Background: After Jules-Emile Pean and Neer, unconstrained prosthetic arthroplasty of the shoulder is widely used for glenohumeral osteoarthritis (OA), rheumatoid arthritis (RA) and trauma. While the debate continues over whether humeral head replacement (HHR) or total shoulder arthroplasty (TSR) is better for OA and RA, hemiarthroplasty is preferred in the trauma situation.

Aims: A retrospective review (1993–2000) of 54 patients with the DePuy Global second generation modular shoulder hemiarthroplasty is presented. We highlight the various complications encountered in the longer term and attempt to use our experience to rationalize treatment choice.

Methods: 49 shoulders in 41 patients were available for review. Case records and radiographs were reviewed and clinical assessment carried out. Functional assessment was done using the Constant (CM) and the American Shoulder and Elbow Society systems (ASES). Patients were also asked to complete a shoulder self assessment questionnaire (Insalata, Hospital for Special Surgery).

Results: The pre-op diagnosis was 20 RA, 12 OA, 11 trauma. The mean age was 63.4Yrs. The mean follow up was 6.8Yrs (4–11 Yrs). The mean Constant scores for the 3 groups were 48.1 RA, 46.3 OA, and 56.6 trauma. The mean ASES functional scores were 39.6 RA, 37.5 OA and 32.9 trauma. Active elevation in the RA group was a mean of 83.5°, and 69.5° for OA and 79° for the trauma group. 65% of the RA group was completely pain free compared to 50% and 54% in the OA and trauma groups respectively. Patient satisfaction was high in the RA and OA groups but poor in the trauma group. 90% of RA patients had evidence of superior migration of the prosthesis (ASM), with the majority of them developing cuff dysfunction. 36% of trauma patients had tuberosity escape/nonunion. 100% of OA patients developed late glenoid erosion. Presence of secondary glenoid arthrosis and ASM co-related with poor function (r =0.5, p< 0.05).

Discussion. Second generation prosthesis can prejudice the shoulder biomechanics. Over-tensioning the joint can lead to both glenoid wear and cuff rupture.

In OA patients, satisfaction was highest, but function was poor. Patients with out cuff tear pre-operatively did not develop it later. They may thus benefit from a TSR compared to HHR. RA patients had the least pain, best function and high satisfaction. Most had evidence of either primary or secondary cuff dysfunction. In them patients HHR may be the preferred option. Trauma patients did poorly in all respects. More anatomical reconstruction and greater emphasis on salvage of the humeral head is required in this group.

Low Contact Stress(LCS) total knee arthroplasty was developed to reduce contact stress on the bearing surface and to minimise stresses at the interface between the host bone and the implant surface leading to long term implant survival. The rotating platform was introduced as the bearing interface when both cruciates are sacrificed. It has a central cone which engages into a matching cone in the tibial tray. This allows unconstrained axial rotation of the bearing surface. However, this potentially leaves the platform susceptible to subluxation/dislocation. The reported rate of this complication varies from 0.5% to 4.65%.

In this study from a single center we report the incidence and highlight the associated causative factors found in our series. There were 1053 Low Contact Stress total knee arthroplasties performed between 1994 and 2003. We reviewed 10 knees in 9 patients who had dislocation of the polyethylene rotating platform. This amounts to 0.95% in our series. All the patients with dislocation were women. Average age was 72 years(range 62–84). Osteoarthritis was the primary diagnosis in 8 patients. One patient was suffering from rheumatoid arthritis was on long term steroid therapy and had bilateral dislocations. One patient with Osteoarthritis with Parkinson’s disease went onto have 2 recurrent dislocations. Pre-operative deformity was varus in 9 knees(range 4–10 degrees) and valgus of 15 degrees was noted in one. Time from index operation to dislocation ranged from as early as 10 days to 10 months. There was history of trauma and acute presentation only in one patient. In one patient the knee dislocated while she twisted her knee in bed and in another while climbing up the stairs. In the remaining seven patients the presentation was subacute with symptoms such as pain, decreased range of motion, swelling and a clunking sensation while walking.

Although manipulation under anaesthesia was successful in 3 patients, all of them had recurrent dislocations and two patients had revision to a deepdish platform. Failed closed reduction led to open reduction in two patients with replacement of the rotating platform to deepdish (12.5 mm) type in one. Following both procedures knee was immobilised in a cast for 6 weeks. Five patients were directly revised without attempting closed reduction to a deepdish rotating platform. At revision in all cases the platform was found to be rotated medialy and posteriorly. Soft tissue imbalance and laxity were seen in all but one. At an average followup of 48.5 months (range 11–84 months) no patient had recurrent instability.

Increasing age, questionable soft tissue integrity and varus deformity were significantly associated with rotating platform dislocation. Closed reduction may be possible but invariably leads to recurrence of dislocation and open reduction with revision of the rotating platform gives reliable results. Replacing the rotating platform with a thicker deepdish bearing provides satisfactory stability at revision surgery. Meticulous surgical technique with accurate soft tissue balancing are important in reducing the incidence of dislocations.