Radiological changes and differences between cemented and uncemented components of Grammont reverse shoulder arthroplasties (DePuy) were analysed at a mean follow-up of 9.6 years (8 to 12). Of 122 reverse shoulder arthroplasties implanted in five shoulder centres between 1993 and 2000, a total of 68 (65 patients) were available for study. The indications for reversed shoulder arthroplasty were cuff tear arthropathy in 48 shoulders, revision of shoulder prostheses of various types in 11 and massive cuff tear in nine. The development of scapular notching, bony scapular spur formation, heterotopic ossification, glenoid and humeral radiolucencies, stem subsidence, radiological signs of stress shielding and resorption of the tuberosities were assessed on standardised true anteroposterior and axillary radiographs.

A scapular notch was observed in 60 shoulders (88%) and was associated with the superolateral approach (p = 0.009). Glenoid radiolucency was present in 11 (16%), bony scapular spur and/or ossifications in 51 (75%), and subsidence of the stem and humeral radiolucency in more than three zones were present in three (8.8%) and in four (11.8%) of 34 cemented components, respectively, and in one (2.9%) and two (5.9%) of 34 uncemented components, respectively. Radiological signs of stress shielding were significantly more frequent with uncemented components (p < 0.001), as was resorption of the greater (p < 0.001) and lesser tuberosities (p = 0.009).

Mechanical loosening which begins with early-onset migration of the prosthesis is the major reason for failure of the Souter-Strathclyde elbow replacement. In a prospective study of 18 Souter-Strathclyde replacements we evaluated the patterns of migration using roentgen stereophotogrammetric analysis. We had previously reported the short-term results after a follow-up of two years which we have now extended to a mean follow-up of 8.2 years (1 to 11.3). Migration was assessed along the co-ordinal axes and overall micromovement was expressed as the maximum total point movement. The alignment of the prosthesis and the presence of radiolucent lines were examined on conventional standardised radiographs.

All the humeral components showed increased and variable patterns of migration at the extended follow-up and four humeral components were revised. The maximum total point movement at two years in the revised prostheses was 1.8 mm (sd 1.0) and in the non-revised 0.7 mm (sd 0.5, p = 0.01). Most humeral components migrated into external rotation resulting in an anterior and varus tilt. The ulnar components remained stable.

In a prospective study between 2000 and 2005, 22 patients with primary osteoarthritis of the shoulder had a total shoulder arthroplasty with a standard five-pegged glenoid component, 12 with non-offset humeral head and ten with offset humeral head components. Over a period of 24 months the relative movement of the glenoid component with respect to the scapula was measured using radiostereometric analysis.

Nine glenoids needed reaming for erosion. There was a significant increase in rotation about all three axes with time (p < 0.001), the largest occurring about the longitudinal axis (anteversion-retroversion), with mean values of 3.8° and 1.9° for the non-offset and offset humeral head eroded subgroups, respectively. There was also a significant difference in rotation about the anteversion-retroversion axis (p = 0.01) and the varus-valgus (p < 0.001) z-axis between the two groups. The offset humeral head group reached a plateau at early follow-up with rotation about the z-axis, whereas the mean of the non-offset humeral head group at 24 months was three times greater than that of the offset group accounting for the highly significant difference between them.

We studied the stability of cemented all-polyethylene keeled glenoid components by radiostereometric analysis (RSA) in 16 shoulders which had received a total shoulder replacement. There were 14 women (one bilateral) and one man with a mean age of 64 years. The diagnosis was osteoarthritis in eight and rheumatoid arthritis in seven. Two of the shoulders were excluded from the RSA study because of loosening of the tantalum markers. Three tantalum markers were inserted in the glenoid socket, two in the coracoid process and two in the acromion. The polyethylene keeled glenoid component was marked with three to five tantalum markers. Conventional radiological and RSA examinations were carried out at five to seven days, at four months and at one and two years after operation. Radiolucent lines were found in all except three shoulders. Migration was most pronounced in the distal direction and exceeded 1 mm in four shoulders. In ten shoulders rotation exceeded 2° in one or more axes with retroversion/anteversion being most common. No correlation was found between migration and the presence of radiolucencies on conventional radiographs

The incidence of loosening of a cemented glenoid componentin total shoulder arthroplasty, detected by means of radiolucent lines or positional shift of the component on true anteroposterior radiographs, has been reported to be between 0% and 44%. Radiolucent lines are, however, difficult to detect and to interpret because of the mobility of the shoulder girdle and the obliquity of the glenoid which hinder standardisation of radiographs. We examined radiolucencies around cemented glenoid components in 48 patients, with a mean follow-up of 5.3 years, and found progressive changes to be present predominantly at the inferior pole of the component. This may hold a clue for the mechanism of loosening of this implant. In five patients we performed an additional analysis of loosening of the glenoid component using digital roentgen stereophotogrammetric analysis (RSA). After three years, three of the five implants had loosened (migration 1.2 to 5.5 mm). In only one, with gross loosening, were the radiological signs consistent with the RSA findings. When traditional radiographs are used for assessment, the rate of early loosening is underestimated. We recommend that RSA be used for this