Reverse total shoulder arthroplasty (RTSA) has demonstrated promising medium-term outcomes1-11. Those results as well as the expansion of indications have been reported to explain the growth in the use of RTSA worldwide12-17, but the long-term results of RTSA have rarely been reported. Although the long-term studies that have been published have highlighted high prosthetic survivorship2,6,18,19, the loss of shoulder function and increase in radiographic findings of complications over time are concerning. In 2007, we reported the medium-term outcomes of 186 patients treated with a total of 191 RTSAs for etiologies associated with rotator cuff deficiency8. The average Constant score and anterior active elevation were significantly increased at the time of that medium-term follow-up. The aim of the current study was to evaluate the clinical and radiographic outcomes of the same cohort after a minimum of 10 years and to document rates of prosthetic survival and complications.
Materials and Methods
The original report included a consecutive series of 186 patients (191 procedures) followed for a mean of 40 months (range, 24 to 118 months) after RTSAs performed by one of two shoulder surgeons (G.W. or L.N.-J.)8. The mean age at the time of surgery was 72.7 years (range, 23 to 86 years). The patients were grouped according to the surgical indications, all of which involved a deficient rotator cuff: (1) rotator cuff tear arthropathy, (2) a failed previous arthroplasty (i.e., the RTSA was a revision arthroplasty), (3) a massive rotator cuff tear without arthritis, (4) posttraumatic glenohumeral arthritis with rotator cuff compromise, or (5) primary osteoarthritis with rotator cuff compromise and with or without severe glenoid bone erosion.
Grammont-design prostheses were used, including 164 Delta-III (DePuy France) and 27 Aequalis (Tornier) systems. All but 3 were implanted through a deltopectoral approach. Six shoulders required a custom glenoid implant, allowing acromial fixation with an upper bar and screws because of glenoid bone loss. All but 1 of the humeral stems were cemented.
Clinical and Radiographic Assessments
This study was approved by our institutional ethics committee, and all patients provided written informed consent to allow their data to be used in the study. Patients were evaluated by an independent examiner (G.B.) at a minimum of 10 years after surgery. The surgical techniques and rehabilitation protocol have been described8.
All 186 patients from the medium-term outcome study were asked to attend a consultation at our institution. Long-term assessments were performed for patients who had not undergone a prosthetic explantation. Patients who could not return for an on-site consultation because of poor health completed a self-administered questionnaire with the assistance of his/her general practitioner, and the responses were finalized via a telephone interview. Radiographs were made at a center close to the patient’s home and sent to our institution. When a patient had died during the follow-up period, his/her general practitioner was asked for the date of death along with data on any prosthetic revisions or removals carried out in other centers.
Clinical assessment was based on the absolute and relative Constant scores20,21 and the active range of motion of the shoulder. The range of motion (elevation and external and internal rotation) was measured using a goniometer. Strength measurements were carried out using a handheld dynamometer. Radiographic studies included anteroposterior views of the glenohumeral joint in neutral rotation and axillary views obtained under fluoroscopic control for all patients reassessed at our institution. Scapular notching was evaluated according to the Sirveaux classification2. Humeral radiolucent lines were classified according to the Sperling system22 modified by Lévigne et al.23. Loosening of implants was defined as described by Melis et al.19. Osteolysis of the humeral tuberosities was assessed by comparing immediate postoperative radiographs with those obtained at the time of final follow-up. The preoperative condition of the humeral tuberosities as well as any intraoperative fractures, nonunions, or excisions of the humeral tuberosities were identified by reviewing the operative reports and analyzed by comparing preoperative and immediate postoperative radiographs.
Postoperative complications were classified as early when they had occurred in the first 2 years following surgery and as delayed when they had occurred afterward.
A Kaplan-Meier survival analysis was performed with revision (i.e., removal or replacement of the prosthesis) for any reason as the end point. All revisions after the initial procedure were included in the survivorship analysis. Log-rank (Mantel-Cox) tests were performed to determine statistical differences in prosthetic survival among etiologies. Given that the paired data are from the same population, clinical outcomes were compared using nonparametric Wilcoxon signed-rank tests. The significance level was set at p < 0.05.
Of the 186 patients (191 prostheses) assessed in the earlier study8, 77 (79 prostheses) died before the long-term assessment, at a mean age of 82 years (range, 64 to 94 years) and at a mean of 6 years (range, 2 to 9 years) after the initial procedure. Of the remaining 109 patients (112 prostheses), 17 (17 prostheses) were lost to follow-up and 8 (8 prostheses) developed a complication that led to removal or revision of the prosthesis before the 10-year follow-up assessment. Long-term clinical data were thus obtained for 84 (77%) of the living patients and 87 (78%) of the prostheses in living patients. Both clinical and valid radiographic assessments were obtained for 64 (59%) of the living patients and 67 (60%) of the prostheses in living patients (Fig. 1). Eight patients (8 prostheses) were unable to return to our clinic and their cases were reviewed at an external institution. Only 2 of the 8 had interpretable radiographs. We also encountered difficulties assessing the grade of scapular notching on radiographs of 14 patients (14 prostheses), which were excluded from the radiographic analysis. We did not find significant differences in the Constant score preoperatively or at the follow-up time points, the sex ratio, or the age at the procedure between the patients evaluated outside and those examined at our center (p > 0.05).
The mean duration of follow-up was 150 months (range, 121 to 241 months), and the average patient age was 83 years (range, 45 to 95 years). The preoperative etiologies included rotator cuff tear arthropathy in 27 of the shoulders, failed arthroplasty in 21, a massive rotator cuff tear in 20, posttraumatic arthritis in 10, and primary osteoarthritis in 9.
Constant Scores and Range of Motion
The mean absolute Constant score at the last follow-up was 55 ± 16 points, which was a significant improvement relative to the preoperative score (Table I). Nonetheless, there was a significant decrease in the absolute Constant scores between the medium and late follow-up evaluations both overall (p < 0.001) and in every etiology group (p < 0.020). Except for the pain score (p = 0.051), all of the components of the Constant score significantly decreased between the medium and long-term follow-up evaluations (p < 0.001). There was also a significant reduction in the relative Constant score (p = 0.025). Similarly, anterior active elevation improved significantly after implantation of the prosthesis but decreased significantly between the medium and long-term follow-up evaluations (p < 0.001). Rotational range of motion did not diminish between the medium and long-term follow-up assessments.
Rotator cuff tear arthropathy and primary osteoarthritis (as the indication for the RTSA) were associated with the highest absolute Constant scores at the time of the long-term follow-up (63 ± 13 points and 62 ± 8 points, respectively). RTSAs due to a failed previous arthroplasty or to posttraumatic arthritis were associated with the lowest Constant scores (45 points for both) and with less anterior active elevation after 10 years of follow-up (Table II). Patients who had had the RTSA because of a failed previous arthroplasty were the only etiological group in which the pain and activity values of the Constant score significantly decreased between the medium and long-term follow-up evaluations (p = 0.012 and p = 0.006, respectively). A failed previous arthroplasty and a massive rotator cuff tear were the only 2 RTSA indications associated with a significant decrease in the Constant score for mobility and in anterior active elevation between the medium and long-term follow-up evaluations. The Constant score for strength decreased significantly between the 2 assessments in all of the etiological groups (Table II).
Radiographic analyses showed a radiolucent line around the glenoid implant in 3 shoulders. The radiolucency was ≥2 mm wide and was located around the glenoid screws and below the baseplate, with implant migration associated with the glenoid loosening.
Preoperative radiographs showed the absence of 1 humeral tuberosity in 5 shoulders and the absence of both tuberosities in 1. In addition, a fracture or confirmed nonunion during RTSA led to excision of both tuberosities in 5 cases and excision of only 1 tuberosity in 7 cases. The combination of preoperative tuberosity absence and intraoperative tuberosity excision resulted in 8 cases with absence of both tuberosities and 8 other cases with absence of only the lesser tuberosity. Postoperative absence of at least 1 tuberosity was significantly associated with a failed previous arthroplasty as the indication for the RTSA (p < 0.001). All except 3 cases of tuberosity absence were related to revision of an arthroplasty that had been carried out for the treatment of a displaced 4-part proximal humeral fracture.
Humeral radiolucent lines in at least 3 zones were seen in 8 shoulders (12%), including 1 with asymptomatic loosening. In 5 of these cases, a failed previous arthroplasty was the original indication for the RTSA (p = 0.016). Postoperative absence of the greater tuberosity was significantly associated with humeral radiolucent lines at the final follow-up evaluation (p < 0.001).
Forty-nine shoulders (73%) had scapular notching; 30 (61%) of these cases were classified as Sirveaux stage 1 or 2 and the other 19, as stage 3 or 4. Since the time of the medium-term follow-up8, 26 new cases of scapular notching (20 of which were Sirveaux stage 1 or 2) had developed (a 39% increase in the rate of scapular notching). Of the 23 cases that were apparent at the earlier follow-up evaluation, 11 progressed. There were no statistically significant differences in long-term Constant scores between patients without notching or with a lower stage of scapular notching (0, 1, or 2) and those with a higher stage of scapular notching (3 or 4) (p = 0.746).
Partial or complete osteolysis of at least 1 tuberosity was seen in 88% of the shoulders. We did not find osteolysis of the greater tuberosity (p = 0.064) or lesser tuberosity (p = 0.702) to have any impact on the long-term Constant score.
A total of 37 complications in 35 patients occurred during the first 2 postoperative years, and 10 in 10 patients occurred afterward (Table III). The 10 delayed complications were identified at a mean of 100 months (range, 37 to 139 months) after the RTSA, and they included 7 cases of unipolar prosthetic aseptic loosening: 4 on the glenoid side and 3 on the humeral side. There were no cases of glenoid loosening secondary to progression of scapular notching. The 4 cases of glenoid loosening were related to the use of a custom-made glenoid implant without a central peg but with acromial screw fixation (3 cases) and/or to technical error (1 case).
Revisions and Prosthetic Survivorship
There were 8 revisions in the first 2 years and 8 after 2 years (Table III). Six of the delayed revisions were secondary to aseptic implant loosening, and 2 were secondary to late infection.
Kaplan-Meier analysis showed a mean implant survival time of 110.3 months (95% confidence interval [CI], 103.9 to 116.7 months), with a 93% survival probability at 120 months and 128 cases at risk (Fig. 2). There was no difference in survival according to etiology (p = 0.30).
The long-term results of RTSA have rarely been reported2,6,18,19, and the average duration of follow-up in the long-term studies that have been published is <120 months (Table IV). In the present study, we evaluated the outcomes, complications, and implant survivorship in a consecutive series of Grammont-style RTSAs followed for a minimum of 10 years. While we found that the overall absolute and the relative Constant scores remained acceptable, both decreased significantly between the 2 evaluations.
Our findings are consistent with the Constant score of 57 points reported by Favard et al.18. The survival rate of 93% at 10 years that we reported confirms that the Grammont-style prosthesis is reliable. Other studies that analyzed the long-term survival of this type of prosthesis demonstrated similar results6,18. In 2004, Sirveaux et al.2 reported better functional outcomes and a higher implant survival rate after a shorter follow-up (mean, 44.5 months) in a study limited to RTSA performed for rotator cuff tear arthropathy. Rotator cuff tear arthropathy has been associated with the best results and long-term survival rates in several studies6,8,18,24. The present study confirms long-term high functional outcome scores in association with this etiology.
In contrast, we previously reported that RTSA for a failed previous arthroplasty or posttraumatic arthritis was associated with limited functional outcomes8. Our 10-year results are in agreement with those findings. Those 2 operative indications were each associated with a mean absolute Constant score of 45 points and mean active anterior elevation of <120°. Previous studies have shown that RTSAs performed for these indications provided less predictable results because of high rates of complications and revisions5,24. In the present study, shoulders that underwent the RTSA because of a failed previous arthroplasty had the lowest mean preoperative Constant score (21 ± 13 points), and each component of the score showed deterioration at the time of the long-term follow-up. It is most likely that a previous shoulder arthroplasty permanently alters shoulder function, as a result of fracture sequelae and/or of the arthroplasty itself.
The decrease in the overall absolute and relative Constant scores between the 2 evaluations is concerning. This degradation was described in 2004 by Sirveaux et al.2, in a study with 8 years of follow-up, and then by Guery et al.6 and Favard et al.18, who reported functional deterioration after 6 and 8 years of follow-up, respectively. Because annual Constant scores were not available in our study, we were unable to determine the pattern of the decrease that we observed. However, as found by Favard et al.18, aging is not the only cause of decreases in the relative Constant score. The reasons are not known. Authors have suggested the influence of occult loosening to explain the deterioration of shoulder function6,18. However, this hypothesis was not confirmed by our study, as the components of the Constant score that decreased most were strength and anterior active elevation, suggesting an impairment of active deltoid power. Muscle contractions of the deltoid tensioned by a lowered and medialized center of rotation correspond to alternating nonphysiological contraction-stretching cycles. Experimental studies have shown that the aging of muscle tissue hinders its adaptation to repetitive contraction-stretching movements and decreases its motor performance25-27. Thus, impaired deltoid efficiency could be the result of muscle senescence coupled with nonphysiological biomechanical requirements.
Functional decline varied among the different etiologies leading to the RTSA. Rotator cuff tear arthropathy, primary osteoarthritis, and a massive rotator cuff tear were associated with the least deterioration, while failed previous arthroplasty and posttraumatic arthritis were associated with a greater decrease. As suggested by Boileau et al.5, previous operations could impact deltoid power. Also, tuberosity malunion could interfere with prosthetic positioning, with disturbed biomechanical behavior and a faster decrease in outcomes over time28. Shoulders that underwent the RTSA because of a massive rotator cuff tear showed the largest decrease in anterior active elevation (17°) between assessments. Others have reported poor early functional outcomes of RTSAs done for the treatment of a massive rotator cuff tear29, but we do not have an explanation for this decrease at the time of long-term follow-up. The destruction of the rotator cuff muscles accompanying massive rotator cuff tears could lead to greater mechanical stress on the deltoid muscle, resulting in faster degradation of its action.
In the present study, 7 of the 10 complications that occurred after the 2-year follow-up time-point were related to mechanical loosening. The 4 cases of glenoid loosening were related to custom implants or a technical error. No glenoid loosening appeared in shoulders after highly advanced scapular notching. The glenoid side, which caused short-term complications in the series reported by Sirveaux et al.2 and Guery et al.6, did not result in frequent issues in our long-term follow-up study. We agree with Favard et al., who specified that once the period of short-term complications was past, the fixation of the glenoid component is stable18.
We noted 3 cases of aseptic humeral loosening and 7 cases of humeral periprosthetic radiolucent lines without loosening, totaling 15% of the prostheses assessed. In 2011, Melis et al.19 reported a similar rate of humeral radiolucent lines (12%). We noticed an association among previous failed arthroplasty (as the indication for the RTSA), perioperative tuberosity alteration, and humeral radiolucent lines. The small number of cases precluded a statistical analysis to explore 2-by-2 the strength of correlation among the diagnosis, the condition of the tuberosities, and humeral radiolucent lines. As shown by Cuff et al.30, the absence of a tuberosity deprives the humeral implant of an epiphyseal bone seat and could generate mechanical stresses with development of radiolucent lines. Although the number of cases with notching had increased by 39%, notching had no impact on the outcomes. The long-term development of notching and slow progression that were reported in previous series18,19,31 was confirmed by the current study. It could be a mixed process, with mechanical damage followed by a biological response31,32. In agreement with Melis et al.19, we did not find any correlation between the tuberosity resorption and the presence of a scapular notch (p = 0.328) and we agree with those authors that a stress-shielding mechanism may explain this tuberosity resorption.
This study has limitations. First, it was a retrospective analysis, with 94 patients (50.5%) who died or were lost to follow-up before the long-term assessment. Second, the patient age range was wide, making the results difficult to generalize to other patients. As the majority of the patients who had been operated on after 76 years of age died before the final assessment, our findings mainly concern patients who were operated on before this age. Our finding of degradation of functional results in patients who underwent the operation before 76 years of age are consistent with those of Guery et al.6 and Favard et al.18, who encouraged caution in using this prosthesis in younger patients.
In conclusion, while we found that RTSA remained an effective therapeutic option with long-term implant survival rates similar to those described in previous reports6,18, it is important to acknowledge that functional outcomes may be impacted by both the etiology of the shoulder dysfunction and the time since implantation.
NOTE: The authors acknowledge A/Prof. Eugene Ek (MBBS, PhD, FRACS) and Mr. Richard Large (MBBS, FRACS[Orth], FAOrthA) for assistance in writing and in translating this manuscript.
Investigation performed at Hôpital Privé Jean Mermoz Ramsay-GDS, Centre Orthopédique Santy, Lyon, France
Disclosure: The authors did not receive any outside funding or grants in support of their research for or preparation of this work. Two of the authors received, in any 1 year, payments or other benefits in excess of $10,000 from a commercial entity (Tornier). On the Disclosure of Potential Conflicts of Interest forms, which are provided with the online version of the article, one or more of the authors checked “yes” to indicate that the author had a relevant financial relationship in the biomedical arena outside the submitted work and “yes” to indicate that the author had other relationships or activities (involvement in the initial biomechanical development of the Aequalis reversed prosthesis design) that could be perceived to influence, or have the potential to influence, what was written in this work (http://links.lww.com/JBJS/A144).
1. Rittmeister M, Kerschbaumer F. Grammont reverse total shoulder arthroplasty in patients with rheumatoid arthritis and nonreconstructible rotator cuff lesions. J Shoulder Elbow Surg. 2001 ;10(1):17–22.
2. Sirveaux F, Favard L, Oudet D, Huquet D, Walch G, Molé D. Grammont inverted total shoulder arthroplasty in the treatment of glenohumeral osteoarthritis with massive rupture of the cuff. Results of a multicentre study of 80 shoulders. J Bone Joint Surg Br. 2004 ;86(3):388–95.
3. Frankle M, Siegal S, Pupello D, Saleem A, Mighell M, Vasey M. The reverse shoulder prosthesis for glenohumeral arthritis associated with severe rotator cuff deficiency. A minimum two-year follow-up study of sixty patients. J Bone Joint Surg Am. 2005 ;87(8):1697–705.
4. Werner CM, Steinmann PA, Gilbart M, Gerber C. Treatment of painful pseudoparesis due to irreparable rotator cuff dysfunction with the Delta III reverse-ball-and-socket total shoulder prosthesis. J Bone Joint Surg Am. 2005 ;87(7):1476–86.
5. Boileau P, Watkinson D, Hatzidakis AM, Hovorka I. Neer Award 2005: the Grammont reverse shoulder prosthesis: results in cuff tear arthritis, fracture sequelae, and revision arthroplasty. J Shoulder Elbow Surg. 2006 ;15(5):527–40.
6. Guery J, Favard L, Sirveaux F, Oudet D, Mole D, Walch G. Reverse total shoulder arthroplasty. Survivorship analysis of eighty replacements followed for five to ten years. J Bone Joint Surg Am. 2006 ;88(8):1742–7.
7. Levy J, Frankle M, Mighell M, Pupello D. The use of the reverse shoulder prosthesis for the treatment of failed hemiarthroplasty for proximal humeral fracture. J Bone Joint Surg Am. 2007 ;89(2):292–300.
8. Wall B, Nové-Josserand L, O’Connor DP, Edwards TB, Walch G. Reverse total shoulder arthroplasty: a review of results according to etiology. J Bone Joint Surg Am. 2007 ;89(7):1476–85.
9. Gohlke F, Rolf O. [Revision of failed fracture hemiarthroplasties to reverse total shoulder prosthesis through the transhumeral approach : method incorporating a pectoralis-major-pedicled bone window]. Oper Orthop Traumatol. 2007 ;19(2):185–208. German.
10. Young AA, Smith MM, Bacle G, Moraga C, Walch G. Early results of reverse shoulder arthroplasty in patients with rheumatoid arthritis. J Bone Joint Surg Am. 2011 ;93(20):1915–23.
11. Mizuno N, Denard PJ, Raiss P, Walch G. Reverse total shoulder arthroplasty for primary glenohumeral osteoarthritis in patients with a biconcave glenoid. J Bone Joint Surg Am. 2013 ;95(14):1297–304.
12. Jain NB, Yamaguchi K. The contribution of reverse shoulder arthroplasty to utilization of primary shoulder arthroplasty. J Shoulder Elbow Surg. 2014 ;23(12):1905–12. Epub 2014 Oct 7.
13. Day JS, Paxton ES, Lau E, Gordon VA, Abboud JA, Williams GR. Use of reverse total shoulder arthroplasty in the Medicare population. J Shoulder Elbow Surg. 2015 ;24(5):766–72. Epub 2015 Feb 18.
14. Schairer WW, Nwachukwu BU, Lyman S, Craig EV, Gulotta LV. National utilization of reverse total shoulder arthroplasty in the United States. J Shoulder Elbow Surg. 2015 ;24(1):91–7. Epub 2014 Oct 29.
15. Kim SH, Wise BL, Zhang Y, Szabo RM. Increasing incidence of shoulder arthroplasty in the United States. J Bone Joint Surg Am. 2011 ;93(24):2249–54.
16. Oppermann J, Celik E, Bredow J, Beyer F, Hackl M, Spies CK, Müller LP, Burkhart KJ. Shoulder arthroplasty in Germany: 2005-2012. Arch Orthop Trauma Surg. 2016 ;136(5):723–9. Epub 2016 Feb 8.
18. Favard L, Levigne C, Nerot C, Gerber C, De Wilde L, Mole D. Reverse prostheses in arthropathies with cuff tear: are survivorship and function maintained over time? Clin Orthop Relat Res. 2011 ;469(9):2469–75.
19. Melis B, DeFranco M, Lädermann A, Molé D, Favard L, Nérot C, Maynou C, Walch G. An evaluation of the radiological changes around the Grammont reverse geometry shoulder arthroplasty after eight to 12 years. J Bone Joint Surg Br. 2011 ;93(9):1240–6.
20. Constant CR, Murley AH. A clinical method of functional assessment of the shoulder. Clin Orthop Relat Res. 1987 ;214:160–4.
21. Constant CR. Age related recovery of shoulder function after injury [Thesis]. Cork, Ireland: University College, Cork; 1986.
22. Sperling JW, Cofield RH, O’Driscoll SW, Torchia ME, Rowland CM. Radiographic assessment of ingrowth total shoulder arthroplasty. J Shoulder Elbow Surg. 2000 ;9(6):507–13.
23. Lévigne C, Boileau P, Favard L, Garaud P, Molé D, Sirveaux F, Walch G. Scapular notching in reverse shoulder arthroplasty. J Shoulder Elbow Surg. 2008 ;17(6):925–35. Epub 2008 Jun 16.
24. Stechel A, Fuhrmann U, Irlenbusch L, Rott O, Irlenbusch U. Reversed shoulder arthroplasty in cuff tear arthritis, fracture sequelae, and revision arthroplasty. Acta Orthop. 2010 ;81(3):367–72.
25. Cutlip RG, Baker BA, Geronilla KB, Mercer RR, Kashon ML, Miller GR, Murlasits Z, Alway SE. Chronic exposure to stretch-shortening contractions results in skeletal muscle adaptation in young rats and maladaptation in old rats. Appl Physiol Nutr Metab. 2006 ;31(5):573–87.
26. Baker BA, Hollander MS, Mercer RR, Kashon ML, Cutlip RG. Adaptive stretch-shortening contractions: diminished regenerative capacity with aging. Appl Physiol Nutr Metab. 2008 ;33(6):1181–91.
27. Faulkner JA, Brooks SV, Zerba E. Muscle atrophy and weakness with aging: contraction-induced injury as an underlying mechanism. J Gerontol A Biol Sci Med Sci. 1995 ;50(Spec No):124–9.
28. Boileau P, Trojani C, Walch G, Krishnan SG, Romeo A, Sinnerton R. Shoulder arthroplasty for the treatment of the sequelae of fractures of the proximal humerus. J Shoulder Elbow Surg. 2001 ;10(4):299–308.
29. Hartzler RU, Steen BM, Hussey MM, Cusick MC, Cottrell BJ, Clark RE, Frankle MA. Reverse shoulder arthroplasty for massive rotator cuff tear: risk factors for poor functional improvement. J Shoulder Elbow Surg. 2015 ;24(11):1698–706. Epub 2015 Jul 11.
30. Cuff D, Levy JC, Gutiérrez S, Frankle MA. Torsional stability of modular and non-modular reverse shoulder humeral components in a proximal humeral bone loss model. J Shoulder Elbow Surg. 2011 ;20(4):646–51. Epub 2011 Jan 13.
31. Lévigne C, Garret J, Boileau P, Alami G, Favard L, Walch G. Scapular notching in reverse shoulder arthroplasty: is it important to avoid it and how? Clin Orthop Relat Res. 2011 ;469(9):2512–20.
32. Nyffeler RW, Werner CM, Simmen BR, Gerber C. Analysis of a retrieved Delta III total shoulder prosthesis. J Bone Joint Surg Br. 2004 ;86(8):1187–91.