In brief
Robotic assistance is being introduced into reverse total shoulder arthroplasty, carried by its use in the knee and hip. It provides a unique and much needed opportunity to investigate the relationship between reverse total shoulder geometry and patient reported outcomes. A number of thought leaders are pioneering these efforts. The role of robotics in the clinical practice of shoulder arthroplasty is being actively explored.
At the International Congress on Shoulder and Elbow Surgery in Vancouver this September we discussed some questions about it, and a paper published since then adds another: whether randomized trials could detect a patient benefit at all [1]. The answers so far:
• In the knee, robots improve alignment. However, three meta-analyses and a masked randomized trial of 339 patients found no clinically meaningful improvement in what patients report.
• In the shoulder, the robot has been shown to be more precise in plastic glenoid models. No study has yet shown that it improves the outcomes patients experience.
• The common causes of reverse arthroplasty failure (instability, infection, aseptic loosening and fracture) may not be primarily problems of component position.
• The robot brings its own complications. For one experienced surgeon, operating time did not reach conventional levels until about the 23rd robotic case; the median US surgeon performs only 2 reverse arthroplasties a year.
• The reduction in revision needed for the robot to pay for itself may be larger than the first-year revision rate for reverse total shoulders performed without the robot.
A masked trial of a few hundred patients answered the question about the benefit the patient receives from the robot in total knee arthroplasty. A similar trial could answer the same question for the shoulder.
The questions, and the evidence that bears on each, follow.
1. Do robots improve outcomes in total knee arthroplasty?
Figure 1. Robotic total knee arthroplasty. The tracker arrays are fixed to the femur and tibia with bone pins.
Knee robotics are often offered as the precedent for shoulder robotics, so the knee evidence is the place to start. Three recent meta-analyses agree. Twelve randomized trials with 2,200 patients showed fewer alignment outliers with robotic total knee arthroplasty but no clinically important gain in function [2]. Twenty-five randomized trials with 5,614 patients showed modest and inconsistent differences in the Knee Society Score and pain, no difference in the Oxford Knee Score, and only two trials at low risk of bias [3]. Twenty prospective studies with 2,421 patients showed about 5 degrees more motion and better WOMAC scores with the robot but no difference in the Oxford Knee Score, KOOS, Forgotten Joint Score or EQ-5D; the authors judged the differences not clinically meaningful [4]. NICE allowed robot-assisted hip and knee surgery in the NHS only on condition that further evidence be collected [5].
The RACER-Knee trial, published in The Lancet in August, randomized 339 patients across ten UK hospitals and 33 surgeons [6]. Patients and assessors were masked, with sham incisions in the conventional group, and both arms had CT planning, so the trial isolated what the robot adds in the operating room. The robot came closer to the planned alignment, by 0.8 degrees on average, and added 10.5 minutes of operating time. At one year the Forgotten Joint Score differed by 1.5 points in favor of conventional surgery (95% CI −7.5 to 4.5), and the confidence interval did not reach the 12-point difference the trial was designed to detect; for pain, the Oxford Knee Score and EQ-5D, the confidence intervals did not reach previously reported minimal important differences. The robot cost more and had at most a 4% probability of being cost-effective at the thresholds tested.
Two other knee studies are instructive. In the Glasgow randomized trial of robotic versus manual unicompartmental knee arthroplasty (139 patients randomized, 104 seen at five years), patient-reported outcomes did not differ, but reintervention occurred in none of the robotic group and in six (9%) of the manual group [7]. This result is often cited for robotics, but the authors found no alignment or component-position errors in any of the six: two were revisions, one for loosening after a fall and one for pain, and four were arthroscopies for pain, a possible meniscal tear or a possible infection. The trial was unmasked. A difference in reoperations that cannot be traced to positioning does not show that precision prevents failure. The trial’s five-year cost-utility analysis found the robot cost-effective in units performing at least 300 robotic cases a year, but that result rested on one manual-group revision for presumed infection, which the authors themselves call arguably a random event; without it, the robot was not cost-effective, and it reached cost neutrality only above 900 cases a year with no consumable costs [8]. The ten-year analysis was similar: the robot was cost-saving above 100 cases a year with all reinterventions counted, but only above 800 once the infected cases were removed [9]. In the Australian registry, with more than 300,000 total knees over a decade, nine-year revision was 5.2% without computer navigation and 4.6% with it, a difference that was not statistically significant overall; a significant reduction appeared only in patients under 65 (7.8% versus 6.3%) [10]. A third of a million knees could not establish an overall difference of half a percentage point.
Outside orthopaedics, robotic inguinal hernia repair spread widely before the RIVAL pilot trial randomized 102 patients to robotic or laparoscopic repair. At 30 days there were no differences in wound events, readmissions, pain or quality of life; the robotic operations took longer (median 76 versus 41 minutes), cost more than twice as much ($3,258 versus $1,421), and frustrated the surgeons more [11].
Proponents answer that patient-reported outcome measures are too blunt to detect the benefits of precision, that mean scores hide outliers, and that the absence of evidence is not evidence of absence [12]. The first two points are fair, and they argue for measuring what happens to individual patients, such as the proportion with a poor result, rather than for adopting the technology. The third point is true, but it cannot justify adoption: the burden of showing benefit rests with the technology that adds cost.
2. How sensitive are reverse arthroplasty outcomes to precision in positioning?
Figure 2. Two possible relationships between component malposition and revision risk. Left: each degree adds risk. Right: risk rises only beyond a threshold.
3. How wide are the goal posts? What is an “outlier”?
Figure 3. How wide are the goal posts? The acceptable window for component position may be narrow or wide, and it is rarely derived from outcomes.
Accuracy studies count “outliers” against thresholds that differ between studies and are rarely derived from outcomes. The most-cited figure is from a series of failed arthroplasties: among 60 reverse arthroplasties with usable radiographs before revision, 56 (93%) had at least one glenoid measurement outside predefined limits, most often superior tilt, with the acceptable inclination window set at 10 degrees [13]. When an independent reviewer was simply asked whether the glenoid was malpositioned, the answer was yes in 54%. No shoulders that were doing well were measured. Without the same measurements in successful shoulders, we cannot tell whether malposition causes failure or is simply common. A test would be straightforward: apply identical criteria to failed and successful shoulders and see where the goal posts actually need to be.
4. In a laboratory model, robotics outperformed patient-specific guides. Do differences of a few degrees or millimeters matter clinically?
Mean deviation from plan, synthetic glenoid models [14]
Deviation from plan | Manual | Patient-specific guide | Robotic |
Version | 9° | 4° | 1° |
Inclination | 9° | 3° | 2° |
AP translation | 2 mm | 2 mm | 0.3 mm |
SI translation | 2 mm | 1 mm | 0.7 mm |
Outliers (>10° or >5 mm) | 11 | 0 | 0 |
36 preparations in synthetic glenoid models (two glenoid designs) by 6 surgeons.
Figure 4. A patient-specific guide on a plastic glenoid model (left) and a shoulder robot (right).
In plastic glenoids prepared by experienced surgeons, the robot was more accurate than freehand preparation by 7 to 8 degrees, and more accurate than patient-specific guides by 1 to 3 degrees and about a millimeter [14]. Freehand preparation produced 11 outliers; patient-specific guides and the robot produced none. If only gross outliers matter (the threshold curve in item 2), the cheaper guide already removed them in this model. This is a surrogate, demonstrated in the laboratory. Whether differences of this size change what patients experience is the question in item 2, and it remains unanswered.
5. Can the dominant failure modes be addressed by more precise transfer of a preoperative plan?
Figure 5. Causes of failure of primary reverse arthroplasty: 2,086 failures among 69,222 arthroplasties in 42 studies [15]. Each percentage is a separately pooled estimate, so they do not add to 100%. Glenoid loosening is part of aseptic loosening.
A meta-analysis of 42 studies, with 2,086 failures among 69,222 primary reverse arthroplasties, attributed failures to instability (21%), aseptic loosening (19%), infection (19%), periprosthetic fracture (11%) and pain (under 1%). Glenoid loosening accounted for 4.6% of all failures (95% CI 2.1% to 9.6%) [15]. Infection is not a positioning problem. Instability depends on soft tissues, tensioning and patient factors as well as component position.
In a CORR Clinical Faceoff, it was estimated that robotic assistance might reduce surgical failures by 25%. An alternate estimate was that it could address at most a fraction of the failures attributed to glenoid loosening, perhaps 2% to 3% of all failures, consistent with the 4.6% that glenoid loosening represents in the meta-analysis above, and only if the difference between freehand and robotic positioning matters to the patient [16]. Neither estimate is data. The difference between them is the question a study should answer.
Figure 6. Oxford Shoulder Score outcome of 12,030 reverse arthroplasties in the UK National Joint Registry. An unsatisfactory score was below 29 [17].
Revision also misses most failures. Of 12,030 reverse arthroplasties in the UK National Joint Registry with a postoperative Oxford Shoulder Score, 27% had an unsatisfactory score (below 29), and fewer than 5% of those were revised. Patients with equally poor function after an anatomic total shoulder or a hemiarthroplasty were revised two to three times as often (11% and 14%) [17]. The threshold for revising a reverse is higher, so its low revision rate partly reflects reluctance to revise rather than success. Revision rate is therefore a poor measure of whether a technology helps.
6. Much of the accuracy data concerns the position of the guide pin. Do we know the ideal position?
Figure 7. A guide pin placed in a plastic glenoid model. Much of the accuracy data concern the position of this pin.
A robot or navigation system transfers a plan faithfully. Whether the plan is the right one is a separate question. In a matched comparison of navigated with non-navigated reverse arthroplasty, the authors themselves noted that a baseplate can be placed exactly as planned while there is no gold standard for the plan [18]. Precise execution of an unvalidated plan is precision without a demonstrated target.
Planning software is now used to choose the target. In a simulation of 49 planned reverse arthroplasties, impingement-free motion depended mostly on implant selection: humeral neck-shaft angle, glenosphere size, eccentricity and lateralization; baseplate version had less consistent effects [19]. The model considered only bone and implant contact, not soft tissues, and was not linked to how the patients actually did. These are choices of what to put in, not how precisely to put it there, and whether the simulated optimum produces better shoulders is unknown.
7. Does robotic arthroplasty introduce complications of its own?
Adverse events reported with robotic arthroplasty, FDA MAUDE database
Reported event | Hip | Knee |
Surgical delay | 146 | 73 |
Unexpected robot-arm movement during bone cuts | 59 | |
Intraoperative hardware failure | 304 | |
Incorrect component placement | 95 |
Hip: 521 robotic total hip reports, 2017–2021 [20]. Knee: 204 robotic total knee reports [21]. MAUDE counts reports, not procedures, so these numbers cannot give a rate.
The FDA MAUDE database records adverse events with robotic hip arthroplasty: among 521 reports from 2017 to 2021, there were 304 intraoperative hardware failures and 95 instances of incorrect acetabular cup placement; the robot was abandoned in 13% of reports, surgery was delayed in 28% (by about 18 minutes on average), and patient injury was described in 14% [20]. MAUDE counts reports, not procedures, so these numbers cannot give a rate, but they show the kinds of harm the technology itself introduces. A second MAUDE analysis of 263 reports across robotic hip, knee and partial knee arthroplasty found that the commonest knee event was unexpected robot-arm movement during bone cuts (59 of 204 knee reports). Surgery was delayed in 99 reports, by 20 minutes on average and up to two hours; 31 cases were converted to manual surgery; and 68 patient injuries were reported, including fractures, a medial collateral ligament laceration, retained registration checkpoints and an electrical burn, 7 of which required reoperation [21]. The authors judged the technology generally safe, with the serious events related more to technique than to the device. In randomized knee trials, robotic cases were converted to conventional technique in 10% versus 2% for conventional cases, and readmissions were more frequent [3]. Tracker pins bring their own risks. A systematic review of 28 studies covering 15,004 robotic or navigated knee arthroplasties found pin-related complications in 0.95%, mostly wound problems and infections; pin-site fractures occurred in 0.16%, usually through the femoral pin site about two months after surgery, and were treated with intramedullary nailing or plating; two arterial injuries were reported [22]. These events are uncommon, but any benefit has to exceed them, and a shoulder robot needs its own pins in the scapula.
8. Does robotics improve the skills of the surgeon?
Figure 8. Does a self-driving car make the driver better?
A robot helps execute a cut or a pin placement. It does not supply judgment about indications, exposure, soft-tissue balance, or what to do when the unexpected happens. Does a self-driving car make the driver better? A surgeon who learns reverse arthroplasty with the robot may become dependent on it, and the day the robot fails or the anatomy does not match the plan, the surgeon’s own skills are what remain.
9. Could robotics help the low-volume surgeon?
Figure 9. A surgeon with a draped robot in the operating room.
This is the strongest argument made for the technology. In Medicare data from 2016 to 2018, 7,097 surgeons performed 90,318 reverse arthroplasties: a mean of 6 per surgeon per year and a median of 2. Ninety-six percent of surgeons performed fewer than 29 a year, and their patients had higher 90-day readmission for infection and for dislocation [23]. Learning curves in robotic knee arthroplasty apply to operating time rather than accuracy: 7 to 11 cases in a meta-analysis [24], but 11 to 43 cases in a series of six surgeons, longer for those with lower volumes [25]. In a multicenter study of 146 surgeons, operating time reached steady state after 15 to 20 cases or more, and only about two-thirds of surgeons ever matched their manual operating times [26]. For robotic reverse arthroplasty, a single fellowship-trained surgeon’s first 10 robotic cases took about 30 minutes longer than conventional cases at the same center, and operating time reached the conventional benchmark at about the 23rd case [27]. At 6 reverse arthroplasties a year, 20 to 25 cases would take three to four years; at the median of 2 a year, more than a decade. And the Glasgow analysis found robotics cost-effective only at high volume [8], the opposite of the setting where it is said to help most.
10. What determines the cost of acquiring and using a robot?
The costs of robotic arthroplasty, including capital, service, disposables, CT scans and personnel, are often set by non-disclosure agreements between the vendor and the hospital. Pricing may depend on anticipated case volumes, which could disadvantage lower-volume centers. Surgeons and patients cannot weigh a cost they are not allowed to see.
11. If there is an adverse outcome, who bears the responsibility?
Figure 10. If a robot-assisted operation goes wrong, who is responsible?
If a robot-guided component is malpositioned, or if there is a robot-related complication leading to a poor patient outcome, who is responsible: the surgeon, the hospital that bought the system, or the company that built it? Is the surgeon the method or the robot the method? These questions have not been tested, and surgeons should know the answers before consenting patients.
12. Can routine reverse arthroplasty be performed well without a CT scan or a robot?
Figure 11. First case on the first day of fellowship: preoperative radiographs (left), with templating (center), and postoperative radiographs (right), performed without CT or robot.
Most reverse arthroplasties are done for cuff tear arthropathy or osteoarthritis with modest glenoid deformity. Careful radiographic templating, good exposure of the glenoid and attention to its anatomy produce reliable component placement in these routine cases, as this first fellowship case illustrates.
13. Can young surgeons become good at reverse arthroplasty without a robot?
Figure 12. Reverse arthroplasties performed by two young surgeons without robotic assistance.
Young surgeons with good training can achieve consistent component placement with standard instruments. The skills they build doing so are the ones they will need when the anatomy is unusual, the exposure is difficult, or the technology is not available.
14. Is robotics the best investment for improving the lives of the millions of people with symptomatic shoulder arthritis?
A break-even analysis frames the question simply. The reduction in revision a robot must achieve to pay for itself equals its added cost per case divided by the cost of a revision. Using an added robotic cost of $1,500 per case, a revision cost of $22,920, and a baseline early revision rate of 2.1%, the robot would need to prevent one revision for every 15 procedures, an absolute reduction of 6.55% [28]. The reported revision rate in the first year after reverse arthroplasty in US hospital billing data is 2.1%, about one revision in every 48 cases, with a mean hospital cost of $22,920 per revision [29]; in the UK National Joint Registry, one-year revision stayed below 2% from 2013 through 2021 [30]. The robot would have to prevent about three times as many revisions as occur using a non-robotic approach. Even a robot that prevented every early revision would repay only about $481 per case (2.1% of $22,920), less than a third of its added cost. The revision cost is a hospital cost that excludes professional fees; but even at the roughly $60,000 hospital cost of a revision for infection, the break-even reduction (2.5%) would still exceed the entire first-year revision rate. The analysis considered only first-year revision; the authors note that any value of more accurate positioning may appear later, in time to revision and in patient-reported outcomes, and that selective use in higher-risk anatomy or in high-volume centers already owning a robot may change the arithmetic [28].
Is spending substantial money on robotics the best way to improve the outcomes for the many patients with symptomatic shoulder arthritis?
15. Is there evidence that robotic assistance will improve the outcomes our patients now experience?
Figure 13. Percentage of 232 patients able to perform each Simple Shoulder Test function before conventional reverse arthroplasty (grey) and at two years (blue). Matsen, unpublished data.
This is the standard robotic arthroplasty must improve upon, and it is a moving one: in the UK registry, the mean improvement in the Oxford Shoulder Score six months after reverse arthroplasty rose from 15.8 points in 2013 to 20.3 in 2021, without robots [30]. The answer here is that without robotics most patients regain comfort and the ability to sleep, reach a shelf and lift a light object; fewer regain overhead lifting and throwing. A robot would need to raise these percentages, and the improvement would need to be large enough for patients to recognize. Published minimal clinically important differences for the ASES score after reverse arthroplasty range from 6.2 to 21 points, the widest range of any shoulder score, with 10.3 the value most often cited [31]. The Hurley analysis modeled robotic benefits of 1 to 5 points [1], below even the lowest of them.
16. Can randomized trials realistically detect a patient benefit from robotic reverse arthroplasty?
A trial designed to detect a small reduction in revision would need to be very large. With a 4% baseline revision rate, detecting a 1% absolute reduction would take 5,301 patients per arm, and a 0.5% reduction 22,663 [1]. The authors conclude that the field should therefore turn to intermediate endpoints, registries and subgroups. That reverses the usual burden of proof: a technology that adds cost needs to show that it helps, and the absence of a feasible trial is not evidence of benefit. An effect that takes tens of thousands of patients to detect is also an effect that few patients will ever experience.
The only comparison of patient outcomes after navigated and non-navigated reverse arthroplasty matched 113 shoulders of each [18]. The navigated patients started with better motion and scores, and their improvement after surgery did not differ on any measure. Revision was 0.9% versus 3.5% (P = .37). The navigated shoulders also received augmented baseplates far more often (96% vs 50%), and no postoperative imaging confirmed that navigation changed component position. This is what an underpowered comparison looks like: it demonstrates neither a benefit nor its absence.
RACER-Knee shows that the question can be answered. With 339 patients, it found no clinically meaningful benefit of robotic knee arthroplasty at one year, with confidence intervals narrow enough to be informative. A shoulder trial of comparable size could do the same, using an outcome patients recognize: the proportion with poor comfort and function. If one patient in five has that outcome after conventional reverse arthroplasty, about 220 patients per arm would give 80% power to detect a halving of that proportion, and would rule out a benefit of that size if none exists.
The question ahead
A trial of fewer than 500 patients may clarify whether the robot delivers a benefit patients would notice. Isn’t it time for this study?
Golden-cheeked Warbler
San Antonio
Follow on twitter/X: https://x.com/RickMatsen
Follow on facebook: https://www.facebook.com/shoulder.arthritis
Follow on LinkedIn: https://www.linkedin.com/in/rick-matsen-88b1a8133//
The author of this post has no financial relationship with any orthopaedic company
References
1. Hurley ET, Frederickson M, Sanchez-Sotelo J, Athwal GS, Menendez ME. Can randomized trials realistically detect patient benefit from robotic-assisted reverse shoulder arthroplasty? J Shoulder Elbow Arthroplasty 2026 (journal pre-proof). doi:10.1016/j.jsea.2026.100106
2. Ruangsomboon P, Ruangsomboon O, Pornrattanamaneewong C, Narkbunnam R, Chareancholvanich K. Clinical and radiological outcomes of robotic-assisted versus conventional total knee arthroplasty: a systematic review and meta-analysis of randomized controlled trials. Acta Orthop 2023;94:60-79. doi:10.2340/17453674.2023.9411
3. Mert Ü, Khasawneh MY, Ghandour M, Al Zuabi A, Horst K, Hildebrand F, Bouillon B, Mahmoud MA, Kabir K. Comparative efficacy and precision of robot-assisted vs. conventional total knee arthroplasty: a systematic review and meta-analysis of randomized controlled trials. J Clin Med 2025;14:3249. doi:10.3390/jcm14093249
4. Klincke V, Huyghe R, Vermue H, Martens M, Victor J. Robot-assisted total knee arthroplasty demonstrates superior radiological but comparable clinical outcomes compared to conventional techniques: a meta-analysis of prospective studies. Knee Surg Sports Traumatol Arthrosc 2026;34(2):618-632. doi:10.1002/ksa.70007
5. National Institute for Health and Care Excellence. Robot-assisted surgery for orthopaedic procedures: early value assessment. HealthTech guidance HTG743 (formerly HTE22). April 17, 2025. https://www.nice.org.uk/guidance/htg743
6. Parsons H, Metcalfe A, Griffin J, Khatri C, et al., on behalf of the RACER-Knee trial team. Robotic-arm-assisted versus conventional total knee replacement (RACER-Knee): a pragmatic, multicentre, participant-masked and assessor-masked, superiority, randomised controlled trial. Lancet 2026;408:924-934. doi:10.1016/S0140-6736(26)00986-4
7. Banger M, Doonan J, Rowe P, Jones B, MacLean A, Blyth MJG. Robotic arm-assisted versus conventional medial unicompartmental knee arthroplasty: five-year clinical outcomes of a randomized controlled trial. Bone Joint J 2021;103-B(6):1088-1095. doi:10.1302/0301-620X.103B6.BJJ-2020-1355.R2
8. Clement ND, Fraser E, Gilmour A, Doonan J, MacLean A, Jones BG, Blyth MJG. Cost-utility analysis of robotic arm-assisted medial compartment knee arthroplasty: five-year data from a randomized controlled trial. Bone Jt Open 2023;4(11):889-899. doi:10.1302/2633-1462.411.BJO-2023-0090.R1
9. Blyth MJG, Clement ND, Choo XY, Doonan J, MacLean A, Jones BG. Robotic arm-assisted medial compartment knee arthroplasty is a cost-effective intervention at ten-year follow-up: a cost-effectiveness analysis from a randomized controlled trial. Bone Joint J 2025;107-B(1):72-80. doi:10.1302/0301-620X.107B1.BJJ-2024-0245.R2
10. de Steiger RN, Liu YL, Graves SE. Computer navigation for total knee arthroplasty reduces revision rate for patients less than sixty-five years of age. J Bone Joint Surg Am 2015;97(8):635-642. doi:10.2106/JBJS.M.01496
11. Prabhu AS, Carbonell A, Hope W, et al. Robotic inguinal vs transabdominal laparoscopic inguinal hernia repair: the RIVAL randomized clinical trial. JAMA Surg 2020;155(5):380-387. doi:10.1001/jamasurg.2020.0034
12. Clement ND, Duckworth AD, Haddad FS. Robotic-assisted arthroplasty paradox: the disconnect between precision and patient-perceived outcomes. Bone Jt Open 2026;7(8):994-996. doi:10.1302/2633-1462.78.BJO-2026-0164
13. Sperling JW, Anderson MB, Jobin CM, Verborgt O, Duquin TR. Humeral and glenoid component malposition in patients requiring revision shoulder arthroplasty: a retrospective, cross-sectional study. J Shoulder Elbow Surg 2025;34:1886-1896. doi:10.1016/j.jse.2024.11.015
14. Athwal GS, Nelson A, Antuna S, Ponce B, Mighell M, St Pierre P, et al. Glenoid preparation in reverse shoulder arthroplasty: robotic arm-assisted preparation compared to manual preparation and patient-specific guides. J Shoulder Elbow Surg 2025;34:2022-2030. doi:10.1016/j.jse.2024.12.007
15. Mason TW, Lauck BJ, Glover MA, Recker AJ, Reynolds AW, Trasolini NA, Waterman BR, van der List JP. Why primary reverse total shoulder arthroplasties fail today? A systematic review and meta-analysis. JSES Rev Rep Tech 2026 (journal pre-proof). doi:10.1016/j.xrrt.2026.100855
16. Menendez ME, Matsen FA 3rd, Athwal GS. Clinical Faceoff: What is the role of robotics in reverse shoulder arthroplasty? Clin Orthop Relat Res 2026;484:646-650. doi:10.1097/CORR.0000000000003884
17. O’Malley O, Davies A, Rangan A, Sabharwal S, Reilly P. Is there a difference in thresholds for revision between shoulder arthroplasty types? A National Joint Registry study. PLoS One 2025;20(8):e0330975. doi:10.1371/journal.pone.0330975
18. Holzgrefe RE, Hao KA, Panther EJ, Schoch BS, Roche CP, King JJ, Wright JO, Wright TW. Early clinical outcomes following navigation-assisted baseplate fixation in reverse total shoulder arthroplasty: a matched cohort study. J Shoulder Elbow Surg 2023;32:302-309. doi:10.1016/j.jse.2022.07.007
19. Parsons KE, Shenoy DA, Lorentz SG, Hurley ET, Navacchia A, Moverman M, Levin JM, Klifto CS. Optimization of implant selection and positioning for reverse total shoulder arthroplasty using three-dimensional computed tomography-guided simulation software with scapulothoracic motion. J Shoulder Elbow Arthroplasty 2026;10:100049. doi:10.1016/j.jsea.2026.100049
20. Graefe SB, Kirchner GJ, Pahapill NK, Nam HH, Dunleavy ML, Haines N. Adverse events associated with robotic-assistance in total hip arthroplasty: an analysis based on the FDA MAUDE database. Hip Int 2024;34(6):688-694. doi:10.1177/11207000241263315
21. Pagani NR, Menendez ME, Moverman MA, Puzzitiello RN, Gordon MR. Adverse events associated with robotic-assisted joint arthroplasty: an analysis of the US Food and Drug Administration MAUDE database. J Arthroplasty 2022;37:1526-1533. doi:10.1016/j.arth.2022.03.060
22. Di Carlo G, Zampogna B, Criseo N, Aragona D, Pugliesi O, Calaciura S, Fenga D, Sanzarello I, Leonetti D. The dark side of precision: pin-related complications in computer-navigated and robotic-assisted knee arthroplasty. J Clin Med 2026;15(10):3793. doi:10.3390/jcm15103793
23. Girdler SJ, Maza N, Lieber AM, Vervaecke A, Kodali H, Zubizarreta N, Poeran J, Cagle PJ, Galatz LM. Impact of surgeon case volume on outcomes after reverse total shoulder arthroplasty. J Am Acad Orthop Surg 2023;31(24):1228-1235. doi:10.5435/JAAOS-D-23-00181
24. Zhang J, Ndou WS, Ng N, Gaston P, Simpson PM, Macpherson GJ, Patton JT, Clement ND. Robotic-arm assisted total knee arthroplasty is associated with improved accuracy and patient reported outcomes: a systematic review and meta-analysis. Knee Surg Sports Traumatol Arthrosc 2022;30(8):2677-2695. doi:10.1007/s00167-021-06464-4
25. Vermue H, Luyckx T, Winnock de Grave P, Ryckaert A, Cools AS, Himpe N, Victor J. Robot-assisted total knee arthroplasty is associated with a learning curve for surgical time but not for component alignment, limb alignment and gap balancing. Knee Surg Sports Traumatol Arthrosc 2022;30(2):593-602. doi:10.1007/s00167-020-06341-6
26. Chen Z, Bhowmik-Stoker M, Palmer M, Coppolecchia A, Harder B, Mont MA, Marchand RC. Time-based learning curve for robotic-assisted total knee arthroplasty: a multicenter study. J Knee Surg 2023;36(8):873-877. doi:10.1055/s-0042-1744193
27. Buac NP, Chhokar M, Menendez ME. Robotic-assisted reverse shoulder arthroplasty achieves operative time neutrality after an initial learning period. Int Orthop 2026;50(4):853-857. doi:10.1007/s00264-026-06774-7
28. Menendez ME, Moverman MA, Schiffman CJ, Matsen FA 3rd. Is robotic-assisted reverse shoulder arthroplasty economically justified? A break-even analysis. J Shoulder Elbow Surg 2026;35(9):2313-2317. doi:10.1016/j.jse.2026.03.010
29. Corso KA, Smith CE, Vanderkarr MF, Debnath R, Goldstein LJ, Varughese B, Wood J, Chalmers PN, Putnam M. Postoperative revision, complication and economic outcomes of patients with reverse or anatomic total shoulder arthroplasty at one year: a retrospective, United States hospital billing database analysis. J Shoulder Elbow Surg 2025;34:e59-e71. doi:10.1016/j.jse.2024.05.009
30. O’Malley O, Davies A, Taghavi Azar Sharabiani M, Rangan A, Sabharwal S, Reilly P. Are we getting better over time? Clinical and patient-reported outcomes for reverse shoulder arthroplasty: a National Joint Registry cohort study. BMJ Open 2025;15:e096084. doi:10.1136/bmjopen-2024-096084
31. Yendluri A, Alexanian A, Lee AC, Megafu MN, Levine WN, Parsons BO, et al. The variability of MCID, SCB, PASS, and MOI thresholds for PROMs in the reverse total shoulder arthroplasty literature: a systematic review. J Shoulder Elbow Surg 2024;33:2320-2332. doi:10.1016/j.jse.2024.03.051
32. Menendez ME, Moverman MA, Namdari S, Matsen FA 3rd, Ring D. Modeling surgeon belief updating under bias: a Bayesian simulation in shoulder arthroplasty. JSES Int 2026;10:101399. doi:10.1016/j.jseint.2025.09.018
