International Orthopaedics (SICOT) DOI 10.1007/s00264-017-3443-0

ORIGINAL PAPER

Is there evidence that the outcomes of primary anatomic and reverse shoulder arthroplasty are getting better? Jeremy S. Somerson 1 & Moni B. Neradilek 2 & Jason E. Hsu 3 & Benjamin C. Service 4 & Albert O. Gee 3 & Frederick A. Matsen III 3

Received: 27 February 2017 / Accepted: 10 March 2017 # SICOT aisbl 2017

Abstract Purpose Have the results of shoulder arthroplasty got better over the last two decades? To answer this question, we sought published evidence that the patient-reported outcomes and reoperation rates have improved in reports of more recently performed anatomic (TSA) and reverse (RSA) total shoulder arthroplasties. Methods We analyzed the arthroplasty results among studies published from 1990 to 2015, adjusting for the fact that the different publications presented patient groups with different

combinations of diagnoses, used various outcome scales, and had different lengths of follow-up. Results The adjusted clinical outcomes (p = 0.048), but not the revision rates (p = 0.3), were significantly better for articles reporting more recent TSA procedures. Neither the clinical outcomes (p = 0.9) nor the revision rates (p = 0.4) were significantly better in articles reporting more recent RSA surgeries. Conclusions Better evidence from reports with greater detail will be necessary to show that patients are realizing progressively better outcomes from shoulder arthroplasty. Level of evidence Level IV

* Frederick A. Matsen, III [email protected]

Keywords Shoulder arthroplasty outcomes . Revision rates . Effect of surgical date . Published evidence

Jeremy S. Somerson [email protected] Moni B. Neradilek [email protected] Jason E. Hsu [email protected] Benjamin C. Service [email protected] Albert O. Gee [email protected] 1

The University of Texas Medical Branch, 301 University Boulevard, Galveston, TX 77555, USA

2

The Mountain-Whisper-Light Statistics, 1827 23rd Avenue East, Seattle, WA 98112, USA

3

Department of Orthopaedics and Sports Medicine, University of Washington Medical Center, 1959 NE Pacific Street, Box 356500, Seattle, WA 98195-6500, USA

4

Orthopaedic Surgery and Sports Medicine, Orlando Health, 1222 S. Orange Ave., 5th floor, Orlando, FL 32806, USA

Introduction Shoulder arthroplasty improves quality of life [1–3] with favourable cost-effectiveness and cost-utility [4–7]. However, the high cost of arthroplasty procedures has made them a focal point of efforts to reduce healthcare expenditures. In 2012, the United States Government Accountability Office reported that spending on orthopaedic implant procedures increased from US$6.1 billion in 2004 to US$9.0 billion in 2009 (8.1% per year) [8]. Of the orthopaedic procedures included in the report, the case volume for shoulder arthroplasty showed the highest rate of growth. Implants account for a substantial portion of shoulder arthroplasty expenditures: the percentage of fouryear costs attributable to prostheses was 38% (mean ± SD, US$6,643 ± 944) for TSA [9] and 54% (US$13,288 ± 2,033) for reverse total shoulder arthroplasty (RSA) [10]. The cost-effectiveness of shoulder arthroplasty is strongly influenced by implant price [11].

International Orthopaedics (SICOT)

Exponentially, more new devices are brought to market each year (Fig. 1) [12]; each prosthesis introduction bringing with it new costs for research, development, marketing, training and price increase that are added to the expenses associated with the prostheses in current use. At present there is little evidence showing that patients receiving the newer technologies realize superior clinical results in comparison to those receiving prior iterations of TSA and RSA [13–15]. In fact, joint registry data indicate that revision rates are substantially higher for some of the more recently introduced implants [16]; these increased revision rates further add to the effective cost of the new implant [17, 18]. The goal of this study was to answer the question, BIs there published evidence that the outcomes of anatomic and reverse shoulder arthroplasties improved over two decades, during which many new prostheses and techniques were introduced?^ Recognizing that this question will never be answered by randomized controlled trials, we sought to analyze the clinical outcomes in published case series of arthroplasties performed by the year in which the surgeries were performed. Understanding that such reports differ with respect to the diagnoses treated and the instruments used to assess the outcome, we developed, validated and applied an approach for adjusting for these confounders so that the reports of cases performed in different years might be compared.

Methods One of the authors (J.S.S.) performed a systematic search of the available published literature using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [19], using two non-mutually-exclusive literature search engines (SCOPUS [20] and PubMed [21, 22]) for studies reporting shoulder arthroplasty outcomes published in English between January 1, 1990 and January 31,

2015. For SCOPUS the search string was TITLE-ABS-KEY ((shoulder OR shoulder* OR glenohumeral OR humeral OR glenoid) AND (arthroplasty OR prosthes* OR arthroplast* OR replac* OR hemiarthroplast* OR prosthes* OR endoprosthes* OR resurfac*) AND (outcome OR outcomes OR result OR results)) AND PUBYEAR > 1989 AND (LIMIT-TO(LANGUAGE,BEnglish^)) AND (EXCLUDE ( S U B J A R E A , BD E N T ^) ) A N D ( E X C L U D E (DOCTYPE,Bre^). For PubMed the search string was (shoulder[mh] OR shoulder joint[mh] OR shoulder[tw] OR glenohumeral[tw] OR humeral[tw]) AND (arthroplasty[mh] OR arthroplast*[tw] OR replace*[tw] OR hemiarthroplasty [tw] OR joint prosthesis[mh] OR prosthesis[tw] OR prostheses[tw] OR endoprosthesis[tw] OR endoprostheses[tw] OR resurface*[tw]) AND (outcome[tw] OR outcomes[tw] OR result[tw] OR results[tw]). Filters: Exclude reviews, Exclude non-Non-English. Year > 1989. This investigation was registered with the International Prospective Register of Systematic Reviews (PROSPERO) as no. CRD42016038736. Conference proceedings and bibliographies without full-text publications were not included due to the need to evaluate methodology and quality in detail for the inclusion criteria. The search identified 7187 publications of which 5718 were unique. An initial evaluation eliminated publications that were non-English, published prior to 1990, or did not report unique results from clinical arthroplasty (Fig. 2). This left 910 articles for full-text appraisal. Articles were excluded on full-text appraisal because of (1) lack of twoyear minimum follow-up, (2) lack of pre-operative and post-operative scores using the same instrument, (3) lack of unique arthroplasty patients, (4) failure to identify which prostheses were implanted, (5) failure to report outcomes individually for each prosthesis when multiple implants were used, (6) case report or small case series (i.e., n < 20), (7) withdrawn study, or (8) methodology that included the use of allograft tissue. Of the 102 remaining articles, we confined our analysis to the 79 that reported TSA or RSA outcomes. The total number of patients in these studies was 6,954. Relevant data were extracted from the publications, including patient factors (mean age, percentage of patients by sex), shoulder diagnoses, treatment characteristics (percentage of surgeries by glenoid type, humerus fixation and implant model as well as the years in which the surgery was performed) and outcome measures (outcome scale used, mean preoperative clinical score, mean clinical score at the latest follow-up and either its standard deviation or standard error, the number of re-operated upon patients and the mean duration of follow-up). Outcomes

Fig. 1 The exponential increase in the number of shoulder arthroplasty 510(k) approvals by year [35]

Two types of results were recorded: the clinical outcome scores and the re-operation rates. The clinical outcome scales

International Orthopaedics (SICOT) Fig. 2 Flowchart showing selection process for articles included in this analysis

used in the analyzed articles included the American Shoulder and Elbow Surgeons (ASES, 21% of the TSA and 43% of the RSA studies), unadjusted Constant (64% TSA, 59% RSA), Disability of the Arm, Shoulder and Hand score (DASH, 5% of the TSA), the pain component of the Neer (7% TSA), Oxford Shoulder Score (14% RSA), Penn (2% TSA), Single Assessment Numeric Evaluation (SANE, 2% TSA, 11% RSA), Simple Shoulder Test (SST, 12% TSA, 19% RSA), University of California Los Angeles (UCLA, 5% TSA, 5% RSA), pain visual analog scale (VAS, 12% TSA, 32% RSA), and the Western Ontario Osteoarthritis Shoulder index (WOOS, 10% TSA, 8% RSA). There were insufficient data with any one scale to allow us to answer our primary question, BAre the clinical outcomes of shoulder arthroplasty improving with time?^ Thus, to enable comparison of studies using different outcome scales, we developed a method for normalizing the pre-operative and follow-up mean values for each

scale to a 0–100% score, with 0% corresponding to the worst and 100% to the best possible value for each outcome scale. The rationale for this normalization is that for each scale, the highest score represents the Bbest possible shoulder^ (i.e., a 100% shoulder), while the lowest score represents the Bworst possible shoulder^ (i.e., a 0% shoulder). We validated this normalization method using data from studies that provided scores from different outcome scales to assess the correlation among the normalized values for these different scales. The standard deviations (SDs) and standard errors of the means (SEs) were also transformed to the normalized score. When neither SDs nor SEs were reported, we imputed them from other studies using the same scale. When values were reported for multiple scales in the same study, their normalized means and SDs were averaged to yield a single mean and SD per study. The SE for the mean was then calculated as SD/√N.

International Orthopaedics (SICOT)

The clinical outcomes were analyzed in two forms: (1) the normalized mean post-operative scores and (2) the percent of maximum possible improvement (%MPI) calculated as [23–26]: ðFollow−up mean% – Preoperative mean%Þ

.

ð100% – Preoperative mean%Þ

The advantage of the analysis of the normalized postoperative scores was that it accounted for the precision (i.e., the SE) of the values. This was not possible for the %MPI outcome for which no standard errors were available. The advantage of the %MPI analysis was that it represented the amount of the patient self-assessed pre-operative functional deficit that was regained after surgery. We determined the annualized re-operation rate in each study, calculated as the percent of all re-operated cases divided by the average duration of follow-up in years. The standard errors were estimated by dividing the normal approximation to the standard error of the percent of re-operated cases [p × (1 - p)/sqrt(N)], where p is the proportion of cases reoperated and N is the sample size divided by the average duration of follow-up. In case p equaled zero, continuity correction was applied to the SE calculation (i.e., p was set to 0.5/ N rather than to 0). Statistical methods Descriptive statistics for the characteristics of the included studies were summarized as mean ± SD (range) for continuous characteristics and as percentages for categorical characteristics. The follow-up scores and the re-operation rates were meta-analyzed using multivariable mixed-effects models with moderators implemented in the meta-analytic R package metafor. The models for the post-operative score were adjusted for the pre-operative means and the particular outcome scales used as well as for the diagnosis. The models for the re-operation rates were adjusted for the average follow-up duration and the diagnosis. The meta-analysis for the %MPI outcome was carried out using linear regression All calculations were carried out using R version 3.2.0 (R Foundation for Statistical Computing, Vienna, Austria). A p value of <0.05 was statistically significant.

Results There were 42 TSA studies with a mean (± SD) of 116 ± 159 (range, 20–705) patients per study with an average follow-up of five ± three years, including 42 ± 21% males with average age 66 ± five years. Sixty-seven percent of the studies were retrospective, 19% prospective, 10% prospective randomized, 2% case controlled, and 2% retrospective review. The levels of evidence were: I, 12%; II, 5%: III, 23%; IV, 60%. The

prostheses used were Tornier Aequalis, 45%; DePuy Global, 24%; Cofield, 9%; Bigliani/Flatow, 9%; Biomet Biomodular, 6%; Mathys Affinis, 6%. There were no statistically significant differences in outcome among the studies of different quality or among papers reporting on different prostheses. There were 37 RSA studies with 56 ± 32 (20–174) patients per study with an average follow-up of four± five years, including 34 ± 15% males with average age 72 ± five years. Seventy-six percent of the studies were retrospective, 21% prospective, and 3% prospective randomized. The levels of evidence were: I, 3%; II, 5%: III, 27%; IV, 65%. The prostheses used were: Tornier Aequalis Reverse, 29%; DePuy Delta III, 29%; DJO Reverse, 29%; Lima SMR reverse, 7%; Zimmer TM reverse, 7%. There were no statistically significant differences in outcome among the studies of different quality or among papers reporting on different prostheses. To evaluate the validity of normalizing scores obtained from different outcome scales, we assessed the correlation between the normalized score values when two or more outcome scales were used in the same study. Comparison between two pairs of scales for TSA and three pairs of scales for RSA could be evaluated using the Spearman correlation (R). All five comparisons showed very strong correlations between the scales: (1) TSA: ASES versus VAS, R = 0.98 (four studies); (2) TSA: ASES versus WOOS, R = 0.93 (four studies); (3) RSA: ASES versus SST, R = 0.91 (six studies); (4) RSA: ASES versus VAS, R = 0.89, (ten studies); (5) RSA: ASES versus Constant, R = 0.94 (three studies). The particular outcome scale used in the different publications had a major effect on the normalized outcome score for shoulder arthroplasty (Fig. 3). For TSA, the mean normalized post-operative scores were ≥80% for reports using the WOOS, UCLA, Penn, and SANE scales and <65% using the DASH or Constant scales. For RSA, the mean normalized postoperative scores were ≥70% for the Oxford, ASES and VAS Pain scales and ≤60% for the SANE and UCLA scales. The distribution of diagnoses represented in the different studies varied widely (Table 1). For reports of TSA, osteoarthritis was most common diagnosis (73% for an average study, but the percentage of patients with this diagnosis varied greatly from study to study (0–100%). For reports of RSA, the most common diagnosis was cuff tear arthropathy (48% for an average study) but, again, the percentage of patients with this diagnosis varied greatly from study to study (0-100%). For both types of arthroplasty, the diagnosis had a significant effect on the outcome (Tables 2 and 3, Fig. 4). For TSA the results were worse in those studies of rheumatoid arthritis and cuff tear arthropathy. Clinical outcomes for RSA were worse in studies of post-traumatic arthritis. For TSA, the adjusted clinical outcomes but not the revision rates were worse for male patients; age did not have a significant effect on the clinical outcomes or revision rate

International Orthopaedics (SICOT)

normalized outcomes for those articles using the DASH scale. For articles on reverse total shoulders using the Oxford scale, the normalized outcomes were substantially higher than the normalized outcomes for those articles using the SANE scale. For both anatomic and reverse total shoulders, the mean normalized Simple Shoulder Test outcome scales were approximately in the middle of the mean scores for the other scales

Fig. 3 The effect of outcome scale used on (a) the normalized reported post-operative score adjusted for the pre-operative score and (b) the improvement expressed as a percent of the maximal possible improvement for both anatomic and reverse total shoulder arthroplasties. Note the wide variations in the normalized outcomes for articles using the different scales. For articles on anatomic total shoulders using the WOOS scale, the normalized outcomes were well over twice the Table 1 Diagnoses represented in the 79 studies used in the metaanalysis of 42 TSA and 37 RSA publications (4,868 and 2,086 patients, respectively)

TSAs N Diagnosis, % per study % CTA % Mixed diagnoses % OA % OA eccentric wear % PHF % PTA % RA % RCT

RSAs Mean ± SD (range)

42

N

Mean ± SD (range)

37 2 ± 15% (0–100%) 19 ± 40% (0–100%) 73 ± 44% (0–100%) 1 ± 7% (0–43%) – – 5 ± 22% (0–100%) –

48 ± 49% (0–100%) 35 ± 48% (0–100%) <0.5 ± 2% (0–15%) – 3 ± 16% (0–100%) 7 ± 24% (0–100%) 3 ± 16% (0–100%) 4 ± 18% (0–100%)

Surgical cases included in this analysis were performed between 1980 and 2012, with a mean study mid-range surgery year of 2002 (range, 1,988–2,009) CTA cuff tear arthropathy, OA osteoarthritis, PHF proximal humeral fracture, PTA post-traumatic arthritis, RA rheumatoid arthritis RCT rotator cuff tear

International Orthopaedics (SICOT) Table 2 Association between study characteristics and TSA outcomes adjusted for the preoperative mean score and outcome scale

Post-operative score

%MPI

Re-operation rate

Coef. (95% CI)

p

Coef.

Coef. (95% CI)

p

% male, per 10%

-2.0 (-3.2, -0.8)

0.001

-2.4

1.0

Average age, per 10 years

3.4 (-0.8, 7.6)

0.12

3.9

-0.01 (-0.23, 0.22) -0.36 (-0.96, 0.25)

0.2

Diagnosis % OA, per 10%

0.0 ref.

–

0.0

0.00 ref.

–

% mixed, per 10%

0.1 (-0.4, 0.7)

0.6

0.2

0.12 (0.03, 0.20)

0.01

% CTA, per 10% % RA, per 10%

-1.4 (-2.6, -0.1) -2.01 (-2.9, -1.0) 1

0.03 <0.0011

–1.6 -2.51

<0.001 0.7

% OA eccentric wear, per 10%

-3.0 (-6.3, 0.2)

0.07

–3.7

Surgery mid-range year, per 10 years

4.1 (0.0, 8.2)

0.048

5.3

0.63 (0.28, 0.97) -0.02 (-0.15, 0.11) -0.02 (-0.37, 0.33) -0.48 (-1.35, 0.39)

0.9 0.3

The effects (coefficients) for diagnosis are presented for 10% increase (corresponding to a 10% increase in the representation of a given group per study compared with a reference group) N = 42 studies for post-operative score and %MPI, except N = 35 (gender), 40 (age) and 39 (surgery mid-range year). N = 29–33 studies for re-operation rate. % MPI percent maximum possible improvement Re-operation rates were annualized. Random effects meta-regressions were used for the post-operative score and the re-operation rate. Linear regression for % MPI. Models for post-operative score and %MPI were adjusted for the pre-operative mean, the scale and the diagnosis. Models for re-operation rate were adjusted for the average duration of follow-up and the diagnosis

(Table 2). For RSA neither patient sex nor age had a significant effect on outcome (Table 3). Over the two decades of this study, there were significantly better clinical outcomes in more recently reported TSA studies (p = 0.048) (Fig. 5), but this was not the Table 3 Association between study characteristics and RSA outcomes adjusted for the pre-operative mean score and outcome scale

case for RSA (Fig. 6). Neither (Table 2) or the revision rate significantly lower in studies during these two decades, respectively. Post-operative score

% male, per 10% Average age, per 10 years Diagnosis % CTA, per 10% % OA, per 10% % mixed, per 10% % RA, per 10% % PTA, per 10% % RCT, per 10% % PHF, per 10% Surgery mid-range year, per 10 years

the revision rate for TSA for RSA (Table 3) were published more recently p = 0.3 and p = 0.4,

%MPI

Re-operation rate

Coef. (95% CI)

p

Coef.

Coef. (95% CI)

p

-0.3 (-2.0, 1.5) 1.8 (-1.8, 5.5)

0.8 0.3

–0.3 2.8

0.17 (-0.25, 0.58) -1.06 (-2.24, 0.12)

0.4 0.08

0.0 ref. 2.6 (-3.5, 8.7) -0.2 (-0.6, 0.2) 0.6 (-0.5, 1.6) -0.9 (-1.6, -0.2) -0.4 (-1.4, 0.5) 0.0 (-1.0, 1.0) -0.4 (-6.3, 5.5)

– 0.4 0.3 0.3 0.009 0.4 0.9 0.9

– 3.7 –0.3 0.6 –1.3 –0.6 0.1 –0.8

0.00 ref. - (-) 0.01 (-0.11, 0.12) 0.34 (-0.04, 0.72) 0.02 (-0.21, 0.24) -0.06 (-0.33, 0.21) 0.32 (-0.02, 0.65) -0.84 (-2.87, 1.19)

– – 0.9 0.08 0.9 0.7 0.06 0.4

The effects (coefficients) for diagnosis are presented for 10% increase (corresponding to a 10% increase in the representation of a given group per study compared with a reference group) N = 37 studies for post-operative score and %MPI, except % MPI = 32 (gender), 35 (age) and 35 (surgery midrange year). N = 27–29 studies for reoperation rate. % MPI percent maximum possible improvement Reoperation rates were annualized. Random effects meta-regressions were used for the postoperative score and the reoperation rate. Linear regression for % MPI. Models for postoperative score and %MPI were adjusted for the preoperative mean, the scale and the diagnosis. Models for reoperation rate were adjusted for the average duration of follow-up and the diagnosis

International Orthopaedics (SICOT) Fig. 4 The effect of diagnosis on (a) the normalized reported post-operative score adjusted for the pre-operative score and (b) the improvement expressed as a percent of the maximal possible improvement for both anatomic and reverse total shoulder arthroplasties. Note that for anatomic total shoulders, the diagnoses of rheumatoid arthritis and cuff tear arthropathy had lower normalized outcome scores than the other diagnoses. For reverse total shoulders, rheumatoid arthritis had the highest normalized outcome scores while post-traumatic arthritis had the lowest scores

Discussion In an attempt to improve the outcomes of shoulder arthroplasty, new prostheses and techniques are being introduced annually, each with increased costs associated with

research and development, patenting, U.S. Food and Drug Administration (FDA) approval, training and marketing. The justification for an incrementally more expensive implant or technique should be made in terms of a demonstrated incremental benefit for the patients receiving it. Currently,

Fig. 5 The adjusted mean clinical outcomes scores (vertical axis) for studies reporting clinical outcomes of total shoulder arthroplasty as a function of the midrange year of surgery (horizontal axis). The graph

on the left shows the post-operative score while the graph on the right shows the improvement expressed as a percent of the maximum possible improvement. The linear regression lines and their p values are shown

International Orthopaedics (SICOT)

Fig. 6 The adjusted mean clinical outcome scores (vertical axis) for studies reporting clinical outcomes of reverse total shoulder arthroplasty as a function of the midrange year of surgery (horizontal axis). The graph

on the left shows the post-operative score, while the graph on the right shows the improvement expressed as a percent of the maximum possible improvement. The linear regression lines and their p values are shown

however, it is unclear whether the clinical outcomes of shoulder arthroplasty have improved over the last two decades, during which many new implants and techniques have been introduced. The purpose of this analysis was to seek evidence from the available literature that patients having shoulder arthroplasty are benefitting from more recently introduced innovations in shoulder arthroplasty. Because of the absence of relevant controlled trials comparing outcomes among patients having surgery in different years, this analysis had to rely on comparing outcomes among publications reporting surgery performed in different years. These comparisons required adjustments for several confounders. Most notably, the specific outcome scale (e.g., ASES, SST, WOOS, etc.) used to assess the clinical outcome had a major effect even when the scores on the outcome scale were normalized to a 0–100% scale. The reasons for this effect are not clear. However, this result emphasizes the importance of consistency in outcome scales if the results of different studies are to be compared. The effects of patient age, sex, diagnosis, and prosthesis used are also potential confounders in the analysis of clinical outcomes and revision rates [16, 23, 27–29]. Shoulder arthroplasty outcome reports that include the age and sex, diagnosis, and prosthesis for each patient will improve the validity of comparisons among these reports. Finally, it is recognized that the clinical outcomes and revision rates change with time after surgery [30]. Thus, in comparing publications of shoulder arthroplasty outcomes, the mean duration of post surgical follow-up needs to be carefully accounted for along with some measure of the variability or uncertainty of the outcomes (standard deviation, standard error or confidence interval). The results of this study should be viewed in light of certain limitations. First, we recognize that the robust comparison of data from different published reports is challenging. This study represents our best effort to answer the important question, BIs there published evidence that the outcomes of anatomic and reverse

shoulder arthroplasties improved over twodecades during which many new prostheses and techniques were introduced?^ It seems unlikely that controlled studies bearing on this question will become available in the foreseeable future. Thus, because we believe the question posed is of great importance to orthopaedics, we have attempted a methodologically rigorous approach to the data that are available: a systematic study comparing the results of studies of surgery performed during different years. Second, we identified substantial variability in outcome scale and diagnosis among the studies analyzed. We found that these variables had a major effect on the results reported and made adjustments for these effects. With respect to the outcome scales, each measures a somewhat different aspect of the condition of the shoulder so that they are not exactly comparable; the logic for comparing outcome scale values normalized to a 0-100% score is that each scale is a reflection of the patient-reported shoulder function in relationship to a ‘normal’ shoulder. While we were able to demonstrate a high degree of correlation among many of these scales, the scale used in the publication still had a substantial effect on the normalized outcome and was therefore used as a covariate in the analysis. Third, while we adjusted the reoperation rate according to the time of follow-up, we recognize that re-operation rates are not a linear function of time after surgery, but rather begin to increase dramatically five or more years after arthroplasty [31, 32]. Fourth, a proper evaluation of the value of an implant should include both its clinical outcomes and its actual selling price. Our attempts to relate average selling price to outcome were frustrated by the substantial differences between the list prices for the implants and their average selling price, and by the observation that the implant prices varied with time, tending to be higher at the time the prosthesis is introduced, and then diminishing with time to levels close to that of the prior implants [33]. These observations prevented the analysis of the incremental benefit of a new prosthesis in relation to the increment in the selling price. Finally, we recognize that the clinical

International Orthopaedics (SICOT)

outcome of shoulder arthroplasty depends on the complex interactions of many characteristics related to the patient (age, sex, overall health, socioeconomic factors), those related to the shoulder (primary and other diagnoses, prior surgeries), those related to the treatment (prosthesis, surgical technique, rehabilitation), and those related to the surgical team (training, experience, annual case volume) [28]. This analysis did not include all of these potentially important variables. We conclude that better evidence will be necessary to show that the clinical outcomes for patients with glenohumeral arthritis are improving with the application of newer shoulder arthroplasty implants and techniques. To that end, we suggest that future studies reporting the results of shoulder arthroplasty should include an appendix containing a set of basic data elements [34] for each patient so that meaningful comparisons can be facilitated. Such a data set should include for each patient the age, sex, diagnosis, the scale used to document the presurgical and postoperative patient self-assessed shoulder comfort and function, and date and reason for any reoperation. While this minimal data set will not capture the full set of potential confounders—such as the degree of shoulder stiffness, the condition of the rotator cuff, radiographic pathoanatomy and the effect of surgical team volume and experience—this degree of standardization will enable a more robust comparison of the outcomes for individual patients treated over time with different therapeutic approaches, so that we can learn whether newer implants and techniques contribute added value to the patient with glenohumeral arthritis.

3.

4.

5.

6.

7.

8.

9.

10.

Compliance with ethical standards Ethical approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. For this type of study formal consent was not required. Funding There was no extramural funding for this investigation. Conflict of interest Outside of this submitted work, Dr. F.A. Matsen has received royalties from the Elsevier Publishing Company for The Shoulder, 4th edition by C.A. Rockwood Jr and F.A. Matsen III; Dr. J.S. Somerson is receiving direct payment from Springer Publishers through a consulting agreement. The other four authors declare that they have no conflicts of interest.

11.

12.

13.

14.

15.

References 16. 1.

2.

Boorman RS, Kopjar B, Fehringer E, Churchill RS, Smith K, Matsen FA III (2003) The effect of total shoulder arthroplasty on self-assessed health status is comparable to that of total hip arthroplasty and coronary artery bypassgrafting. J Shoulder Elbow Surg 12(2):158–163. doi:10.1067/mse.2003.18 Castricini R, Gasparini G, Di Luggo F, De Benedetto M, De Gori M, Galasso O (2013) Health-related quality of life and functionality

17.

after reverse shoulder arthroplasty. J Shoulder Elbow Surg 22(12): 1639–1649. doi:10.1016/j.jse.2013.01.020 Collins DN, Harryman DT, Wirth MA (2004) Shoulder arthroplasty for the treatment of inflammatory arthritis. J Bone Joint Surg Am 86(11):2489–2496 Coe MP, Greiwe RM, Joshi R, Snyder BM, Simpson L, Tosteson ANA, Ahmad CS, Levine WN, Bell J-E (2012) The costeffectiveness of reverse total shoulder arthroplasty compared with hemiarthroplasty for rotator cuff tear arthropathy. J Shoulder Elbow Surg 21(10):1278–1288. doi:10.1016/j.jse.2011.10.010 Mather RC III, Watters TS, Orlando LA, Bolognesi MP, Moorman CT III (2010) Cost effectiveness analysis of hemiarthroplasty and total shoulder arthroplasty. J Shoulder Elbow Surg 19(3):325–334. doi:10.1016/j.jse.2009.11.057 Renfree KJ, Hattrup SJ, Chang Y-HH (2013) Cost utility analysis of reverse total shoulder arthroplasty. J Shoulder Elbow Surg 22(12): 1656–1661. doi:10.1016/j.jse.2013.08.002 Steen BM, Cabezas AF, Santoni BG, Hussey MM, Cusick MC, Kumar AG, Frankle MA (2015) Outcome and value of reverse shoulder arthroplasty for treatment of glenohumeral osteoarthritis: a matched cohort. J Shoulder Elbow Surg 24(9):1433–1441. doi:10. 1016/j.jse.2015.01.005 GAO U.S. Government Accountability Office (2012) Medicare: lack of price transparency may hamper hospitals’ ability to be prudent purchasers of implantable medical devices. http://www.gao. gov/products/GAO-12-126. Accessed January 10 2016 Virani NA, Williams CD, Clark R, Polikandriotis J, Downes KL, Frankle MA (2013) Preparing for the bundled-payment initiative: the cost and clinical outcomes of total shoulder arthroplasty for the surgical treatment of glenohumeral arthritis at an average 4-year follow-up. J Shoulder Elbow Surg 22(12):1601–1611. doi:10. 1016/j.jse.2012.12.028 Virani NA, Williams CD, Clark R, Polikandriotis J, Downes KL, Frankle MA (2013) Preparing for the bundled-payment initiative: the cost and clinical outcomes of reverse shoulder arthroplasty for the surgical treatment of advanced rotator cuff deficiency at an average 4-year follow-up. J Shoulder Elbow Surg 22(12):1612– 1622. doi:10.1016/j.jse.2013.01.003 Teusink MJ, Virani NA, Polikandriotis JA, Frankle MA (2012) Cost analysis in shoulder arthroplasty surgery. Adv Orthop:e692869. doi: 10.1155/2012/692869 Harmer L, Throckmorton T, Sperling JW (2016) Total shoulder arthroplasty: are the humeral components getting shorter? Curr Rev Musculoskelet Med 9(1):17–22. doi:10.1007/s12178-0169313-3 Churchill RS, Kopjar B, Fehringer EV, Boorman RS, Matsen FA (2005) Humeral component modularity may not be an important factor in the outcome of shoulder arthroplasty for glenohumeral osteoarthritis. Am J Orthop 34(4):173–176 Mileti J, Sperling JW, Cofield RH, Harrington JR, Hoskin TL (2005) Monoblock and modular total shoulder arthroplasty for osteoarthritis. J Bone Joint Surg (Br) 87(4):496–500. doi:10.1302/ 0301-620x.87b4.15558 Schoch B, Werthel J-D, Schleck C, Sperling JW, Cofield RH (2015) Does an increase in modularity improve the outcomes of total shoulder replacement? comparison across design generations. Int Orthop 39(10):2053–2060. doi:10.1007/s00264-015-2874-8 Australian Orthopaedic Association (2015) National joint replacement registry annual report 2015—shoulder arthroplasty. AOA, Adelaide Bhat SB, Lazarus M, Getz C, Williams GR Jr, Namdari S (2016) Economic decision model suggests total shoulder arthroplasty is superior to hemiarthroplasty in young patients with end-stage shoulder arthritis. Clin Orthop Relat Res 474(11):2482–2492. doi: 10.1007/s11999-016-4991-0

International Orthopaedics (SICOT) 18.

Makhni EC, Swart E, Stainhaus ME, Mather RC 3rd, Levine WN, Bach BR Jr, Romeo AA, Verma NN (2016) Arthroplasty versus arthroscopic rotator cuff repair for symptomatic large and massive rotator cuff tears. Arthroscopy 32(9):1771–1780. doi:10.1016/j. arthro.2016.01.063 19. Moher D, Liberati A, Tetzlaff J, Altman DG (2009) Preferred reporting items for systematic reviews and meta-analyses: the PRISMA StatementThe PRISMA Statement. Ann Intern Med 151(4):264–269. doi:10.7326/0003-4819-151-4-20090818000135 20. Scopus (2016) Elsevier B.V. https://www.scopus.com/. Accessed January 25 2016 21. U.S. National Library of Medicine (2016) PubMed. http://www. ncbi.nlm.nih.gov/pubmed. Accessed 25 January 2016 22. Harris JD, Quatman CE, Manring MM, Siston RA, Flanigan DC (2014) How to write a systematic review. Am J Sports Med 42(11): 2761–2768. doi:10.1177/0363546513497567 23. Gilmer BB, Comstock BA, Jette JL, Warme WJ, Jackins SE, Matsen FA III (2012) The prognosis for improvement in comfort and function after the ream-and-run arthroplasty for glenohumeral arthritis: an analysis of 176 consecutive cases. J Bone Joint Surg Am 94(14):e102. doi:10.2106/JBJS.K.00486 24. Matsen FA III, Russ SM, Vu PT, Hsu JE, Lucas RM, Comstock BA (2016) What factors are predictive of patient-reported outcomes? A prospective study of 337 shoulder arthroplasties. Clin Orthop Relat Res 474(11):2496–2510. doi:10.1007/s11999-016-4990-1 25. McElvany MD, McGoldrick E, Gee AO, Neradilek MB, Matsen FA III (2014) Rotator cuff repair: published evidence on factors associated with repair integrity and clinical outcomes. Am J Sports Med 43:491–500. doi:10.1177/0363546514529644 26. Somerson JS, Sander P, Bohsali K, Tibbetts R, Rockwood CA Jr, Wirth MA (2016) What factors are associated with clinically important improvements after shoulder hemiarthroplasty for cuff tear arthropathy? Clin Orthop Relat Res 474(12):2682–2688. doi:10. 1007/s11999-016-5037-3 27. Fevang BT, Lie SA, Havelin LI, Skredderstuen A, Furnes O (2009) Risk factors for revision after shoulder arthroplasty: 1,825 shoulder

arthroplasties from the Norwegian Arthroplasty Register. Acta Orthop 80(1):83–91 28. Matsen FA III, Li N, Gao H, Yuan S, Russ SM, Sampson PD (2015) Factors affecting length of stay, readmission, and revision after shoulder arthroplasty: A population-based study. J Bone Joint Surg Am 97(15):1255–1263. doi:10.2106/jbjs.n.01107 29. Werner BC, Burrus MT, Begho I, Gwathmey FW, Brockmeier SF (2015) Early revision within 1 year after shoulder arthroplasty: patient factors and etiology. J Shoulder Elbow Surg 24(12):e323– e330. doi:10.1016/j.jse.2015.05.035 30. Raiss P, Bruckner T, Rickert M, Walch G (2014) Longitudinal observational study of total shoulder replacements with cement: fifteen to twenty-year follow-up. J Bone Joint Surg Am 96(3):198– 205. doi:10.2106/JBJS.M.00079 31. Buckingham BP, Parsons IM, Campbell B, Titelman RM, Smith KL, Matsen FA III (2005) Patient functional self-assessment in late glenoid component failure at three to eleven years after total shoulder arthroplasty. J Shoulder Eblow Surg 14(4):368–374. doi:10. 1016/j.jse.2004.10.008 32. Walch G, Young AA, Boileau P, Loew M, Gazielly D, Mole D (2012) Patterns of loosening of polyethylene keeled glenoid components after shoulder arthroplasty for primary osteoarthritis: results of a multicenter study with more than five years of followup. J Bone Joint Surg Am 94(2):145–150. doi:10.2106/JBJS.J. 00699 33. Mendenhall Associates, Inc (2016) Orthopedic network news. https://www.orthopedicnetworknews.com. Accessed January 25 2016 34. Rasmussen JV, Brorson S, Hallan G, Dale H, Aarimaa V, Mokka J, Jensen SL, Fenstad AM, Salomonsson B (2016) Is it feasible to merge data from national shoulder registries? a new collaboration within the Nordic Arthroplasty Register Association. J Shoulder Eblow Surg 25(12):e369–e377. doi:10.1016/j.jse.2016.02.034 35. U.S. Food and Drug Administration: Releasable 510(K) Database (2016) U.S. Food and Drug Administration, U.S. Department of Health and Human Services. http://www.fda.gov/MedicalDevices/ ProductsandMedicalProcedures/DeviceApprovalsandClearances/ 510kClearances/ucm089319.htm. Accessed 25 January 2016