In order to compare the outcome from surgical repair and physiotherapy, 103 patients with symptomatic small and medium-sized tears of the rotator cuff were randomly allocated to one of the two approaches. The primary outcome measure was the Constant score, and secondary outcome measures included the self-report section of the American Shoulder and Elbow Surgeons score, the Short Form 36 Health Survey and subscores for shoulder movement, pain, strength and patient satisfaction. Scores were taken at baseline and after six and 12 months by a blinded assessor. Nine patients (18%) with insufficient benefit from physiotherapy after at least 15 treatment sessions underwent secondary surgical treatment.

Analysis of between-group differences showed better results for the surgery group on the Constant scale (difference 13.0 points, p − 0.002), on the American Shoulder and Elbow surgeons scale (difference 16.1 points, p < 0.0005), for pain-free abduction (difference 28.8°, p = 0.003) and for reduction in pain (difference on a visual analogue scale −1.7 cm, p < 0.0005).

Which patients with rotator cuff tears should have surgical treatment and which should have physiotherapy remains unclear. Early surgical repair is essential for younger, active patients with acute tears and severe functional deficit.1 In other cases, however, surgery is not indicated and non-operative treatment may be recommended. This has the advantage of less treatment-related morbidity, but should only be preferred if the short- and long-term results are comparable to those of surgical repair.

Earlier studies regarding either tendon repair25 or physiotherapy68 have shown benefit for both approaches. However, the studies are difficult to compare because of differing study populations, methods of treatment and evaluation, and periods of follow-up. Most studies of non-operative treatment are retrospective,6,911 present pre-selected groups, and exclude from analysis those patients who were operated upon after failed conservative treatment, the effect of which may have been overestimated.

The purpose of this study was to present a controlled, randomised trial comparing operative repair with physiotherapy in the treatment of small and medium-sized full-thickness tears of the rotator cuff.

Patients and Methods

The study was a single-centre, randomised clinical trial with a minimum follow-up of 12 months. Between September 2004 and October 2007, 103 patients with symptomatic small (< 1 cm) or medium-sized (1 cm to 3 cm) tears of the rotator cuff were randomly allocated to operative treatment or physiotherapy. The primary outcome measure was the Constant score.12 Secondary outcome measures included the self-report section of the American Shoulder and elbow surgeons (ASES)13 score, the short form (SF-36) Health survey,14 measuring results for pain, strength and pain-free mobility of the shoulder, and treatment satisfaction. The null hypothesis was that there would be no significant difference in treatment benefit between the surgical and the non-surgical group. The study was approved by our regional health ethics board and written informed consent obtained from all participants before randomisation.

Patients from the Oslo region deemed by their general practitioners to have a rotator cuff lesion and referred to our hospital were screened for inclusion by one orthopaedic surgeon (SM). The necessary criteria were lateral shoulder pain at rest or with exercise, a painful arc, positive impingement signs and a passive range of movement of at least 140° for abduction and flexion. Further inclusion criteria included the demonstration of a full-thickness tear by sonography and MRI, a tear size of not more than 3 cm on short and long axis ultrasound scans and muscle atrophy on MRI not exceeding stage 2, according to the classification of Thomazeau et al.15 Traumatic and atraumatic tears were included. Exclusion criteria included age < 18 years, tears with an absolute indication for surgery such as those involving substantial parts of the subscapularis tendon, the presence of other local or systemic diseases affecting shoulder function, previous tendon surgery on the relevant shoulder, medical comorbidities and an inability to comply with follow-up.

A computer-generated randomisation list (block length 20, ratio 1:1) was drawn up by our statistician. Sequentially numbered, sealed envelopes were used to assign treatment according to the participants’ study number, given at baseline assessment. The randomisation sequence was concealed from the study’s collaborators until treatment was assigned. Only the outcome assessor (TH) remained blinded throughout the study.

The operations were performed in a standard manner by mini-open (nine patients) or open (42 patients) tendon repair. All were performed in the deck-chair position under interscalene block regional anaesthesia and total intravenous anaesthesia without the use of inhalational agents, by one of three experienced orthopaedic surgeons (SM, KTN, NK). Following diagnostic arthroscopy and through a deltoid splitting approach, an anteroinferior acromioplasty was performed as described by Neer.16 With the arm at the side, the rotator cuff was mobilised until the tear was fully exposed. The footprint was prepared to bleeding bone and tendon repair performed with a combination of tendon-to-tendon and tendon-to-bone techniques by passing sutures through bone tunnels in the greater tuberosity. The deltoid was repaired to the acromion through drill holes. Tenodesis of the long head of biceps tendon was performed in 18 patients in whom arthroscopy had shown inflammation or a partial tear. Mini-open tendon repair differed from open repair by a shorter incision and arthroscopic acromioplasty.

Post-operatively the arm was immobilised in a sling and passive range-of-movement exercises commenced. Active-assisted movements were initiated after six weeks, and supplemented by strengthening exercises 12 weeks after surgery.

Patients randomised to the non-surgical group received outpatient treatment at our hospital by one of four physiotherapists experienced in conservative shoulder management (GL, US, BH, IS). A rehabilitation programme describing treatment goals and methods was planned before the study. Non-surgical treatment was given on the basis of this programme in a non-standardised manner according to clinical findings and progress. Treatment sessions of 40 minutes were given on average twice weekly for 12 weeks, and with increasing intervals during the following six to 12 weeks. Particular attention was directed towards correction of upper quarter posture and the restoration of scapulothoracic and glenohumeral muscular control and stability.

Local glenohumeral control was addressed by exercises to centre the humeral head in the glenoid fossa. Isometric exercises and exercises against eccentric and concentric resistance for shoulder rotators were given. When local glenohumeral control was achieved, exercises were given with increasing loads and progressed from neutral to more challenging positions. During all exercises, scapular stability had to be maintained. Additional exercises were given for specific demands in work, sports and leisure activities.1719 Patients who did not improve after at least 15 sessions of physiotherapy were re-examined by an orthopaedic surgeon (SM), and additional testing with outcome scores was performed. If inadequate improvement was confirmed (Table I), secondary surgical treatment was offered. The results of scoring from additional testing were considered as final results from non-surgical treatment and carried forward to the six- and 12-month analyses. After secondary surgical treatment these patients were followed as a separate secondary surgery group.

View this table:
Table I.

Scoring results for the secondary surgery group at baseline, from additional pre-operative measuring, and at six and 12 months’ follow-up

Primary and secondary study scores were performed at baseline and six and 12 months after intervention. At the 12-month follow-up, satisfaction was assessed on a visual analogue scale (VAS) and, for the surgical treatment group, tendon healing was assessed on MRI. This was performed on a 1.5 T scanner (Signa, GE Medical Systems, Waukesha, Wisconsin). All images were interpreted by the same musculoskeletal radiologist (RT). A re-tear of the rotator cuff was diagnosed if a fluid-like signal extending from the articular to the bursal surface of the tendon was shown on T2-weighted images. A partial-thickness tear was diagnosed if the fluid signal did not traverse the full thickness of the tendon.2023

The Constant score12 is a shoulder-specific outcome instrument. Its inter-observer reliability was established in the original paper and a more complete assessment of its reliability in patients with shoulder pathology has been performed by Conboy et al.24 The system uses a combination of subjective and examiner-derived components to assess shoulder function. A maximal score of 100 points is achievable. All examiner-derived measurements were performed with the patient wearing a T-shirt that covered the shoulder.

The self-report section of the ASES score13 consists of parts for pain and shoulder function, each contributing 50 points to a maximum score of 100. The system’s validity, reliability and responsiveness have been demonstrated.25

The SF-36 is a generic measurement of health-related quality of life. It comprises eight component scales and two summary scales for physical and mental health. Each scale ranges from 0 to 100 points, with 100 indicating the best possible health condition. The validity of the parameters has been documented.26 Sensitivity and responsiveness have been demonstrated for measurement of the relative impact of shoulder pathologies and the benefit of treatment.2731

Subscores for shoulder movement were measured with a goniometer and shoulder strength was measured with a hand-held spring balance according to Constant.12

Statistical analysis.

For sample size calculation, a between-groups difference of 12 points on the Constant scale at 12 months was assumed to represent clinical relevance.3234 On condition of a within-group sd of 20 points on the Constant scale, 102 patients would be needed to obtain a power of 0.85 and a significance level of 0.05.

All patients in the study were analysed on the basis of an intention-to-treat principle. This means that for patients in the physiotherapy group who changed treatment, the final score before secondary surgery was carried forward to the six- and 12-month analyses. Patient characteristics at baseline were compared between groups by t-tests, chi-squared tests and Mann-Whitney U tests. Testing of our null hypothesis (no difference in treatment benefit between groups) was performed by a linear mixed model with an unstructured covariance matrix. The time of follow-up and choice of treatment were used as categorical variables, and an interaction term between time and treatment was included.

Between-group differences are given with 95% confidence intervals and p-values. The results of patient satisfaction at 12 months were compared between groups by a Mann-Whitney U test.


A total of 52 patients were allocated to surgical treatment and 51 to physiotherapy (Fig. 1). At baseline, demographic data (Table II) and study scores (Tables III and IV) were comparable between the groups (p = 0.12 to 1.00). Patients allocated to the physiotherapy group completed a mean of 24 (9 to 55) training sessions. Nine patients (18%) in the physiotherapy group showed no improvement after a mean of 24 (15 to 34) sessions and underwent surgery. One patient, allocated to surgical treatment, withdrew from the study before treatment due to a sudden improvement of symptoms and his baseline results were carried on to all further analyses.

View this table:
Table II.

Patient characteristics at time of enrolment

View this table:
Table III.

Scoring results for the two treatment groups at baseline, and at six and 12 months’ follow-up

View this table:
Table IV.

Results from the Short Form 36 Health Survey (SF-36) for the two treatment groups at baseline, and at six and 12 months’ follow-up

Fig. 1

Flow diagram showing the patients through the study.

The mean Constant score improved from baseline to 12 months by 41.4 points (sd 19.6) in the surgery group and 28.4 points (sd 21.9) in the physiotherapy group (Figure 2, Table III). The between-group difference of 13.0 points was statistically significant (95% confidence interval (CI) 4.9 to 21.1, p = 0.002). The mean ASES score improved by 47.1 points (sd 20.0) in the surgery group and 31.0 points (sd 20.2) in the physiotherapy group (Table III) with a statistically significant between-group difference of 16.1 points (95% CI 8.2 to 23.9, p < 0.0005). At six months, differences in improvement of the scores between groups reached significance only on the ASES scale (11.4 points, 95% CI 3.6 to 19.1, p = 0.005).

Fig. 2

Plot showing mean Constant scores (with 95% confidence intervals) at baseline and at six and 12 months for the surgery group (closed triangles) and the physiotherapy group (closed rectangles).

Sub scores (Table III).

Analysis of between-group differences showed a significantly larger increase of abduction in the surgery group by 28.8° (95% CI 10.1 to 47.5, p = 0.003) and a significantly larger reduction in pain by 1.7 on a VAS scale (95% CI −2.6 to −0.9, p < 0.0005). For shoulder flexion, a trend for a larger increase in the surgery group was found (difference 17.3°, 95% CI −2.3 to 36.8, p = 0.08). For shoulder strength the between-group difference in improvement in the score was not significant (difference 0.2 kg, 95% CI −2.0 to 2.3, p = 0.89).

General health score (Table IV).

In both treatment groups, within-group differences from baseline to 12 months were largest for the component scales, with the highest validity for physical health (physical function, physical role, bodily pain) and the physical component summary score, but between-group differences did not reach significance (p-values ranging from 0.10 to 0.84). For the component scales with moderate (general health, vitality) and high validity for mental health (mental health scale) and the mental component summary score, within-group increases were small and between-group differences were not significant (p-values ranging from 0.29 to 0.92). Emotional role, a component score strongly related to mental health, showed a substantial baseline difference between groups. This finding was interpreted as coincidental, and as a consequence was not followed up further.

Patient satisfaction.

Mean values for patient satisfaction after 12 months as measured on a VAS scale (0 = very unsatisfied, 10 = very satisfied) were 9.0 (1.0 to 10.0) for the surgery group and 7.2 (0.0 to 10.0) for the physiotherapy group. Comparison of the results using a Mann-Whitney U test showed significance for the between-groups difference with p < 0.0005.

Secondary surgery group (Table I).

The patients derived insufficient benefit from physiotherapy as reflected by small changes of score from baseline to the pre-operative measurements (mean change in Constant score −0.3 points (sd 13.0), and ASES score 6.8 points (sd 12.9)). After secondary surgery, the mean Constant scores increased by 33.9 points (sd 19.7), which is less than that achieved in the primary surgery group (41.4 points). On the ASES scale, the corresponding mean increase was 40.0 points (sd 25.2) compared to an interval of 47.1 points in the primary surgery group.

Assessment of tendon repair by MRI.

Assessment of the integrity of the tendon 12 months after surgical repair was possible in 50 of 51 patients in the primary surgery group (one patient had artificial heart valves) and in all nine patients of the secondary surgery group. In the primary surgery group MRI diagnosed an intact rotator cuff in 38 cases (76%), a partial-thickness tear in six cases (12%) and a full-thickness tear in four (8%). In two cases (4%) assessment of the tendon was not conclusive because of artefacts. In the secondary surgery group, eight of nine tendons were intact and there was a further full-thickness tear in one.

Unexpected events.

Three months post-operatively one patient in the surgery group sustained a traumatic fracture of the proximal humerus. This was treated in a sling, and routine MRI at 12 months’ follow-up showed an intact tendon repair. One patient from the physiotherapy group was diagnosed with polymyalgia rheumatica four months after inclusion and was treated with steroids. Both patients remained in the study and outcome was assessed as randomised.


In the treatment of small and medium-sized tears of the rotator cuff, although the effect of physiotherapy or tendon repair has been demonstrated in case studies, comparison studies are lacking.35,36 In studies that used the Constant score, the mean improvement ranged from 13.2 to 30.0 points for physiotherapy8,3739 and from 24.0 to 43.1 points after open4042 or arthroscopic4347 tendon repair. Comparison between such studies and our findings is difficult because of differences in study design and patient selection. In the treatment of these patients the superiority of physiotherapy or surgical repair should be established by randomised studies comparing the two approaches.

In our study, there was an improvement in the mean Constant score in both groups, but the change of score was 13.0 points larger in the surgery group (Fig. 2). The validity of this result was supported by the finding of a significantly better treatment effect in the surgery group on the ASES scale and by the subscores for shoulder abduction, pain and patient satisfaction. In agreement with other studies, we believe that a between-groups difference of 13.0 points on the Constant scale is not only statistically significant but also clinically relevant.3234

In our study surgical treatment and physiotherapy were given on the basis of defined treatment protocols. However, as rotator cuff tears are not uniform, therapeutic approaches have to be adapted to the individual characteristics of the tear. Depending on the location of a tear, its size and the condition of the long head of the biceps tendon, surgery was performed by a mini-open instead of an open approach whenever possible, and a tenodesis of the long head of biceps carried out when the tendon was found to be pathological. This reflects accepted surgical standards. In the same way, physiotherapy had to be tailored individually. Before the study, a pool of therapeutic measures was defined, but the selection of exercises, the number of repetitions and the duration of treatment was individually adjusted according to the clinical findings and the responses to treatment as has been described for the treatment of these patients.1719

In order to avoid interpretation bias and loss of power it was decided to perform all outcome analyses ‘as randomised’ by following an intention-to-treat principle. As a consequence, results from patients who withdrew (one from the surgery group) and those who changed treatment (nine from the physiotherapy group) were assessed as originally randomised. For the patient who withdrew from the surgery group, the baseline data carried forward gave him a result far below the mean of his group. The nine patients who changed treatment after failed physiotherapy had all performed at least 15 treatment sessions, and it seemed adequate to interpret their final score pre-operatively as the best estimate for the final result from physiotherapy. Following these patients without further treatment would have been unethical, and elimination of their results from analysis would have led to an overestimation of the effect of physiotherapy.

As shown in Table III, there are small, non-significant baseline differences between the groups (a mean difference in the Constant score: 3.1 points, p = 0.26, differences in the mean ASES score: 2.7 points, p = 0.36). Following the as randomised principle, no baseline correction was performed at analysis. The inclusion of a correction for baseline differences would have reduced the between-groups differences by their baseline differences, but p-values for between-groups differences would still have been significant on the Constant (p = 0.009) and on the ASES scores (p = 0.001).

Tendon repair should be performed prior to the development of tendon retraction and muscle atrophy. A period of failed physiotherapy will delay surgical treatment and may influence the result of secondary surgery. The mean operative delay in our secondary surgery group compared to the primary group was five months, and treatment benefit, as expressed by the mean Constant score, was 7.5 points below the result from the primary surgery group. Interpretation of this difference in outcomes, however, is difficult because of the small sample size in the secondary surgery group.

The intention of this study was to compare treatment effects from surgical treatment and physiotherapy. By carrying forward the last outcome result prior to secondary surgery as the final result for patients with failed physiotherapy, our analysis considered physiotherapy as an isolated treatment regimen. In clinical practice, however, physiotherapy will be offered together with secondary surgery as a rescue procedure for patients who fail conservative treatment. The final result in the nine patients with failed physiotherapy is the 12-month result after secondary surgery. Replacing the results that were carried forward by those of secondary surgery in these nine patients, and excluding the result of the patient who withdrew from the surgical group before treatment, would have reduced the difference between groups on the Constant scale from 13.0 to 7.7 points (95% CI 0.39 to 15.1, p = 0.04). This size of difference may be considered as of small relevance clinically. Together with the large number of patients (42 of 51, 82%) who accepted the result achieved by physiotherapy as their final result, this may support physiotherapy as an initial option for small and medium-sized rotator cuff tears. However, by this way of analysis, surgery is no longer compared with physiotherapy but with the results of physiotherapy combined with secondary surgery for those with insufficient response to physiotherapy.

A limitation of this study is the follow-up period of only one year. In particular, the consistency of our results after non-operative treatment may be questioned.6,7 Tears that are not repaired may enlarge, become more symptomatic and be more difficult to treat. Further follow-up is necessary and is underway.

In this study larger treatment benefits in terms of outcome scores after one year were found for surgical treatment of rotator cuff tears compared to physiotherapy. Differences with statistical and clinical significance were demonstrated on different scoring instruments. At the same time, 42 of 51 non-surgically treated patients accepted the result from physiotherapy as their final treatment result, and nine patients with insufficient effect from physiotherapy were treated effectively by secondary surgery. In the short term, both approaches can be considered in the treatment of patients with small and medium-sized rotator cuff tears but better results can be expected after primary surgical repair.


  • The authors would like to thank Mr B. Haldorson, Department of Physiotherapy for his role in conservative treatment and Drs N. Kise and K. Nerhus for their help in the surgical management.

  • Although none of the authors has received or will receive benefits for personal or professional use from a commercial party related directly or indirectly to the subject of this article, benefits have been or will be received but will be directed solely to a research fund, foundation, educational institution, or other nonprofit organisation with which one or more of the authors are associated.

  • Received March 17, 2009.
  • Accepted August 14, 2009.


View Abstract