Sports Medicine

Rotator cuff repair in patients over 70 years of age

Early outcomes and risk factors associated with re-tear

  1. P. M. Robinson, MBChB(Hons), BMedSci, MRCS(Eng), Specialty Registrar Trauma and Orthopaedics1 Author Profile;
  2. J. Wilson, MRCSEd, MSc, Specialty Registrar Trauma and Orthopaedics2 Author Profile;
  3. S. Dalal, FRCSEd(Tr & Orth), Consultant Orthopaedic Surgeon3 Author Profile;
  4. R. A. Parker, BSc, MSc, Medical Statistician4;
  5. P. Norburn, MBChB, MRCP, FRCR, Consultant Radiologist5; and
  6. B. R. Roy, MSc (Bioeng), FRCS(Tr & Orth), DMI, Consultant Orthopaedic Surgeon5

+Author Affiliations

  1. 1Addenbrooke’s Hospital, Cambridge University Hospitals NHS Foundation Trust, Hills Road, Cambridge CB2 0QQ, UK.

  2. 2North West Deanery, 3 Piccadilly Place, Manchester M1 3BN, UK.

  3. 3Sunderland Royal Hospital, Kayll Road, Sunderland, Tyne & Wear SR4 7TP, UK.

  4. 4Centre for Applied Medical Statistics, University of Cambridge, Department of Public Health & Primary Care, Institute of Public Health, Forvie Site, Robinson Way, Cambridge CB2 0SR, UK.

  5. 5Trafford General Hospital, Moorside Road, Davyhulme, Manchester M41 5SL, UK.
  1. Correspondence should be sent to Mr P. M. Robinson;


This study reports the clinical and sonographic outcome of arthroscopic rotator cuff repair in patients aged ≥ 70 years and aimed to determine factors associated with re-tear. A total of 69 consecutive repairs were performed in 68 patients with a mean age of 77 years (70 to 86). Constant-Murley scores were collected pre-operatively and at one year post-operatively. The integrity of the repair was assessed using ultrasound. Re-tear was detected in 20 of 62 patients (32%) assessed with ultrasound. Age at operation was significantly associated with re-tear free survival (p = 0.016). The mean pre-operative Constant score was 23 (SD 14), which increased to 58 (SD 20) at one year post-operatively (paired t-test, p < 0.001). Male gender was significantly associated with a higher score at one year (p = 0.019).

We conclude that arthroscopic rotator cuff repair in patients aged ≥ 70 years is a successful procedure. The gender and age of the patient are important factors to consider when planning management.

Cite this article: Bone Joint J 2013;95-B:199–205.

The incidence of rotator cuff tears (RCTs) increases with age. Cadaveric studies have demonstrated full thickness tears in 30% of subjects aged > 60 years1and MRI studies have shown full-thickness tears in 28% of those aged > 60 years.2In patients aged > 70 years, the prevalence of asymptomatic full thickness RCTs is about 38% as diagnosed by ultrasound scan.3

The United Nations uses the age of 60 years to describe ‘older’ people. In developed countries elderly may be considered to be older than this, with the retirement age moving towards 68 years. In this study we have taken elderly to mean age ≥ 70 years. However, we acknowledge that chronological ageing is not a precise marker of biological ageing.

Arthroscopic rotator cuff repair (ARCR) is increasingly used because it is minimally invasive and causes less soft-tissue damage, potentially allowing a faster recovery than open repair. Rotator cuff repair in elderly patients is, however, controversial for several reasons. They are more likely to have osteoporosis of the proximal humerus and poorer tendon quality4 and those aged > 65 years have a higher proportion of technically challenging, massive (> 5 cm) RCTs,5 which are associated with greater tendon degeneration, reduced cellular activity and poorer potential for healing.68 Increasing age is also associated with a high rate of re-rupture following repair.9,10 This prospective study aimed to report the clinical and sonographic outcome of patients aged ≥ 70 years who underwent arthroscopic repair of a full-thickness RCT and to determine the factors associated with clinical outcome and re-tear.

Patients and Methods

Between September 2006 and December 2009, a total of 69 ARCRs were performed in 68 consecutive patients with a symptomatic full-thickness RCT. Their mean age was 77 years (70 to 86). There were 33 males and 35 females (one female with bilateral tears repaired at different times).

The median duration of symptoms was 14 months (1 to 100). Of the 67 RCTs for which these data were collected, 29 (43%) were associated with a history of injury involving the shoulder. The American Society of Anesthesiologists (ASA) grade11 was recorded for 61 patients: it was grade 1 in seven (11%), 2 in 48 (79%), 3 in five (8%) and 4 in one (2%).

Consecutive patients aged ≥ 70 years who underwent ARCR were included in the study. Before consideration for surgery, all patients received shoulder physiotherapy in primary or secondary care settings. Patients who had failed non-operative treatment, with a symptomatic full-thickness rotator cuff tear were offered ARCR. Exclusion criteria were partial thickness RCT, cuff tear arthropathy and comorbidities precluding a general anaesthetic. All procedures were performed by the senior author (BRR).

Patients were assessed pre-operatively and at one year post-operatively using the Constant-Murley score.12 This scoring was performed by members of the senior author’s team, trained in assessment of the shoulder, in the outpatient clinic. Strength was measured using a digital dynamometer (Myometer 500N; Atlantech Medical Devices Ltd., Nottingham, United Kingdom). Measurements were performed with the shoulder abducted to 90° in the scapular plane (30° in front of the coronal plane). The loop was placed over the forearm, with the elbow in extension and the forearm pronated with the palm facing the floor. This was repeated three times and the average was recorded. The patient scored zero if he or she was unable to abduct to 90° or if the procedure was painful. Details of symptoms, duration and history of shoulder injury were collected. A total of 39 RCTs (57%) were diagnosed pre-operatively on ultrasound (USS) and nine (13%) were diagnosed on pre-operative MRI. In three shoulders (4%) no cuff tear was seen on USS. The remaining 18 shoulders (26%) were not assessed pre-operatively using USS or MRI.

The operations were performed under general anaesthesia and interscalene block with intravenous cefuroxime antibiotic prophylaxis. Hypotensive anaesthesia was used when safe, aiming for a systolic blood pressure of 100 mmHg. Patients were placed in the beach chair position with the arm held in a Spider limb positioner (Smith & Nephew UK Ltd, London, United Kingdom). A fluid management system (FMS duo+; DePuy Mitek EMEA, Ascot, United Kingdom) was used, with a standard inflation pressure of 40 mmHg.

Subacromial decompression (SAD) was performed in 61 repairs (88%), using a Helicut burr (Smith & Nephew UK Ltd). Tears were classified according to the tendons involved and by size using the system of Cofield13 and described as either crescent or ‘U’ shaped.14 The latter often required a side-to-side suture for ‘margin convergence’.

Mobilisation was performed for all but the small tears as described by Burkhart.15,16 The anterior interval slide released tissue at the rotator interval from the supraspinatus. The coracohumeral ligament was released to improve excursion of the tendon for a tension-free repair. A posterior interval slide was added for massive contracted crescent shaped tears. Footprint preparation comprised soft-tissue debridement. Lateral fixation was performed with Fastin RC anchors double loaded with Orthocord (both DePuy Mitek EMEA). Single-row repair was performed in all cases. Anchors were placed at the margin of the articular cartilage. There were 35 single-anchor repairs and 34 double-anchor repairs. Simple sutures were used when possible, a modified Mason-Allen suture configuration was used for poorer quality tissue.17,18 The Duncan loop19 was used as the standard sliding knot. Simple knots were used if a sliding knot was not possible. In 41 cases (59%), the repair was judged to be ‘watertight’. The repair was assessed as ‘not watertight’ in 22 cases (32%) and this assessment was not recorded in six (9%). Wounds were dressed with Opsite dressings (Smith & Nephew UK Ltd) and the limb was then placed in a Polysling (Mölnlycke Health Care Ltd., Dunstable, United Kingdom). A range of tear sizes and tendons were operated on (Tables I and II). Long head of biceps (LHB) pathology was noted in 25 shoulders (36%). LHB procedures consisted of debridement in six cases, tenodesis in 11 and tenotomy in six.

Rehabilitation was supervised by a specialist physiotherapist and began ten to 14 days post-operatively. During this phase, passive movements and closed kinetic chain exercises were performed with the aim of restoring neuromuscular control and maintaining the ROM whilst protecting the repair. Polyslings were removed at four weeks and active assisted exercises started. The mid phase from six weeks involved progression to active exercises. The late phase from eight weeks onwards included full active movements throughout all ranges with rotator cuff strengthening and proprioceptive exercises. Specific functional activities were also addressed in the late stage. Progressions through each phase not only relied on the timeframe post-operatively but also pain and movement pattern.

Post-operatively the integrity of the repair was evaluated by USS, which was performed by one of two musculo-skeletal radiologists using a 12 MHz transducer (Toshiba Aplio XG; Toshiba medical systems Ltd., Crawley, United Kingdom). With the patient sitting on a stool the long head of biceps tendon (LHBT), subscapularis, supraspinatus and infraspinatus tendons were assessed in their longitudinal and transverse axes. A re-tear was diagnosed if there was a focal full thickness defect in the rotator cuff tendon, or if it could not be visualised as a result of retraction beneath the acromion.

Statistical analysis

A paired t-test was used to investigate the changes in the Constant-Murley scores at one year post-operatively compared with the pre-operative scores. A regression test for trend was then performed on the change in scores, with RCT as the explanatory variable, to determine if the change in score was related to the size of the RCT. A chi-squared test for trend was used to determine if the probability of re-tear varied with the size of RCT.

A Kaplan-Meier plot was produced to show how the cumulative re-tear free survival probability changes with time after surgery. Further plots were stratified by history of injury, age, gender, size of tear and duration of symptoms.

Cox’s proportional hazards regression was used to investigate predictors of survival time until re-tear. The variables of interest were history of injury, age, gender, pre-operative shoulder score, size of tear, duration of symptoms, SAD procedure, and watertight repair. Univariate Cox regression models were fitted with each of these variables, entering the models as single explanatory variables. A log10 transformation was applied to the duration of symptoms in order to reduce the influence of outliers in this variable. Time-dependent covariates were used to check the proportional hazards assumption.

For the Constant-Murley score at one-year linear regression modelling was used after adjusting for pre-operative score. Linear regression models were fitted with score at one-year as the outcome variable and with each of the variables of interest entering the models as single explanatory variables with pre-operative score.

A forwards model selection method (F-tests) was used in the context of multiple linear regression. Beginning with a linear regression model with no explanatory variables included, variables were added to the model in turn and only remained if they gave a significant improvement to the model goodness-of-fit. Sensitivity analysis involved using a stepwise model selection method to check the results.

The SAD variable could not be included in the multiple linear regression method, because after exclusion of patients with missing values, only one of the remaining patients did not undergo SAD.

R software v2.13.0 (R Foundation for Statistical Computing, Vienna, Austria) was used to construct the Kaplan-Meier plots. SPSS version 18 was used for all other analyses (SPSS Inc., Chicago, Illinois). The significance level was set at 5% throughout.


Josh Sandberg

Josh Sandberg is the President of Ortho Spine Partners and Partner for The De Angelis Group. He also serves as Co-Founder and Editor of OrthoSpineNews.

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button