Clinical Identifiers for Early-Stage Primary/Idiopathic Adhesive Capsulitis: Are We Seeing the Real Picture?

Sarah Walmsley, Peter G. Osmotherly, Darren A. Rivett


Background Adhesive capsulitis is often difficult to diagnose in its early stage and to differentiate from other common shoulder disorders.

Objective The aim of this study was to validate any or all of the 8 clinical identifiers of early-stage primary/idiopathic adhesive capsulitis established in an earlier Delphi study.

Design This was a cross-sectional study.

Methods Sixty-four patients diagnosed with early-stage adhesive capsulitis by a physical therapist or medical practitioner were included in the study. Eight active and 8 passive shoulder movements and visual analog scale pain scores for each movement were recorded prior to and immediately following an intra-articular injection of corticosteroid and local anesthetic. Using the local anesthetic as the reference standard, pain relief of ≥70% for passive external rotation was deemed a positive anesthetic response (PAR).

Results Sixteen participants (25%) demonstrated a PAR. Univariate logistic regression identified that of the proposed identifiers, global loss of passive range of movement (odds ratio [OR]=0.26, P=.03), pain at the end of range of all measured active movements (OR=0.06, P=.02), and global loss of passive glenohumeral movements (OR=0.23, P=.02) were associated with a PAR. Following stepwise removal of the variables, pain at the end of range of all measured active movements remained the only identifier but was associated with reduced odds of a PAR.

Limitations The lack of a recognized reference standard for diagnosing early-stage adhesive capsulitis remains problematic in all related research.

Conclusions None of the clinical identifiers for early-stage adhesive capsulitis previously proposed by expert consensus have been validated in this study. Clinicians should be aware that commonly used clinical identifiers may not be applicable to this stage.

Adhesive capsulitis is a diagnostic label attributed to a disorder of the glenohumeral joint capsule that has been reported to affect up to 5% of the population.1,2 Primary adhesive capsulitis is due to an unknown cause, as opposed to secondary adhesive capsulitis, which results from a known cause or extrinsic event.3 The condition is generally described as consisting of 3 stages.3 These have been identified as the painful stage (first), the adhesive stage (second), and the resolution stage (third).4 The first (or painful) stage, which was considered in this study, is generally considered to last 3 to 9 months.4 Although the later stages are easily recognized, often due to marked restriction of movement, the early stage of this disorder is commonly difficult to identify and correctly diagnose.5 It has been proposed, however, that treatment in the early stage of adhesive capsulitis may decrease the overall morbidity,1 arguably suggesting that early recognition of this disorder is desirable.

Musculoskeletal health care frequently relies on recognition of patient-reported and physical examination findings, together with special tests and medical imaging, to inform diagnosis and direct management. Determining the clinical features considered necessary to establish a diagnosis frequently is achieved through research using various types of consensus methods.68 Several studies using this approach have attempted to identify clinical characteristics of adhesive capsulitis in general,9,10 as well as clinical characteristics specific to the early stage5; however, validation of these characteristics is lacking. As well as routine clinical assessment, musculoskeletal assessment often relies on a gold standard that may include a particular diagnostic test, imaging procedure, or even surgical findings with which to determine a diagnosis. Because surgery is not indicated and imaging procedures in the early stage of adhesive capsulitis have yet to be systematically explored,11 a gold standard for diagnosis remains problematic in this population. Clinical tests have recently been described that may assist the diagnosis of adhesive capsulitis12,13; however, the duration of symptoms of participants in these studies was not reported, resulting in difficulty determining the stage of the disorder and whether the findings are valid for patients in the early stage.

A set of clinical identifiers considered necessary and sufficient by a group of experts to diagnose early-stage adhesive capsulitis5 (Fig. 1) has been proposed as a framework with which to begin the process of addressing this diagnostic dilemma. The identifiers established in that study by our research group included both patient-reported and physical examination findings and, interestingly, clustered into 2 discrete dimensions of pain and movement. As pain is reportedly a significant feature of the early stage,1 it was not surprising that several dimensions in pain were reported to achieve consensus. Night pain, a marked increase of pain with rapid or unguarded movements, discomfort lying on the affected shoulder, and pain easily aggravated by movement were all identified as required to achieve diagnosis. These descriptors were suggested to reflect the inflammatory nature of the disorder in the early stage.14

Figure 1.

Clinical identifiers achieving consensus.

Although often unquantified, recognition of the later stages of adhesive capsulitis through marked movement restriction, in particular external rotation, has been reported.15 Conversely, there is a lack of description of movement dysfunction in the early stage of the disorder. Physical examination findings achieving consensus in our Delphi study5 similarly lacked quantification, but it was suggested global loss of both active and passive ranges of movement and pain at the end of range in all directions were necessary characteristics. Although the clinical identifiers proposed for early-stage adhesive capsulitis by expert consensus5 were suggested as a starting point for future validation studies, it was recognized that they could not at this time be regarded as a gold standard or provide a certain differential diagnosis, but potentially could be used to assist in clinical decision making.

The aim of this study, therefore, was to validate any or all of the 8 clinical identifiers previously proposed for the early stage of adhesive capsulitis.5

Materials and Method


Participants were recruited from a private upper limb physical therapy clinic in Newcastle, Australia, over a 3-year period between May 2010 and April 2013. Patients clinically diagnosed with adhesive capsulitis by various health care practitioners, including orthopedic surgeons, shoulder physicians, general practitioners, and physical therapists, were invited to participate in the study.

To be considered for inclusion, potential participants were required to have been referred for an intra-articular glenohumeral joint corticosteroid and local anesthetic injection using radiological guidance to confirm correct placement of the needle as part of routine clinical care. Consistent with the reported duration of the early stage of adhesive capsulitis,4,16 potential participants were excluded from the study if they had a symptom duration of greater than 9 months. As primary/idiopathic adhesive capsulitis was being investigated, individuals with a history of previous major trauma or surgery on the affected shoulder also were excluded. Reported minor trauma was not an exclusion criterion. Potential participants were required to have had a recent unremarkable radiographic examination to eliminate glenohumeral osteoarthritis, calcific deposits, or other potentially confounding diagnoses. They also were required to have had a recent ultrasound examination that excluded a full-thickness rotator cuff tear. Potential participants who had undergone an intra-articular corticosteroid injection into the glenohumeral joint in the preceding 6 weeks or had a history of inflammatory arthropathies or of cervical spine pathology that may refer into the shoulder joint also were excluded from the study. Because the contralateral shoulder was being used to determine percentage loss of range of movement, the presence of pain or restriction of movement in that shoulder was a further exclusion criterion. All participants signed an informed consent form prior to entering the study.


Immediately prior to the injection, each participant attended the clinic to complete routine assessment, including measurement of active and passive ranges of movement and pain at the end of ranges of movement. Additional questions were asked to determine the presence of the 8 clinical identifiers being validated. To provide baseline measurements of shoulder pain and disability, the Shoulder Pain and Disability Index (SPADI)17,18 was administered. This instrument is a validated questionnaire measuring shoulder pain and impairment and has a high level of internal consistency and good test-retest reliability.19 General health status was measured using the 36-Item Short-Form Health Survey (SF-36).20 This instrument is easy to administer, has been demonstrated to be reliable and valid,20 and has been used previously to describe study samples with adhesive capsulitis.21,22 Upon completion of the assessment, participants attended a radiology practice to undergo the intra-articular glenohumeral corticosteroid and local anesthetic injection under radiological guidance. Within 1 hour of administration of the injection,23 participants returned for reassessment, which included measurement of active and passive ranges of movement and pain at the end of ranges of movement. Following the measurement of range of movement and recording of postinjection pain levels, the participants continued with routine clinical management.

Shoulder movement measurement.

A comprehensive series of shoulder active and passive ranges of movement were evaluated. Seated upright in a chair to limit trunk extension, measurement of the following ranges of movement were performed based on the method described by Green et al24: total shoulder flexion (TSF), glenohumeral flexion (GHF), total shoulder abduction (TSA), and glenohumeral abduction (GHA). The starting position for each of these movements was with the palm facing medially to ensure consistent rotation. The elbow was extended, and the inclinometer was placed along the shaft of the humerus.24 As GHF and GHA were being measured, a device was constructed to limit movement of the acromion in order to provide consistent scapular stabilization (Fig. 2).

Figure 2.

Device to stabilize the scapula for measurement of glenohumeral joint movement.

Each of the following movements was performed in the supine position based on previously described methods2426: external rotation in neutral abduction (ERN), external rotation in 90 degrees of abduction (ERA), and internal rotation in 90 degrees of abduction (IRA). A towel was placed under the shaft of the humerus to ensure it was parallel to the plinth, with the elbow flexed to 90 degrees, and the inclinometer was placed on the dorsal surface of the participant's forearm. For ERA and IRA, the arm was abducted to 90 degrees, or, if this was not possible, it was taken to the limit of movement. Internal rotation in abduction was measured based on a method previously described whereby the end range was determined as the point at which the posterolateral acromion was visualized to rise off the plinth.27 In addition, hand behind back (HBB) was measured in standing using the distance between the spinous process of T1 and the spinal level reached by the radial styloid process with the arm taken behind the back.28

All movements, with the exception of HBB, were measured in degrees using a Baseline digital inclinometer (Fabrication Enterprises Inc, Irvington, New York). Prior to each measurement, the digital inclinometer was reset to zero after placement on the participant to ensure consistency. Digital inclinometery has been demonstrated to have a measurement error of ±1 degree.29 Hand behind back was measured with a tape measure and recorded in millimeters. The order of measurement was standardized (TSF, GHF, TSA, GHA, ERN, ERA, IRA, HBB), and all active movements were performed prior to any passive movements.

The instruction to participants for all active movements was to move the arm as far as possible until they were no longer able to tolerate the movement due to pain or they were unable to move the arm any farther. For passive movements, the researcher performed each of the movements to the point of resistance or when the participant reported the pain was intolerable. To determine percentage of loss of active and passive ranges of movement, contralateral shoulder range of movement also was measured prior to the injection of corticosteroid and local anesthetic in an identical manner to the affected shoulder. In the absence of any documented deficit, a loss of range of movement of 10% or greater with respect to the contralateral shoulder was determined to constitute loss of movement. Such a loss exceeds the measurement error of shoulder range of movement of less than 7% previously reported,26 as well as that reported for the commonly used universal goniometer (5°–7°),30 thus affording some translation of the findings to the clinical setting.

Calculation of postinjection pain intensity.

In the absence of a gold standard for the diagnosis of early-stage adhesive capsulitis, the response to the local anesthetic (administered concurrently with the corticosteroid injection) was used as the reference test standard. Local anesthetic injection has been previously proposed as a method of determining diagnosis.31,32 To determine the anesthetic response, each participant was required to record their level of pain at the end of active and passive ranges of movement on a 100-mm visual analog scale (VAS) with 0 mm=“no pain” and 100 mm=“worst pain imaginable.” The percentage change in pain intensity from before to after the injection was calculated for each active and passive movement. Pain relief of ≥70% for ERN was considered a positive anesthetic response (PAR). External rotation in neutral abduction was chosen because it is generally recognized as the most frequently affected movement in adhesive capsulitis.33 The required ≥70% of pain relief obtained was chosen because it is considered clinically relevant and has been used in previous research.34

Data Analysis

Descriptive statistics were used to summarize the characteristics of the participants and presence of the 8 clinical identifiers. The participant characteristics together with the 8 identifiers were analyzed against anesthetic response using univariate logistic regression. As the clinical identifier describing pain at the end of range in all directions was nonspecific about whether this was active or passive range of movement; both dimensions were included in the analysis. Furthermore, although only global loss of passive glenohumeral joint movement was proposed as a clinical identifier, for completeness, active range of movement also was included in the model. Movements of the glenohumeral joint include GHF, GHA, ERN, ERA, and IRA. All factors with a P value of ≤.2 were included in a multiple logistic regression model. Outcomes were expressed as odds ratios (ORs) with 95% confidence intervals. A P value of <.05 was considered to be statistically significant. Data were analyzed using Stata 12.0 statistical software (Stata Corp, College Station, Texas).


The flow of participants through the study is shown in Figure 3. In total, 255 patients were assessed for inclusion in the study, and 191 were excluded for either not meeting the inclusion or exclusion criteria (n=150) or being unwilling or unable to participate (n=41). Sixty-four participants were included in the study, and participant demographic characteristics are reported in Table 1.

Figure 3.

Design and flow of participants through the study.

Table 1.

Characteristics of the Study Participants (n=64)a

The prevalence of the 8 clinical identifiers is presented in Table 2. All of the participants were aged over 35 years. Global loss of active and passive ranges of movement were the least prevalent of the 8 criteria (65% and 67%, respectively).

Table 2.

Prevalence of the 8 Clinical Identifiers (n=64)

Sixteen participants (25%) demonstrated a PAR. The relationship between the demographic characteristics and the proposed 8 clinical identifiers of the participants with a positive PAR is reported in Table 3. Univariate logistic regression identified that none of the patient demographic characteristics were associated with a PAR. Of the 8 proposed clinical identifiers, pain at the end of range of all measured active movements (OR=0.06, P=.02), global loss of passive range of all measured movements (OR=0.26, P=.03), and global loss of passive glenohumeral movements (OR=0.23, P=.02) were associated with a PAR. Following stepwise removal of the variables, pain at the end of range of all measured active movements remained the only identifier but was associated with reduced odds of a positive response (OR=0.06, P=.018).

Table 3.

Relationship Between Participant Characteristics and the 8 Clinical Identifiers and Positive Anesthetic Response (n=64)a


This is the first study that has attempted to validate a set of clinical identifiers for the early stage of primary/idiopathic adhesive capsulitis. It is unique in that it used clinical identifiers previously established by expert consensus5 and investigated only patients with symptoms for less than 9 months. Although the identifiers established by this consensus method have frequently been recognized in the literature,3537 none were validated in this study. Interestingly, of the 8 clinical identifiers, pain at the end of all active ranges of movement has emerged as the least likely to indicate a diagnosis of early-stage adhesive capsulitis. These results may suggest expert opinion, and possibly clinical practice may not be recognizing the appropriate clinical identifiers of patients in the early stage of this disorder. This study highlights the difficulty in quantitatively determining an exclusive set of criteria for the early stage of adhesive capsulitis.

Using the effect of intra-articular local anesthetic injection and associated pain relief of ≥70% in external rotation as the diagnostic reference standard, 25% of the participants in this study were determined to have early-stage adhesive capsulitis. This percentage was less than anticipated; however, it is in line with the proposal that this disorder is overdiagnosed, and the true incidence is much lower than generally reported.38 A further consideration is that every patient with a painful shoulder and apparent limitation of motion may not necessarily indicate a diagnosis of early-stage adhesive capsulitis.39 It is likely that the clinicians assessing the patients in the current study used clinical identifiers similar to those used by the experts in the Delphi study,5 given the specialist nature of the practice from which the participants were recruited. It is not surprising, therefore, that the prevalence of the identifiers in the participants was generally high, as demonstrated in Table 2.

Our results suggest that using these criteria may not be appropriate to identify the early stage of this disorder. The differences of opinion and lack of understanding of adhesive capsulitis in its early stage, as well as the general appreciation of the specific diagnostic criteria that distinguish it at this stage from other shoulder disorders, have been reported previously.40 Furthermore, there is no consensus as to the exact range-of-motion restriction required for a patient to qualify for a diagnosis of early-stage adhesive capsulitis.41 Although consensus exists regarding the presence of 3 phases of the disorder, controversy still arises regarding the diagnostic criteria that distinguish these stages.42 The findings of this study are consistent with this confused picture.

Recent understanding of the pathology of adhesive capsulitis has suggested that the behavior of the symptoms throughout the stages of the disorder may be explained by the underlying pathological process of initial inflammation followed by subsequent contracture.14 In particular, inflammation of the anterior glenohumeral joint capsule43,44 has been implicated in early adhesive capsulitis. It may be reasonable, therefore, to expect pain or restriction of movement to not be global in the early stage of adhesive capsulitis, given this reported pathology.14 Despite these findings, consensus studies on diagnostic criteria or clinical identifiers previously reported (with the exception of the Delphi study5) notably omit consideration of the stages described when proposing diagnostic criteria.9,33 Furthermore, the degree and directions of restriction required to constitute adhesive capsulitis have not been identified as necessary to determine appropriate diagnosis.45 The fact that each of the 8 measured active and passive movements stresses various aspects of the glenohumeral joint capsule may provide an explanation for none of the clinical identifiers involving physical assessment being validated. It also may suggest that a “one size fits all” approach to diagnosis has been taken and, as the later stages reportedly are characterized by global restriction of movement and end-range pain,37,46 this approach is likely to be similarly assumed in the early stage of the disorder. Potentially, it is the global rather than specific nature of these clinical identifiers that resulted in reduced odds of a PAR. The suggestion that limitation of external rotation may be the most recognizable feature33 may warrant specific further exploration in a similar population.

The early stage of adhesive capsulitis has been reported to be frequently confused with impingement syndrome, with differentiation between the 2 disorders often difficult.47,48 Compounding the confusion between these 2 disorders, impingement tests used clinically have been reported to lack specificity.49 As well as recognition of groups of physical examination findings, the use of local anesthetic as a diagnostic tool in shoulder disorders has been reported previously.50 The confusion between early-stage adhesive capsulitis and impingement syndrome may be better addressed with injection of local anesthetic into the subacromial space51 to facilitate the diagnosis of adhesive capsulitis by exclusion.

The aim of musculoskeletal health care is to provide effective treatment of patients with various disorders. However, the lack of strong evidence for successful treatment of shoulder disorders reported in a systematic review52 has been suggested to be a result of the lack of uniformity of the use of diagnostic labels or that the criteria used in determining diagnostic subgroups are not related to treatment success.53 Establishing diagnostic criteria or clinical identifiers for various shoulder disorders allows identification of a homogeneous subgroup of patients with which to evaluate treatment outcomes and make comparisons across trials more meaningful.5 However, in the shoulder, the validity of various shoulder examination procedures has recently been challenged,54 with the lack of diagnostic accuracy possibly explained by the lack of anatomical validity of most shoulder tests.55 Schellingerhout et al53 proposed that alternate methods should be used to classify patients with shoulder disorders. The shoulder symptom modification procedure approach proposed recently to address rotator cuff tendinopathy/subacromial impingement syndrome56 may be worthy of further exploration in the group of patients with presumed early-stage adhesive capsulitis.

There are a number of limitations that require consideration in this study. First, the lack of an agreed-on reference standard for early-stage adhesive capsulitis makes any validation investigation problematic. The selection of intra-articular local anesthetic, however, was based on its previously reported diagnostic utility as a method of determining the source of patient symptoms.31,32 Although an alternative reference standard may be to provide follow-up of patients in the long term to confirm the diagnosis of adhesive capsulitis (as the characteristic loss of motion becomes evident), this approach was not feasible in the present study because the participants were concurrently clinically treated with a corticosteroid injection and stretching exercises.

Second, as this study investigated patients undergoing normal clinical management, it was not ethically possible to administer a local anesthetic injection without the simultaneous corticosteroid component. In some patients, the corticosteroid may have resulted in a reaction that was not sufficiently negated by the local anesthetic,57 although all participants were remeasured within 1 hour.

A further limitation of this study was the large number (n=191) of potential participants who were excluded. The requirement to use strict inclusion and exclusion criteria to obtain a homogeneous sample resulted in recruitment being slower than projected and the sample size accordingly modest. Interestingly, earlier authors reported similar recruitment difficulties,21,58 perhaps supporting recent opinions that the incidence of the disorder is overestimated.38 Although intrarater reliability was not specifically determined for the measurements due to the ethical consideration of patient pain provocation, previous published reports support the reliability of the method on which it was based.2426,59

Finally, the study might have been strengthened if participants had been randomly sampled over a wider area. The generalizability of the findings, therefore, may be limited if these patients are not representative of other areas.

In conclusion, the early diagnosis of adhesive capsulitis remains problematic. Clinicians should be aware that commonly used clinical identifiers may not be applicable to this stage, which also may explain some of the poor reported outcomes of treatment to date. Recognition that the features of adhesive capsulitis in its early stage are likely to differ from those of the later stages also is needed to correctly diagnose this disorder. This study raises 2 issues that may warrant exploration in future research. First, given the reported confusion with impingement syndrome,47,48 it may be worthwhile to include patients with “general” shoulder pain and assess the presence of any of the agreed identifiers in a heterogeneous group. Second, analysis of subgroups of movement deficit and pain at the end of range of groups of movements, rather than global movement, also may be worthy of further exploration.


  • All authors provided concept/idea/research design and writing. Ms Walmsley provided data collection. Ms Walmsley and Dr Osmotherly provided data analysis. Dr Osmotherly provided facilities/equipment. Dr Osmotherly and Dr Rivett provided project management and consultation (including review of manuscript before submission).

  • The Human Research Ethics Committee of The University of Newcastle granted ethical approval for this study.

  • Preliminary results of this study were presented at the 10th Congress of the International Federation of Orthopaedic Manipulative Therapists; September 30–October 5, 2012; Quebec, Ontario, Canada, and at Musculoskeletal Physiotherapy Australia 18th Biennial Conference; October 2013; Melbourne, Victoria, Australia.

  • Received August 28, 2013.
  • Accepted March 13, 2014.


View Abstract