Successful total shoulder arthroplasty (TSA) relies on the accurate placement of implants, which can be difficult in some patients, especially in the setting of bone loss or deformity. Achieving solid fixation of the glenoid component while simultaneously maintaining accurate alignment is one of the most challenging aspects of TSA. If performed poorly, improper implant positioning can potentially lead to inferior clinical outcomes, complications, and early implant failure [1].

Traditionally, surgeons have relied on their own knowledge and experience when performing TSA, estimating glenoid version and inclination with no clear reference to the orientation of the scapula. This, combined with the limited and sometimes challenging exposure during TSA, can hinder accurate placement of the glenoid component. It has been reported that more than 70% of TSA procedures worldwide are performed by surgeons who participate in fewer than 10 cases per year, increasing the risk of component malposition [2]. A recent registry study identified associations between surgeons who averaged more than 10.4 shoulder replacements per year and lower rates of revision surgery and reoperation, lower risk of serious adverse events, and shorter hospital stays [3].

Various technological innovations have recently become available to provide intraoperative assistance to better execute the preoperative plan. These include patient-specific instrumentation (PSI), computer navigation, mixed reality, and robot-assisted surgery. The hope for these technologies is that they will improve accuracy in cases of more complicated pathology, in addition to providing better surgical uniformity and precision. Improved placement of implants will not only potentially enhance implant survivorship but may also decrease complication rates and improve patient outcomes.

The purpose of the study was to review patient-reported outcome measures (PROMs), complication, revisions, and reoperations in a group of patients undergoing TSA using computer navigation and compare these with a group of patients utilizing the same modular TSA system using conventional (non-navigated) surgical techniques. Lack of radiographic outcome data (implant version, inclination, rotation, seating) limits the ability of this study to fully understand the benefit of computer navigation in TSA, and studies are ongoing.

This study was deemed exempt from requiring Ethical Committee Approval from the New Zealand Health and Disability Ethics Committee. Written informed consent was obtained from patients.

A retrospective analysis of a prospectively collected outcome database of the same implant type was performed, adhering to the STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) guidelines for observational studies. A total of 180 patients was included in this study, comprising those with anatomic TSA (aTSA) or reverse TSA (rTSA) implanted by a single fellowship-trained subspecialist shoulder surgeon with more than 20 years of TSA experience. There were 40 aTSA cases and 140 rTSA cases with a minimum follow-up period of 2 years (range, 24–66 months). All procedures were performed using the Equinoxe TSA system (Exactech Inc.). Glenoid implant options included standard and 8° posterior augments for aTSA and standard, 8° posterior, 10° superior, and combined posterior superior augments for rTSA. In all cases, the Equinoxe Preserve platform short stem was used, and all stems for both aTSA and rTSA were implanted to replicate the patient’s native humeral version. For aTSA, an uncemented cage peg glenoid was used in all patients.

The inclusion criterion was primary TSA (anatomic or reverse) for osteoarthritis, rheumatoid arthritis, osteonecrosis, massive irreparable rotator cuff tear, or rotator cuff tear arthropathy. Patients who underwent revision TSA or shoulder arthroplasty for acute fracture or fracture sequelae were excluded. PROMs were obtained using either in-person or on-line data collection with the Socrates 360 Orthopaedic Outcomes Software (360 Med Care Pty Ltd.). These were collected and analyzed by an independent research assistant preoperatively and at 6 months, 1 year, and 2 years postoperatively. Outcome metrics included patient satisfaction, American Shoulder and Elbow Surgeons (ASES), Oxford Shoulder Score, patient function and pain score, and Veterans RAND 12 Item Health Survey (VR-12) assessments of physical and mental function. These were compared between the navigated and non-navigated TSA groups. In addition, we compared postoperative complications, reoperations, and revisions.

The senior author began using the Equinoxe TSA system in June 2018. Initially, only two-dimensional (2D) planning was used to prepare for TSA, consisting of an analysis of preoperative radiographs, computed tomography (CT) scans, and/or magnetic resonance imaging. Subsequently, 3D planning software became available with the Equinoxe Planning App version 1.5 (Exactech; Blue Ortho). Preoperative planning using this system was then integrated into the chosen procedure, but it did not include the use of any PSI or selective guides. There was a gradual transition to the use of intraoperative computer navigation using the Exactech GPS Total Shoulder Application V1.4.1 (Exactech; Blue Ortho). Initially, only the more complex aTSA and rTSA cases were performed using GPS navigation; subsequently, all cases were performed with navigation.

Computer navigation patients underwent a preoperative CT scan of their affected shoulder, performed according to a specific protocol. CT scans were performed fewer than 6 months prior to the surgery, with images acquired in axial format with no rotation (gantry tilt 0°). Slice thickness and maximal slice spacing distance was 1.25 mm, with a recommended distance of 0.625 mm. Both thickness and spacing were constant for the entire exam, with no overlap. The shoulders were scanned to capture the entirety of the scapula and proximal portion of the humerus, with a display field of the view of approximately 25 to 30 cm for a matrix size of 512×512.

After segmenting the scapula from the humerus, the images were used to 3D reconstruct the scapula, and the scapular axis (or Friedman’s axis [4]) was determined by drawing a line from the centre of the glenoid (determined by the average of the most superior, inferior, anterior, and posterior boundaries of the glenoid) to the trigonum (determined by the average of 3 points along the medial border of the scapula where the spines converge in a triangular shape). The planning software allowed the surgeon to virtually implant a glenoid component onto the 3D model of the shoulder and preoperatively assess which component and position would be most appropriate for each patient (Fig. 1).

Preoperative assessment for both conventional and navigated TSA cases followed the same principles. Glenoid version and inclination were planned for correction with either eccentric reaming, use of glenoid augment, or both. The glenoid component was aimed to be implanted to best fit the glenoid surface with at least 90% backside coverage, centered in aTSA or inferior in rTSA. The aim was to position the central cage of the component in the best position in the glenoid vault to avoid penetrating its cortex. For the non-navigated cases, conventional instrumentation was used to attempt to replicate this preoperative plan. For navigated cases, the navigation system was calibrated intraoperatively with a tracker placed on the coracoid process, and a probe was used to register bony landmarks that correlated the anatomy to the preoperative 3D model. GPS navigation was then used intraoperatively to precisely drill, ream, and position the glenoid component (Fig. 2). For both 2D- and 3D-planned surgeries, the aim in all cases was to correct the version to between 0° and 5° of retroversion while minimizing the amount of bone removal. Although all patients underwent routine postoperative radiographs of the affected shoulder at 6 weeks, 1 year, and 2 years, a detailed analysis of implant positioning was not performed. Given the purported benefit of computer navigation in improving implant alignment, this does limit the study’s ability to fully assess the benefit of computer navigation over conventional techniques in TSA.

Of the 40 aTSA performed, 12 used conventional instrumentation and 28 used navigation. Of the 140 rTSA performed, 80 were undertaken using conventional instrumentation and 60 with navigation. Patient demographics are summarized in Table 1. The mean age in the rTSA group was on average of 10 years older than in the aTSA group. Other than patient sex in the aTSA group, both navigated and non-navigated groups were otherwise very similar in terms of sex, side involved, and prior non-arthroplastic surgery to the involved shoulder.

Of the 180 patients, not all were available for review at 2 years. Of the 40 aTSA cases, one patient moved overseas and was lost to follow-up, two patients were unable to complete the questionnaires for medical reasons, and two patients elected not to be involved in the study. This left 35 patients who were available for review (87.5%), 11 in the non-navigated group and 24 in the navigated group. Of the 140 rTSA cases, two patients moved overseas and were lost to follow-up, three patients died due to causes unrelated to TSA, two patients were unable to complete the questionnaires for medical reasons, and 12 patients elected not to be involved in the study. This left 121 patients who were available for review (86.4%), 72 in the non-navigated group and 49 in the navigated group. Because the number of patients in the non-navigated aTSA group was small, analysis of this cohort was underpowered, limiting the findings.

Preoperatively, although the navigated cohorts for both aTSA and rTSA had better ASES scores, Oxford scores, pain scores, and VR-12 mental scores, only the VR-12 mental score in the aTSA group was significant (Table 2). Postoperatively, of the patients available for review at 2 years, significant improvements were seen in all PROMs in both cohorts for both aTSA and rTSA. All outcome metrics for both the non-navigated and navigated TSA cohorts achieved the minimum clinically important difference (MCID) and substantial clinical benefit thresholds [5], where relevant.

At the 2-year assessment, the navigated aTSA cohort demonstrated significant improvements compared to the non-navigated aTSA cohort in ASES score, Oxford score, and patient pain score. These did reach the MCID for ASES and Oxford scores, but the numbers were small. For the navigated rTSA cohort, significant improvements were seen in the Oxford score (Table 3). The remaining outcome scores demonstrated no significant difference at the 2-year follow-up. There was a not significant sex-based trend in both cohorts, with females having slightly better outcomes at 2 years compared to males for both aTSA and rTSA, better outcomes in patients who had not had prior surgery to the involved shoulder for both aTSA and rTSA, and better outcomes in rTSA patients younger than 65 years compared to older patients.

In a multivariate analysis of all the 2-year outcome scores related to covariates of age, sex, surgery type (anatomic versus reverse), primary versus prior non-arthroplasty surgery, and preoperative baseline values, all outcomes were more favorable for the navigated procedures (Table 4). The ASES, Oxford, and patient function scores all were significant but did not reach MCID thresholds. Radiographic data are essential for validating the precision of computer navigation over conventional techniques (implant version, inclination, rotation, seating). Analysis of this was outside the scope of the current study, which was primarily to assess clinical outcomes. This does limit the study ability to fully understand the benefits of computer navigation in TSA, and further studies are ongoing.

In the current study, there was a significant increase in the use of augments favoring the navigation group for both anatomic and rTSA (Table 5). When using conventional instrumentation, 30 of 92 cases involved the use of glenoid augments (32.6%). With navigation, this number increased to 79 of 88 cases (89.8%), a significant difference (P<0.001). In 51.1% of cases, the augmentation was 8° posterior. For rTSA, a superior augment was used in 23.9% of cases and a posterosuperior augment in 14.8% of cases.

Complications were more common in the non-navigated cohort for both aTSA and rTSA. Unfortunately, there were some implant-specific complications that were not related to TSA navigation. These included one case of polyethylene dissociation from the cage peg in aTSA that required revision to rTSA and three cases of liner dissociation from the baseplate in rTSA patients that all required component exchange. These implant problems have been addressed by the company with the introduction of the laser cage glenoid for aTSA and a new locking mechanism for the baseplate for rTSA. Excluding these implant-related issues, there were no other complications, reoperations, or revisions in the aTSA group. However, there was an additional nine complications in the rTSA group (6.4%), one in the navigated cohort and eight in the non-navigated cohort. The overall complication rate in the navigated rTSA group was 1.7% versus 10.0% in the non-navigated rTSA group.

The one complication in the navigated rTSA group was a prosthesis dislocation that occurred when the patient was performing pull-ups against medical advice. The eight complications in the non-navigated rTSA group included one nerve injury, three acromial fractures, and four prosthesis dislocations. All acromial fractures were successfully treated nonoperatively. Two of the dislocations required revision for recurrence. All three revisions in the rTSA cohort (2.1%) involved exchange of components only, with the humeral stem and glenoid baseplate left in situ. The differences in acromial fracture and dislocation rates favored the navigated cohort, as did that of the revision rate.

This study showed the value of computer navigation over conventional techniques when performing anatomic or rTSA, based on complications, implant survivorship, and PROMs. Previous studies have shown that navigation results in more accurate and precise implant placement (inclination and version), even in the hands of expert surgeons [6-10]. It is also more reproducible than conventional techniques, with a high degree of concordance between preoperative planning and final component selection [11,12], an important consideration given that patients who experience intraoperative deviations in their preoperative plan have inferior clinical and radiographic outcomes after TSA [13].

Translating the benefits of computer navigation in TSA into decreased complication rates and improved patient outcomes has previously not been reported. Greene et al. [14] in 2022 reported on 2-year minimum outcomes for patients following TSA performed with navigation and showed excellent results (as good or better) compared to non-navigated patients of similar age, sex, and follow-up matched cohort. Non-significant reductions in postoperative complications, revision rate, and adverse events were observed in the navigated aTSA patients. Significant reductions in postoperative complications, revision rate, and adverse events were observed in the navigated rTSA patients.

Another recent 2-year multi-centre outcome report comprising more than 500 rTSAs found that patients who underwent a navigation-guided procedure had more satisfactory IR and ER. They also found equivalent or better outcome scores and lower postoperative complication and dislocation rates with navigated rTSA, regardless of glenoid morphology [15]. They concluded that the use of computer navigation in both aTSA and rTSA was safe and produced, at minimum, similar outcomes at 2 years to standard instrumentation, without any increased risk of complications. With longer follow-up, That study hypothesized that longer follow-up would show computer navigation to improve patient outcomes and minimize complication rates.

However, Gaj et al. [16] reported no significant differences in range of motion (ROM), PROMs, and satisfaction between patients receiving navigated and standard rTSA at short-term follow-up. They reported that, despite more severe preoperative glenoid erosion in the navigated group, all patients were able to achieve an appropriate neutral axis postoperatively, although navigation did better optimize inclination and version. Their study size was small (33 patients), however, and the average length of follow-up was only 16 months.

Radiographic data are essential for validating the precision of computer navigation in TSA (implant version, inclination, rotation, seating), and the lack of radiographic analysis in this study does limit a more complete understanding of the benefit of computer navigation in TSA. However, there are plans to address this in future analysis. In another study, Holzgrefe et al. [17], in a matched cohort of 226 shoulders, found that both navigated and non-navigated rTSAs yielded similar improvement in ROM and function at early follow-up. Improvements did trend towards favoring the navigated group but did not reach significance. Notching and re-operation were more common in non-navigated shoulders. That study concluded that longer follow-up and larger cohort size are required to determine if navigation lengthens the durability of rTSA and reduces the incidence of complications.

In the current study, benefits were seen when navigation was used across all PROMs for both anatomic and rTSA, with the values for ASES, Oxford, and patient function scores all reaching significance when using a multivariate analysis. Although encouraging, the differences did not achieve MCID thresholds, highlighting the uncertainty of clinical impact of these differences. For rTSA, the complication, reoperation, and revision rates were also reduced in the navigation patients compared to patients where navigation was not used. The unrelated implant-specific complications was very unfortunate as they affected overall complication and revision rates for this implant.

It is interesting to speculate on the reasons for this improvement in patient outcome when using navigation for TSA. One factor may lie in the use of augments. Rosenthal et al. found that preoperative planning with 3D software changed our understanding of the glenoid anatomy [18]. Glenoid shape is illustrated differently, and version is calculated differently in 2D planned versus 3D planned surgeries, with the level of measurement used to determine version and inclination based on the average of multiple points. As a result, navigation has seen a dramatic increase in use of augmented glenoids, with one study reporting use in 54% to 76% of cases compared to 15% to 19% of cases utilizing 2D planning [18] and others reporting similar significant increases in the use of augmented baseplates [19].

Kida et al. [20] reported that, while planning software and navigation have increased the use of augmented glenoids, they have allowed less reaming of the glenoid to achieve the desired correction. Although there is a lack of data to support the correct version (neutral versus premorbid) [9], it is possible that this increased use of augments to correct the version to between 0° and 5° of retroversion resulted in more ideal reconstruction of the glenoid. This may have allowed better tensioning of the remaining soft tissues, which may have contributed to the improved outcomes seen with navigation.

The main advantage of navigation over other currently available techniques to assist with performing TSA is that it allows intraoperative feedback at each step of the procedure, allowing real-time modification of the preoperative plan; guidance of instrument and implant version, inclination, and rotation; and control of depth of glenoid reaming and positioning. It allows the best part of the glenoid vault to be harnessed for fixation, reducing the chance of glenoid-sided failure. Other advantages include improved rTSA baseplate screw trajectory to maximize screw length, placement, and configuration [19,21,22]. The navigated steps are identical for every surgery regardless of complexity, increasing consistency and confidence of the entire surgical team. This consistency creates a more streamlined workflow for more complicated implant preparation techniques such as when using augmented glenoids that require separate reaming and drilling axes.

As glenoid deformity increases, surgeon perception and ability with conventional instrumentation to execute accurately a surgical plan decreases [10,23]. Without the feedback of navigation, surgeons are less likely to consistently implement the planned position of the glenoid component. Even with PSI, which aims to increase the accuracy of initial guide pin placement, reaming depth cannot be assessed, off axis reaming and asymmetric seating during implantation can still occur, and there is no real-time feedback or control of implant insertion. One recent study reported no improvement in short-term outcome or adverse events after either anatomic or rTSA with PSI [24]. Another reported no significant 2-year differences in any clinical or radiological outcomes between standard and PSI rTSA [25], and another showed similar short-term functional outcomes following rTSA regardless of use of a standard or PSI guide [26].

Whether mixed reality or robot-assisted surgery can produce similar levels of improvement to computer navigation in patient outcomes, complications, and implant survivorship remain to be determined. Mixed reality is limited to 3D models, cadavers, and patients to a limited extent, but there is promise for more widespread application [27,28]. Although not yet commercially available, robot-assisted TSA confers the theoretical advantage of precise humeral head cuts, more accurate glenoid preparation, and improved soft tissue assessment in limited early studies [29,30]. More studies by authors without conflicts of interest are needed to clarify the place of these and other new technologies in the practice of TSA.

Despite the promising results shown in the current study, the use of navigation does have some limitations. These include a learning curve, added surgical time and cost, intraoperative navigation malfunctioning, and lack of portability [19,23]. While initial operating time can be longer, studies report that the learning curve flattens rapidly, with similar times after 8 to 25 procedures [18,31]. There are a few additional steps, but many steps in the conventional procedure can be eliminated, including estimation of proper version and inclination, placement of guide pins, use of drill guides or patient-specific blocks in PSI cases, and screw measurement. One recent study reported high intraoperative efficacy and a low complication rate when using navigation. They described the use of navigation as user friendly, with an intraoperative success rate of 98% and a low coracoid fracture rate of 0.05% [32]. The learning curve for the senior author was estimated to be between 10 and 15 cases, and the coracoid fracture rate was 1.1% after training. The cost difference for the senior author when using computer navigation was only NZ$575.00, and surgical times for both anatomic and rTSA were very similar to conventional times (±5 minutes).

This study does have several limitations. Follow-up is short, at a minimum of only 2 years, but this does capture the biggest initial improvement in PROMs and the immediate and early complications seen with TSA. However, some late complications including rotator cuff tears and implant loosening in aTSA and late dislocations and implant loosening in rTSA, will need longer-term follow-up. Additional limitations relate to the retrospective nature of this study and the absence of randomization of cases to receive either a non-navigated or navigated TSA. Although patient data were collected prospectively, the study aim was developed retrospectively. Selection bias may be present; for example, when computer navigation was first available and being utilized by the senior author, it was only used for the more complicated cases. This bias may have led to an underestimation of computer navigation benefit. There is also a potential single-surgeon bias. Although this was a single surgeon’s experience using the same implant type and navigation manufacturer, representing a gradual transition from non-navigated to navigated TSA in very similar (although not matched) patient populations, it is not known whether there is generalizability to other implant systems or other surgical teams. The cost-benefit of these changes with respect to mid- and long-term clinical outcomes and implant survival has not been determined. Finally, given that the aim of computer navigation is to optimize implant positioning, the lack of radiographic outcome data in this study is a significant limitation to the conclusions and will need to be addressed in future research.

This study shows that an integrated system of 3D CT planning, intraoperative computer navigation, and augmented glenoid components may decrease complication rates and improve patient reported outcomes in TSA compared to conventional techniques. Computer navigation lends itself nicely to TSA, allowing users to understand the 3D shape of the scapula in real time and receive instantaneous feedback about the position of the instrumentation and components during glenoid implantation. Ongoing developments including the addition of humeral navigation may see these benefits improved further. However, more research with longer-term follow-up and radiographic analysis is required to clarify the place of this and other new technology in the practice of TSA.