Revision of reversed total shoulder arthroplasty. Indications and outcome

Data were statistically tested for normal distribution using the D’Agostino and Pearson omnibus normality test before applying either two tailed paired Students-t-test or Wilcoxon test for normally and not-normally distributed data for intra-sample comparison. Values are reported as mean and standard deviations or 95% confidence interval where appropriate. All statistical tests were performed using a commercial statistical software package, with significance levels set at p < 0.05.

Baseline characteristics are summarized in Table 1. Sixty-seven patients with a mean age of 69 ± 9 years (male to female ratio of 0.5 to 1) needed at least one additional intervention to treat a complication of RTSA, thirty of them needed a second, eleven a third and four a forth intervention. Additional interventions after RTSA were needed at between the same day of the RTSA and eight years thereafter, with a mean of 17 months postoperatively. Overall, the most common reason for the first intervention was instability resulting in subluxation or dislocation (n = 12, (18%)) occurring at a mean of 17 months after initial RTSA (5 days to 72 months). However this complication was seen in 11 of 44 shoulders with Delta-III prostheses and in 1 of 23 patients with Anatomical Reverse prostheses (significant difference with confidence of >90% by a two tailed z-test of proportions). It was definitely treated by closed reduction in those five cases if no further component re-dislocation was radiologically evident and no recurrent subluxation was reported. A change of components of the prosthesis was necessary in seven cases: Selection of a higher humeral polyethylene inlay was used in three, revision of a glenoid component in two and revision of a humeral component in two cases. The second most common reasons for revision were hematoma and superficial wound problems (e.g. dehiscence) (n = 10, (15%)) occurring after a mean of 38 days. Hematoma was treated by simple evacuation in seven cases. Other frequent reasons for revision were complications of the glenoid component (n = 8, (12%)) and acromial (n = 5) or coracoid fractures (n = 2, (10%)). Complications of the glenoid component were treated in four cases by change of the glenoid (1 Delta and 3 Anatomical) component and in three cases by conversion (1 Delta and 2 Anatomical) to a hemi-prosthesis. In one case infection was suspected and a spacer was implanted (Table 2). Infections, deficiency in external rotation, acromio-clavicular pathologies and humeral component loosening were followed by other infrequent reasons such as wear of components, metallosis and other rare events (Table 3). The two most common reasons for a second intervention were instability (n = 7, 23% of all reasons for the second intervention) and infections (n = 6, 20%). The most common reason for the third (n = 5, 45%) and forth (n = 2, 50%) intervention was again instability.

Thirty-seven patients could be reviewed at no less than two years with a mean follow-up of 47 (range 24 to 150) months after the first revision. Twenty-one of these 37 patients needed a second, nine a third and four a forth additional intervention to treat complications associated with RTSA. The most common cause for an re-intervention in these 37 patients was similar to the total cohort of 67 patients. Instability (n = 9, (24%)) was the most common reason for an intervention followed by hematoma or superficial wound complications (n = 6, (16%)) and complications of the glenoid component (n = 3, (8%)). However, an external rotation deficit perceived as disabling and correctable was also a relevant reason for revision in this subgroup (n = 5, (14%)). The most common indication of an additional intervention consistently was instability in the second (n = 5, (24% of all revision causes)), third (n = 4, (44%)) and forth additional intervention (n = 2, 50%)). The total Constant-Murley score of the 31 personally examined patients increased by 24 points (95% CI: 17, 31) (p < 0.0001) from the time-point immediately before implantation of the RTSA to the last follow-up at an mean of 49 months (95% CI: 1382, 1777 days) after RTSA. There was a significant increase in SSV, relative Constant-Murley score, abduction strength and a substantial decrease in pain (Table 4). There was an increase in ability to abduct the arm, but a loss of active internal rotation (Table 4).

The indication for the first closed or open revision was instability with twelve of 67 cases. These twelve patients did not differ from the entire cohort of 67 patients with regard to age (70 ± 7 years vs 69 ± 9 years) or gender (male to female: 0.5 to 1 in both groups) but to the type of prosthesis used for primary RTSA (Delta to Anatomical: 11:1 compared to 44:23). Early occurrence of instability (mean 101 days) was treated by closed reduction under general anesthesia (n = 5), later occurrence of instability (mean 781 days) was treated operatively (n = 7). From those five patients who were treated with closed reduction initially, in two patients a second intervention was performed subsequently; one was treated with change of the inlay (two days after reposition) and the other patient needed a closed reduction one day after initial reposition and a change of the humeral component was performed 3 weeks after.

In the other seven patients in whom instability occurred later, lengthening the prosthesis using a thicker inlay was performed in three cases, a change of the glenoid component in two and exchange of the humeral component in other two cases. One patient in whom the glenoid component was changed needed a second revision with change of the humeral component 25 days after. This was also the case in one of the patients with initial change of inlay who also needed a third intervention (conversion to a Hemi-prosthesis, 27 days later) followed by removal of the prosthesis (172 days after conversion) because of a chronic subluxation with compression of the brachial plexus and denial of the patient to undergo arthrodesis. Another patient also with initial change of inlay showed recurrent dislocation treated by closed reduction 91 days after followed by change of the prosthesis 171 days after.

One of the two patients who were revised because of a dislocation that could not be reduced closed, was treated with an exchange of the stem including correction of version. She developed an incomplete plexus palsy two days postoperatively which was attributed to tension, was treated conservatively with a splint maintaining the arm in flexion and recovered only partially.

In the cohort of 67 patients with 112 revisions, the total amount of indication for a revision was infection in ten cases, from which three were first revisions, six were second revisions and one third revision. Patients with infections were treated mostly with implantation of a spacer (n = 6) or with either debridement (n = 2) or total explantations of the prosthesis (n = 2). One of the patients was revised two times for infection. The most common pathogen was Propionibacterium acnes (n = 4). Broad antibiotic therapy was initiated intraoperatively and was replaced by antibiotics specific to the pathogen once results of bacteriological analyses were available and continued during a mean of 4 months (range 45 to 360 days). One of the patients with total explantation of the prosthesis died of an unrelated reason before 24 months follow-up after explantation. The outcome was not different in patients with revision for infection if compared with the whole cohort of 31 fully followed patients. The total Constant-Murley score of patients revised for infection increased also by 23 points (95% CI: 10.2, 37.6) from the time before RTSA to final follow-up as did the relative Constant-Murley score with a gain of 32% (95% CI: 24.7, 50.2) from a baseline that was very similar to the whole cohort (Table 4). In six cases there were no sings of a recurrent infection, in three cases the patients are still with an antibiotic spacer and do not wish to undergo re-implantation of a RTSA and one patient died of an unrelated cause with a resection arthroplasty for treatment of infection.

With the data available, we were unable to identify specific risk factors for infection in our material, specifically superficial hematoma and wound dehiscence were surprisingly not associated with a statistically higher infection rate.