Manipulative Therapy in Addition to Usual Medical Care for Patients with Shoulder Dysfunction and Pain: A Randomized, Controlled Trial


Manipulative Therapy in Addition to Usual Medical Care for Patients with Shoulder Dysfunction and Pain: A Randomized, Controlled Trial

Bergman GJ, Winters JC, Groenier KH, Pool JJ, Meyboom-de Jong B,
Postema K, van der Heijden GJ

University of Groningen and University Hospital of Groningen,
Groningen, The Netherlands.

BACKGROUND:   Dysfunction of the cervicothoracic spine and the adjacent ribs (also called the shoulder girdle) is considered to predict occurrence and poor outcome of shoulder symptoms. It can be treated with manipulative therapy, but scientific evidence for the effectiveness of such therapy is lacking.

OBJECTIVE:   To study the effectiveness of manipulative therapy for the shoulder girdle in addition to usual medical care for relief of shoulder pain and dysfunction. DESIGN: Randomized, controlled trial.

SETTING:   General practices in Groningen, the Netherlands. PATIENTS: 150 patients with shoulder symptoms and dysfunction of the shoulder girdle. INTERVENTIONS: All patients received usual medical care from their general practitioners. Only the intervention group received additional manipulative therapy, up to 6 treatment sessions in a 12-week period.

MEASUREMENTS:   Patient-perceived recovery, severity of the main complaint, shoulder pain, shoulder disability, and general health. Data were collected during and at the end of the treatment period (at 6 and 12 weeks) and during the follow-up period (at 26 and 52 weeks).

RESULTS:   During treatment (6 weeks), no significant differences were found between study groups. After completion of treatment (12 weeks), 43% of the intervention group and 21% of the control group reported full recovery. After 52 weeks, approximately the same difference in recovery rate (17 percentage points) was seen between groups. During the intervention and follow-up periods, a consistent between-group difference in severity of the main complaint, shoulder pain and disability, and general health favored additional manipulative therapy.

LIMITATIONS:   The sample size was small, and assessment of end points was subjective.

CONCLUSIONS:   Manipulative therapy for the shoulder girdle in addition to usual medical care accelerates recovery of shoulder symptoms.

From the FULL TEXT Article:

Manipulative Therapy

According to the International Federation of Orthopedic Manipulative Therapists, “orthopedic manipulative (manual) therapy is a specialization within physical therapy and provides comprehensive conservative management for pain and other symptoms of neuro-musculo-articular dysfunction in the spine and extremities.” Our approach to manipulative therapy focused on manual manipulation and mobilization techniques used in western Europe, North America, and Australia, including those described by Cyriax [17], Greenman [18], and Lewit [19]. In our trial, manipulative therapy included specific manipulations (low-amplitude, high-velocity thrust techniques) and specific mobilizations (high-amplitude, low-velocity thrust techniques) to improve overall joint function and decrease any restrictions in movement at single or multiple segmental levels in the cervical spine and upper thoracic spine and adjacent ribs. The manual therapist chose the applied techniques on the basis of the location of the dysfunction and the therapist’s technique preferences. Within the boundaries of the protocol, treatment could be reassessed and adapted to the patient’s condition.

A maximum of 6 treatment sessions could be given over a 12-week period. Eight experienced physiotherapists who were members of the Dutch Association of Manual Therapy and registered by the Royal Dutch Society for Physical Therapy (a member of the International Federation of Orthopedic Manipulative Therapists) provided the manual therapy. To minimize variations in manipulative therapy, therapists received a special training session to familiarize them with the protocol’s mobilization and manipulation techniques for treatment of the cervicothoracic spine and the adjacent ribs. Other interventions (for example, exercises, massage, advice about posture, and treatment of the shoulder joint) were considered deviations from the treatment protocol and were therefore discouraged throughout the trial. Specific treatment characteristics and protocol deviations were recorded at each visit.


Outcome measures were recorded at baseline, at 6 weeks (during the intervention period), and at 12 weeks (at completion of the intervention period). The primary outcome measure was patient-perceived recovery. Patients were considered recovered if they reported being “completely recovered” or “very much improved” on a 7-point ordinal scale. In addition, patients were asked whether they felt “cured” according to the following definition: “You are considered cured if your shoulder symptoms are improved to such an extent that you no longer perceive them as inconvenient.” Secondary outcomes included the severity of 3 individual main complaints [20], shoulder pain [21], functional disability [22], general health [23], and costs (costs data not yet available) [24]. The main complaint is defined as an unavoidable painful or limited functional activity during daily life in which the shoulder is involved. It is a patient-specific or individualized approach for measuring limitation of shoulder function during daily activities. During each session, manual therapists and general practitioners documented the treatment content on a standardized registration form.


In our study, manipulative therapy for the cervicothoracic spine and the adjacent ribs in addition to usual medical care by a general practitioner accelerated recovery of shoulder symptoms. At 12 weeks after randomization, we found a statistically significant difference in recovery rate (43% vs. 21%; difference, 22 percentage points [CI, 6.9 to 35.4 percentage points]) in favor of additional manipulative therapy. Other outcome measures, such as shoulder pain and shoulder disability, consistently favored additional manipulative therapy, supporting our main finding. These favorable effects were maintained during the follow-up period. At 52 weeks, we found a statistically significant difference in recovery rate (52% vs. 35%; difference, 17 percentage points [CI, 0 to 31.4 percentage points]) in favor of additional manipulative therapy. Adjustment for important prognostic factors (for example, age, sex, treatment preference, and duration of symptoms) did not change our results. However, we found a differential effect of individual manual therapists. The patient recovery rates for individual therapists varied from 14% to 67% at the end of the intervention period, regardless of prognostic status or the number of patients treated. Analysis of the treatment registration forms showed that all therapists treated within the boundaries of the protocol; therefore, this variation is probably due to unfavorable prognostic status among treated patients. Although such differential effects across therapists reflect daily health care practice, they probably caused us to underestimate the overall effectiveness of manipulative therapy.

To our knowledge, this is the first trial to focus on the effectiveness of adding manipulative therapy for the cervicothoracic spine and the adjacent ribs to usual medical care for treatment of shoulder symptoms. We did not deviate from the original study design, which was published independently of the study results [15]. At randomization, the treatment groups were similar in demographic and patient characteristics and putative prognostic indicators. Our sample is comparable to those of other studies examining shoulder symptoms in general practice [3, 14]. To improve the transparency of the contrasted treatments, we used specific protocols for both usual medical care and manipulative therapy. Protocols for manipulative therapy focused on a limited number of manipulative and mobilizing techniques that target prespecified bones and joints.

Because of the open nature of manipulative therapy, blinding of patients, general practitioners, and physical therapists was not possible. Lack of blinding among patients could have caused ascertainment bias. Patients’ treatment preferences could have influenced their responses regarding subjective outcome measures [25]. Therefore, patients who were a priori unwilling to adhere to allocated treatments and those who had an absolute preference for or against manipulative therapy were excluded. In addition, our analyses showed that treatment preferences did not affect patient-perceived recovery. Lack of blinding of general practitioners and manual therapists could have reduced the comparability of usual medical care. However, the number and content of general practitioner sessions were similar for both groups.

Because recruitment yield was lower than expected, we decided to extend the inclusion period by 6 months, which allowed us to include 150 patients instead of the intended 250 patients. We decided to stop recruitment because of time and budget constraints. Neither decision was supported by interim analysis. Before the start of the study, we considered a difference of 20 percentage points in favor of manipulative therapy to be clinically relevant [15]; the reported difference in recovery of 22 percentage points is in line with our expectations. However, we anticipated a recovery rate of 50% in patients who received usual medical care, twice as high as the actual rate observed. In addition, although we expected that 10% of patients would be lost to follow-up, only 3 patients discontinued the allocated manipulative therapy. The amount of missing data due to attrition is limited and appears to be completely random. Because fewer patients in the control group recovered and adherence to allocated treatments and follow-up was high, we reached sufficient statistical power with a smaller-thanplanned study sample.

Our trial was necessarily designed using open-label treatment. Therefore, discontinuation of treatment and attrition may have biased our results [26]. However, patients with missing values were equally distributed between treatment groups, and there were no indications that treatment discontinuation and attrition were related to prognostic status or treatment allocation or outcome. Imputation of missing values according to the last-observation-carriedforward method and the baseline-value-carried-forward method yielded similar results. However, missing values for outcome measures may have made our results less precise.

Manipulative therapy for treatment of shoulder symptoms is rarely studied. Our findings corroborate the findings of the previous study by Winters and colleagues [13, 14], which found that manipulative therapy for the shoulder girdle yielded considerable benefit compared with physiotherapy. We demonstrated that manipulative therapy for the shoulder girdle in addition to usual medical care by a general practitioner accelerated recovery of shoulder symptoms and reduced their severity. These effects were sustained at 52 weeks of follow-up. Compared with the study by Winters and colleagues [13, 14], our study included only patients with shoulder symptoms and dysfunction of the shoulder girdle. We also included more patients, had nearly complete follow-up, and restricted manipulative therapy to avoid bias due to treatment contamination.

We believe that general practitioners should include a short physical examination of the shoulder girdle in their structured medical examinations. For patients with shoulder symptoms in whom dysfunction of the cervicothoracic spine and adjacent ribs is found, referral to a manual therapist should be considered.