adhesive capsulitis. It seems reasonable to state that US is not required for diagnosing frozen shoulder, but it allows examination of the rotator cuff in patients with symptoms that may suggest a tendon problem. US examination of shoulders with frozen shoulder has shown an increase in the coracohumeral ligament (CHL; 1.4 vs. 3 mm) and inferior capsule (1.3 vs. 4 mm) thicknesses between subjects without shoulder pathology and patients with frozen shoulder. Kim et al. demonstrated a significant increase in inferior capsule thickness (4.4 vs. 2.2 mm; P < .001) and a limitation of subacromial gliding in patients with frozen shoulder when compared with the nonaffected shoulders. Other findings in US examination include the presence of a hypoechoic soft tissue with hypervascularity on power Doppler within the rotator interval, reflecting an area of inflammation.


The arthrographic findings of the capsule and coracoclavicular ligament thickening present in frozen shoulder were first described by Kernwein et al. Later, Neviaser described decreased capsular distension (usually less than 10 mL), decreased volume of the axillary recess, and early contrast extension to the biceps tendon sheath. In patients with frozen shoulder, these injections may be very painful compared with routine glenohumeral joint injections. Currently, conventional arthrography is rarely used and has largely been replaced by magnetic resonance arthrography (MRA).

Magnetic resonance imaging

Establishing the diagnosis of primary idiopathic shoulder stiffness does not require an MRI. However, MRI may be useful for diagnosis in the early stages of the disease when symptoms are subtle, and patients may be misdiagnosed with rotator cuff pathology. Some consistent MRI findings have been reported in patients with frozen shoulder. MRA or MRI with intravenous contrast medium is less commonly performed because they are invasive, but they provide higher sensitivity and specificity. ,

One of the most consistent findings in MRI of patients with frozen shoulder is the thickening of the rotator interval capsule and CHL ( Fig. 18.3 ). In the study by Mengiardi and colleagues, a rotator interval capsule thickness of 7 mm or greater and a CHL thickness of 4 mm or greater were indicative of adhesive capsulitis. More recent data proposes that a 3-mm threshold of CHL thickness provides the highest accuracy for adhesive capsulitis diagnosis by MRA. These findings seem highly specific for frozen shoulder, but their sensibility is low. In the early stages of the disease, synovitis of the rotator interval may also be seen.

Fig. 18.3

Sagittal oblique magnetic resonance imaging of a patient with frozen shoulder, showing thickening of the rotator interval tissue (yellow arrow) and a very thick anterior, inferior, and posterior capsule (white arrows) .

(Courtesy Drs. Tapia and Bernabeu.)

Inferior glenohumeral ligamentous complex thickening is another typical finding of MRI in frozen shoulder ( Fig. 18.4 ). Capsular thickening as quantified in the axillary pouch relates to the loss of external rotation in elevation. , As previously stated, the diagnosis of primary idiopathic stiffness should not be based exclusively on MRI findings because there are conflicting data concerning the sensitivity and specificity of capsular and synovial thickening. , Nevertheless, a capsular and synovial thickness greater than 4 mm visualized on nonarthrographic coronal oblique T2-weighted images grants a sensitivity of 70% and a specificity of 95% for the detection of frozen shoulder. Using MRA, Jung et al. established that an axillary pouch capsule thickness greater than 3 mm, together with an obliterated axillary recess, generates a diagnostic accuracy of 89% for frozen shoulder.

Fig. 18.4

Coronal T1 magnetic resonance imaging of a patient with frozen shoulder, showing thickening of the inferior glenohumeral ligament (large arrow) . Note the close vicinity of the axillary nerve branches (small arrow) to the capsule.

Signal hyperintensity of the inferior glenohumeral ligamentous complex on standard MRI T2-weighted fat-suppressed sequences was both highly sensitive (85%) and specific (88%) for the presence of adhesive capsulitis. The presence of high signal in the axillary pouch correlates with pain intensity. , , It should be noted that the entire glenohumeral joint capsule and glenohumeral ligaments may be involved during the course of the disease, but the rotator interval capsule and the axillary pouch are the most commonly affected areas at imaging.

One additional MRI sign suggestive of frozen shoulder is the obliteration of the normal-appearing fat between the coracoid process and the CHL on the sagittal oblique images. This subcoracoid triangle sign was not found to correlate well with clinical impairment, , but it is more commonly observed early in the disease.

According to a recent meta-analysis performed by Suh and colleagues, the MRI findings with the highest diagnostic accuracy for frozen shoulder diagnosis are the rotator interval and axillary joint capsule enhancement and the inferior glenohumeral ligament hyperintensity. However, there are available data comparing findings on MRA in patients with and without frozen shoulder that did not identify any specific diagnostic findings for frozen shoulder.

Natural history

Frozen shoulder is considered a self-limiting disease by many authors. , , In his original description, Codman stated that full recovery could be confidently expected. However, we know this is not true because there are some patients with residual mild shoulder pain and stiffness and others show very mild or no improvement over time.

As described in the previous section on general principles of shoulder stiffness, three phases in the evolution of a frozen shoulder have been classically described: the freezing , frozen, and thawing phases ( Fig. 18.5 ). , In the thawing phase, functional ROM slowly returns and pain resolves. Complete resolution of symptoms may take 5 to 25 months. Often some degree of motion restriction may remain much longer. , Patients should be advised about this potential residual mild shoulder stiffness after a frozen shoulder.

Fig. 18.5

The traditional stages of primary, idiopathic frozen shoulder with symptom progression and resolution.

Ten percent of the 78 patients followed by Meulengracht and Schwartz had referred persistent pain up to 3 years from the beginning of symptoms. Furthermore, Reeves found significant residual stiffness in 12% of their patients. Patients tend not to complain of limited motion even if there is an objectionable mild restriction in motion, but they will be unsatisfied if there is residual pain, especially if they have concomitant comorbidities. Patients with insulin-dependent diabetes for more than 10 years tend to have a worse outcome. Hand and colleagues followed 269 frozen shoulders after a mean follow-up of 4.4 years and found that only 59% considered their shoulders normal or near normal. The rest had mild symptoms. In contrast, Vastamäki and colleagues reported 94% of normal shoulders with benign neglect, with a mean duration of symptoms of 15 months.

In a very well-designed recent systematic review on the natural history of untreated frozen shoulder, Wong et al. concluded that the literature does not support the idea of frozen shoulder as a self-limiting condition going through predetermined progressive phases. Although there is some evidence that patients usually show an early recovery in pain and function, moderate-quality evidence from longitudinal studies describe that improvement slows over time and may result in limitations that can last for long periods. We agree with this statement, especially in diabetic patients, in whom residual stiffness and pain when sudden capsular stretching is performed are commonly observed if left untreated.

Management of frozen shoulder

Frozen shoulder can be a very disabling condition. Significant pain and reduced function may influence the ability to develop daily living tasks. Patients commonly become frustrated because they cannot work, participate in leisure activities, or even sleep comfortably. Although the literature seems to support that frozen shoulder should be considered a self-limiting disease, the reality is that it may take up to 2 years before resolution occurs. A recent study by Kim et al. found that 28% of their patients still complained of some discomfort at a mean follow-up of 42 months. The longer the patient had symptoms, the more likely there was residual pain.

Understandably, most patients are reluctant to accept a “wait and see” attitude. We believe that once the diagnosis is made, some form of treatment must be offered immediately to the patient. Very importantly, the first message that patients should receive before initiating treatment is that their involvement in understanding and complying with their management is of paramount importance.

Many factors may influence the selected treatment of a patient with frozen shoulder when first seen: phase of the disease, duration of symptoms, prior treatments, degree of functional disability, or significant sleep disturbance. Although there is no consensus on an optimal treatment, there appears to be a consensus that some form of treatment, nonoperative or operative, is indicated in any patient with a frozen shoulder.

Most patients with frozen shoulder, independently of the stage in which they are first seen, will initially undergo a trial of nonoperative treatment, including a combination of physical therapy, oral medication, or injections. Operative treatment is considered when there is no improvement after 6 to 8 weeks of conservative treatment and may include manipulation under anesthesia (MUA), arthroscopic release and, very rarely, open surgical release.

Nonoperative management


Nonsteroidal antiinflammatory drugs (NSAIDs) usually represent the first line of treatment for patients presenting with a painful, stiff shoulder. NSAIDs can reduce pain levels in all phases of the disease due to both their analgesic and antiinflammatory effects. However, there is no strong evidence that they modify significantly the course of the disease. Typically, NSAIDs are combined with other analgesics such as acetaminophen or narcotics. Some authors have shown that standard analgesia is more effective than NSAIDs in controlling pain levels.

Oral steroids have been used for the treatment of frozen shoulder, especially during the freezing phase. However, oral administration has been less effective than an intra-articular injection. Buchbinder et al. compared the short-term results of a daily dose of 30 mg of prednisolone against placebo for patients with adhesive capsulitis. Clinical outcome favored the use of an oral steroid for pain, function, and self-perception scores at 3 weeks, but the benefit was not maintained beyond 6 weeks. Similar results were found by other authors, with short-term night pain relief but no improvement in ROM or daily pain up to 8 months. Even with short-term use, oral medication has side effects that should be discussed with the patient, including gastrointestinal, renal, and cardiac events, or even osteonecrosis with the use of low doses of oral steroids. ,

We favor the use of selective cyclooxygenase-2 inhibitors initially, together with physical therapy, because they seem to be more effective than other NSAIDs in reducing night pain. When NSAIDs are contraindicated, we may elect to prescribe a short course of an oral steroid (deflazacort). Usually, these patients present with intense shoulder pain and significant sleep disturbance, have not improved with other analgesia, and are not willing to undergo an injection.

Physical therapy

Physical therapy is routinely prescribed in all patients with a frozen shoulder when pain and stiffness have been present for less than 3 months or when they have not received any previous treatment. The mainstay of physical therapy consists of gentle active-assisted and stretching exercises in all restricted planes of motion for at least 3 months. Patients may do the exercises on their own or under direct supervision. In both scenarios, patients must assume that they are, to a great degree, responsible for their improvement. Typically, patients will combine physical therapy with NSAIDs or other analgesics, including narcotics or topical dimethyl sulfoxide. It is of paramount importance to avoid any forceful stretching, especially during the most inflammatory phase of the disease, because this may aggravate symptoms and prolong the inflammatory response. ,

There is scarce information on the ideal protocol concerning the type, length, and frequency of therapy sessions. It is probably better to increase the number of short repetitions instead of undergoing prolonged sessions. We usually recommend performing stretching exercises for a few minutes, six to eight times a day, avoiding any significant pain but trying to achieve small increments in ROM within the tolerated limit of discomfort. Patients should be advised on the expected slow progression of improvement so they do not become discouraged. In addition, strengthening exercises should be postponed until functional and painless ROM has been recovered. A well-executed therapy program leads to significant improvement in both pain and ROM in more than 90% of cases, but it may take several months before patients achieve a comfortable situation.

Regarding the most effective therapy program for frozen shoulder, Nicholson recommended adding passive stretching exercises in abduction, and Griggs et al. reported a 90% success rate on a group of 75 patients treated with a four-direction stretch program. In specific situations, a group hospital class might have a benefit over individual or home physiotherapy.

In our practice we recommend doing multiple short sessions per day, starting with heat application and performing 1 to 2 minutes of low-stress stretching exercises, maintaining the arm in maximum elevation and external rotation within a tolerated limit of discomfort. The exercises are better done with the patient seated or lying down. We advise patients to avoid pain greater than 5 on a 10-point visual analog score when performing the exercises. If sore, they can apply ice at the end of the session. Regaining ROM is usually a slow process, and it is crucial to reinforce patients’ compliance.

Most well-executed therapy programs achieve long-term success in 9 out of 10 patients, with significant improvements in pain, ROM, and quality of life. , , Dudkiewicz and colleagues reported an excellent long-term outcome in 54 patients treated with physical therapy and NSAIDs after a mean follow-up of 9.2 years, with improvement in all planes of motion.

In a Cochrane systematic review, Page and colleagues reviewed the existing literature on the results of manual therapy alone or associated with other therapies in patients with frozen shoulder. Thirty-two trials were included in the analysis, but none of them compared physical therapy with no intervention, limiting the possibility of carrying out a proper meta-analysis. Overall differences in outcome after successful interventions were observed only in the short term (up to 7 weeks). A combination of manual therapy and home exercises for 6 weeks was less successful (46% of significant clinical improvement) than an intra-articular steroid injection (77% of clinical improvement). However, the lack of high-quality trials limits our ability to make conclusions on the real effect of manual therapy or exercises when performed alone.

In 2019, Lowe et al. tried to clarify the current evidence on physical therapy modalities for frozen shoulder. They reviewed 30 trials, but only four with a low risk of potential bias were included in their analysis. These trials showed that joint mobilization in combination with stretching was more effective than stretching alone. Another finding was that adding US or a multimodal program that included mobilization, heat, and stretching was more successful than stretching alone. However, data from these studies should be interpreted with caution because they lack long-term follow-up.

In a recent systematic review by Alsubheen et al. it was shown that, despite the low quality of the available series, physical therapy in combination with joint mobilization techniques was also effective in diabetic patients with frozen shoulder.

Inappropriately performed physical therapy can be deleterious and make frozen shoulder patients worse. It is not uncommon to see patients complaining of significant worsening after a few sessions of aggressive physical therapy. Rizk and colleagues found that only 60% of their patients achieved pain-free sleeping at 5 months after receiving physical therapy. Hazleman reported a similar experience, with only 50% of patients showing a clinical benefit after physical therapy and one-third of them showing an increase in pain. In their prospective study comparing intensive therapy against light home exercises in patients with frozen shoulder, Diercks and Stevens reported that more than one-third of patients treated with intense therapy failed to achieve a Constant score greater than 80 at 2 years, and 90% of patients undergoing “supervised neglect” had normal shoulders.

Increased activity of the upper trapezius compared with the lower part of the muscle has been reported in patients with frozen shoulder. This muscle imbalance causes scapular substitution in compensation for reduced glenohumeral motion. Shih and colleagues have found that a single session of heat and manual muscle release of the trapezius decreased pain and increased motion and muscle performance, suggesting that trapezius rehabilitation should be included in therapy protocols of frozen shoulder.

Associated physical therapy modalities such as microwave, short wave, extracorporeal shock wave therapy, laser therapy, electromagnetic therapy, US, magnetic field therapy, hyperbaric oxygen, and heat application have been used in different studies with little success. , A Cochrane systematic review on electrotherapy modalities for frozen shoulder showed that application of low-level laser therapy for 6 days provided some benefit to patients with frozen shoulder, with 80% of patients reporting success versus 10% in the placebo group. One other study failed to show a clear benefit of electromagnetic field therapy compared with placebo. Chen et al. found that patients treated with shock wave therapy obtained short-term pain relief, with clinical improvement comparable with an intra-articular corticosteroid injection.


Intra-articular steroid injections

Intra-articular steroid injections are used to interrupt the initial inflammatory and fibrogenic tissular phase of frozen shoulder. Crisp and Kendall, in 1955, were the first to report on the efficacy of intra-articular injections of hydrocortisone during the acute and chronic phases of frozen shoulder. They found that the effect was faster in patients treated acutely.

The reported benefit of steroid injections varies. Some authors have found little or no benefit of intra-articular injection, whereas others have shown improvement after steroid injection, occurring mainly during the first 7 weeks and mostly due to decreased pain rather than improvement in ROM. However, when compared with other forms of treatment, intra-articular steroid injections do not provide significant improvement beyond 4 to 6 months. One difficulty in elucidating the efficacy of steroid injections derives from the fact that they are almost always combined with a stretching program.

The effectiveness of an intra-articular injection has been compared with other treatment options. Tveita et al. found similar results to those obtained after intra-articular injection when hydrodilation was added. Several authors have concluded that intra-articular steroid administration provides faster improvement than oral intake. , Williams et al. compared repeated intra-articular injections with stellate ganglion blocks and did not find any significant difference in outcomes.

Administration of an intra-articular injection in the shoulder without radiographic guidance is not always an easy procedure. In a cadaveric study, Patel and colleagues showed a 92% accuracy when a US-guided technique was used compared with 72.5% when the injection was performed freehand. Raeissadat et al. performed a prospective study comparing the clinical outcome of 20 patients treated with a US-guided injection and 21 patients in whom the shoulder was injected without imaging aid and found that guided injections provided better clinical improvement. In their meta-analysis on 506 glenohumeral injections, Amber et al. found that US-guided were slightly more accurate than fluoroscopy-guided injections with the advantage of avoiding radiation. We believe it seems reasonable to use image guidance when performing an intra-articular injection. Because it is cheap, fast, and avoids radiation, we favor freehand or US-guided injections in our practice through an anterior approach. The role of the surgeon’s experience in improving the accuracy of injections is controversial. While some authors have shown no effect of experience on accuracy, others have shown that surgeons with experience in shoulder arthroscopy seem to have better accuracy output.

There is some evidence that questions the necessity of applying the steroid into the joint to obtain a good result. Sun and colleagues have shown that injecting the steroids into the rotator interval during the freezing phase of the disease provided better pain relief, ROM, and patient satisfaction than intra-articular or subacromial injections. No difference in outcomes has been shown using a low-dose versus a high-dose preparation of steroids.

Injection of the shoulder joint is not a harmless procedure. Fatal clostridial myonecrosis and severe chronic sepsis have been reported after intra-articular injections. In addition, the deleterious effect of steroids on cartilage and adjacent soft tissues has been well documented. , It is important to inject under a meticulous sterile technique and, if required, try to delay the next injection at least 3 months. , , If there are any known risk factors, such as immunosuppression or diabetes, avoiding the injection should be seriously considered.

In 2019, Kitridis et al. performed a systematic review and network meta-analysis on the efficacy of intra-articular steroids in primary idiopathic stiffness. They found robust evidence supporting the efficacy of intra-articular corticosteroids and distention of the shoulder at the 0- to 2-month time frame. Rotator interval corticosteroids provided significant results regarding pain relief in the short and intermediate term. The short-term benefits of steroids dissipated over time. Multiple-site corticosteroid injections yielded clinical advantage over placebo for short- and intermediate-term composite outcome assessments.

In our practice, the ideal candidate for an intra-articular injection has not responded to previous physical therapy or, more commonly, presents with severe pain that precludes effective initiation of physical therapy, usually during the freezing phase of the disease.

Injection of steroids into the subacromial space is rarely performed as an isolated treatment for frozen shoulder. However, similar results on pain relief and ROM gain have been reported when compared with intra-articular injections. Dacre and colleagues suggested that local injection of steroids was more cost-effective than physical therapy, but this statement should be accepted with caution because only low-quality information is available, precluding our ability to make definite conclusions.

Trigger point injections

Painful muscle areas described as trigger points have been the focus of multiple-site, local injections of anesthetics, with or without a corticosteroid. Only short-term pain relief has been reported with this technique. Steinbrocker and Argyros reported on a group of 42 patients with frozen shoulder that underwent injection at multiple sites (supraspinatus, subacromial, biceps tendon, and articular capsule) with significant functional improvement in 95% of cases. Unfortunately the study did not include a control group.

Other injections

Based on their theoretical benefit on synovial fluid metabolism, intra-articular injections of hyaluronic acid (HA) have been used by several authors. Although these studies showed some benefit, unfortunately they lacked a control group. HA injections have also been used in combination with triamcinolone and physiotherapy, showing improved results when compared with triamcinolone and physiotherapy alone. In a recent meta-analysis, Lee and colleagues concluded that intra-articular HA administration alone was not superior to conventional treatments, and neither was its addition to conventional therapies. Anecdotally, intra-articular injection of proteases such as α-chymotrypsin and hylase or ketorolac have provided results similar to those of steroids. However, the value of these studies is controversial because they also lacked a control group.

In an attempt to lyse the anterior shoulder capsule, Badalamente and Wang have reported good results with the clinical use of a collagenase Clostridium histolyticum injection into the rotator interval area in patients with frozen shoulder. On the contrary, Schydlowsky and colleagues performed a randomized study investigating the effect of a subcutaneous injection of adalimumab (antitumor necrosis factor agent) and found no benefit.

Hydrodilation (capsular distention)

Hydrodilation involves the injection of fluid into the joint at a pressure high enough to distend and stretch the joint capsule. Although it was first described by Payr in the German literature as capsular distention, it then became popularized as distention arthrography or brisement by Andren and Lundberg. The procedure has received wide attention because it can be easily done in the clinic or radiology department. However, hydrodilation is often poorly tolerated due to the painful nature of the distention. Progressively higher volumes of fluid are injected into the glenohumeral joint until the pressure is high enough to disrupt the capsule. At some point, the pressure of the injection decreases, indicating when capsular disruption occurs, usually through the biceps tendon sheath of the subacromial bursa.

Buchbinder and colleagues performed a randomized controlled study comparing hydrodilation with placebo. They demonstrated a significant short-term improvement in pain and ROM, but this was not maintained beyond 2 months. Hydrodilation is commonly performed in conjunction with a steroid injection, which makes evaluation of its effectiveness very difficult. Three studies have compared hydrodilation associated with steroid to intra-articular steroid injection alone. , , All three series failed to demonstrate any benefit of hydrodilation at any outcome point.

Khan et al. compared hydrodilation followed by physiotherapy with physiotherapy alone in a small group of patients and found significant improvements in motion at 8 weeks but no differences in pain relief. Robinson et al. did not find any difference in clinical outcomes between two groups of patients who underwent hydrodilation and were enrolled either in a supervised therapy program or a home-based exercise program. Buchbinder and colleagues performed a systematic Cochrane review of hydrodilation for frozen shoulder and found some evidence that there was an improvement in pain, function, and ROM at the short-term follow-up. In a later meta-analysis, the same group conferred silver evidence to this treatment but was unable to confirm superiority over other techniques.

Most surgeons recommend capsular distension in patients with failed prior physical therapy and moderate restriction in ROM because patients with less severe loss of ROM and those who achieved capsular distension prior to capsule rupture benefited most. This procedure is effective for pain relief, but improvement in ROM is not as predictable, and up to one-third of patients require further treatment. Piotte and colleagues showed that repeated distention arthrography with steroids and a home exercise program improved shoulder function.

MRI findings are useful for evaluation but fail to correlate with the outcome after a capsular distension procedure. Quraishi and colleagues reported improved pain control without differences in ROM in patients undergoing hydrodilation compared with MUA with triamcinolone. Gallacher et al. recently performed a randomized trial on 50 patients comparing hydrodilation and arthroscopic capsular release. Although both groups showed significant improvements in the Oxford Shoulder Score, the capsular release group showed higher increments in ROM and long-lasting pain relief.

Suprascapular nerve block

Wertheim and Rovestine used suprascapular nerve blocks to treat shoulder pain in 1941. The suprascapular nerve has motor, sensitive, and sympathetic fibers, providing approximately two-thirds of sensory fibers to the glenohumeral joint capsule. The available evidence has shown that suprascapular nerve blockade provides pain relief without significant improvement in ROM or function. Repeated suprascapular nerve blockade has shown to improve passive ROM in frozen shoulder associated with reflex sympathetic dystrophy. , ,

Dahan and colleagues performed a randomized case-control study with a placebo control group and found that a bupivacaine block provided short-term benefit with no improved function. Ozkan and colleagues have also shown improved results in diabetic patients with frozen shoulder refractory to intra-articular steroids. As with any other injection technique, accuracy is variable and influenced by experience, body habitus, or previous interventions. Near-nerve EMG assisted injection has shown to be superior to injections guided by anatomic landmarks.

Other techniques


Some authors have found a short-term benefit of adding acupuncture to other forms of treatment, such as physical therapy or regional nerve blocks. Recently, acupuncture has been compared with acupotomy, a novel device, and technology derived from the combination of the acupuncture needle with the surgical knife to treat soft tissue, achieving improved pain relief and increase in ROM.


Although several authors reported some benefit of radiation therapy in frozen shoulder in the past, it has been basically abandoned, and it is only very rarely used nowadays for cases associated with heterotopic ossification. ,

Operative treatment

Patients who do not demonstrate adequate pain relief or progress in regaining ROM after 3 to 6 months of nonoperative treatment are candidates for operative intervention. The decision to proceed with a more aggressive approach may sometimes be difficult because this is considered a self-limiting disease by many surgeons. We inform our patients that any surgical intervention is intended to speed the recovery so they can be functional earlier. However, if they are reluctant to undergo a more aggressive approach, they should know that they still have a reasonable probability of improving their motion and decreasing their pain.

Manipulation after anesthesia

Historically, MUA has been considered an effective mode of treatment for frozen shoulder that has provided improvement in both pain and ROM at short- and long-term follow-up. , MUA is ideally indicated for patients in phase II when pain at rest is decreasing, it is elicited mainly at the extremes of motion, and there is significant stiffness. It seems that MUA should be delayed until the inflammatory phase of frozen shoulder is resolving; it has been shown that clinical improvement after MUA was more significant in patients who had symptoms for more than 6 months. Some authors have reported significant worsening when the shoulder was manipulated early, while the initial inflammatory phase was taking place. It is important to avoid this procedure in severely osteopenic patients and those with shoulder arthritis. Most patients undergoing MUA have usually not responded to a prior course of nonoperative treatment, including an intra-articular corticosteroid injection.

Manipulation is performed with the patient supine and under general anesthesia. Complete muscle relaxation (e.g., succinylcholine) is required before any mobilization is initiated to reduce the risk of complications. We prefer to complete the anesthetic procedure with an interscalene block that provides postoperative analgesia. A painless shoulder after the manipulation allows immediate mobilization that can be witnessed by the patient in the early postoperative period. This will increase the motivation to participate more actively in the subsequent therapy program. We usually administer a single block that provides analgesia for 12 hours, but it could be repeated or a continuous infusion of anesthetic administered if required.

Several techniques have been described to perform the MUA. We follow, with slight modifications, the steps described by Copeland in 1995 ( Fig. 18.6 , ). This technique involves the use of a short lever arm to hold the affected arm at the humerus midshaft area, with stabilization of the scapula to prevent a brachial plexus traction injury. The manipulation is carried out in four stages: (1) elevation-abduction to tear the inferior capsule; (2) external rotation at the side, with the arm in slight adduction, and then progressively bringing the arm back to abduction position while maintaining the shoulder in external rotation (which tears the anterior capsular restriction); (3) forced adduction with the arm elevated to 90 degrees; and (4) internal rotation to tear the posterior capsule. All these maneuvers are performed by applying a constant controlled force and avoiding any sudden movement or a significant increase in the force applied.

Aug 21, 2021 | Posted by in ORTHOPEDIC | Comments Off on Primary idiopathic shoulder stiffness: Frozen shoulder
Premium Wordpress Themes by UFO Themes