Clinical and radiological outcomes following complete release of the long head of triceps in reverse shoulder arthroplasty Authors: Hean Wu Kang, Kristine Italia, Rohit Mahesh Sane, Rachel Sarah Marie Price, Helen Ingoe, Jashint Maharaj, Sarah L Whitehouse, Ashish Gupta Published: 2026 ABSTRACT AbstractBackgroundComplete long head of the triceps release (LHTR) during reverse shoulder arthroplasty (RSA) aims to eliminate inferior impingement and ensures tether-free range of motion (ROM). However, it remains infrequently performed due to concerns such as triceps strength deficits and instability. This study aimed to present the clinical and radiological outcomes following RSA with LHTR, focusing on the impact on triceps strength.MethodsThis was a retrospective review of patients who underwent primary RSA with LHTR. The primary outcome was 2-year triceps strength compared with the preoperative strength. Secondary outcomes were the improvement of shoulder ROM, Constant Score, American Shoulder and Elbow (ASES) Score, and Visual Analog Score (VAS) at 2-year postoperative time point, and radiological outcomes including notching and heterotopic ossification (HO).ResultsEighty patients were included with mean age of 71.1 years. Two-year triceps strength was comparable to the preoperative strength (6.3 kg vs. 6.2 kg, p = .703). Active ROM, VAS, ASES, and Constant scores significantly improved at 2 years postoperatively (p < .001). Notching was present in 15% of patients while HO was present in 18.75%. There were no dislocations or neurovascular injuries.DiscussionComplete LHTR during RSA appears safe with no adverse effect on triceps strength, patient-reported outcomes, or stability.Keywords: Triceps release, RSA, scapular notching, heterotopic ossification, grammont-style prosthesis, anatomic RSA Introduction The incidence of reverse shoulder arthroplasty (RSA) has increased exponentially over the last two decades since the introduction of Grammont's reverse prosthesis in 1985.1,2 The rate of RSA implantation is projected to continue to rise worldwide, and with that follows the pursuit of better clinical and functional outcomes for patients. 3 Several concerns still exist, including scapular notching, internal rotation deficit, and instability. Instability is the leading cause of revision in RSA's. 2 Along with various improvements in implant designs and surgical techniques, one of the key strategies contributing to optimal patient outcomes and lowering the risk of impingement and instability is the use of soft tissue releases during shoulder arthroplasty. These releases often include the long head of the biceps tendon, labrum, and capsule circumferentially. In addition to these structures, the complete long head of the triceps release (LHTR) is also believed to be one of the necessary steps during RSA. Inadequate release of these structures can cause soft tissue impingement of the medial humeral component at the inferior capsular recess, which could lead to construct instability and abnormal coaptation during humeral rotation in adduction. By completely releasing these structures, particularly the long head of the triceps, tether-free shoulder range of motion (ROM) in adduction and rotation can be achieved, allowing the humerus to move freely under the scapular neck without soft tissue impingement. The absence of an inferior tether aids in better abduction and forward flexion ROM post RSA. In addition, LHTR is particularly important to attain adequate glenoid exposure in RSA, especially inferiorly. Better visualization of the inferior bony glenoid margin and its continuity into the proximal lateral pillar of the scapula allows for more accurate inferior baseplate positioning and adequate fixation. This reduces the likelihood of scapular notching, given the sufficient glenosphere overhang with a low baseplate placement.4–6 Despite its benefits, complete LHTR is infrequently carried out and not widely reported in the literature. There remains a cause for concern regarding the neurovascular structures within the quadrangular space, as well as the reduction in triceps strength and function. Extensive soft-tissue release in other joints has been associated with a higher incidence of heterotopic ossification (HO). 7 It is, therefore, plausible that soft-tissue release within the axillary pouch could similarly increase this risk. Hence, this study aimed to present the clinical and radiological outcomes of patients who underwent primary RSA with complete LHTR. In particular, the triceps strength on the operated side was evaluated in comparison to the preoperative strength to assess the effect of LHTR. Materials and methods A retrospective review of prospectively collected data was performed on patients who underwent primary RSA with a complete long head of triceps release by a single fellowship-trained shoulder surgeon (AG) from January 2018 to December 2021. Patients who had RSA for fracture or revision surgery, with neurological injury to the upper limb, less than 2-year follow-up, or incomplete clinical outcome data were excluded. Informed consent was obtained from all patients, and ethical approval was obtained from the Ramsay Human Ethics Review Committee (HREC 19/32). The primary outcome was the difference in triceps strength assessed at the preoperative visit and at the 2-year postoperative review. Triceps strength was measured using a handheld digital dynamometer (Commander Echo MMT, J Tech, USA) during active elbow extension with the shoulder in 0° elevation and neutral rotation, as in this position, the long head of the triceps has been shown to contribute the most force to elbow extension. 8 The secondary outcome measures included shoulder active ROM, patient-reported outcome measures (PROMs), and radiological outcomes. Active ROM included forward elevation (FE), lateral elevation (LE), external rotation in adduction (ER), and internal rotation at 90-degree abduction (IR). PROMs included Constant score, American Shoulder and Elbow Score (ASES), and Visual Analog Score (VAS). As part of standard clinical practice, these outcomes were collected preoperatively and at 6, 12, and 24 months postoperatively. Radiological outcomes included the presence of HO and scapular notching evaluated on the true anteroposterior (AP) view of x-rays obtained at least 2 years postoperatively. HO was classified using the modified Brooker system for HO of the shoulder, which defines grade 1a as islands of bone within the soft tissues of the shoulder; grade 1b as bone spur on the inferior scapular neck; grade 1c as incomplete ankylosis of the shoulder; grade 2 as a combination of bone spur on the inferior scapular neck and bone islands within the soft tissues of the shoulder; and grade 3 as complete ankylosis. 9 Scapular notching was evaluated using the Sirveaux classification, which defines grade 1 as a defect confined to the scapular pillar; grade 2 as a defect reaching the lower screw; grade 3 as a defect extending over the lower screw; and grade 4 as a defect extending under the baseplate. 10 Grades I and II were considered as low grade, while grades III and IV were considered high grade. 11 Radiological evaluation was performed independently by two fellowship-trained shoulder surgeons, and any discrepancies were resolved by a third fellowship-trained shoulder surgeon. Surgical technique After induction of general anesthesia with interscalene peripheral nerve block and with 100% muscle relaxation, the patient is safely positioned in a beach chair with the operated arm placed in an arm positioner (Spider Limb Position, Smith & Nephew, London, England). A deltopectoral approach or a superolateral approach with subscapularis peel is used. Adequate subperiosteal release of humeral capsular attachment until the inferior and posterior neck is performed. Adequate release of the humeral capsular insertion inferiorly is confirmed with visualization of the latissimus dorsi tendon insertion anteriorly and the teres minor insertion posteroinferiorly. The humeral head is then delivered and resected at 15° retroversion using an intramedullary guide and extramedullary referencing to the long axis of the forearm. A Fukuda humeral head tractor is then placed into the joint to retract the humeral head posteriorly. Anterior glenoid capsulectomy is carried out first, as well as the release of the superior and middle glenohumeral ligaments off the anterior scapular neck. The long head of biceps and glenoid labrum are resected off the glenoid rim circumferentially using a long-handed scalpel blade, taking care to remain close to the bony surface at all times. Inferior glenoid capsulectomy is performed by sharply dissecting the anterior and posterior bands of the inferior glenohumeral ligament complex with a scalpel. The axillary nerve is not necessarily visualized at this stage; instead, a Cobb elevator is used to retract the inferior capsular complex caudally, thereby protecting the axillary nerve together with the inferior capsule while the capsulectomy is completed along the inferior glenoid rim. Any remaining inferior capsular attachments along the inferior glenoid vault are then peeled off bluntly with the Cobb elevator. While maintaining the Cobb elevator inferiorly to retract and protect inferior structures, the long head of triceps origin at the infraglenoid tubercle is visualized (Figure 1a), and its lateral tendinous origin is released sharply using Metzenbaum scissors under direct vision (Figure 1b). The cranial part of the lateral scapula pillar is visualized and palpated. An audible snip is usually heard during the release, and a “blow for freedom” can be felt by the assistant holding the retracted humeral head. This normally happens in tight arthritic shoulders or those with chronically superiorly migrated humeral heads. The muscular origin of the long head of triceps, more medial along the lateral pillar, is then released bluntly by blunt dissection using the Cobb elevator (Figure 1c). No sharp dissection is carried out medial to the muscular fibers to avoid injury to the circumflex scapular artery and vein within the triangular space. A hydrogen peroxide-soaked sponge can be inserted into this space to tamponade any bleeding should it occur. Lastly, posterior glenoid capsulectomy is carried out by removing the posterior labral capsular complex. After circumferential 360° capsulectomy, the lateral pillar is easily visualized and palpated along the first 4 to 5 cm. Scapular neck length can be easily ascertained to differentiate short neck to long neck scapulae, which could influence the lateralization needed for each patient. Preparation of the glenoid and implantation of the glenoid components is performed. Glenoid bone grafting is performed, as necessary, to restore native glenoid joint line in cases of significant glenoid bone defects. 12 The glenoid baseplate is positioned as low as possible to the inferior glenoid rim with 5° anteversion and 10° inferior tilt. Grammont-style RSA prosthesis (Aequalis Reversed System Wright-Tornier, Memphis, TN, USA and Delta Xtend, Depuy Synthes, Warsaw, IN, USA) implanted following the hybrid humerus technique (inlay or onlay as required and inserted laterally to medialise the humerus) was used in all patients. 13 The joint is reduced and ROM is tested to check for impingement (Figure 2). Posterosuperior cuff and subscapularis are repaired meticulously in all patients. If posterosuperior cuff cannot be repaired, latissimus dorsi transfer is performed. Complete release of the long head of the triceps. (a) The long head of triceps origin at the infraglenoid tubercle is visualized (Tr) after capsulectomy. (b) Its lateral tendinous origin is released sharply using Metzenbaum scissors under direct vision. (c) The muscular origin of the long head of triceps more medial along the lateral pillar is released bluntly by blunt dissection using the Cobb elevator. Postreduction ROM for impingement check. (a) Without the release of the long head of the triceps, soft tissue impingement can occur inferiorly, which can decrease range of motion and increase the risk of instability. (b) Complete release of the long head of the triceps facilitates impingement-free motion inferiorly, allowing the humerus to completely slide inferiorly to the glenoid. ROM, range of motion. Statistical analysis Descriptive statistics were presented as means, standard deviations, ranges or frequencies, and percentages, as appropriate. Wilcoxon signed rank test and paired t-tests were performed to assess statistical differences between preoperative and 2-year postoperative triceps strength, ROM, and PROMs. Fisher's Exact test was performed for categorical variables such as patient satisfaction. Mann-Whitney U-test and paired t-test were performed to compare the clinical outcomes between patients with and without notching or HO, depending on whether the data were nonparametric or normally distributed, respectively. A p-value of less than 0.05 was considered statistically significant. All statistical analyses were performed using SPSS (version 29.0; IBM Corp, Armonk, NY, USA). Results A total of 80 patients, with a mean age of 71.1 ± SD 6.3 years (range 52–86), met the inclusion criteria and were included in the study. There was an equal distribution between male and female patients (Table 1). Most of the patients were operated on for cuff tear arthropathy (CTA) followed by osteoarthritis. At the 2-year post-operative time point, the mean triceps strength of the operated shoulder was comparable to the preoperative strength (6.3 ± 2.1 kg vs. 6.2 ± 2.0 kg, respectively, p = .703). All clinical outcomes showed significant improvement from preoperative to 2 years postoperatively (Table 2). Satisfaction rate was 93.8%. There were no injuries to the neurovascular structures inferior to the glenoid and no dislocations of RSA prostheses were noted. Table 1.Patient demographics.VariablesMean ± SD, or n (%)Age at surgery, y71.1 ± 6.3Sex Male40 (50%) Female40 (50%)Affected side Right49 (61.3%) Left31 (38.8%)Diagnosis Cuff tear arthropathy49 (61.25%) Osteoarthritis17 (21.25%) Inflammatory arthropathy4 (5%) Irreparable cuff tear3 (3.75%) Dislocation3 (3.75%) Dislocation arthropathy2 (2.5%) Fracture malunion sequelae1 (1.25%) Avascular necrosis1 (1.25%)Prosthesis Wright Aequalis77 (96.25%) Depuy Delta Xtend3 (3.75%)Approach Deltopectoral58 (72.5%) Superolateral22 (27.5%)Glenoid bone grafting With38 (47.5%) Without42 (52.5%) Patient demographics. Table 2.Clinical outcomes.OutcomePre-opMean (SD, range)2 years postopMean (SD, range)Mean differenceMean (SD, range)p-valueTriceps strength (kg)6.2(2.0, 3.2−15.4)6.3(2.1, 2.5−14.7)0.11(1.2, −2.7−5.4)p = .703FE (degrees)111(48, 10−170)164(17, 90−180)54(47, −60−150)p < .001LE (degrees)99(43, 0−180)158(19, 60−180)59(44, −30−170)p < .001ER (degrees)33(21, 0−80)53(16, 0−80)20(23, −30−70)p < .001IR (degrees)32(24, 0−80)62(11, 30−90)31(25, −40−80)p < .001VAS4.5(1.8, 0−10)0.3(0.9, 0−5)−4.3(2.9, −10−4)p < .001Constant Score34.9(14.8, 0−63)77.0(9.4, 39−98)42.8(15.0, 10−80)p < .001ASES Score55.7(17.0, 5−83)93.0(9.7, 50−100)38.1(18.7, −7−95)p < .001Satisfaction, n (%)SatisfiedNot satisfied4 (5.0%)76 (95.0%)75 (93.8%)5 (6.3%)Note. ASES, American Shoulder and Elbow Surgeons; ER, external rotation; FE, forward elevation; IR, internal rotation; kg, kilogram; LE, lateral elevation; VAS, visual analog scale. Clinical outcomes. Note. ASES, American Shoulder and Elbow Surgeons; ER, external rotation; FE, forward elevation; IR, internal rotation; kg, kilogram; LE, lateral elevation; VAS, visual analog scale. Radiographic outcomes Grade 1 and 2 notching were present in 12 patients (15%) at a minimum of 2 years postoperatively (Table 3). HO was present in less than 20% of cases, the majority of which consisted of small free-floating bony fragments (type 1a). Among these patients with HO, only six patients (7.5%) had concomitant scapular notching, the majority of which was low-grade and associated with free-floating bony fragments. Table 3.Radiographic parameters.Notching*No HO (n = 65)With HO⬨ (n = 15)n (%)01a1b1c23068 (85%)5953010110 (12.5%)62110022 (2.5%)02000030 (0%)00000040 (0%)000000Total8065 (81.25%9 (11.25%)4 (5%)1 (1.25%)1 (1.25%)0 (0%)Note. HO, heterotopic ossification.*Notching classified according to Sirveaux classification.⬨HO classified according to modified Brookers classification. Radiographic parameters. Note. HO, heterotopic ossification. *Notching classified according to Sirveaux classification. HO classified according to modified Brookers classification. Association of HO and notching with clinical outcomes At 2-year follow-up, there were no significant differences in the clinical outcomes among those with notching and HO compared to those who had none, except for active LE which was noted to be decreased in patients with notching (p < .001) (Table 4). Table 4.Association of HO and notching with 2-year clinical outcomes.NotchingMean (SD, range)HOMean (SD, range)OutcomeNo Notching(n = 68)Notching(n = 12)p-valueNo HO(n = 65)HO(n = 15)p-valueTriceps strength (kg)6.3(2.0, 2.5–14.7)6.4(2.7, 3.6–13.8)p = .6326.2(2.0, 2.5–14.7)7.0(2.4, 3.2–13.8)p = .094FE (degrees)166(13, 100–180)152(27, 90–170)p = .053162(18, 90–180)168(7, 160–180)p = .451LE (degrees)161(15, 110–180)140(32, 60–160)p < .001158(21, 60–180)160(10.0, 140–180)p = .616ER (degrees)54(15, 0–80)50(18, 10–80)p = .52753(16, 0–80)51(18, 10–80)p = .615IR (degrees)62(11, 30–90)61(13, 30–80)p = .59763(11, 30–90)61(13, 30–80)p = .701VAS0.2(0.7, 0–4)0.8(1.8, 0–5)p = .5190.3(1.0, 0–5)0.1(0.5, 0–2)p = .459Constant Score77.4(7.9, 49–98)75.2(15.7, 39–91)p = .53576.5(9.3, 39–98)79.2(9.4, 49–91)p = .167ASES Score93.6(8.1, 65–100)89.4(16.2, 50–100)p = .84392.3(10.1, 50–100)95.7(7.2, 72–100)p = .134Satisfaction SatisfiedNot satisfied64 (94.1%)4 (5.9%)11 (91.7%)1 (8.3%)p = .56660 (92.3%)5 (7.7%)15 (100.0%)0 (0.0%)p =.578Note. ASES, American Shoulder and Elbow Surgeons; ER, external rotation; FE, forward elevation; IR, internal rotation; kg, kilogram; KO, heterotopic ossification; LE, lateral elevation; VAS, visual analog scale. Association of HO and notching with 2-year clinical outcomes. Note. ASES, American Shoulder and Elbow Surgeons; ER, external rotation; FE, forward elevation; IR, internal rotation; kg, kilogram; KO, heterotopic ossification; LE, lateral elevation; VAS, visual analog scale. Discussion The results of this study show that completely releasing the long head of triceps during glenoid exposure did not affect the 2-year postoperative triceps strength, with no negative effect as well on the patient-reported outcomes and postoperative stability. Glenoid exposure during arthroplasty is crucial to visualize glenoid implant position.12,13 However, there remains paucity of literature about the release of the inferior glenoid capsule with the long head of the triceps. The importance of inferior glenoid exposure is often highlighted, but the release of the inferior glenohumeral capsule stops short at the long head of triceps origin. Only a few publications briefly mentioned about releasing the long head of triceps with none reporting direct clinical or radiological outcomes. Verhofste et al. 9 in their retrospective study of 132 consecutive RSA examining HO postoperatively mentioned a complete release of the triceps, however, they did not go into detail with regards to their operative technique. Hatzidakis et al. 14 stated in their technical note the release of the origin of the long head of the triceps from the inferior glenoid during glenoid exposure. Similarly, Simovitch et al. 15 reported their routine complete release of the long head of the triceps along the scapular neck. Additionally, Boileau et al., 16 Ross et al., 17 and Chawla et al. 18 mentioned only partial triceps release when exposing the inferior glenoid. Chawla et al. 18 emphasized releasing the inferior capsule and some triceps origin until 1 cm of medial exposure along the bottom of the glenoid is ensured. The limited discussion and execution of this step in RSA glenoid exposure may be influenced by several factors. Historically, anatomical descriptions often identified the long head of triceps originating solely from the infraglenoid tubercle without acknowledging its integration with the inferior glenohumeral capsule or the scapular neck in the lateral pillar. This limited understanding may have contributed to uncertainty regarding the functional implications of its release and the impact of complete release on triceps strength and joint stability. Additionally, perceived risks of injury to the axillary nerve and posterior circumflex vessels within the quadrangular space could further discourage dissection medial and inferior to the inferior glenoid rim. Lastly, there is a potential assumption that this step has little benefit in reducing scapular notching and increased risk of HO. We shall address these assumptions in the following discussion. In 2008, the 40th edition of Gray's Anatomy described the long head of triceps as originating from the infraglenoid tubercle and blending with the inferior glenohumeral capsule. 15 Before that, Eiserloh et al. 19 reported the same anatomical findings in their cadaveric analysis. Handling et al. 20 followed suit with similar cadaveric observations. More recently, Nasu et al. 21 looked in detail at the long head of triceps origin in 64 cadaveric specimens, 18 of which were also histologically examined. They reported that the long head of triceps has bony origins from the infraglenoid tubercle, with other bony contributions from the dorsal and costal surfaces of the lateral scapula pillar cranially. From the dorsal and costal surfaces, the lateral part of the long head of triceps tendon joins with the axillary pouch of the inferior glenohumeral joint capsule, and the medial part extends along the lateral scapula pillar medially. This inferior capsular-tendon complex also integrates with soft tissue fibers that descend from the posterior glenoid rim and posterior glenoid labrum, agreeing with Huber and Putz's early study of the glenoid labrum. 22 With the long head of triceps origin structurally integrated with the inferior glenohumeral capsule, its release is often necessitated in tight shoulders with inferior capsular contractures such as in arthritis or cuff tear arthropathy. Failure to do so can impede post-operative shoulder ROM in abduction and forward flexion. 23 An inferior tether also contributes to RSA construct instability by impeding full humeral extension in adduction. When the humerus externally rotates, there is abnormal superior coaptation at the glenosphere-polyethylene interface and anterior subluxation of the humeral component, creating a dislocating force anteriorly. This poses a dislocation risk to elderly patients as they rise and push off with their arms from a seated position. Conversely, when the humerus tries to internally rotate, there is a lack of functional internal rotation due to the lack of humeral extension. 24 There is not enough space inferior to the scapular neck to allow for the humeral component to adduct, extend, and internally rotate. This has been supported by Simovitch et al. 15 stating that releasing the long head of the triceps eliminates the impingement of the humeral prosthesis against soft tissue as it allows clearance for the humeral metaphyseal component, facilitating adduction. 18 Although this study included only patients who underwent RSA using Grammont-style prosthesis applied in an onlay fashion following the hybrid humerus technique, 25 we believe that these principles, as well as the findings of this study, are also applicable for prostheses with 135° neck-shaft angle. Despite the theoretical advantage of 135° neck-shaft angle in reducing scapular notching and inferior impingement because of better clearance during adduction, extension, and internal rotation compared to the Grammont-style prosthesis, 135° prostheses can still encounter soft tissue impingement and inferior tethering by the long head of the triceps when the scapula tilts on ROM. Hence, releasing the long head of the triceps in these cases mitigates these effects and improves scapular kinematics. The lack of space inferior to the scapular neck may also cause axillary nerve neuropraxia manifesting as postoperative pain. The axillary nerve is a mobile structure and traverses freely from anterior to posterior through the quadrangular space during glenohumeral movements. Intraoperatively, this can be palpated as the humerus is cycled through its ROM (RSA axillary nerve glide test). With the humerus adducted and fully flexed, the nerve can be palpated to be slack just anterior to the glenoid. As the flexion decreases and the humerus enters a neutral position, it becomes taut as it stretches over the anterior glenoid rim. Once the humerus passes neutral into full extension, it then passes under the glenoid neck and onto the lateral pillar where it becomes completely slack again as it has now cleared the anterior glenoid rim, as well as the absence of soft tissue impingement for the nerve in this position because of LHTR. Post RSA, the axillary nerve is anatomically further away from the glenosphere caudally and closer to the posterior humeral metaphysis or humeral component. 26 It is also known that distalization, rather than lateralization, causes more stretch on the axillary nerve in RSA implantation. 27 With that, it is important to ensure that the nerve is able to glide freely within the quadrangular space during humeral movements, especially in terminal extension. Without complete release of the inferior capsule and long head of triceps within the quadrangular space, crowding can happen, and the nerve can be caught within the soft tissue contractures. Literature has also shown that releasing the long head of the triceps has contributed to decreasing the risk of scapular notching. Releasing the long head of the triceps allows adequate exposure of the inferior glenoid, helping surgeons to achieve low positioning of the glenoid baseplate and avoiding superior tilt, which subsequently reduces the risk of scapular notching by allowing clearance for the humeral metaphyseal component in adduction.15,18,28 Our results have shown scapular notching in 15% of our cohort, with majority having low-grade notching only. This is lower than the published rate of notching of 42.5% after RSA using Grammont style prosthesis. 11 Similarly, Simovitch et al. 15 showed 44% rate of inferior scapular notching after RSA using Grammont-style prosthesis despite performing complete triceps release. On the other hand, Verhofste et al., 9 who also performed complete triceps release in RSA using Grammont-style prosthesis, showed lower rate of scapular notching at 28%. The lower incidence of scapular notching in our cohort, however, could be brought about by several other factors, in addition to releasing the long head of the triceps and subsequent positioning of the glenoid components. On the humeral side, our technique involves the hybrid humerus technique, as described by Pareyon et al., 25 wherein an undersized Grammont-style humeral prosthesis is implanted in an onlay technique with medial-lateral offset adjusted depending on tension and patient anatomy. We believe that this technique also contributes to decreasing the risk of scapular notching. By keeping the stem slightly proud in an onlay fashion and allowing adjustable medial-lateral offset to achieved desired tension, the inherent risk of scapular notching associated with Grammont-style prosthesis is decreased while avoiding the risk of over-lateralization and/or over-distalization associated with onlay design stems. 25 The use of the Grammont design, in turn, provides more stability as demonstrated by no cases of instability in our cohort. Our cohort has also shown that HO occurred in 18.75% of cases, which was much less than what was reported by Ko et al. 29 and Melis et al. 30 who showed HO in 61.6% and 75% of cases, respectively. HO is a common radiographic finding after RSA. It has been postulated that HO formation is related to the increased tension when the arm is lengthened during RSA. 29 It has also been previously suggested that HO presenting as traction osteophytes or bony scapular spurs could result from incomplete release of the triceps tendon, 9 which was in contrast to other literature that postulates that extensive release of the inferior capsule and triceps tendon are the factors that trigger HO formation. 29 Our technique involves the latter wherein complete release of the inferior capsule and triceps tendon is done, which is visually confirmed intraoperatively. Similarly, Verhofste et al. reported performing complete release of the triceps tendon routinely and reported lower rate of HO at 29.5% only. While our results show low rate of HO formation after complete triceps release, further studies with a comparison group of no or incomplete triceps tendon release are needed to confirm the causal relationship of the extent of triceps tendon release to the development of HO. Other factors postulated to contribute to the formation of HO were the intrinsic factors related to the disease process of CTA9,29,31 or the soft tissue response to constant mechanical irritation of the humeral tray, which were not investigated in this study. 29 Of the patients who developed HO in our study, only 7.5% had concomitant scapular notching. In the study of Ko et al., 29 14.6% of cases had HO with notching, whereas Verhofste et al. 9 had 30.8% of combined HO and notching, which were both higher than our cohort. Majority of these cases of combined notching and HO in our cohort had only small, free-floating HO and low-grade notching. As pointed out by Ko et al., 29 it was important to carefully distinguish HO from scapular notching so as not to incorrectly classify HO as notching. This was meticulously done in this study, wherein notching was assessed to be an erosion of the native glenoid neck whereas HO is excess bone formation below the native glenoid neck. Despite this combined notching and HO in our cohort, results have shown that the functional scores such as ASES, constant score, and UCLA remain satisfactory and did not differ from the patients without notching and/or HO. Interestingly, postoperative LE of patients with notching have been shown to be statistically significantly less compared to those without notching. Literature on the impact of scapular notching on outcomes after RSA is conflicting, with some authors reporting no influence of notching on clinical outcomes, 32 while others reported its association with poorer outcomes.15,33,34 Although our results show that the patients with notching had statistically significantly less LE than those without notching, this group of patients still achieved clinically good LE of 140°. This study is not without any limitations. As pointed out earlier, the cohort lacks a comparison group of patients who did not have or had incomplete triceps release. While this is necessary to confirm the causal relationship of triceps release with the various secondary outcomes presented, this is a limitation as this control group is not available in the surgeon's cohort since it is part of standard surgical technique to perform the complete triceps release in all patients. Moreover, this study did not account for other potential confounding factors that may contribute to the development of scapular notching, such as implant positioning, body mass index, and adduction range. Hence, the presence or absence of notching cannot be attributed solely to the complete release of the long head of the triceps. Conclusion Complete LHTR during RSA appears safe with no adverse effect on triceps strength or patient-reported outcomes. At 2 years postoperatively, triceps strength was comparative to preoperative values and clinical outcomes showed improvement. Importantly no cases of instability or axillary nerve-related injury or pain were observed. Acknowledgements We would like to thank Dr. Levent Efe for the illustrations. Footnotes ORCID iDs: Hean Wu Kang https://orcid.org/0000-0002-6216-5607 Kristine Italia https://orcid.org/0000-0001-8408-3197 Rohit Mahesh Sane https://orcid.org/0000-0001-6445-1159 Helen Ingoe https://orcid.org/0000-0003-1623-8489 Jashint Maharaj https://orcid.org/0000-0002-8869-5518 Sarah L Whitehouse https://orcid.org/0000-0002-3260-4736 Ethical approval: Ethical approval for this study was obtained from the Ramsay Human Ethics Review Committee (HREC 19/32). Informed consent: Written informed consent was obtained from all subjects before the study for their anonymized data to be included in this article. Guarantor: HWK, AG Contributorship: Hean Wu Kang wrote the first draft and revisions of the manuscript. Kristine Italia was involved in writing the first draft and revisions, and editing the manuscript, data curation, investigation, project administration; Rohit Mahesh Sane and Rachel Sarah were involved in investigation and review and editing of manuscript; Helen Ingoe was involved in review and editing; Jashint Maharaj was involved in data curation, investigation, project administration, review and editing; Sarah L Whitehouse was involved in formal analysis, review and editing; Ashish Gupta was involved in the conceptualization, investigation, methodology, supervision, review and editing. All authors reviewed and edited the manuscript and approved the final version of the manuscript. Funding: The authors received no financial support for the research, authorship, and/or publication of this article. The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. References