This was a randomized controlled study approved by the institutional review board of National Taiwan University Hospital (ClinicalTrials.gov ID: NCT01022827; Protocol ID: 200905041R). Patients evaluated as having glenohumeral internal rotation limitation in our outpatient clinic were eligible for participation in the study. The inclusion criteria were: (1) limitation of internal rotation ROM compared to the sound side at least 10%; (2) tightness in the posterior shoulder region. Posterior shoulder tightness was defined as more tightness measurement values at least 10% compared to the sound side. Measurement of posterior shoulder tightness was based on horizontal flexion ROM (cross-chest adduction) measurement [15
]. Because we measured the transverse tightness of the muscles by myotonometer, skin/subcutaneous tissue thickness may affect the validity of the measurement. Thus, subjects with body mass index (BMI) (less than 19 or more than 24) were expected to have confounding factor of skin/subcutaneous tissue thickness on the muscle tightness measurement and were excluded from the study. The BMI was calculated by dividing his or her body weight in kilograms by the square of the body height in meters. The other exclusion criteria were: (1) surgery on the particular shoulder, (2) rheumatoid arthritis, (3) stroke with residual shoulder involvement, or (4) fracture of the shoulder complex.
Based on the judgment of what constitutes clinically meaningful differences and variability estimates from previous studies [7
], a sample size of 25 subjects per group provided 80% power to detect differences of 15 degrees internal rotation ROM between the pre- and post-intervention as well as between the 2 groups of interest at an alpha level of .05 with a two-tailed test. They received a written and verbal explanation of the purposes and procedures of the study. If they agreed to participate, they signed informed consent forms approved by the Human Subjects Committee of NTUH.
A total of 69 patients were recruited, of whom 9 were excluded by the criteria. Sixty patients were randomized by computer generated permuted block randomization of 15 by sequentially numbered, sealed, opaque envelopes to massage and control groups: 43 women and 17 men, with an average age of 54 years (range 43-73 years) (Table ). The permutation lists were MMCC, MCMC, MCCM, CCMM, CMCM, CMMC (M: massage; C: control). Patients signed an informed consent form before participating in the study. Figure presents a CONSORT diagram that summarizes the flow of activities and participants through the clinical trial.
CONSORT diagram of enrollment and flow of activities through the clinical trial.
Muscle tightness measurement
Muscle tightness, defined as the change in passive tension per unit change in length, is an indication of a muscle's passive tension to length change. The assessment of the muscle tightness can be longitudinal or transverse to the muscle [21
]. A computerized myotonometer (Neurogenic Technologies, Inc) was used to measure the transverse tightness of the muscles. The myotonometer measures tissue tightness by quantifying the amount of tissue displacement (± 0.1 mm) as compared to the constant applied pressure as a probe is pushed downward onto the muscle and underlying tissue. The tissue displacement values were recorded at eight force probe pressures (0.25, 0.50, 0.75, 1.00, 1.25, 1.50, 1.75, 2.00 kg). The force-displacement curves were generated from these data. Thus, the slope for each force-displacement curve was calculated (Figure ). Less penetration of the probe and a sharp slope of the force-displacement curve indicate higher resistance (more tightness). Myotonometer measurements of muscle tightness has been demonstrated to be valid and reliable [21
]. Jenkyn et al. [25
] have pointed out that transverse tightness could be correlated with muscle tension. Based on our pilot study on 8 shoulders, high intrarater within-session (20 minutes time lapsed) reliability (intraclass correlation coefficient = 0.98) of this measurement was observed. Additionally, construct validity of this measurement was observed. More posterior muscle tightness was proposed to occur in end-range position. As expected, less penetration of the probe was observed in end-range internal rotation compared to neutral internal rotation in our pilot study (P
Force-displacement curves of 3 muscle tightness. Slopes of 3 muscles are demonstrated pre- and post-massage for one subject.
The self-reported Flexilevel Scale of Shoulder Function (FLEX-SF) was used to present functional disability from symptoms [26
]. In this scale, respondents answer a single question that grossly classifies their level of function as low, medium, or high. They then respond to only the items that target their level of function. This scale covers the entire continuum assessment of shoulder functions and has been satisfactorily tested for appropriate psychometric properties of reliability, validity, and responsiveness to clinical change [26
]. Scores were recorded from 1, with the most limited function, to 50, without any limited function in the subject. Each patient was asked to indicate functional disability at the baseline and at a 4-week follow up. The percentage change in FLEX-SF was calculated (final score - initial score)/initial score × 100). To develop a prediction method, we need to justify that the two subgroups are responsive and nonresponsive. If the change was > 20%, the patient was categorized in the responsive group. If change was < = 20%, the patient was categorized in the nonresponsive group. We chose 20% change in FLEX-SF as the responsive criterion because the patients generally felt satisfied with 20% improvement from our investigation in the clinic [27
After signing the informed consent form, the subjects were examined by a physical therapist to establish the clinical conditions of their shoulders, including glenohumeral internal ROM, 3 muscle tightness measurements [posterior deltoid, infraspinatus, and teres minor muscles] and the FLEX-SF questionnaire.
In a prone position, the subject's arm was moved passively to the cessation of movement (firm end-feel) of internal rotation with the arm held in 90 degrees abduction by the tester. The recorder who was blinded to group allocation placed a hand-held goniometer (Ever Prosperous Instrument, Inc.) with two arms parallel to the forearm and trunk, respectively, and documented glenohumeral internal rotation ROM. During the test, the scapula was palpated at the lateral border and stabilized by hand. These measurements were aborted and restarted if the subject was unable to relax or if the scapula could not be stabilized effectively.
Subsequently, the tightness of the 3 posterior shoulder muscles was evaluated by the assessor. Each patient was tested while maintained in internal rotation end-range prone position, and the patient was told to expose the shoulder area undergoing the testing. The patient was asked to relax the shoulder. A surface electromyography was used to monitor the muscle tone and to confirm muscular activity at rest (less than mean activity plus 2 standard deviation at rest for 1 minute with shoulder neutral rotation in prone position) during muscle tightness measurement. The head of the myotonometer probe was placed over the 3 posterior shoulder muscles in Latin square order (posterior deltoid: two fingerbreadths caudad to the posterior margin of the acromion; infraspinatus: two fingerbreadths below the medial portion of the spine of the scapula; teres minor: one-third of the way between the acromion and the inferior angle of the scapula along the lateral border). The placements of probe head were between 2 electrodes of EMG of each muscle to confirm resting muscular activity during muscle tightness measurement. According to the software manual, each muscle was tested in three trials (each trial had 4 measurements) (Figure ). Each muscle was tested in three trials (each trial consisted of 4 measurements). Myotonometer data recordings of all eight force increments were acquired in approximately 1 second. The intrarater/interrater reliabilities are high (ICC = 0.99) on muscle tightness measurements [28
]. Therefore, the mean of 3 trials for each muscle was calculated for data analysis.
Figure 3 The stiffness measurement sites for the 3 muscles using myotonometer probe placement over the 3 posterior shoulder muscles. Posterior deltoid: two fingerbreadths caudad to posterior margin of the acromion; infraspinatus: two fingerbreadths below medial (more ...)
For the massage group, 2 physical therapists with at least 8 years of clinical experience in manual therapy provided the massage on the posterior deltoid, infraspinatus, and teres minor of the involved shoulder for 18 minutes (about 6 minutes for each muscle with Latin square order) two times a week for 4 weeks. The techniques of massage including petrissage for 3 minutes and rolling for 3 minutes of soft tissues were applied to the patients with prone position and arm by side [29
]. For the control group, same therapists applied light hand touch on the muscles (placebo control) 10 minutes two times a week for 4 weeks. After 4 weeks, the glenohumeral internal rotation ROM and 3 muscle tightness measurements at the pre-massage internal rotation position (posterior deltoid, infraspinatus, and teres minor muscles) were evaluated by the same blinded assessor for each patient.
Data were analyzed using SPSS 15 software (SPSS Inc., Chicago, IL). Baseline variables were compared between groups using independent t tests. To test whether a difference of treatment efficacy existed between 2 groups, 2-factor ANOVA mixed models with factors of group (control group, massage group) and time (the initial data and follow up data at 4 weeks) were performed on each of the outcomes. Bonferroni follow-up analyses were used to adjust for multiple pair-wise comparisons where appropriate. Intention-to-treat analysis was performed by including the drop-out data carrying the last data point forward into analysis. Additionally, Pearson correlations between tightness slope for each muscle and BMI were calculated to evaluate the potential skin/subcutaneous tissue thickness effect on the muscle tightness measurement.
We evaluated the potential predictors for the massage treatment. Responders versus non-responders within the massage group were compared with the chi-square or t test for all potential predictor variables (sex, age, BMI, duration of symptoms, glenohumeral internal rotation, muscle tightness in each muscle, and FLEX-SF score), as appropriate. Predictor variables that had a difference with a p-value ≤ .10 were entered into a logistic regression model. The variables with the least predictive value were then removed, one by one, in a backwards stepwise fashion until all predictors in the model had p-values ≤ 0.05.