Manual therapy and exercise for adhesive capsulitis: a systematic review with meta-analysis

ABSTRACT

Background

Adhesive capsulitis (AC) affects approximately 1% of the general population. Current research lacks clear guidance on the dosage of manual therapy and exercise interventions.

Objective

The purpose of this systematic review was to assess the effectiveness of manual therapy and exercise in the management of AC, with a secondary aim of describing the available literature present on the dosage of interventions.

Methods

Eligible studies were randomized clinical/quasi-experimental trials with complete data analysis and no limits on date of publication, published in English, recruited participants >18 years of age with primary adhesive capsulitis, that had at least two groups with one group receiving manual therapy (MT) alone, exercise alone, or MT and exercise, that included at least one outcome measure of pain, disability, or external rotation range of motion, and that had dosage of visits clearly defined. An electronic search was conducted using PubMed, Embase, Cochrane, Pedro, and clinicaltrials.gov. Risk of bias was assessed using the Cochrane Collaboration Risk of Bias 2 Tool. The Grading of Recommendations Assessment, Development, and Evaluation was used to provide an overall assessment of the quality of evidence. Meta-analyses were conducted when possible, and dosage was discussed in narrative form.

Results

Sixteen studies were included. All meta-analyses revealed non-significant effects of pain, disability, and external rotation range of motion at short- and long-term follow-up, with an overall level of evidence ranging from very low to low.

Conclusion

Non-significant findings with low-to-very-low-quality of evidence were found across meta-analyses, preventing seamless transition of research evidence to clinical practice. Lack of consistency in study designs, manual therapy techniques, dosing parameters, and duration of care impedes the ability to make strong recommendations regarding optimal dosage of physical therapy for individuals with AC.

Introduction

Adhesive capsulitis (AC), also known as frozen shoulder, affects approximately 1% of the general population, most commonly women between 40 and 60 years of age [1]. Risk factors, such as diabetes mellitus or thyroid dysfunction, may increase the likelihood (10–40%) of developing primary (idiopathic) AC [1,2]. In addition, one-third of individuals will experience AC on the contralateral side [1,2]. Theories have suggested that the condition is self-limiting and will be resolved on its own [3], however a recent systematic review by Wong et al. [2] determined that this notion is unfounded in the absence of strong scientific supporting evidence.

While specific pathophysiology is elusive, research has suggested a generalized inflammatory reaction with elevated cytokine levels that eventually leads to scarring of the glenohumeral joint capsule, resulting in pain, stiffness, and decreased quality of life [4]. Clinically, most patients progress through sequential phases [5] of the disease (freezing, frozen, and thawing) with variable presentations ranging from high irritability and pain to predominant stiffness of the shoulder. While numerous interventions exist to help manage patients with AC, no clear guidance or treatment algorithms are supported by strong evidence [6].

The current clinical practice guidelines from the Academy of Orthopaedic Physical Therapy summarize the overall evidence for various therapeutic interventions for AC, including joint mobilization, modalities, stretching, patient education, and corticosteroid injections [5]. These guidelines demonstrated a higher level of evidence for patient education and stretching (Grade B) but weaker evidence for modalities and joint mobilization (Grade C).

Conversely, corticosteroid injections received the highest level of evidence (Grade A) in agreement with a recent systematic review by Challoumas et al., [7] identifying corticosteroid injections as the single best intervention for patients with AC. While this systematic review demonstrated both statistical and clinical superiority of corticosteroid injection for pain and function when compared to no treatment or placebo and physical therapy, the results of subgroup and network meta-analyses suggest that the addition of a home exercise program (HEP), stretching, and physical therapy, including joint mobilization, to the injection may provide added benefit between 12 weeks and 12 months. While the methodological designs of this review along with the clinical practice guidelines and a systematic review conducted by the Cochrane Collaboration in 2014 [8] were rigorous, their findings suggesting the inferiority of physical therapy may be limited by the lack of the literature surrounding the dosage of physical therapy interventions.

Specifically, there are no clinical trials assessing the role of dosage in manual therapy or exercise in the management of AC, making it difficult to assess the pragmatic and clinical applicability of these interventions. As the theories behind the mechanisms of manual therapy continue to develop [9], a stronger emphasis on dosage must be incorporated in the design of clinical trials in order to ascertain the true effect of musculoskeletal pathologies. Therefore, the primary purpose of this systematic review and meta-analysis was to assess the effectiveness of manual therapy and exercise in the management of AC. The secondary aim was to describe the available literature related to dosage parameters of manual therapy and exercise, including dosage of technique, single-session duration, frequency of intervention, total number of sessions, and duration of care.

Methods

Protocol and registration

This systematic review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines [10] (Manual Therapy and Exercise for Adhesive Capsulitis: A Systematic Review with Meta-Analysis) and registered in PROSPERO (CRD42022354502). One minor change was made to the registration protocol. No statistical analysis on the dosage of interventions was performed as the available data did not permit this. The dosage was instead presented in narrative form.

Inclusion criteria

Studies had to meet the following inclusion criteria: (1) randomized clinical/quasi-experimental trials with completed data analysis and no limits on date of publication; (2) published in English; (3) recruited participants >18 years of age with primary (idiopathic) adhesive capsulitis; (4) had at least two groups with one intervention group receiving manual therapy (MT) alone, exercise alone, or MT and exercise, and the other group receiving any of the previously listed criteria or was considered a control group; (5) included at least one outcome measure of pain, disability, or external rotation range of motion; and (6) had dosage of visits clearly defined.

Exclusion criteria

Case studies, case series, retrospective studies, and studies that included participants with secondary adhesive capsulitis were excluded. In addition, studies were excluded if they implemented any interventions other than manual therapy or exercise, such as modalities, medication, injection, surgery, or specific devices as a method of treatment. The inclusion of hot or cold packs was permitted.

Search strategy and study selection

An electronic search was conducted by MO, LB, and RTP in September 2022 using PubMed, Embase, Cochrane, Pedro, and clinicaltrials.gov identifying all relevant articles without date limitation. Clinicaltrials.gov was included to capture grey literature not published due to non-significant findings. The search strategy can be found in Appendix 2. In addition, KK hand-searched reference lists of related articles. Three authors (MO, LB, and RTP) independently examined all titles, abstracts, and screening for eligibility. Full-text articles were assessed to determine final eligibility (MM and KK). If a discrepancy arose, it was resolved through discussion among all authors.

Intervention

The interventions of interest in this systematic review were MT and exercise. Across included studies, mode and dosage parameters of MT and exercise interventions varied considerably with details provided in Appendix 3. For patients with adhesive capsulitis, MT often includes but is not limited to glenohumeral joint (GHJ), scapulothoracic joint (STJ), acromioclavicular joint (ACJ), sternoclavicular joint (SCJ), cervicothoracic non-thrust mobilization, as well as soft tissue mobilization and passive stretching of the shoulder girdle. Exercise encompasses a variety of interventions, such as active range of motion (ROM), self-stretching of the capsule and surrounding soft tissue, clinician-assisted neuromuscular reeducation, and resistance exercise. For this review, MT and exercise were compared in several combinations: MT versus exercise, MT plus exercise versus exercise alone, MT plus exercise versus a different form of MT plus exercise, MT versus no treatment/control, exercise versus no treatment/comparison, and MT versus MT (different MT techniques). There were no included studies that compared MT plus exercise versus no treatment/control or one exercise to another exercise.

Outcomes

Primary outcomes for this review were pain and disability, with external rotation (ER) ROM and dosage as secondary outcome measures. Pain was measured using the Visual Analog Scale [11] (VAS; 0–100 mm/0-10 cm), Numeric Pain Rating Scale [12] (NPRS; 0–10 pts), and McGill Pain Questionnaire [13,14] (0–100). Disability was measured using the Disabilities of the Arm, Shoulder, and Hand [15,16] (DASH; 0–100%), Quick-DASH [12,15,16] (0–100%), Shoulder Pain and Disability Index [17,18] (SPADI; 0–100%), Constant Murley Score [19–22] (CMS; 0–100 pts), Oxford Shoulder Score [23–25] (OSS; 12–60 pts), Shoulder Rating Questionnaire [26–28] (SRQ; 17–100 pts), and Shoulder Disability Questionnaire [29,30] (SDQ; 0–100). The psychometric properties of all the outcomes are found in Appendix 4.

Timing of outcome assessment

Outcomes were assessed in the short term (<3 months) and long term (6–12 months). When multiple time points existed, the outcomes closest to 3-month and 12-month follow-up were used in data analyses [31].

Risk of bias

Risk of bias was assessed using the Cochrane Collaboration Risk of Bias 2 (RoB 2) Tool [32] that examines risk of bias across five domains: (1) bias arising from the randomization process, (2) bias due to deviations from intended interventions, (3) bias due to missing outcome data, (4) bias in measurement of the outcome, and (5) bias in selection of the reported result. An overall risk of bias judgment (low risk of bias, some concerns, and high risk of bias) was provided for each outcome measure for each article included in the meta-analysis as per the Cochrane Review Group Starter Pack [33]. Two authors (MM and KK) independently scored each study, with discrepancies resolved through discussion until consensus was reached.

The Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach was used to provide an overall assessment of the quality of evidence across five domains: risk of bias, inconsistency of results, indirectness, imprecision, and publication bias [31,34] (Appendix 5). The GRADE provides a summary rating of the quality of the body of evidence for the effect of an intervention on a particular outcome measure, providing a recommendation that may guide clinicians’ decision-making when choosing interventions. Following evidence appraisal, outcomes are classified by the level of evidence [31,34].

• High-quality evidence: further research is very unlikely to change confidence in the estimate of the effect, as all domains are met.

• Moderate-quality evidence: further research is likely to have an important impact on confidence in the estimate of the effect and may change the estimate, if one of the domains is not met.

• Low-quality evidence: further research is very likely to have an important impact on confidence in the estimate of the effect and is likely to change the estimate if two of the domains are not met.

• Very low–quality evidence: any estimate of the effect is very uncertain if three of the domains are not met.

• No evidence: no RCTs were identified that addressed this outcome.

Data collection

Data extraction was performed by MO, LB, MM, and KK and included study details and design, patient demographics, interventions, timing of assessment, outcome measures, and results (Table 1). Study authors were contacted in the event of missing data. For the purposes of this review, intervention groups were labeled by whether they received MT, exercise (EX), or a combination.

Table 1.

Description of studies.

Study	Participant characteristics	Intervention group	Comparison group	Follow-up time points	Outcomes	Summary of results
Agarwal et al. 2016 [35], India	n = 30 13F, 15 M Age: MTA: 48.7y ±6.4y MTB: 52.5y ±9.6y Symptom Duration: MTA: 4.6mo (median) MTB: 5mo (median)	Manual Therapy A (MTA) n = 15 Reverse distraction GHJ + STJ mobilization	Manual Therapy B (MTB) n = 15 Kaltenborn GHJ mobilization	6 weeks	VAS ER AROM ER PROM	Statistically significant between group differences in VAS favoring MTA. No statistically significant between group differences for ER AROM or PROM.
Ali and Khan, 2015 [36], Pakistan	n = 44 Age: MT + EX: 51.31y EX: 51.71y Symptom duration: >3 months	Manual Therapy + Exercise (MT + EX) n = 22 Maitland GHJ mobilization Shoulder stretching HEP	Exercise (EX) n = 21 Shoulder stretching HEP	5 weeks	VAS SPADI ER ROM	No statistically significant between group differences.
Bulgen et al., 1984 [37], United Kingdom	n = 42 28F, 14 M Age: 55.8y Symptom Duration: 4.8mo	Manual Therapy (MT) n = 11 Maitland GHJ mobilization Education: pendulums	Exercise (EX) n = 12 PNF Education: pendulums No treatment n = 8 Education: pendulums	1 week 2 weeks 3 weeks 4 weeks 5 weeks 6 weeks 2 months 3 months 4 months 5 months 6 months	VAS for night pain, pain with movement and rest pain ER PROM	No statistically significant between group differences.
Celik and Mutlu, 2016 [38], Turkey	n = 30 18F, 8 M Age: MT + EX: 54.2y ±7.9y EX: 54.8y ±6.4y Symptom duration: MT + EX: 16 wk ±2.2 wk EX: 15.4 wk ±2.0 wk	Manual Therapy + Exercise (MT + EX) n = 15 GHJ mobilization Shoulder stretching HEP	Exercise (EX) n = 15 Shoulder stretching HEP	6 weeks 1 year	VAS DASH CMS ER ROM	Statistically significant between group differences in Constant Score and ER ROM favoring MT + EX. No statistically significant between group differences in VAS or DASH.
Chauhan et al., 2011 [39], India	n = 26 15F, 11 M Age: NA Symptom duration:2mo to 1y	Manual Therapy (MT) n = 13 Deep transverse friction massage Inferior capsular stretching Passive ROM HEP	Comparison n = 13 HEP	Day 1 Day 3 Day 5 Day 7 Day 9 Day 11	VAS SPADI ER ROM	Statistically significant between group difference in VAS and SPADI favoring MT. No statistically significant between group differences in ER ROM.
Chen et al., 2009 [40], Australia	n = 90 64F, 26 M Age: MT + EX: 64.7y ±12.5y EX: 65.5y ±12.7y Symptom duration: MT + EX: 9.3mo ±11.1mo EX: 11.1mo ±13.6mo	Manual Therapy + Exercise (MT + EX) n = 45 Maitland GHJ, ACJ, SCJ mobilizations Neuromuscular reeducation and stabilization for rotator cuff and scapular muscles	Exercise (EX) n = 45 Neuromuscular reeducation and stabilization for rotator cuff and scapular muscles	1 month 6 months	SPADI	No statistically significant between group differences.
Deshmukh et al., 2014 [41], India	n = 30 Age: NA Symptom duration: NA	Manual Therapy A + Exercise (MTA + EX) n = 15 Maitland GHJ, ACJ, SCJ, STJ mobilization Myofascial release arm pull Shoulder AAROM Capsular stretching HEP	Manual Therapy B + Exercise (MTB + EX) n = 15 Maitland GHJ, ACJ, SCJ, STJ mobilization Shoulder AAROM Capsular stretching HEP	1 week 2 weeks 3 weeks	VAS SPADI ER ROM	Statistically significant between group differences in VAS, SPADI, and ER ROM favoring MTA + EX.
Horst et al., 2017 [42], Poland	n = 72 25F, 41 M Age: EX: 44y ±16y MT + EX: 47y ±17y Symptom duration: NA	Manual Therapy + Exercise (MT + EX) n = 36 Structural-oriented manual therapy with mobilization based on examination findings PNF	Exercise (EX) n = 36 Activity-oriented facilitation with manual guidance	2 weeks 3 months	McGill Pain Questionnaire ER ROM	Statistically significant between group differences in pain at 3 months and ER ROM at 2 weeks and 3 months favoring EX.
Iqbal et al., 2020 [43], Pakistan	n = 60 39F, 21 M Age: MTA: 45.05y ±6.46y MTB: 46.63y ±5.22y Symptom duration: >3 months	Manual Therapy A (MTA) n = 30 Spencer Muscle Energy Technique GHJ	Manual Therapy B (MTB) n = 30 Shoulder stretching	2 weeks 4 weeks	NPRS Quick-DASH SPADI ER ROM	Statistically significant between group differences in NPRS, SPADI, Quick-DASH, and ER ROM favoring MTA.
Kumar et al., 2012 [44], India	n = 40 14F, 26 M Age: MT: 47.9y EX: 47.1y Symptom duration: >2 months	Manual Therapy + Exercise (MT + EX) n = 20 Maitland GHJ mobilization Shoulder stretching and AAROM	Exercise (EX) n = 20 Shoulder stretching and AAROM	4 weeks	VAS SPADI ER ROM	Statistically significant between group differences in VAS, SPADI, and ER ROM favoring MT + EX.
Nicholson, 1985 [45], USA	n = 20 10F, 10 M Age: MT: 51y ±12.16y EX: 55y ±16.43y Symptom duration: MT: 27.6 wk ±33.41 wk EX: 30.8 wk ±31.28 wk	Manual Therapy + Exercise (MT + EX) n = 10 GHJ mobilization Active exercises in restricted range Resistance exercises HEP	Exercise (EX) n = 10 Active exercises in restricted range Resistance exercise HEP	1 week 2 weeks 3 weeks 4 weeks	Pain ER AROM	No statistically significant between group differences.
Pragassame et al., 2019 [46], India	n = 30 11F, 19 M Age: MTA: 51.73y ±7.70y MTB: 51.40y ±7.37y Symptom duration: 30% 3-6mo 33.33% 6-9mo 36.67% >9mo	Manual Therapy A (MTA) n = 15 GHJ capsular stretching STJ mobilization HEP	Manual Therapy B (MTB) n = 15 GHJ capsular stretching HEP	Post10-day treatment	NPRS CMS ER AROM	Statistically significant between group differences in NPRS, ER AROM, and Constant Score favoring MTA.
Russell et al., 2014 [47], UK	n = 75 35F, 40 M Age: 51.1y ±6.84y Symptom Duration: 5.79mo ±1.48 months	Manual Therapy (MT) n = 24 Maitland mobilization Massage/trigger point release Stretching HEP Exercise (EX) n = 25 Exercise class circuit, including ROM and stretching HEP	Comparison n = 26 HEP	6 weeks 6 months 1 year	CMS OSS	Statistically significant between group differences in Constant Score and Oxford Shoulder Score favoring EX compared to MT and HEP. Statistically significant between group differences in Constant Score and Oxford Shoulder Score favoring MT compared to HEP.
Tariq et al., 2021 [48], Pakistan	n = 28 20F, 8 M Age: MT: 53.5y EX: 53y Symptom duration: NA	Manual Therapy A (MTA) n = 14 GHJ mobilization Inferior capsular stretch	Manual Therapy B (MTB) n = 14 GHJ mobilization	After 8 sessions	VAS SPADI ER ROM	Statistically significant between group differences in VAS, SPADI, and ER ROM favoring MTA.
Vermeulen et al., 2006 [49], the Netherlands	n = 100 66F, 34 M Age: MTA: 51.6y ±7.6y MTB: 51.7y ±8.6y Symptom duration: MTA: 8 months MTB: 8 months	Manual Therapy A (MTA) n = 49 Maitland GHJ and STJ mobilization at end range	Manual Therapy B (MTB) n = 51 Maitland GHJ mobilization in pain-free range	3 months 6 months 12 months	VAS SRQ SDQ ER AROM ER PROM	Statistically significant improvements in ER AROM, ER PROM, the Shoulder Rating Questionnaire, and the Shoulder Disability Questionnaire favoring MTA at 1 year, but not at 3 or 6 months. Statistically significant between group differences for the number of treatment sessions, favoring the MTA group. No statistically significant between group differences in pain.
Yang et al., 2007 [50], Taiwan	n = 28 24F, 4 M Age: MTA: 53.3y ±6.5y MTB: 58y ±10.1y Symptom duration: MTA: 18 wk ±8wk MTB: 22 wk ±10wk	Manual Therapy A (MTA) n = 14 A-B-A-C: GHJ mobilizations A= Mid-range (MRM) B= End range (ERM) C= with Movement (MWM)	Manual Therapy B (MTB) n = 14 A-C-A-B: GHJ mobilizations A= Mid-range (MRM) B= End range (ERM) C= with Movement (MWM)	3 weeks 6 weeks 9 weeks 12 weeks	ER shoulder complex kinematics using the FASTRAK Motion Analysis system	No statistically significant between group differences between ERM and MWM after the first 6 weeks (A-B versus A-C).

Abbreviations: ACJ, acromioclavicular joint; AROM, active range of motion; ADD, adduction; CMS, Constant Murley Score; DASH, Disabilities of the Arm, Shoulder, and Hand questionnaire; ERM, end-range mobilization; EX, exercise group; ER, external rotation; F, females; GHJ, glenohumeral joint; HEP, home exercise program; IR, internal rotation; M, males; NPRS, Numeric Pain Rating Scale; MT, manual therapy group; MRM, mid-range mobilization; MWM, mobilization with movement; mo, months; OSS, Oxford Shoulder Score; PROM, passive range of motion; ROM, range of motion; STJ, scapulothoracic joint; SF-36, Short Form-36; SDQ, Shoulder Disability Questionnaire; SPADI, Shoulder Pain and Disability Index; SRQ, Shoulder Rating Questionnaire; STJ, sternoclavicular joint; VAS, Visual Analog Scale; wk, weeks; y, years.

Data analysis

Data analyses were conducted using Revman 5.4. When available, posttest means and standard deviations (SD) were used for meta-analysis. Change scores were alternately used in the absence of posttest measures or when studies within the same meta-analysis only reported change scores. When SDs were not provided, the authors calculated them for meta-analysis purposes. A random-effects model with inverse variance was used to calculate standardized mean differences (SMD) and 95% confidence intervals (CI) for pain, disability, and ER ROM [31,51]. Although calculation of mean differences would have allowed for clinically meaningful discussion through comparison with minimal clinically important differences (MCID), heterogeneity of measurement tools, scales of measure, and reporting of data with posttest versus change scores were a concern. Statistical heterogeneity was evaluated using the I² [2] statistic, with values greater than 50% indicating high heterogeneity [52]. Effect sizes were presented in forest plots and interpreted based on previous research: 0.2 represents a small effect, 0.5 represents a moderate effect, and 0.8 represents a large effect [53].

Separate meta-analyses were conducted comparing MT and EX groups in isolation and in combination for their effect on pain, disability, and ER ROM in the short term and long term when data were available. Where statistical pooling was not possible, findings were presented in narrative form.

Results

Study selection

The search identified 1953 studies. There were 106 full-text articles assessed for eligibility, with 16 [35–50] meeting inclusion criteria, and eight [35,36,38,40,44–46,49] included in meta-analysis (Figure 1).

Figure 1.

PRISMA flow diagram.

Characteristics of included studies

A total of 745 participants were included across 16 studies, with 58% female and 42% male participants among the studies that reported a distribution (14). All the studies included participants with adhesive capsulitis. Of the 16 studies, 13 [35–46,48,49] assessed pain (VAS, NPRS, and McGill), 11 [36,38–41,43,44,46–49] assessed disability (DASH, Quick-DASH, SPADI, CMS, OSS, SRQ, and SDQ), and 14 [35–50] assessed ER ROM. Additional details of the included studies are found in Tables 1 and 2, and Appendix 3.

Table 2.

Dosage of interventions.

Study	Dosage of intervention group	Dosage of comparison group	Results
Manual therapy vs. exercise
Bulgen et al., 1984 [37]	Visits per week: 3 Number of weeks: 6 Total visits: 18 Session Duration: NR	Visits per week: 3 Number of weeks: 6 Total visits: 18 Session Duration: NR	Pain: NS Disability: NS ROM: NS
Russell et al., 2014 [47]	Visits per week: 2 Number of weeks: 6 Total visits: 12 Session Duration: 50 min	Visits per week: 2 Number of weeks: 6 Total visits: 12 Session Duration: 50 min	Disability: Favors EX
Manual Therapy + Exercise vs. Exercise Only
Ali and Khan, 2015 [36]	Visits per week: 3 Number of weeks: 5 Total visits: 15 Session Duration: 45 min	Visits per week: 3 Number of weeks: 5 Total visits: 15 Session Duration: 45 min	Pain: NS Disability: NS ROM: NS
Celik and Mutlu, 2016 [38]	Visits per week: 3 Number of weeks: 6 Total visits: 18 Session Duration: 50 min	Visits per week: 3 Number of weeks: 6 Total visits: 18 Session Duration: 20 min	Pain: NS Disability: Favors MT + EX ROM: Favors MT + EX
Chen et al., 2009 [40]	Visits per week: 1–2 Number of weeks: 8 Total visits: 10 Session Duration: 30 min	Visits per week: 1–2 Number of weeks: 8 Total visits: 10 Session Duration: 30 min	Disability: NS
Horst et al., 2017 [42]	Visits per week: 2 Number of weeks: 5 Total visits: 10 Session Duration: 30 min	Visits per week: 2 Number of weeks: 5 Total visits: 10 Session Duration: 30 min	Pain: Favors EX ROM: Favors EX
Kumar et al., 2012 [44]	Visits per week: 3 MT, 5 EX Number of weeks: 4 Total visits: 12 MT, 20 EX Session Duration: NR	Visits per week: 5 Number of weeks: 4 Total visits: 20 Session Duration: NR	Pain: Favors MT + EX Disability: Favors MT + EX ROM: Favors MT + EX
Nicholson, 1985 [45]	Visits per week: 2–3 Number of weeks: 4 Total visits: 8–12 Session Duration: NR	Visits per week: 8–12 Number of weeks: 4 Total visits: 8–12 Session Duration: NR	Pain: NS ROM: NS
Manual Therapy + Exercise vs. Manual Therapy + Exercise
Deshmukh et al., 2014 [41]	Visits per week: 3 Number of weeks: 3 Total visits: 9 Session Duration: NR	Visits per week: 3 Number of weeks: 3 Total visits: 9 Session Duration: NR	Pain: Favors MT Disability: Favors MT ROM: Favors MT
Manual Therapy vs. No Treatment/Control
Bulgen et al., 1984 [37]	Visits per week: 3 Number of weeks: 6 Total visits: 18 Session Duration: NR	Visits per week: 0 Number of weeks: 0 Total visits: 0 Session Duration: NR	Pain: NS ROM: NS
Chauhan et al., 2011 [39]	Visits per week: 3 Number of weeks: 2 Total visits: 6 Session Duration: 60 min	Visits per week: NR Number of weeks: 2 Total visits: NR Session Duration: NR	Pain: Favors MT Disability: Favors MT ROM: NS
Russell et al., 2014 [47]	Visits per week: 2 Number of weeks: 6 Total visits: 12 Session Duration: 50 min	Visits per week: NR Number of weeks: NR Total visits: NR Session Duration: NR	Disability: Favors MT ROM: Favors MT
Exercise vs. No Treatment/Comparison
Bulgen et al., 1984 [37]	Visits per week: 3 Number of weeks: 6 Total visits: 18 Session Duration: NR	Visits per week: NR Number of weeks: NR Total visits: NR Session Duration: NR	Pain: NS ROM: NS
Russell et al., 2014 [47]	Visits per week: 2 Number of weeks: 6 Total visits: 12 Session Duration: 50 min	Visits per week: NR Number of weeks: NR Total visits: NR Session Duration: NR	Disability: Favors EX ROM: Favors EX
Manual Therapy vs. Manual Therapy
Agarwal et al., 2016 [35]	Visits per week: 3 Number of weeks: 6 Total visits: 18 Session Duration: NR	Visits per week: 3 Number of weeks: 6 Total visits: 18 Session Duration: NR	Pain: Favors MTA ROM: NS
Iqbal et al., 2020 [43]	Visits per week: 3 Number of weeks: 4 Total visits: 12 Session Duration: NR	Visits per week: 3 Number of weeks: 4 Total visits: 12 Session Duration: NR	Pain: Favors MTA Disability: Favors MTA ROM: Favors MTA
Pragassame et al., 2019 [46]	Visits per week: 1×/day Number of weeks: 10 days Total visits: 10 Session Duration: NR	Visits per week: 1×/day Number of weeks: 10 days Total visits: 10 Session Duration: NR	Pain: Favors MTA Disability: Favors MTA ROM: Favors MTA
Tariq et al., 2021., [48]	Visits per week: 3 Number of weeks: 3 Total visits: 8 Session Duration: 15-20 min	Visits per week: 3 Number of weeks: 3 Total visits: 8 Session Duration: 15-20 min	Pain: Favors MTA Disability: Favors MTA ROM: Favors MTA
Vermeulen et al., 2006 [49]	Visits per week: 2 Number of weeks: 6–12 Total visits: 18.6 ± 4.9 (mean) Session Duration: 30 min Treatment ended when normal ROM achieved	Visits per week: 2 Number of weeks: 6–12 Total visits: 21.5 ± 2.5 (mean) Session Duration: 30 min Treatment ended when normal ROM achieved	Pain: NS Disability: Favors MTA ROM: Favors MTA
Yang et al., 2007 [50]	Visits per week: 2 Number of weeks: 12 Total visits: 24 Session Duration: 30 min 3 weeks per intervention: 6 MRM, 6 ERM, 6 MRM, 6 MWM	Visits per week: 2 Number of weeks: 12 Total visits: 24 Session Duration: 30 min 3 weeks per intervention: 6 MRM, 6MWM, 6 MRM, 6 ERM	ROM: NS

Abbreviations: ERM, end range mobilization; EX, exercise; MT, manual therapy; MTA, manual therapy A group; MRM, mid-range mobilization; MWM, mobilization with movement; NR, not reported; NS, not significant; ROM, range of motion.

Risk of bias

Of the nine RCTs included in meta-analysis, 11% had low risk of bias, 67% had some concerns, and 11% had high risk of bias (Table 3).

Table 3.

Risk of Bias [32,38,38,39,39,41,41,43,47,47,48,48,49,49,52,52].

Abbreviations: !, some concerns; -, high risk of bias; +, low risk of bias; EX, exercise; ER, external rotation; GHJ, glenohumeral joint mobilization; MT, manual therapy; ROM, range of motion; STJ, scapulothoracic joint mobilization.

Bias domains: 1, Randomization process; 2, Deviations from intended interventions; 3, Missing outcome data; 4, Measurements of the outcome; 5, Selection of the reported result.

Outcomes

Manual therapy versus exercise

Two studies [37,47] included 72 participants, 54% female. One study, Russell et al. [47] demonstrated statistically significant between-group differences in disability in the short- and long term favoring exercise compared to MT, while Bulgen et al. [37] reported no significant between-group differences on pain or ER ROM in the short or long term.

Manual therapy plus exercise versus exercise only

Six studies [36,38,40,42,44,45] included 296 participants, 54% female. Meta-analysis of four studies [36,38,44,45] (n = 133) revealed a non-significant effect on pain (SMD −0.54; 95% CI: −1.16, 0.08; I [2] = 67%; p = 0.09) at short-term follow-up (Figure 2a). Meta-analysis of four studies [36,38,40,44] (n = 203) revealed a non-significant effect on disability (SMD −0.48; 95% CI: −1.01, 0.05; I [2] = 69%; p = 0.08) at short-term follow-up (Figure 2b). Meta-analysis of two studies [38,40] (n = 120) revealed a non-significant effect on disability (SMD −0.45; 95% CI: −1.61, 0.72; I [2] = 86%; p = 0.45) at long-term follow-up (Figure 2c). Meta-analysis of four studies [36,38,44,45] (n = 133) revealed a non-significant effect on ER ROM (SMD 0.21; 95% CI: −0.43, 0.84; I [2] = 69%; p = 0.53) at short-term follow-up (Figure 2d).

Figure 2.

Meta-analyses of manual therapy plus exercise versus exercise only for (a) pain in the short term, (b) disability in the short term, (c) disability in the long term, and (d) external rotation range of motion in the short term.

Manual therapy plus exercise versus a different form of manual therapy plus exercise

One study [41] included 30 participants (no female-to-male distribution reported). Deshmukh et al. [41] demonstrated statistically significant between-group differences on pain, disability, and ER ROM in the short term favoring the addition of myofascial release to GHJ, STJ, SCJ, and ACJ mobilization and exercise compared to the group that did not receive myofascial release.

Manual therapy versus no treatment/control

Three studies [37,39,47] included 96 participants, 55% female. Two studies demonstrated statistically significant between-group differences on disability in the short term [39,47] favoring MT compared to a home exercise program (HEP), with Russell et al. [47] maintaining significant differences in the long term. One study, Chauhan et al., [39] demonstrated statistically significant between-group differences on pain in the short term favoring MT over HEP, while another Bulgen et al. [37] demonstrated non-significant findings. Neither of the studies that measured ER ROM [37,39] reported statistically significant between-group differences in ER ROM.

Exercise versus no treatment/comparison

Two studies [37,47] included 71 participants, 54% female. One study, Russel et al., [47] demonstrated statistically significant between-group differences on disability in the short- and long term favoring exercise compared to HEP, while another Bulgen et al., [37] reported no statistically significant between-group differences on pain or ER ROM.

Manual therapy versus manual therapy

Six studies [35,43,46,48–50] included 276 participants, 63% female. Of the six studies that compared one manual therapy intervention to another, three [35,46,49] implemented GHJ and STJ mobilization techniques compared to only GHJ mobilization techniques and were, therefore, the only studies included in meta-analysis. Meta-analysis of three studies [35,46,49] (n = 160) revealed a non-significant effect on pain (SMD −0.98; 95% CI: −2.50, 0.54; I [2] = 93%; p = 0.21) at short-term follow-up (Figure 3a). Meta-analysis of two studies [46,49] (n = 130) revealed a non-significant effect on disability (SMD 0.67; 95% CI: −0.46, 1.79; I [2] = 84%; p = 0.24) at short-term follow-up (Figure 3b). Of the other three studies [43,48,49], two demonstrated statistically significant between-group differences on pain and disability in the short term favoring muscle energy technique to the GHJ compared to passive stretching [43] and GHJ mobilization with inferior capsular stretch compared to GHJ mobilization only [48], while the third [49] demonstrated statistically significant between-group differences on disability in the long term but not disability in the short term or pain in the short or long term.

Figure 3.

Meta-analyses of glenohumeral joint and scapulothoracic joint mobilization versus glenohumeral joint mobilization only for (a) pain in the short term, (b) disability in the short term, and (c) external rotation range of motion in the short term.

Meta-analysis of three studies [35,46,49] (n = 160) revealed a non-significant effect on ER ROM (SMD 0.31; 95% CI: −0.01, 0.62; I [2] = 0%; p = 0.05) at short-term follow-up (Figure 3c). Of the other three studies that assessed ER ROM [43,48,50], two demonstrated statistically significant between-group differences on ER ROM in the short term favoring muscle energy technique [43], and GHJ mobilization with inferior capsular stretch [48], while the third [50] reported no significant between-group differences when comparing end range mobilization to mobilization with movement.

Dosage

Manual therapy technique

Across the 16 included studies, all implemented MT as an intervention and all included joint mobilization to the GHJ and capsule. The majority of the studies implemented inferiorly [35,36,38,39,44,46,49] and posteriorly [35,36,38,41,44,46,49] directed GHJ mobilizations, with three [38,46,49] also including anterior mobilization, three [42,45,50] reporting a pragmatic design based on examination findings, and four [37,40,47,48] that did not specify the direction of mobilization. In addition to GHJ mobilization, four studies [35,41,46,49] included STJ mobilization, and two [40,41] included both ACJ and SCJ mobilization. Three studies [39,41,47] utilized soft tissue mobilization techniques, and one study, Iqbal et al. [43] implemented a muscle energy technique (Table 1, Appendix 3).

Dosage of manual therapy technique

Seven studies [35,36,38,41,43,44,50] defined a specific dosage of repetitions and/or duration of MT technique delivered, with considerable variation in both dosage and language of reporting, i.e. 15 bouts of 1 minute, 5 bouts of 30 seconds, 3 bouts of 15 repetitions, 2–3 oscillations per second for 1–2 minutes repeated for 3–4 bouts. Six studies [35,36,38,40,45,49] specified the grade of mobilization delivered, with only one study [49] comparing the implementation of grades III-IV to grades I-II (Table 1 , Appendix 3). Of the other five studies, the MT + EX group of two studies [36,40] received grades II-III, the MT + EX group of one study [38] received grades I-II, and the MT group of one study [35] received grade III. The last study reported that the force and amplitude of the mobilization varied, with all subjects in the MT + EX group eventually tolerating grade IV mobilization [45].

Dosage of exercise interventions

Two studies defined a specific dosage of repetitions of exercise interventions [38,44]. Celik and Mutlu [38] describe cyclic intermittent stretching for 10 repetitions of 20 seconds in each desired direction for a total of 20 minutes, whereas Kumar [44] et al. prescribed 10 repetitions of 10 seconds in each direction. Specific exercise interventions varied from active-assisted ROM, active ROM, stretching, and resistance exercise (Appendix 3).

Single session duration

Nine studies [36,38–40,42,47–50] reported the single session duration, ranging from 15 to 60 minutes, which included all MT and exercise interventions within the session (Table 2 , Appendix 3).

Frequency of intervention

All 16 studies [35–50] reported the frequency of intervention, ranging from 1–2 times per week to 1 time per day (for 10 days), with the most common frequency reported as 3 times per week (10 studies [35–39,41,43–45,48] (Table 2 , Appendix 3).

Total number of sessions and duration of care

The total number of sessions ranged from 6 to 18 visits [35–49]. Duration of the episode of care ranged from 10 days to 8 weeks [35–49]. One study, Yang et al., [50] represented an outlier by implementing an A-B-A-C versus A-C-A-B study design with 6 visits per intervention for 3 weeks totaling 24 visits over 12 weeks. Finally, in one study design, Vermeulen et al., [49] terminated treatment when normal ROM was achieved, resulting in a mean of 18.6 ± 4.9 visits and 21.5 ± 2.5 visits for MT groups A and B, respectively, over 6–12 weeks (Table 2, Appendix 3).

GRADE evidence profile

Two comparisons were assessed across outcomes of pain, disability, and ER ROM: MT plus exercise versus exercise alone, and GHJ and STJ mobilization versus GHJ mobilization alone. The overall level of evidence ranged from very low to low and is presented in Table 4. Criteria for downgrading of clinical trials are presented in Appendix 5. Studies were downgraded most often due to risk of bias, as well as inconsistency, and imprecision due to heterogeneity and small sample sizes of the included studies.

Table 4.

GRADE evidence profile³¹.

Outcome (n=studies)	Participants	Risk of bias	Inconsistency	Indirectness	Imprecision	Publication Bias	Level of Evidence
Manual Therapy + Exercise versus Exercise Alone
Pain <6 weeks (n = 4)	133	Serious*	Serious†	Not serious	Serious‡	None	⨁OOO Very Low
Disability <6 weeks (n = 4)	203	Serious*	Serious†	Not serious	Serious‡	None	⨁OOO Very Low
Disability 6–12 months (n = 2)	120	Not serious	Serious†	Not serious	Serious‡	None	⨁⨁OO Low
ER ROM <6 weeks (n = 4)	133	Serious*	Serious†	Not serious	Serious‡	None	⨁OOO Very Low
GHJ and STJ Mobilization versus GHJ Mobilization Alone
Pain <3 months (n = 3)	160	Serious§	Serious†	Not serious	Serious‡	None	⨁OOO Very Low
Disability <3 months (n = 2)	130	Serious§	Serious†	Not serious	Serious‡	None	⨁OOO Very Low
ER ROM (n = 3)	160	Serious§	Not serious	Not serious	Serious‡	None	⨁⨁OO Low

Abbreviations: ER ROM, external rotation range of motion; GHJ, glenohumeral joint; STJ, scapulothoracic joint.

*Risk of bias associated with selection bias, performance bias, detection bias, and attrition bias.

†Studies demonstrate heterogeneity I² > 50%.

‡Studies contain small sample sizes.

§Risk of bias associated with selection bias, performance bias, and detection bias.

Discussion

To our knowledge, this is the first systematic review on AC that assessed the effectiveness of MT and exercise on outcomes, incorporating a discussion related to the dosage of these common physical therapy interventions. This systematic review included 16 studies [35–50] with data permitting meta-analyses on eight studies [35,36,38,40,44–46,49]. All meta-analyses revealed non-significant effects for pain, disability, and ER ROM on short- and long-term follow-up. Unfortunately, the use of SMDs, in addition to non-significant statistical results prevents further discussion on the clinically meaningful effects of these interventions. However, because the overall level of evidence for the non-significant results ranged from very low to low, the certainty of the findings is likely to change with further research.

While all meta-analyses demonstrated non-significant effects, 11 of the 16 included studies demonstrated significant between-group differences across a variety of outcomes. Individual studies have demonstrated the effectiveness of MT, exercise, and MT and exercise combined. When comparing MT to exercise, one study favored EX for its effect on disability [47]. When comparing MT plus exercise to exercise only, two studies [38,44] favored MT plus exercise for their effects on pain, disability, and ER ROM, while a third study [42] favored exercise for pain and ER ROM. When comparing MT to an HEP, two studies [39,47] favored MT for their effects on disability, while one of them [39] also showed significant effects on pain. Finally, when comparing exercise to an HEP, one study [47] favored exercise for its effect on disability. Additional significant findings of individual studies demonstrated the superiority of one MT technique over another [35,43,46,48,49].

While individual studies have demonstrated support for both MT, exercise, and combined MT and exercise for AC, overall recommendations for specific techniques, intensity, frequency of therapy, and duration of the episode of care remain absent in the literature. Such recommendations are difficult to ascertain from the available literature given the lengthy nature of AC and the many variables to consider throughout the duration of this pathology, including, but not limited to, stage of the condition, severity of pain, patient irritability, range of motion, joint mobility, and joint involvement. Based on a previous systematic review [54], several factors (MT technique, repetition, and duration of the technique, duration of a single session, frequency of sessions, total number of sessions, and total duration of care) were considered in an attempt to better understand the available literature on dosage of MT for AC, demonstrating several trends when evaluating the data on MT dosing parameters. First, out of the 16 studies that implemented MT, only three described a pragmatic approach to address participants’ impairments based on examination findings, while four neglected to mention the direction of joint mobilization. Second, as expected, the majority of studies implemented GHJ mobilization, however, considering that efficient scapulohumeral rhythm functions optimally as coordinated movement between the GHJ, STJ, SCJ, and ACJ, only five studies mentioned intervention to restore motion to these essential structures. Previous research, albeit in the form of a case series, has demonstrated the positive effects of a pragmatic approach to AC, based on theories of regional interdependence [55]. Third, the variability as it pertains to the dosage and application of MT techniques and the language used to describe the interventions was quite broad and made it very difficult to subgroup studies in order to provide more specific dosage recommendations. A standardized nomenclature for manual therapy has been recommended by the American Academy of Orthopedic Manual Physical Therapists, suggesting that MT techniques be described using the following six characteristics: rate of force application, location within the available range of motion, direction of force application, target of force application, intended structural movement, and patient positioning [56]. When considering the potential for variability in patient presentation for those with AC, standardized language, and implementation into study design would allow for a more precise assessment of the literature, which may lead to more conclusive guidelines for managing patients within different stages of the condition.

Additional trends were observed as it pertains to the dosage of physical therapy care over time. The majority of studies assessed short-term effects, with only four out of 16 included studies carrying out follow-up assessments to at least 6 months, with the maximum number of prescribed visits being 24 visits over 12 weeks. Given the nature of AC, with the potential for patients to experience symptoms beyond 12 months of the condition, studies that discontinue intervention after a maximum of 3 months may be missing the potential for more significant findings with longer episodes of care. Furthermore, as reimbursement via in-network physical therapy intervention continues to be a challenge [57], it is vital that more definitive guidelines be explored for the ideal duration of an episode of care for AC to ensure we are prioritizing patients who will benefit from physical therapy services. Similarly, the duration of treatment sessions and frequency of visits varied considerably across studies as well, with total session duration lasting from 15 to 60 minutes, and frequency of sessions anywhere from one to seven times per week. Coincidentally, variation in these two dosage parameters is actually most generalizable to current clinical practice, where some practices may see patients one-on-one for only 10–15 minutes and others have the financial means to spend a full hour with patients. With more conclusive recommendations for other dosage parameters (duration, intensity, and variability of the technique) there may be potential to modify the efficiency of session duration to fit multiple practice models.

While this systematic review is innovative in design, it is not without limitations. The design of the included studies did not allow for comparison between dosages of MT or exercise interventions, making it impossible to determine the ideal dosage of physical therapy for patients with AC. Furthermore, AC of different stages was analyzed together across studies, and therefore, conclusions about effective management strategies within specific stages of the pathology could not be drawn. Even though most studies included a home exercise program, details regarding frequency, intensity, duration, and adherence were insufficient and not taken into account when analyzing the data in any of the included studies. Finally, and perhaps most importantly, the large heterogeneity present in the study designs and data reported prevented further meta-analysis.

Conclusions

Non-significant findings with low to very low quality of evidence were found across meta-analyses, preventing seamless transition of research evidence to clinical practice. However, many individual studies demonstrated positive results, suggesting that further research may support the use of MT and exercise in the management of patients with AC. The lack of consistency in study designs, manual therapy techniques, dosing parameters, and duration of care make it difficult to make strong recommendations regarding the optimal dosage of physical therapy for individuals with AC.

Biographies

Kaitlin Kirker is a physical therapist who holds a Doctorate of Physical Therapy (DPT), is board certified in orthopedics (OCS), and completed her fellowship training in orthopedic manual physical therapy at Bellin College. She is a staff physical therapist at Masefield and Cavallaro Physical Therapy in Brooklyn, NY, and is an Adjunct Assistant Professor in the DPT program at the Brooklyn Campus of Long Island University.

Melanie O’Connell is a physical therapist who holds a Doctor of Philosophy (PhD) and is board certified in pediatrics (PCS). She is an Assistant Professor and Director of Student Advisement and Engagement in the DPT program at the Brooklyn Campus of Long Island University and is a per diem physical therapist at the New Jersey Institute for Disabilities in the Early Intervention Program.

Lisa Bradley is a physical therapist who holds a Doctorate of Physical Therapy (DPT) and a Masters in Social Work (MSW). She is an Associate Professor and the Director of Clinical Education in the DPT program at the Brooklyn Campus of Long Island University.

Rosa Elena Torres-Panchame is a physical therapist who holds a Doctorate of Physical Therapy (DPT), is board certified in orthopedics (OCS), and is a PhD candidate at Nova Southeastern University. She is the Assistant Program Director and Director of Clinical Education in the DPT program at St. John’s University, and a per diem physical therapist at Brooklyn Body Works Physical Therapy.

Michael Masaracchio is a physical therapist who holds a Doctorate of Physical Therapy (DPT) and Doctor of Philosophy (PhD).

Funding Statement

The authors have no funding to disclose.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary data

Supplemental data for this article can be accessed online at https://doi.org/10.1080/10669817.2023.2180702

Author contributions

KK and MM developed the concept idea, analyzed all data, and wrote the initial draft of the manuscript. MO, LB, and RTP conducted the literature search, collected data, and revised the manuscript. KK, MM, MO, and LB scored all included articles for their methodological quality. All authors approved the final version.