Citation
Sande, M. A. J. van de. (2008, February 14). Rotator cuff degeneration in the rheumatoid shoulder : 'the issue is soft tissue'. Retrieved from https://hdl.handle.net/1887/12603
Version: Corrected Publisher’s Version
License: Licence agreement concerning inclusion of doctoral thesis in the Institutional Repository of the University of Leiden
Downloaded from: https://hdl.handle.net/1887/12603
Shoulder arthroplasty
and rotator cuff surgery:
an overview of results
Chapter 2
Shapter 2.1
Indications, complications, and results of shoulder arthroplasty
M.A.J. van de Sande, R. Brand, P.M. Rozing.
Scan J Rheumatology 2006; 35: 426-436
Summary
To provide estimates of patient outcomes following shoulder arthroplasty using Neer-II type humeral prosthesis and to examine variation in outcomes due to patient and prosthesis characteristics. North American and Western European published articles were identified through a computerized literature search and bibliography review. Studies were included if they enrolled 15 or more patients, discriminated between hemi- arthroplasty (HEMI) and total shoulder arthroplasty (TSA) and measured pain relief, gain in range of motion (ROM), radiographic follow-up (> 2 years), short- and long-term complications, and revision surgery. A total of 40 studies satisfied the inclusion criteria.
The total number of patients enrolled was 3584. The mean follow-up was 59 months.
The mean patient age was 62 years, 65% of patients were women and 73% underwent TSA. All reports showed relevant pain relief, increase in ROM, and high satisfaction rates for HEMI and TSA in both osteoarthritis (OA) and rheumatoid arthritis (RA). The overall rate of revision was 8%. Significant differences between HEMI and TSA for both diagnoses were found for all outcome parameters.
Shoulder arthroplasty is a safe and effective procedure for OA and RA patients. The diagnosis, shoulder pathology, and prosthesis specifics were significant predictors of outcomes. We therefore emphasize that conclusions on the outcome of shoulder arthroplasty can only be made if differentiated between these patient and prosthesis specifics. Limitations in the reporting style of these articles severely constrain the ability to explore variation in outcomes due to study, patient, or prosthesis characteristics and restrict their generalisability.
Introduction
In the aetiology of rheumatoid arthritis (RA), shoulder involvement generally occurs late in the disease process and usually after other joints has manifested arthritic change. In osteoarthritis (OA) or degenerative joint disease, shoulder involvement is most commonly seen with increased age or after shoulder injury, disease, or trauma. Any of three shoulder articulations can be involved, but the glenohumeral joint is most frequently symptomatic.3; 45Although the incidence of shoulder involvement in patients with OA and RA is reported to be as high as 91%, only a small percentage is treated surgically, including synovectomy, arthroplasty, and arthrodesis.46; 47Over the past 30 years, short- and long-term follow-up studies have reported good to excellent pain relief and improvement in range of motion (ROM) with the Neer-II type unconstrained total shoulder prostheses.25; 26; 48-58; 58-68 Only a few studies compared shoulder hemi-
20
arthroplasty (HEMI) with total shoulder arthroplasty (TSA), and even less in a prospective study design. Recent studies reported slightly favouring results for TSA in pain relief and improvement of ROM.46; 69-83These studies concluded that both provide comparable levels of functional motion and pain relief. Some recommended TSA for RA except in cases with inadequate glenoid bone stock, but HEMI for OA.73; 77They also warned of the risk of glenoid loosening in the long-term follow-up, and increasing pain caused by glenoid erosion and sclerosis in HEMI.
In 1996 Rodosky and Bigliani reviewed 51 studies, published between 1955 and 1995, discussing shoulder arthroplasties, comparing TSA with HEMI.30Because most series included patients with a mixture of diagnoses and duration of follow-up varied between 0.7 and 9.5 years, only conditional conclusions could be drawn. This report provides an updated systematic review on shoulder arthroplasty in OA and RA using the Neer-II type prosthesis covering the past three decades. It presents the results for HEMI and TSA subdivided by diagnoses. We believe this updated systematic review offers new insights into shoulder joint replacement for OA and RA patients.
Materials and methods
Over 700 articles on shoulder arthroplasty have been published or reported on over the past 30 years. Studies needed to match the following criteria in order to be included in this systematic review:
• Study starting not earlier than 1973 (Neer II).
• Full distinction between HEMI and TSA and a description of the total group of patients included (n > 15).
• Surgical indication, diagnosis: OA and RA, prosthesis, and technique used should be described and the follow-up period should be > 2 years.
• Pain relief, gain in ROM, radiographic follow-up (2 years postoperative), short- and long-term complications, and revision surgery should be clearly evaluated for all included shoulders.
• Research group situated in North America or Western Europe.
Articles were retrieved using the Medline®database. The following search terms were used: Shoulder, Arthroplasty, Prosthesis. Of 273 articles the abstract was carefully read and if the abstract was not primarily excluded, the article was retrieved as a whole. The bibliography sections in all articles reviewed were examined and missed citations were retrieved. Twenty-eight articles were excluded after they were retrieved and checked.
Other than Neer-II type prostheses, designs such as humeral head resurfacing and reversed or bipolar prostheses are not included in this analysis, as they are not yet common practice or are only used as a salvage procedure. Forty articles fulfilled our inclusion criteria and were carefully analysed defining preoperative status and postoperative clinical results (pathology and rating criteria and endpoints presented in Tables 1 and 2). Included series were analysed as a whole and, if possible, separately analysed for each diagnosis. If the results presented in a report could not be separated by diagnosis, the article was only reviewed once for the whole group (mixed diagnoses). If articles did not distinguish between primary and secondary OA, both are reported on in one group, OA. (Secondary
Indications, complications, and results of shoulder arthroplasty
OA was defined as post-traumatic, post-injury, and post-avascular necrosis.) A total of 3584 shoulders were included of which 2583 (73%) received a TSA (Table 3). Included studies are summarized in Tables 4–6. Not all rating criteria or endpoints could be analysed for each individual diagnosis, as there were insufficient data. For every rating criteria or endpoint, the number of shoulders reviewed is indicated.
Table 1.‘Pathology’ and ‘rating’ criteria.
* For each operative and long-term complication: fracture, nerve injury, deep infection, instability, dislocation, proximal migration of the humeral head, and glenoid erosion.
† ROM: AFE to 90° or more has a positive correlation with the ability to use the hand with the arm at shoulder level.92
‡ Clinical or patient satisfaction, based on the rating system of Neer et al and Cofield.89
Table 2. Endpoints.
* Distinction between cemented and uncemented components.
† Complete RLL > 2 mm, or positional shift of component(s).
‡ Revision surgery for symptomatic aseptic loosening of one or two components, not all ‘loose’ prosthesis were revised.
No relevant differences were found between HEMI and TSA groups for sex, age, or average duration of follow-up. The mean age for HEMI was 62 years, and 63 years for TSA, 65% of all patients were female and the mean follow-up was 59 months (range 24–150). Surgical approach and operative technique were comparable, implant material of course varied, although all 16 different prosthetic designs are unconstrained and based on the original Neer-II prostheses.
Rotator cuff tears were found in all diagnoses, but were significantly more frequent in patients with RA (p = 0.016) (Table 7). The HEMI group showed twice as many cuff tears as the TSA group (p < 0.05). Preoperative pain or pain scores were comparable for HEMI
Diagnoses
Surgical technique, implant material Status of the rotator cuff (preoperative) Complications*
Proximal migration (postoperative) Pain relief
Range of motion (ROM), active forward elevation (AFE) and active external rotation (AER) (pre-/postoperative)†
Satisfaction‡
Radiolucent line (RLL)*
Radiographic aseptic loosening†
Revision surgery for aseptic loosening‡
Revision surgery for increasing pain
Re-surgery for dislocation of the humeral component Removal of prosthesis for deep infection
22
and TSA, but the preoperative ROM was not. The preoperative active forward elevation (AFE) was lower in the TSA group for all diagnoses (HEMI 76˚, TSA 68˚) (p < 0.01). AFE in the RA group was significantly lower than in the OA group (HEMI 60˚, TSA 56˚) (p = 0.02). The active external rotation (AER) did not significantly differ between HEMI and TSA, or between the diagnoses (HEMI 20˚, TSA 19˚) (p = 0.7).
Table 3. Description of shoulders included in all 40 presented series.
* If not specified between primary and secondary osteoarthritis, both are reported as one group;
OA. Secondary osteoarthritis was defined as post-traumatic or post-infection osteoarthritis.
Table 4. Clinical results for HEMI.
Statistical analysis
Odds ratio (OR) calculations for HEMI and TSA were corrected for diagnosis using SPSS version 11.0 (SPPS Inc, Chicago, IL, USA). An OR is calculated by dividing the odds for an event (rating criteria or endpoint) in the TSA group by the odds in the HEMI group.
An OR of 2 would mean that a patient in the TSA group would have twice the chance of having this event, compared to a patient in the HEMI group. The OR was considered statistically significant if the 95% confidence interval did not contain the value 1.
Homogeneity was analysed using the Breslow–Day homogeneity test: if the ORs for a diagnosis did not significantly differ, a Mantel–Haenszel common odds ratio estimate (MHcORe) was calculated, combining the risk estimates for all diagnoses into one weighted average. Percentages for every diagnosis for HEMI and/or TSA were calculated using SPSS and Microsoft Excel spreadsheets. The independent samples and one-sample t-test were used to compare the prevalence of outcome between HEMI and TSA or between the diagnoses. Univariate analysis of variance was used to estimate the marginal means for the ROM. For all tests p < 0.05 (two-sided) was considered significant.
Indications, complications, and results of shoulder arthroplasty
Mixed diagnoses RA OA* Total
HEMI 452 296 217 965 (27%)
TSA 1306 820 457 2583 (73%)
Total 1758 (49%) 1116 (31%) 674 (19%) 3584
Study Implant Diagnosis n Follow Cuff Pain AFE Revision Satisfied
type -up tears relief
(%) (%) (%) (%) (%)
Arntz et al59 1 Mixed 18 35 100 83 185 22 100
Capelle and
Visser93 12 RA 38 45 39 95 136 3 –
Cofield et61 1 Mixed 67 108 9 91 131 18 49
Levine et al64 1 OA 30 29 0 – – – 94
Pfahler et al94 12 Mixed 469 NA 32 – – 21 85
There are several assumptions that underlie a systematic review of non-randomized (retrospective as well as prospective) studies. These should be recognized in order to have a better understanding of the strengths and weaknesses of this analysis.
Attentiveness to the effect of unmeasured or unreported confounders is required at all times. However, it was not always possible to correct for possibly confounding factors such as duration of follow-up, patient selection, and description and measurement of endpoints. While analysing the acquired data, we have tried to incorporate these shortcomings in our conclusions. Moreover, causal conclusions can be drawn only on the basis of randomized studies or a meta-analysis of such studies, which are seldom found on this subject.
Table 5. Clinical and radiographic results for TSA.
AFE, active forward elevation; RLL, radiolucent lines; Rad., radiographically; cem, cement.
Study Implant Diag- n Follow Cuff Pain AFE RLL Rad. Rad. Revision Satisfied type nosis -up tears relief (%) glenoid loose loose (%) (%)
(%) (%) glenoid glenoid
cem (%)
Sneppen et al62 13 RA 62 92 – 98 170 – 40 40 5 –
Stewart and Kelly66 1 RA 37 114 14 89 142 62 27 27 – –
Barret et al 54 1 Mixed 50 42 0 92 345 74 10 10 8 94
Figgie III et al56 1 RA 50 60 24 96 – – 6 6 0 82
McCoy et al50 1 RA 29 37 24 – 125 86 4 4 – 83
Barret et al49 9 RA 140 60 44 93 161 76 9 5 0 –
Hawkins et al51 1 Mixed 70 48 – 97 185 – 7 7 4 91
Figgie et al71 4 RA 27 60 48 – 167 22 7 7 0 –
Fenlin et al60 5 Mixed 47 54 11 87 134 47 11 11 6 –
Frich et al57 1 Mixed 52 27 – 42 – – 4 4 2 –
Cofield and Daly58 10 Mixed 32 41 6 100 – 47 10 – 13 93
Roper et al95 6 Mixed 25 60 20 100 175 16 4 – 8 –
Boileau et al80 7 OA 20 37 20 – 158 85 25 25 0 –
primary
Boileau et al80 7 OA 20 38 5 – 160 25 22 – 25 –
primary
Amstutz et al96 8 Mixed 46 42 – – 182 93 2 2 11 –
Friedman et al52 15 RA 24 54 38 92 188 42 8 8 0 –
Thomas et al26 8 RA 30 46 40 100 155 100 17 17 7 –
Brenner et al97 9 RA 51 69 14 100 150 59 32 32 4 90
and OA primary
Neer et al89 1 Mixed 273 37 18 – – 30 0 0 2 88
Torchia et al65 1 Mixed 113 144 16 73 152 84 44 44 12 70
Wallace et al68 1 RA and 58 64 0 – – 33 2 0 14 –
OA primary
Sperling et al78 10 Mixed 62 58 6 – – 42 2 – 16 –
Norris and
Lachiewicz63 1 Mixed 38 60 37 – 153 53 3 3 0 97
Gristina et al98 12 Mixed 100 38 – 90 338 67 3 3 4 94
Pfahler et al94 7 Mixed 705 NA 15 – – 68 11 10 6 92
Ranawat et al99 14 Mixed 24 20 – 100 191 96 8 8 13 81
24
Indications, complications, and results of shoulder arthroplasty
StudyImplantDiagnosisHEMITSA(n)Follow-upCufftearsPainreliefAFERLLRad.RevisionSatisfied type(n)(%)(%)(%)glenoidloose(%)(%) (%)glenolid HEMITSAHEMITSAHEMITSAHEMITSAHEMITSA Boydetal771Mixed641464420209896154167–12––8996 Gartsmanetal792OA24273500––143149–7130–– NorrisandIanotti832OAprim.329429221–––––735–– Sperlingetal731Mixed7434150721––––561320125557 Claytonetal691Mixed784800100100121158–000–– Pollocketal721RA1911411001009591187102––009591 Sperlingetal821OAsec.102184007067––432230386038 VandeSandeand Rozing7516RAandOA22384729109192164136500009286 TrailandNuttall742RA654061400––13012453380–– Collinsetal762RA3625383661––120168–200––
Table6.ClinicalandradiographicresultsforHEMIandTSA 1NeerII,3M,Minnesota. 2GlobalModularShoulderProsthesisDepuy,Warsaw,IN. 3McnabTSA. 4CustomTSA;TheHospitalforSpecialSurgery,NewYork,NY. 5Zimmer,FenlinTSsystem. 6Roper–DayTSA. 7Aequalisshouldersystem. 8DanaTSA,Howmedica,Rutherford,NJ. 9NeerIITSAandGristinaTSA. 10Cofield,SmithandNephewRichards,Memphis,TN. 11Biomet,Warsaw,IN. 12GristinaMonospherical,HowmedicaInc.,Rutherford,NJ. 13NeerIITSA,3MHealthCareLimited,Rotherham,UK. 14O’Leary-WalkerandNeerITSA. 15NeerIITSA,3M,Minnesota&Monospherical,Howmedica,NJ. 16Multiplextotalshouldersystem,ESKAimplantsGmbH&Co,Lubeck,Germany.
Table 7. Cuff tears found during surgery (as percentage of shoulders evaluated for each group (n)), attenuation of the cuff was not counted as a tear.
Not all series reported cuff tears found during surgery; 605 rotator cuff tears were reported. The HEMI group contained significantly more cuff tears in the mixed diagnoses group.
* Significant difference HEMI vs. TSA.
† No MHcORe was calculated, the Breslow–Day homogeneity test was significant for all diagnoses (p = 0.03).
Results
Peroperative complications
Periprosthetic fractures of the humeral shaft or glenoid (reported with an average value of 1.2%, range 0–8%) and nerve injury (0.9%; 0–8%) were most frequently reported on (31 series). No significant difference was found for the risk of a periprosthetic fracture or nerve injury between the different diagnoses.
Complications
Deep infection or sepsis following shoulder arthroplasty is rarely seen. Of 1919 shoulders (548 HEMI), only 16 infections were reported (0.8%). TSA showed a much higher chance of infection (0.9%), compared to HEMI (0.4%) (MHcORe = 3.8; p > 0.05). The probability of infection did not significantly differ among the individual diagnoses.
Instability
Instability is one of the most common complications in shoulder arthroplasty. It can occur in different directions and is often difficult to manage. Insufficiency of glenoid bone stock and disruption of Subscapularis tendon repair are known causes, but prevention relies largely on surgeon experience and can usually be avoided by precise soft tissue balancing and proper prosthetic positioning.84-86Reviewing 1467 cases of TSA, we found 79 patients (5%) with either complaints of instability or a radiographically diagnosed subluxation. Seventy-nine of 402 cases of HEMI (20%) reported instability. Calculating a common OR for all diagnoses, including these series, showed that HEMI presented three times more postoperative instability than TSA (MHcORe 0.3; p < 0.05). Although prevalence and morbidity are serious, only a few series report on postoperative glenohumeral instability. The higher incidence of rotator cuff tears in the HEMI group is thought to be largely responsible for this difference. However, we were unable to identify
Mixed diagnoses RA OA prim. OA Total N
HEMI 26% 45% 13% 9% 28% 928
TSA 16% 30% 4% 14% 18% 2168
OR 0.5* 0.5* 0.3* 1.7* †
Total 19% 33% 6% 13% 20% 3309
26
the incidence of rotator cuff tears in the reported instability groups; therefore, factors influencing instability, such as rotator cuff tears, could not be identified.
Proximal migration
Prevalence of proximal migration of the humeral head was more than three times greater in HEMI (MHcORe 0.3, p = 0.02) than in TSA and was different according to diagnoses (Table 8). This difference might be explained by the relationship between rotator cuff tears and proximal migration. We found that the prevalence of proximal migration increased significantly with the increasing prevalence of rotator cuff tears in both HEMI and TSA. Series with less than 10% tears showed a 32% and 11% prevalence of proximal migration for HEMI and TSA, respectively, whereas series with more than 10%
tears showed a 53% and 16% prevalence for HEMI and TSA, respectively (p = 0.037).
Table 8. Proximal migration found using standard supine anterior–posterior radiographs, with the arm externally rotated, as a percentage of shoulders evaluated for each group (n).
* Significant difference between the HEMI and TSA groups (p < 0.05).
Not all series reported proximal migration as measured on postoperative radiographs; 1259 shoulders were evaluated, 295 shoulders were reported to have a proximal migration (superior subluxation) of the humeral head.
Glenoid erosion
Analysing series with specific attention to the occurrence of glenoid erosion after HEMI demonstrated that 93 out of 432 shoulders (22%) showed progressive glenoid erosion, of which 38 shoulders (41%) needed revision surgery for increasing pain, associated with glenoid erosion. When we divided all series into two groups, one group containing all series with a mean follow-up under 60 months and one group with an average follow-up over 60 months, we encountered different results for short- and long-term follow-up studies. The prevalence of glenoid erosion in the former group was 5% whereas the prevalence in the latter was 31%, showing a marked increase in prevalence with time.
The RA patients showed the lowest prevalence in glenoid erosion, with only 4%
compared with 34% in the mixed diagnoses group. Series reporting glenoid erosion were unfortunately scarcely found; therefore, no significant differences were found, nor could a causal relation between follow-up and glenoid erosion be demonstrated.59; 61;
70; 72; 82; 87 Extensive research is needed to provide a better risk estimate for the revision rate caused by glenoid erosion in HEMI.
Indications, complications, and results of shoulder arthroplasty
Mixed diagnoses RA OA prim. OA Total N
HEMI 38% 33% 10% 47% 31% 521
TSA 11% 30% 10% – 19% 738
OR 0.2* 0.4* 0.9 0.3*
Total 22% 30% 10% 47% 23% 1259
Pain relief
It is common use to analyse pain relief as a primary outcome of arthroplasty. We believe that the number of patients still reporting severe pain after arthroplasty is equally important. Out of 1147 total shoulders, 108 (9%) still experienced severe pain after TSA.
In 491 HEMIs, 43 still experienced severe pain (9%). Only 5% of the RA patients reported severe postoperative pain. For OA the prevalence was more than 10%. Pain experience, pain medication, or anti-inflammatory drug treatment might play an important role in the pain assessment of rheumatoid patients. Neither data nor literature was found to confirm this supposition.
Table 9. Good pain relief after shoulder arthroplasty, as a percentage of shoulders evaluated for each group (n).
* Significant difference between the HEMI and TSA groups (p < 0.05).
Analysing significant pain relief after arthroplasty, we found no difference in prevalence for HEMI (87%) or TSA (85%), although we did find a substantial difference in the risk estimates (Table 9). For RA we found a significantly better result in pain relief for TSA (p
< 0.05). Length of study did not significantly change the ORs for good pain relief (p = 0.34). Pain relief is considered the most important indication for shoulder arthroplasty and therefore also put forward as the most important outcome parameter. The similar percentage of patients presenting good/significant pain relief might conceal the difference in satisfaction and functional outcome between HEMI and TSA and the different diagnoses.
Range of motion (ROM)
Motion data after shoulder arthroplasty have not been reported or measured; therefore, we only discuss the results for AFE and AER.
Active forward elevation (AFE). Having the greatest influence on the functional outcome of shoulder arthroplasty, AFE is consequently reported on in 28 series including 1202 TSA and 350 HEMI.54For all diagnoses combined, postoperative AFE showed a slightly better result for TSA (111˚) than HEMI (103˚), despite the lower preoperative AFE reported in the TSA group. Therefore, the average improvement in AFE in TSA (43˚) was significantly better than in HEMI (31˚) (p = 0.017). The improvement in AFE in the RA group, compared to the average improvement in the other groups, was significantly lower, improving only 34˚ for TSA and 20˚ for HEMI to 90˚ and 78˚, respectively (p = 0.016). However, a significantly better result in improving the AFE for TSA was calculated (p = 0.052), compared to HEMI. Both study duration and percentage of cuff tears reported were not identified as confounders for improvement in ROM.
Mixed diagnoses RA OA Total N
HEMI 93% 74% 97% 87% 416
TSA 87% 84% 97% 85% 1448
OR 0.5* 1.7* 1.3 (0.3)* –
Total 88% 81% 97% 85% 1864
28
Active external rotation (AER). No significant difference was found between HEMI and TSA for the improvement in AER, although TSA improved with an average of 27˚ and HEMI improved only 23˚. The improvement in AER for the RA group again was significantly lower than that reported in the other diagnoses (p = 0.03), improving 27˚
for TSA (from 18° to 38˚) and 15˚ for HEMI (from 5° to 20˚).
Radiographic data
Comparing the different series by analysing signs of radiographic loosening of the components proved to be very difficult. This is mainly because of non-standardized methods of measurement, inconsistent definitions, and limitations of conventional radiography. We have compared the prevalence of radiolucent lines (RLL) and their progression (PRLL) for the different diagnoses and fixation techniques. Progression, component shift, a complete lucent line measuring more than 2 mm, and cement and component fracture were considered indicative findings for component loosening or failure and, if present, the component was marked as loose.
Glenoid component loosening
The prevalence of RLL, PRLL, and aseptic loosening is reported in Table 10. Despite the prevalence of 59% for RLL, only 13% progressed and eventually 9%, based on the reported radiographic findings, were considered loose. Of all reported glenoid components, 2% were revised for aseptic loosening.
Table 10. Radiographic signs of aseptic loosening of the glenoid component, using standard radiographs (as a percentage of shoulders evaluated for each group). Presenting all short-term (< 60 months) and long-term (> 60 months) follow-up series.
* Revision surgery reported, caused by aseptic glenoid component loosening, as percentages of all TSA. Of 134 revised TSAs, 38 were reported revised for aseptic glenoid component loosening (28%). No difference was found between short- and long-term follow-up studies.
† The OR is calculated for the chance of aseptic loosening for cemented glenoid components versus press- fitted components.
Indications, complications, and results of shoulder arthroplasty
Mixed RA OA OA OA Total
diagnoses primary primary
RLL 57% 69% 55% 43% 71% 59%
RLL < 60 58% 78% 55% — — 59%
RLL > 60 52% 66% — 43% 71% 59%
PRLL 11% 22% 16% – – 13%
Loose 7% 13% 7% 12% – 9%
Loose < 60 7% 9% 7% – – 7%
Loose > 60 9% 17% – 19% – 14%
Revised* 2% 1% 4% – 7% 2%
OR† 0.3 – 0.8 – – 0.5
Humeral component loosening
In our analysis we found a 3.6% prevalence of humeral component loosening, but no difference could be found between the cemented or non-cemented technique, nor between the different diagnoses. Of all 277 revised shoulders, only seven (2.5%) were reported as being caused by aseptic loosening of the humeral component.
Table 11. Revision surgery as a percentage of shoulders evaluated for each group (n).
* Significant difference for the OR between HEMI and TSA (p < 0.05).
Revision surgery
In total, 277 (8%) shoulders underwent revision surgery. The differentiation between the diagnoses, HEMI, and TSA are presented in Table 11. RA patients showed the lowest prevalence of revision surgery (2%). The mixed diagnoses group showed the highest risk for revision surgery with a prevalence of 10%. The risk for revision surgery was three times as high in the HEMI group (p < 0.001). Using a limited number of data, we found that 29% of all revision surgery in TSA was performed because of glenoid loosening. We also found that patients with a HEMI had six times the risk for revision caused by pain than those with a TSA. Aseptic loosening and glenoid erosion with severe pain are responsible for an important part (30%) of all revision surgery performed in the reported series. Not all series reported on the specific cause of revision surgery, therefore it was difficult to analyse the influence of glenoid erosion, RLL, and other rating criteria on the risk for component failure and revision surgery.88
Patient satisfaction
Neer et al89described an excellent result as ‘No significant pain, full use of shoulder and the patient is happy about the operation’, and a satisfactory result as ‘No more than occasional pain, good use of the shoulder for daily use from the top of the head down (90–135 degrees of elevation, and rotation to 50% of that of the normal side) and satisfied about the operation’. Many authors have used this description, or adjusted it somewhat to the present time.53We have included all descriptions that matched this gradation in our analysis. The results are presented in Table 12.
Dividing the series into short- and long-term follow-up revealed that with an increasing study duration, the average prevalence of satisfied patients for HEMI and TSA combined significantly dropped from 86% to 62% (t = 0.032). Therefore, duration of follow-up was a relevant confounder in the analysis of patient satisfaction.
Again, as with pain relief, we believe the unsatisfied patient carries an important message. Out of 1796 patients (9%), 178 reported an unsatisfactory outcome after TSA,
Mixed diagnoses RA OA Total N
HEMI 20% 2% 10% 15% 978
TSA 6% 1,5% 11% 6% 2453
OR 0.3* 0.7 0.8 0.3*
Total 10% 1.5% 10% 8% 3431
30
and 187 of 922 patients reported an unsatisfactory result after HEMI (20%). The risk of an unsatisfactory result was 2.5 times higher for HEMI in the mixed diagnoses group (OR = 2.5; p < 0.05).
Table 12. Satisfactory results after shoulder arthroplasty, as a percentage of shoulders evaluated for each group; both excellent and satisfactory were counted as satisfactory.
A total of 2706 were evaluated (888 HEMI); 2230 (82%) patients were reported as satisfactory (716 HEMI; 81%).
* Significant difference for the OR between HEMI and TSA (p < 0.05).
† Significant difference with other diagnoses.
‡ Secondary OA differed significantly from the other ORs (t = 0.008), MHcORe was computed without the secondary OA group.
Discussion
A general consensus on glenoid resurfacing has not yet been reached since the first shoulder prosthesis of the modern era was implanted in 1952.90In analysing shoulder arthroplasty, the orthopaedic surgeon is confronted with many different types and designs of prostheses, surgical techniques, postoperative rehabilitation, and scores for the outcome of pain, ROM, and function. Various shoulder centres around the world present a different perspective on all of these specifics and series presented are therefore difficult to compare. It should also be remembered that the local indications for glenoid resurfacing have caused a higher prevalence of rotator cuff tears and preoperative glenoid erosion in the HEMI group, permitting most of these patients only limited rehabilitation goals. Analysing 3584 patients with an average follow-up of 5 years also contains many uncertainties for the long-term follow-up results of shoulder arthroplasty, especially in glenoid component loosening, superior subluxation, and glenoid erosion, causing pain, loss in ROM, and sometimes even the need for revision surgery.
Furthermore, interest and subjects of the individual researcher can be relevant confounders for the prevalence of pathology and the endpoints measured.
Adjusted pre- and postoperative parameters showed significant differences between HEMI and TSA but also between the different diagnoses. This underexposes the
Indications, complications, and results of shoulder arthroplasty
Mixed RA OA OA Total N
diagnoses
HEMI 79% 62% 82% 66% 53% 80%
> 60 months 51% 52% 62% 52% 53% 54%
TSA 84% 79% 90% 43% 60% 83%
> 60 months 71% 68% – 32% 59% 67%
OR 1.3* 2.2* 1.8 0.4*† 1.3 1.4*‡
> 60 months 2.3* 2.0* 1.4 0.4 1.2 2.1*‡
Total 82% 75% 86% 57% 57% 82%
calculated difference in pain relief and satisfaction between HEMI and TSA, if not adjusted for these diagnoses. We therefore emphasize that conclusions on these parameters can only be made if they are subdivided into these patient and prosthesis specifics.
The differences in outcome between HEMI and TSA, adjusted for diagnosis, rotator cuff tear, and duration of follow-up, support the consensus on glenoid resurfacing, as it shows better results for TSA in patients with an intact rotator cuff. Glenoid bone stock was insufficiently reported on; therefore no adjustment for this parameter could be made.
The ‘long-term’ studies in our analysis have shown a significant decrease in patient satisfaction and a significant increase in glenoid erosion and glenoid component loosening with time.91However, no difference was found between long- and short-term follow-up for pain relief, ROM, or revision surgery. Long-term follow-up reports on both HEMI and TSA might provide information on this matter.
Taking into account this vast diversity of pathology, patients, surgeons, techniques, prostheses, and follow-up, this systematic review presents an overview of shoulder arthroplasty series presented over the past three decades. By dividing all patients into HEMI and TSA groups, then analysing them separately for each diagnosis and correcting for confounding where possible, we present a clearer picture of indications, complications, and successes in shoulder arthroplasty. We are therefore confident that this systematic review, making use of 3584 well-documented shoulder arthroplasties, presents new and valuable information, facilitating many shoulder surgeons not only in their decision making on glenoid resurfacing but also in assessing whether there is indeed a need for randomized studies, in order to infer the correct causal conclusions leading to evidence-based choices of approaches and materials. It also emphasizes that early referral to an orthopaedic surgeon might present the patient in a less destructive stage of the disease, resulting in a better condition of the rotator cuff and more glenoid bone stock to work with. In turn, this might improve the long-term clinical results of shoulder arthroplasty in the RA and OA patient.
I would like to gratefully acknowledge the statistical expertise, patience and time of Dr. Ronald Brand in the preparation and writing of this article.
