Over the long term, the results of the insertion of a Grammont inverted shoulder prothesis are unknown. The present study reports survivorship curves and the role of the initial aetiology in patients re-examined after 5 to 10 years.

Patients and methods Eighty prostheses (for 77 patients) were implanted between 1992 and 1998: 66 cases of massive cuff tear arthropathy (MCTA) and 24 cases for another aetiology (mainly rheumatoid arthritis and revision).

At the time of follow-up, 18 patients had died and 2 could not be traced. The remaining patients (57 representing 60 prostheses) were seen by an independent examiner. The minimum follow-up was 5 years. The average follow-up was 69,6 months.

Kaplan-Meir survivorship curves for the 60 prostheses were established in order to show the probability of failure as defined by: revision of the prothesis, glenoid loosening, and a functional level< 30 points according to the Constant score.

Results The survivorship curves were as follow:

- for non revision of the prosthesis at 10 years: 91% overall; after 9 years: 95% for MCTA, and 77 % for the others aetiologies. This difference was statistically significant (p< 0,01) ; 6 implants were revised: 3 for MCTA and 3 for other aetiologies.

Discussion

According to revision of the prosthesis, there is a clear rupture in the survivorship curve about 3 years after insertion in aetiologies other than MCTA. This suggests that Grammont inverted total shoulder arthroplasty is not appropriate in these aetiologies (particularly in cases of rheumatoid arthritis).