Anatomical versus reverse total shoulder arthroplasty for primary glenohumeral osteoarthritis with an intact rotator cuff Authors: Paras P. Shah, Rashini Jayawardena, Ryan Wong, Charles Poff, Nicholas H. Maassen, Lewis Shi Published: 2026 ABSTRACT AimsAnatomical total shoulder arthroplasty (aTSA) has been the traditional treatment for glenohumeral osteoarthritis (OA) with an intact rotator cuff. However, reverse total shoulder arthroplasty (rTSA) is now being more frequently performed, with its utility in consistent fixation and severe bony deformity. This study systematically reviews available literature comparing aTSA and rTSA in patients with OA with an intact rotator cuff.MethodsA literature review was conducted to identify studies comparing aTSA and rTSA in patients with glenohumeral OA and intact rotator cuffs. Primary postoperative outcomes were analyzed, including range of motion, patient-reported outcomes, complications, revisions, cost, radiological changes, and satisfaction.ResultsIn all, 18 articles met the inclusion criteria. aTSA demonstrated better range of motion with no difference in American Shoulder and Elbow Surgeons scores and pain outcomes. While complication and revision reasons varied slightly, there was no difference in overall complication and revision rates. Of four studies reporting patient satisfaction, two favoured aTSA.ConclusionIn our population, aTSA and rTSA yield comparable postoperative outcomes. While aTSA may offer a superior range of motion, its clinical value remains uncertain. These findings support the growing consideration of rTSA as a viable alternative to aTSA in this population, although longer follow-up time is required.Cite this article: Bone Jt Open 2026;7(8):1056–1063. Introduction Primary glenohumeral osteoarthritis (OA) is characterized by progressive degeneration of the glenohumeral joint, resulting in functional limitation, stiffness, and pain.1 OA of the shoulder is increasing in prevalence with the ageing population in the USA. Anatomical total shoulder arthroplasty (aTSA) and reverse total shoulder arthroplasty (rTSA) are both common operations used in the treatment of glenohumeral OA, and the incidence of TSA is expected to increase nine-fold in the general population (to 184.8 cases per 100,000 people) and three-fold in patients aged younger than 55 years (to 10.1 cases per 100,000 people) in the USA by 2030.2 Though rTSA has traditionally been considered a treatment for glenohumeral OA in a cuff-deficient shoulder, both rTSA and aTSA have been described in the treatment of primary glenohumeral OA with an intact rotator cuff. The indications for rTSA do extend beyond rotator cuff deficient OA and include a variety of degenerative and traumatic conditions, including rotator cuff tear arthropathy, complex shoulder trauma, and glenoid bone loss.3 Glenoid morphology also represents an important factor in implant choice, as excessive retroversion (> 15° to 25°) has been shown to compromise aTSA outcomes and favour rTSA. The incidence of rTSA has increased in the last two decades.4 From 2011 to 2020, the proportion of rTSA accounting for total TSA use increased from 31.4% to 74.9%.5 The aim of this review is to characterize outcomes for aTSA and rTSA in the treatment of primary glenohumeral OA in the setting of an intact rotator cuff. The authors aimed to systematically review the literature in order to assess postoperative patient reported outcomes (PROs), range of motion (ROM), pain scores, complication rates, revision rates, cost, radiological outcomes, and patient satisfaction between the two procedures in this specified patient population. The authors hypothesized that all postoperative outcomes would be similar between the two groups besides ROM, which would be improved in the aTSA group. To this end, the authors used electronic databases to identify high-yield articles to produce a comprehensive summary of aTSA compared with rTSA in glenohumeral OA with an intact rotator cuff. Methods A literature search was conducted adhering to the PRISMA protocol guidelines. The review was not registered. A requirement for informed consent and ethical permission was waived due to the retrospective, de-identified nature of the data collection process. Literature search The literature search was conducted using electronic databases including PubMed, Cochrane Library, OVID, and Embase. The search was performed for studies that included any comparison between aTSA and rTSA for patients with an intact rotator cuff and glenohumeral OA. The search was performed using a combination of keywords and medical subject headings (MeSH) terms: (“anatomical total shoulder replacement” OR “anatomical total shoulder arthroplasty” OR “ATSA” OR “TSA”) AND (“reverse total shoulder replacement” OR “reverse total shoulder arthroplasty” OR “RTSA”) AND (“osteoarthritis”) AND (“intact rotator cuff”). The time range was from January of 2015 to April of 2024. Eligibility criteria Studies were eligible for inclusion if they met the following criteria: 1) included adult patients with primary glenohumeral OA and an intact rotator cuff; 2) directly compared outcomes between primary aTSA and rTSA; 3) reported at least one postoperative outcome of interest, including ROM, PRO measures, pain scores, complication rates, revision rates, radiological outcomes, cost, and patient satisfaction; and 4) were original full-text research articles. Randomized controlled trials and observational comparative studies, including prospective and retrospective cohort studies, registry analyses, and matched or propensity score adjusted studies, were eligible for inclusion. Studies were excluded if they: 1) involved shoulder arthroplasty performed for indications other than primary OA (including acute fracture, rotator cuff tear arthropathy, inflammatory arthropathy, or tumour); 2) focused on revision arthroplasty; 3) did not allow for separation of outcomes for patients with intact rotator cuffs; 4) lacked a comparative aTSA and rTSA cohort; or 5) were case reports, review articles, editorials, letters, conference abstracts, or non-peer-reviewed publications. A flow diagram further depicting our inclusion and exclusion criteria is shown in Figure 1. PRISMA protocol. aTSA, anatomical total shoulder arthroplasty; rTSA, reverse total shoulder arthroplasty. Study selection Study selection was performed independently by two authors (PS, RJ) in a two-stage process, consisting of title and abstract screening followed by full-text review. Discrepancies were resolved through discussion, with adjudication by a senior author when necessary. The level of evidence for each included study was assigned independently by two reviewers using the Oxford Centre for Evidence-Based Medicine’s Levels of Evidence framework (2011).6 Discrepancies were resolved by consensus discussion. In brief, randomized controlled trials were classified as level 1, prospective comparative studies as level 2, retrospective comparative cohort or case control studies as level 3, and case series as level 4, although all the studies ultimately included were level 3. Results Summary of included publications In all, 18 articles met our review’s criteria, outlined in Figure 1, and were included in the analysis. Two publications assessed revision rates only,1,7 while two focused on cost only.8,9 The remaining publications featured more than one primary outcome in their analysis. This is summarized in Table I, along with follow-up times, which varied. Table I.Study characteristics.StudyProcedure and no. of patientsAge, yrsFollow-up timeOutcomes assesedFriedman et al10aTSA 370rTSA 37073 (mean)≥ 2 years41 months (mean)SST, UCLA, ASES, Constant score, SPADI, SAS, ROMHaritinian et al11rTSA 12aTSA 39rTSA 71 (SD 11)aTSA 68 (SD 7.5)≥ 2 yearsROM, Constant–Murley score and strength in external rotation, SSVKarimi et al12rTSA 57aTSA 59rTSA 72.0 (SD 8.2)aTSA 65.4 (SD 10.6)≥ 1 yearROM, ASES, SSV, VASMahylis et al13rTSA 149aTSA 187rTSA 70.9 (SD 7.0)aTSA 66.3 (SD 7.7)rTSA 40.6 months (SD 22.9)aTSA 62.0 months (SD 37.8)ROM, Constant score, UCLA, SPADI, SASMarigi et al14rTSA 367aTSA 367rTSA 72.9 (SD 7.4)aTSA 71.5 (SD 5.4)≥ 2 yearsROM, ASES, Constant score, SST, SASNazzal et al15+RCrTSA 24−RCrTSA 69TSA 93+RCrTSA 71.1−RCrTSA 72.4TSA 66.0≥ 1 year+RCrTSA 16.2 months−RCrTSA 19.7 monthsVAS, ASES, SSV, Active ROMOeding et al8rTSA 10,680TSA 6,16572.2 (mean)≥ 1 year4.3 years (mean)QALY, ICEROrvets et al7rTSA 685TSA 3,106≥ 7075.8 (SD 4.6)Date of death, date of membership termination, or study end dateCumulative revision probability by treatmentHones et al16Non-stiff aTSA 85Stiff aTSA 85Non-stiff rTSA 74Stiff rTSA 74Non-stiff aTSA 65.7 (SD 7.1)Stiff aTSA 65.0 (SD 7.2)Non-stiff rTSA 71.4 (SD 6.3)Stiff rTSA 71.7 (SD 7.0)≥ 2 yearsROM, strength of shoulder, SST, Constant score, ASES, SPADI, UCLATurnbull et al17TSA/TSA 46rTSA/TSA 9TSA/TSA 67 (SD 7)rTSA/TSA 71 (SD 4)≥ 2 yearsSPADI, SST, ASES, UCLA, Constant score, AROMPolisetty et al9rTSA 63TSA 252rTSA 74 (SD 7.1)TSA 73 (SD 5.8)≥ 2 yearsACER, ICER, SST, ASES, VAS, SANEKim18rTSA 26TSA 41rTSA 75 (SD 5)TSA 76 (SD 4)≥ 2 yearsASES, Constant–Murley score, AROMMerolla et al3TSA 47rTSA 36TSA 70rTSA 74≥ 2 yearsROM, VASWright et al19rTSA 33TSA 102rTSA 78TSA 77≥ 2 years85 months (mean)Patient reported outcomes, VAS, ASES, WOOS, ROMGill et al1TSA 3,795rTSA 8,878TSA 69 (SD 8)rTSA 74 (SD 8)TSA 2.6 (SD 1.4) yearsrTSA 2.1 (SD 1.4) yearsASES, SST, ROMTrammell et al20aTSA 85rTSA 74 (stiff cohorts)Matched cohorts≥ 2 yearsROM, SPADI, SST, ASES, UCLA, Constant score, revisionsMenendez et al21aTSA 2,693rTSA 1,758aTSA 65.1 (SD 8.9)rTSA 70.9 (SD 8.6)5 yearsASES, SANE, VAS, ROMKirsch et al22aTSA 67rTSA 67aTSA 67 (SD 4.7)rTSA 67 (SD 3.5)30 months (mean)ASES, SANE, VAS, ROM, complicationsASES, American Shoulder and Elbow Surgeons; aTSA, anatomical total shoulder arthroplasty; ICER, incremental cost-effectiveness ratio; QALY, quality-adjusted life years; +RC, with rotator cuff; −RC, rotator cuff deficient; ROM, range of motion; rTSA, reverse total shoulder arthroplasty; SANE, Single Assessment Numeric Evaluation; SAS, Shoulder Arthroplasty Smart ; SPADI, Shoulder Pain and Disability Index; SST, Simple Shoulder Test; SSV, Subjective Shoulder Value; UCLA, University of California, Los Angeles; VAS, visual analogue scale; WOOS, Western Ontario Osteoarthritis of the Shoulder. Study characteristics. ASES, American Shoulder and Elbow Surgeons; aTSA, anatomical total shoulder arthroplasty; ICER, incremental cost-effectiveness ratio; QALY, quality-adjusted life years; +RC, with rotator cuff; −RC, rotator cuff deficient; ROM, range of motion; rTSA, reverse total shoulder arthroplasty; SANE, Single Assessment Numeric Evaluation; SAS, Shoulder Arthroplasty Smart ; SPADI, Shoulder Pain and Disability Index; SST, Simple Shoulder Test; SSV, Subjective Shoulder Value; UCLA, University of California, Los Angeles; VAS, visual analogue scale; WOOS, Western Ontario Osteoarthritis of the Shoulder. The review’s primary outcomes were postoperative PROs, ROM, pain scores, complication rates, revision rates, cost, radiological outcomes, and patient satisfaction. The Simple Shoulder Test (SST),23 University of California, Los Angeles (UCLA) score,24 American Shoulder and Elbow Surgeons (ASES) score,25 Constant score, Western Ontario Osteoarthritis of the Shoulder score,26 Shoulder Arthroplasty Smart (SAS) score,27 and Subjective Shoulder Value28 accounted for the PROs across included articles. The visual analogue scale (VAS)29 and Shoulder Pain and Disability Index (SPADI) scores30 were used to assess postoperative pain. Active and passive ROM were included in the analysis. Incremental cost-effectiveness ratios (ICERs) and average cost-effectiveness ratios were incorporated to assess cost. Single Assessment Numeric Evaluation (SANE), the Mann-Whitney questionnaire, Anchor Satisfaction, and quality-adjusted life years (QALYs) were used to quantify patient satisfaction. Patient reported outcomes/range of motion Ten publications featured data on PROs and ROM. Nine out of those ten studies included ASES in their analysis and therefore this was used as a surrogate for functional outcomes across studies. All nine studies found no significant difference in ASES score at multiple timepoints during postoperative follow-up.10-15,18,19,21,22 SST scores were reported by three studies, and none of these analyses demonstrated a significant difference at any point in follow-up between aTSA and rTSA. SAS was reported in three articles. In all of these, SAS was found to be significantly improved in aTSA cohorts relative to rTSA cohorts (p < 0.001 to 0.017).10,13,14 Related to this finding, all ten studies reported greater passive/active internal and external rotation in aTSA cohorts versus rTSA cohorts postoperatively.10-15,18,19,21,22 Three studies reported a significantly higher increase in aTSA forward elevation compared with rTSA.15,16,22 In contrast to these findings, Trammell et al20 compared outcomes in patients with limited preoperative forward elevation (≤ 105°). In this ‘stiff’ cohort, rTSA demonstrated significantly greater postoperative active and passive forward elevation and active abduction compared with aTSA. Furthermore, this study found superior PROs (SPADI, SST, ASES, UCLA, and Constant scores) in the rTSA group compared with the aTSA group, contradicting the equivalent PROs found in the broader population. Of the four studies that reported glenohumeral abduction values postoperatively, only two reported a significantly higher increase in aTSA groups versus rTSA groups.13,14,16,18 Pain Postoperative pain scores were measured and reported using VAS or SPADI metrics in 11 included studies. One study found superior SPADI scores in patients who underwent rTSA compared with aTSA with minimum two years’ follow-up (p = 0.042).3 The remaining nine studies found no significant difference in pain scores between rTSA and aTSA at multiple timepoints postoperatively. Complication/revision rate Nine studies included complication rate in their analysis comparing aTSA and rTSA.7,10-12,14-16,22 The most common complications in aTSA cohorts were determined to be glenoid component loosening14,16 and rotator cuff injury.7,11,12,15 The most common complications in rTSA cohorts were infection,10,11,14,19 glenoid component loosening,7,10,16 periprosthetic fracture,7,10,16 and dislocation.7,10 There were no significant differences between aTSA and rTSA regarding total complication rates in the included articles. Seven studies compared revision rates following rTSA and aTSA. Seven of nine studies documented no difference in total revision rate between the two procedures. Marigi et al14 demonstrated a significantly higher revision rate in aTSA (4.1%) than in rTSA (0.5%) (p = 0.002). Similarly, Trammell et al20 reported a significantly higher rate of revision surgery in stiff patients treated with aTSA compared with those treated with rTSA (p = 0.007). Gill et al1 did find a significantly higher revision rate of rTSA compared with aTSA within the first three months postoperatively in men only; however, this increase did not persist after three months. In the same study, the authors found a higher revision rate for women undergoing aTSA versus rTSA after three months’ follow-up, but not within the first three months of surgery.1 The most common reason for rTSA revision was glenoid component loosening, followed by dislocation, whereas the most common reasons for aTSA revision were rotator cuff injury followed by shoulder dislocation. Finally, Polisetty et al9 found no significant differences between rTSA and aTSA in the likelihood of 90 day Emergency Department visits and 90 day readmissions. Cost Two articles featured cost analyses regarding aTSA and rTSA.8,9 Oeding et al8 performed a Markov model cost-utility analysis. For a baseline 72-year-old patient, the authors published an ICER comparing rTSA with aTSA of $188,222.94/QALY. They determined aTSA to be the more cost-effective treatment for this age group. Polisetty et al9 also performed a cost analysis. Total cost was found to be similar between operations ($9,389 vs $9,829, p = 0.116). Implant cost was greater in the rTSA cohort (p < 0.001); however, operating room material cost, anaesthesia cost, and cement cost were greater in the aTSA group (p < 0.0001).9 Radiological outcomes Radiological outcomes were only assessed in four articles in this review. Friedman et al10 demonstrated no significant difference between the incidence of humeral radiolucent lines between the two procedures at two years’ follow-up. In contrast, several other studies noted radiological advantages of rTSA over aTSA. Polisetty et al9 determined that the presence of radiolucent lines was significantly greater in the aTSA cohort 30 months postoperatively. They determined radiolucency in at least one area around the all-polyethylene glenoid component in 21% of their aTSA cohort compared with no patients with evidence of radiolucency in their rTSA cohort (p < 0.0001). Similarly, Kirsch et al22 observed radiolucent lines in 14.9% of aTSA patients at minimum two years’ follow-up, whereas the rTSA cohort demonstrated no evidence of glenoid loosening. Similarly, Merolla et al3 demonstrated an increased rate of peri-implant radiolucent lines (p < 0.001) and cortical thinning (p < 0.001) in the aTSA group when compared with the rTSA group. Patient satisfaction Five studies reported compared patient satisfaction following the two procedures. All studies used different methods for measuring patient satisfaction, including the SANE score,18,22 Mann-Whitney questionnaire,11 and Anchor Satisfaction.19 Two studies reported a higher patient satisfaction metric associated with aTSA compared with rTSA,10,18 whereas the other three studies reported slightly higher patient satisfaction with rTSA that was not significant.11,19,22 Discussion Patients with glenohumeral OA with an intact rotator cuff have historically been treated with aTSA. As indications for rTSA continue to expand, it remains unclear whether aTSA offers any advantages in this population. In this study, the authors hypothesized that there would be no consensus differences in postoperative outcomes aside from ROM. This hypothesis was largely accepted based on cumulative review of the included articles. In all articles that assessed internal and external rotation, each one demonstrated significantly improved values after aTSA. This is an anticipated finding. Given that aTSA requires preservation of native rotator cuff including the subscapularis and teres minor/infraspinatus, it is expected to observe significantly increased postoperative internal and external rotation, respectively. SAS, an outcome score which considers both subjective shoulder function and objective ROM, was also found to be improved in all three of the publications which assessed it. In aggregate, these findings add to the existing body of literature to suggest that aTSA results in improved motion when compared with rTSA.13 However, preoperative stiffness represents a critical variable that may shift the advantage towards rTSA. Trammell et al20 challenged the paradigm that aTSA yields superior ROM by isolating patients with limited preoperative forward elevation (≤ 105°). In this ‘stiff’ population, rTSA provided superior overhead motion and functional scores compared with aTSA. The authors posit that in stiff shoulders, the predictable mechanics of the reverse prosthesis may overcome soft-tissue contractures more effectively than anatomical resurfacing. This suggests that while aTSA is preferable for ROM in the mobile shoulder, rTSA may be the preferred choice for regaining motion in the stiff, osteoarthritic shoulder with an intact cuff. ASES and SST scores were not statistically different between groups. These findings together certainly bring into question the clinical significance of the reported improvements in ROM in aTSA compared with rTSA. In this review, pain was measured postoperatively using SPADI or VAS scores. Whereas VAS scores stratify patients based on their level of pain alone, SPADI scores incorporate both pain and ‘subjective disability’. All but one study that reported on either of these two variables found no difference in outcomes between rTSA and aTSA postoperatively at all timepoints. The only study17 which demonstrated a decreased pain metric for rTSA in the postoperative period, reported SPADI values rather than VAS values, which does make it possible that subjective disability is what accounted for the difference between procedures rather than pain itself, as nine out of 10 other papers reported no difference in postoperative pain between procedures. These findings add to the line of reasoning that clinical outcomes may be largely equivalent between the two procedures when treating primary glenohumeral OA with an intact rotator cuff. In the seven studies that assessed total complication rates between procedures, there were no differences between aTSA and rTSA in any study.7,10-12,14-16 These results suggest that rTSA and aTSA are reliable even in elderly populations. There have been multiple studies which have demonstrated survivorships as high as 91% in rTSA at ten years.31-33 The complication profiles themselves did differ between procedures. Whereas rTSA complications ranged broadly from infection to glenoid loosening and periprosthetic fracture, aTSA complications were mostly accounted for by rotator cuff dysfunction. Previous reports have estimated the rate of secondary rotator dysfunction after aTSA to be as high as 16.8%.34 Despite increased rotator cuff dysfunction in aTSA patients, there was no significant difference in revision rates between aTSA and rTSA within the limits of this review. One important consideration is that the demographic for rTSA includes older patients with deficient rotator cuffs, and thus these patients may not outlive the need for revision surgery. Although there is no current indication for rTSA due to age alone, recent literature comments on this issue and favours the use of rTSA over aTSA in elderly populations.31,35,36 Of the 15 studies included in our review, four compared aTSA and rTSA after controlling for age or BMI.7,9,10,18 These studies provide a glimpse into a controlled setting for assessing aTSA and rTSA in patients with an intact rotator cuff. A value analysis by Polisetty et al9 found the procedures to be equivalent. Friedman et al,10 Kim,18 and Orvets et al7 reported mostly equivalent outcomes between the two procedures. Further research comparing the two operations in patients of similar ages, perhaps via a prospective and blinded study design, is needed to fully elucidate any clinically significant differences between groups. Several important limitations should be considered when interpreting the findings of this systematic review. First, there were no randomized controlled trials directly comparing aTSA and rTSA in patients with primary glenohumeral OA and an intact rotator cuff. As a result, all included comparative evidence was observational in nature, limiting causal inference and increasing susceptibility to confounding. A major concern is confounding by indication. Patient characteristics that influence implant selection, including age, functional demand, bone quality, glenoid morphology and eccentric wear, preoperative stiffness, medical comorbidity, and perceived risk of late rotator cuff failure, were inconsistently reported and variably adjusted for across studies. In particular, many included studies did not adequately characterize glenoid anatomy, including the degree of retroversion or eccentric wear, despite the known influence of these factors on the decision to perform aTSA or rTSA. Failure to account for such determinants may bias comparisons between procedures and obscure true differences in outcomes. Provider- and centre-level factors may also contribute to bias. It is possible that one procedure is preferentially performed by higher-volume shoulder arthroplasty surgeons or at specialist centres, or alternatively selected for more-complex clinical presentations. Most studies did not report surgeon experience, case volume, or institutional characteristics, limiting assessment of the direction and magnitude of any provider-related bias. The available literature is further limited by relatively short follow-up. No included study reported outcomes beyond five years, which is of particular relevance when interpreting revision and reoperation rates for both procedures. Additionally, there was substantial heterogeneity in study size, ranging from small single-centre cohorts to large national registry datasets (e.g. Gill et al1 and Oeding et al8). In the absence of a meta-analysis, each study contributes equally to the qualitative synthesis regardless of sample size, meaning that the apparent consensus across studies may not fully reflect the statistical weight or large registry-based analysis. Finally, although we feel that our criteria for study inclusion were thorough, this review does not include a meta-analysis, which would bolster the result’s clinical utility and relevance for future research. In conclusion, patients with glenohumeral OA and an intact rotator cuff have traditionally been treated with aTSA. As the pendulum has swung more in favour of rTSA in patients with rotator cuff deficiency and in trauma, it is worthwhile considering if rTSA may be as good or superior in patients with OA with an intact rotator cuff. This systematic review demonstrates that patient-reported functional outcomes, postoperative pain scores, total complication and revision rates, patient satisfaction, and cost have been shown to be similar across multiple studies assessing aTSA and rTSA in patients with an intact rotator cuff. Rotational motion remains the only postoperative outcome that is consistently shown to be different between the two procedures. It remains unclear if the difference in postoperative rotational motion amounts to clinically significant differences between the two operations. Take home message - For patients undergoing reverse total shoulder arthroplasty (rTSA) or anatomical total shoulder arthroplasty (aTSA), both aTSA and rTSA demonstrate similar overall complication rates, pain relief, and patient satisfaction. - While aTSA demonstrates consistently better rotational range of motion, the clinical significance is uncertain.