Comparing clinical and radiographic outcomes of robotic-assisted, computer-navigated and conventional unicompartmental knee arthroplasty: A network meta-analysis of randomized controlled trials

Abstract

Introduction

The purpose was to compare robotic assisted (RA), computer navigated (CN), and conventional UKA techniques.

Methods

Databases were queried for data on study characteristics, UKA systems, complications, and tibiofemoral alignment.

Results

Four RA and six CN RCTs were identified. No significant differences were found in operative time, tibiofemoral alignment, and reoperation rates when comparing RA or CN to conventional UKA. RA UKA resulted in a significantly lower risk of complications compared to conventional UKA.

Conclusions

RA UKA results in fewer complications than conventional UKA with a clinically significant increase in operative time. All groups were similar in remaining evaluated parameters.

1. Introduction

Unicompartmental knee arthroplasty (UKA) has been increasingly accepted as an alternative to total knee arthroplasty (TKA) for the appropriately indicated patient with isolated medial or lateral compartment osteoarthritis.1, 2, 3 In young patients with a long life expectancy, UKA can be regarded as a temporizing procedure, delaying the need for a TKA. However, in older and more sedentary patients with advanced, isolated unicompartmental arthritis and correctable deformity, UKA can sometimes obviate the need for TKA.⁴ Recent studies have demonstrated that outcomes of UKA are at least comparable to TKA in terms of improved pain and function with the additional benefit of faster recovery times.^5,6 The faster recovery associated with the inherent joint-preserving nature of the procedure has likely contributed to the popularity of UKA and broadening of its initially strict indications. However, despite these benefits, some studies have reported revision rates approaching 30% with the use of older implant designs,⁷ while a recent systematic review found that UKA had a relative risk of 5.4 for all-cause revisions in comparison to TKA at a mean 4.8-year follow-up.⁶

As the number of UKA procedures continue to increase annually, it is imperative to implement methods or surgical techniques which may increase the survivorship of UKA, especially in the context of alternative payment models given the often-high cost associated with complications. Robotic-assisted (RA) and computer navigation (CN) assisted techniques have gained popularity because of the proposed benefits of increased accuracy and consistency of component placement and implantation.⁸ Further, improved component positioning may reduce adverse events such as aseptic loosening, polyethylene wear and ultimately avoid the high costs associated with revision arthroplasty and readmissions. However, it is debated whether the potential benefits of CN and RA techniques actually improves long term survivorship and whether the increased relative costs of the advanced technology are justifiable based on these advantages in the long-term.9, 10, 11 The conflicting opinions may be partially due to confounding factors and bias present in retrospective or lower-quality case-control studies inherent to these study designs.11, 12, 13, 14 Thus, a meta-analysis of the high quality, randomized controlled trials (RCTs) is necessary to adequately compare results between conventional, RA, and CN assisted UKA.

Utilizing all the RCTs on the topic, the purpose of the current meta-analysis was to compare operative times, early adverse events, and postoperative tibiofemoral alignment between RA, CN, and conventional UKA techniques.

2. Methods

2.1. Article identification and selection

This systematic review and meta-analysis was performed in accordance with the 2009 Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) guidelines.¹⁵ Queries were performed in August 2020 utilizing the PubMed, OVID/Medline, and Cochrane Databases of Systematic Reviews and Central Register of Controlled Trials. The search query can be found in Supplement I.

All identified articles were subjected to the following inclusion criteria: (1) publication in the English language and (2) RCT study design with outcomes pertaining to the use of robotic or computer-navigated UKA. Exclusion criteria consisted of (1) non-RCTs, (2) cadaveric studies; (3) animal studies; (4) basic science articles; (5) editorial articles; (6) surveys or case reports; and (7) RCTs that did not report one of the following outcomes: operative time, adverse events including complications and reoperations, postoperative lower extremity alignment and component position measures. When authors published on identical study cohorts at various follow-up times, the study with longer-term follow-up was included and the study with shorter-term follow-up was excluded. Additionally, we did not restrict the follow-up to any minimum time period in order to capture all potential complications and revisions, and based on the fact that operative time and initial component position and lower extremity alignment are not dependent on follow-up time. Two investigators independently reviewed the abstracts from all identified articles. Full-text articles were obtained for review to allow further assessment of inclusion and exclusion criteria when necessary. Bibliographies of relevant systematic reviews and included articles were manually searched for additional references.

2.2. Quality assessment measures

The JADAD Scale¹⁶ was used to assess the methodological quality of all included randomized controlled trials (RCTs). The JADAD Scale consists of a five-point questionnaire used to critically evaluate the methodological quality of RCTs. The scale is graded from 0 to 5 (a score of greater than or equal to 3 indicates a high-quality study, whereas a score less than 3 is considered to be low quality). The inter-observer reliability for the two independent graders (blinded for review) was excellent at 0.92 (95% Confidence interval, 0.84–0.97).

2.3. Network meta-analysis

The primary outcomes were (1) operative time, (2) all-cause complications, (3) all-cause reoperations, and (4) postoperative tibiofemoral alignment. All events which were classified as complications by the authors of each RCT were included as complications in this meta-analysis, regardless of severity. A frequentist framework network meta-analysis¹⁷ was conducted for all outcomes of interest that were reported in a minimum of three studies and if the outcome was reported in each of the three treatment arms. A network meta-analysis is useful in that it is capable of comparing multiple treatments through considering both direct (comparisons from studies that include both treatment arms) and indirect evidence (comparisons that did not include both treatment arms within a study). The assumption of consistency from direct and indirect evidence under the random-effects model was assessed using the Q-statistic.¹⁸ Study heterogeneity was assessed using I-squared (I²) tests.

The three competing treatments were “ranked” in the network meta-analysis by using point estimates and standard errors to calculate P-scores.¹⁹ P-scores represent a measure of certainty that a specific treatment is superior to other treatments for a given outcome. The higher the P-score for a given treatment, the more certainty it can be concluded that the treatment is superior in comparison to the other treatments for a given outcome. We utilized the term “treatment ranking” to order the treatment arms by P-score from best to worst treatment for each outcome.

If outcomes did not satisfy these indications, they were reported descriptively. Furthermore, due to variability in outcome reporting, some data could not be analyzed quantitatively and were reported descriptively. These included (1) axial, sagittal, and coronal plane alignment of implant position. Demographic data was reported using means and ranges or frequency with percentages where appropriate. All statistical analyses were performed using R Project for Statistical Computing software (RStudio software version 1.2.1335, R Foundation for Statistical Computing, Vienna, Austria).

3. Results

3.1. Characteristics of included studies

A total of ten RCTs were included in the final qualitative and quantitative analysis (Fig. 1).20, 21, 22, 23, 24, 25, 26, 27, 28, 29 Of these studies, a total of six (60%) investigated computer-navigation versus conventional UKA, while the remaining four (40%) investigated robotic-assisted versus conventional UKA. In three studies, robotic-assisted UKA was performed with the MAKO Robotic System (MAKO Surgical Corporation, Fort Lauderdale, FL, USA), while one study used the Acrobat (The Acrobot Co. Ltd., London, UK) robotic system (Table 1). Various computer-navigation systems were utilized in their respective studies (Table 2). For conventional UKA, a wide variety of implant systems were utilized: five (50%) studies used only mobile bearing implants, three (30%) studies used fixed bearing, and two (20%) studies used both.

Fig. 1.

Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) guidelines for included studies.

Table 1.

Included studies using robotic-assisted UKA.

Author, year	Patients (conventional)	Patients (robotic)	Robotic system	Inclusion Criteria	Exclusion Criteria
Cobb et al., 2006	14	13	Acrobot System (The Acrobot Co. Ltd., London, UK)	•18 years or older •Limitation of disease to medial compartment •Intact anterior cruciate ligament •Varus deformity which could correct fully in 10° of flexion	•Any factors which would jeopardize completion of the trial
Bell et al., 2016	58	62	MAKO Robotic System (MAKO Surgical Corporation, Fort Lauderdale, FL, USA)	•Patients planned to undergo UKA for isolated medial compartment osteoarthritis	•Radiological evidence of osteoarthritis affecting the lateral compartment or lateral facet of the patellofemoral compartment •Deficient anterior cruciate ligament •Fixed flexion deformity of more than 10° •Varus deformity of more than 10°
Blyth et al., 2017	62	64	MAKO Robotic System (MAKO Surgical Corporation, Fort Lauderdale, FL, USA)	•Considered suitable for UKA by the senior authors. •Capable of providing written informed consent •Capable of complying with the study follow-up regime.	•Any contraindications detailed by device manufacturer •Any tibial deformity requiring tibial component augmentation •Requirement for a total knee arthroplasty •Inflammatory polyarthritis •Disorder of contralateral knee, feet, ankles, hips, or spine •Neurological condition affecting movement •Any condition in opinion of investigator that would compromise participation and follow-up
Gilmour et al., 2018	54	58	MAKO Robotic System (MAKO Surgical Corporation, Fort Lauderdale, FL, USA)	•Considered suitable for UKA by the senior authors. •Capable of providing written informed consent •Capable of complying with the study follow-up regime.	•Ligament insufficiency •Inflammatory arthritis •Deformity requiring augmentation •Neurological movement disorders •Pathology of the feet, ankles, hips, or opposite knee causing significant pain or gait alterations •Patients requiring total knee arthroplasty

Table 2.

Studies using computer navigation-assisted UKA.

Author, year	Patients (conventional)	Patients (CN)	Navigation/Imaging system	Inclusion Criteria	Exclusion Criteria
Keene et al., 2006	10	10	Ci Preservation System (Munich, Germany)	•Bilateral medial compartment osteoarthritis awaiting bilateral simultaneous UKA	•Not reported
Jean-Yves Jenny 2008	30	•30: Conventional Open Navigated Technique •30: MIS Experimental Navigated Technique •30: MIS Navigation-dedicated Technique	Orthopilot Navigation System (B. Braun Aesculap, Tuttlingen Germany)	•Medial compartment osteoarthritis	•Not reported
Lim et al., 2009	21	30	Orthopilot Navigation System (B. Braun Aesculap, Tuttlingen Germany)	•Participate in low demand activities •Range of motion of at least 115° flexion •Less than 10° flexion contracture •Less than 10° of varus/valgus malalignment •Stable knee joint with absence of subluxation, varus or valgus thrust on ambulation •Minimal involvement of contralateral tibiofemoral and patellofemoral compartments •Absence of underlying tibiofemoral osseous pathology	•Not reported
Ollivier et al., 2016	30	30	Patient specific information using MRI protocol (Materialise™, Leuven, Belgium)	•Isolated symptomatic medial femorotibial knee arthritis •Varus deformity •Between 50 and 85 years of age •Acceptance of a new technology protocol (including delay between MRI and surgery)	•ROM below 0°–100° (extension to flexion) •Unstable knees in frontal and/or sagittal planes •Personal history of trauma, sepsis, tumor, inflammatory or skeletal disease (that could influence gait parameters) •Previous lower limb joint (ankle to hip) surgery that could lead to an artifact effect on imaging •Any contraindication to MRI
Zhang et al., 2016	41	40	Vector Vision Version 1.52 (BrianLAB, Munich, Germany)	•Pain in a single compartment secondary to osteoarthritis or necrosis •>60 years age •Weight <82 kg •Sedentary lifestyle •Range of motion >90° •Flexion contracture <5° •Angular deformities <10–15°	•Systemic or inflammatory arthritis •Knee instability or subluxation •Fixed flexion contracture •Loss of anterior or posterior cruciate ligaments •Intraoperative finding of eburnated bone in either the patella or the opposite compartment
Alvand et al., 2018	22	23	“Signature” System (Zimmer Biomet Inc, Warsaw, IN, USA)	•Both cruciate ligaments functionally intact •Full thickness cartilage in lateral compartment •Correctable intra-articular varus deformity (based on clinical assessment) •Full-thickness cartilage loss in the medial compartment	•Contraindications for MRI •All forms of inflammatory arthritis

Seven studies (70%) were published within the past ten years,^21,22,24,25,27, 28, 29 while the remaining three between 2006 and 2008.^20,23,26 One study did not report follow-up as they only reported immediate postoperative component position during the hospital admission.²⁴ The mean follow-up across the remaining nine studies was 15.6 (range, 1.4–48) months.20, 21, 22, 23^,25, 26, 27, 28, 29 The mean follow-up for robotic-assisted RCTs was 16.8 (range, 3–48) months, which was not statistically different in comparison to the mean follow-up of 15.3 (range, 1.4–48) months for computer-navigation RCTs (p = 0.91).

3.2. Methodological quality of included studies

The mean JADAD score among all ten studies was 3.6 ± 1.3 (Table 3), indicating high quality study methodology on average. The JADAD score ranged from one to five, with a total of three (30.0%) studies scoring a 5/5 and indicating no methodological flaws in their RCT study designs and reporting. The mean JADAD score for the robotic studies was 4.3 ± 0.96, while the mean JADAD score for the computer-navigation studies was 3.2 ± 1.5 (p = 0.23).

Table 3.

JADAD scores for included studies.

Author, year	JADAD Score
Robotic
Cobb et al., 2006	5
Bell et al., 2016	3 (blinding method not reported, reasons for withdrawal not reported)
Blyth et al., 2017	4 (reasons for withdrawal not reported)
Gilmour et al., 2018	5
Computer-navigation
Keene et al., 2006	3 (blinding not addressed)
Jean-Yves Jenny 2008	2 (randomization method not reported, blinding not addressed)
Lim et al., 2009	1 (inappropriate randomization, blinding not addressed)
Ollivier et al., 2016	4 (blinding method not reported)
Zhang et al., 2016	4 (blinding method not reported, reasons for withdrawal not reported)
Alvand et al., 2016	5

3.3. Operative time

Operative time was reported in four (40%) studies.^20,22,25,26 Three studies compared conventional and computer navigation-assisted UKA directly (n = 146),^20,22,25 while one study compared conventional and robotic-assisted UKA directly (n = 27).²⁶ When operative times were pooled for all RCTs, the mean operative time for was 104 min for robotic-assisted UKA, 66.7 min for conventional UKA, and 68.2 min for computer navigation-assisted UKA.

Results from the network-meta analysis demonstrated that the mean difference in operative time was 16.0 min longer for robotic-assisted UKA when compared to conventional UKA, and 4.38 min longer for computer-navigation UKA compared to conventional UKA. Treatment rankings based off of the network meta-analysis were as follows: (1) conventional UKA (P-score = 0.82), (2) computer navigation-assisted UKA (P-score = 0.54), and (3) robotic-assisted UKA (P-score = 0.15), indicating that conventional UKA on average had shorter operative times (Fig. 2). However, this trend did not reach statistical significance (Table 4).

Fig. 2.

Forest plot displaying the mean difference (MD) and 95% confidence interval (CI) in operation length in minutes for robotic and navigation-assisted UKA relative to conventional UKA. I²⁼96%. Note positive values indicate the added time required to perform the operation compared to conventional UKA.

Table 4.

Pairwise comparison of UKA methods in the network meta-analysis.

Outcome	Treatment Comparison	MD estimate (95% CI)	P-value
Operation Length
	Robotic vs Conventional	16.0 (−7.28, 39.28)	0.18
	Navigation vs Conventional	4.38 (−9.83, 18.58)	0.55
	Navigation vs Robotic	−11.62 (−38.90, 15.65)	0.40
Tibiofemoral Alignment Angle
	Robotic vs Conventional	1.49 (−6.23, 9.22)	0.71
	Navigation vs Conventional	−1.88 (−6.31, 2.55)	0.40
	Navigation vs Robotic	−3.37 (−12.27, 5.53)	0.46
		OR estimate (95% CI)	P-Value
Complications
	Robotic vs Conventional	0.39 (0.18. 0.85)	0.018
	Navigation vs Conventional	0.98 (0.13, 7.7)	0.99
	Navigation vs Robotic	0.40 (0.04, 3.61)	0.41
Revisions
	Robotic vs Conventional	0.22 (0.02, 2.33)	0.21
	Navigation vs Conventional	0.99 (0.06, 16.21)	0.99
	Navigation vs Robotic	0.22 (0.01, 8.61)	0.42

Reference group is listed second in each respective row of the treatment comparison column.

Bold values indicate statistical significance (P < 0.05).

Abbreviations: OR, odds ratio; MD, mean difference; CI, 95% confidence interval.

3.4. Complications

Complications were reported in six (60%) studies.20, 21, 22^,25,26,29 Three studies directly compared conventional and computer navigation-assisted UKA (n = 152),^20,22,25 while three studies directly compared conventional and robotic-assisted UKA (n = 265).^21,26,29 None of the studies reported an intraoperative complication. In studies directly comparing conventional and computer navigation-assisted UKA, one of 73 patients who underwent conventional UKA experienced a complication (surgical site infection), while one of 79 patients who underwent computer navigation-assisted UKA experienced a surgical site infection (Table 5). In studies directly comparing conventional and robotic-assisted UKA, 53 of 130 patients who underwent conventional UKA experienced a complication, while 28 of 135 patients who underwent robotic-assisted UKA experienced a complication (Table 5). Treatment rankings based off of the network meta-analysis were as follows: (1) robotic-assisted UKA (P-score = 0.89), (2) computer navigation-assisted UKA (P-score = 0.36), and (3) conventional UKA (P-score = 0.25), indicating that robotic-assisted UKA was associated with fewer complications (Fig. 3). The network-meta analysis also indicated that performing robotic-assisted UKA was associated with a decreased likelihood of complications compared to conventional UKA (Odds ratio [OR]: 0.39, p = 0.018). Robotic-assisted UKA did not result in fewer complications compared to computer navigation-assisted UKA (OR: 0.40, p = 0.41). Likewise, computer navigation-assisted UKA was not associated with fewer complications when compared to conventional UKA (Table 4).

Table 5.

Revision and complication rates and events among studies stratified by THA approach.

Outcome	Robotic	Computer Navigation	Conventionala	Conventionalb	Conventionala,b
Revisions	Total = 0 •No revisions	Total = 0 •No revisions	Total = 4 •Aseptic loosening (N = 1) •Infection (N = 1) •Dislocation (N = 1) •Pain (N = 1)	Total = 0 •No revisions	Total = 4 •Aseptic loosening (N = 1) •Infection (N = 1) •Dislocation (N = 1) •Pain (N = 1)
Complication	Total = 28 •Wound dehiscence/leakage (N = 14) •Surgical site infection (N = 4) •Acute urinary retention (N = 1) •Swollen ankle related to surgical procedure (N = 1) •Swollen leg related to surgical procedure (N = 1) •Stich abscess (N = 2) •Erythema/swelling in dry wound (N = 5)	Total = 1 •Surgical site infection (N = 1)	Total = 53 •Wound dehiscence/leakage (N = 17) •Surgical Site Infection (N = 10) •Dislocation (N = 1) •Nerve Palsy (N = 1) •Blood transfusion (N = 1) •Myocardial Infarction (N = 1) •Stich abscess (N = 12) •Erythema/swelling in dry wound (N = 5)	Total = 1 •Surgical site infection (N = 1)	Total = 54 •Wound dehiscence/leakage (N = 22) •Surgical Site Infection (N = 11) •Dislocation (N = 1) •Nerve Palsy (N = 1) •Blood transfusion (N = 1) •Myocardial Infarction (N = 1) •Stich abscess (N = 12) •Erythema/swelling in dry wound (N = 5)

^aConventional treatment arms in robotic studies.

^bConventional treatment arms in navigation studies.

Fig. 3.

Forest plot displaying the odds ratio (OR) and 95% confidence interval (CI) of complications for robotic and navigation-assisted UKA relative to conventional UKA. I² = 23.6%. Note a value less than one indicates a lower risk of complications relative to conventional UKA.

3.5. All-cause reoperations

Reoperation rates were reported in four (40%) studies.^21,22,25,29 Two studies directly compared conventional and computer navigation-assisted UKA (n = 126),^22,25 while two studies directly compared conventional and robotic-assisted UKA (n = 238).^21,29 In studies directly comparing conventional and computer navigation-assisted UKA, no patients underwent a reoperation procedure during the follow-up period (Table 5). In studies directly comparing conventional and robotic-assisted UKA, six patients who underwent conventional UKA required reoperations, while no patients who underwent robotic-assisted UKA required a reoperation (Table 5). These included two polyethylene exchanges, one revision for aseptic loosening, one revision for dislocation, one revision for infection, and one revision for persistent pain. Treatment rankings based off of the network meta-analysis were as follows: (1) robotic-assisted UKA (P-score = 0.88), (2) computer navigation-assisted UKA (P-score = 0.34), and (3) conventional UKA (P-score = 0.28), indicating that robotic-assisted UKA was the best treatment in terms of P-Score to reduce the incidence of reoperations after UKA (Fig. 4). However, no statistically significant differences were found in reoperation rates between RA, CN and conventional UKA (Table 4).

Fig. 4.

Forest plot displaying the odds ratio (OR) and 95% confidence interval (CI) of postoperative revision for robotic and navigation-assisted UKA relative to conventional UKA. I²⁼0%. Note a value less than one indicates a lower risk of revision relative to conventional UKA.

3.6. Postoperative tibiofemoral alignment

The tibiofemoral alignment angle was measured in four (40%) studies. However, one study included patients who underwent UKA for both medial and lateral compartment osteoarthritis and therefore was excluded,²⁵ leaving three studies to be included in the meta-analysis. A negative tibiofemoral angle indicated valgus alignment, while a positive angle indicated varus alignment. Two studies compared conventional and computer navigation-assisted UKA directly (n = 171),^23,24 while one study compared conventional and robotic-assisted UKA directly (n = 27).²⁶ The mean difference in the tibiofemoral alignment angle was 1.49° for robotic-assisted UKA compared to conventional UKA and −0.09° for computer navigation-asssited UKA compared to conventional UKA. Treatment rankings based off of the network meta-analysis were as follows: (1) robotic-assisted UKA (P-score = 0.96), (2) conventional UKA (P-score = 0.30), and (3) computer navigation-assisted UKA (P-score = 0.26), indicating that based on P-scores robotic-assisted UKA was the best way to avoid overcorrection in to valgus alignment. However, no statistically significant differences were observed among the three UKA methods (Table 4).

Zhang et al.²⁵ reported postoperative tibiofemoral alignment angles for patients who underwent either conventional or computer navigation-assisted UKA for both medial and lateral compartment osteoarthritis. This group reported that the postoperative tibiofemoral alignment in patients who underwent computer navigation-assisted UKA was significantly more varus than those who underwent conventional UKA at minimum two-year follow-up (p = 0.033). Furthermore, they reported that computer navigation-assisted UKA resulted in significantly less internal rotation of the femoral component compared to conventional UKA (p = 0.025).

3.7. Component position

Other measures of component orientation varied greatly and were not amenable to meta-analysis. These included tibial and femoral component position in the axial, sagittal, and coronal planes, the mechanical axis, and achievement study-specific correction goals. Bell et al.²⁸ found that a significantly greater proportion of patients who underwent robotic-assisted UKA had component implantation within 2° of planned target positions when compared with the group who underwent conventional UKA for the sagittal, coronal, and axial positions of the femoral component, as well as the sagittal and axial positions of the tibial component.

Ollivier et al.²⁷ reported that they did not observe any differences in mechanical axis alignment between the conventional (178° ± 4°) and PSI groups (178° ± 3°, p = 0.24) and that there were no significant differences in implant position on mediolateral and anteroposterior radiographs between the two cohorts (p > 0.05). Jean-Yves Jenny²³ compared the femoral and tibial sagittal and coronal implant positions among conventional UKA, conventional open-navigated UKA, minimally invasive, experimentally navigated UKA, or minimally invasive, standard navigation UKA and found that in all cases, the navigated techniques outperformed the conventional UKA technique in achieving a higher proportion of component positioning within 2° of the planned target ranges. Alvand et al.²² compared conventional and computer navigation assisted UKA and found that both techniques resulted in similar component alignment and positioning. Keene et al.²⁰ reported that the mean variation between the pre-operative plan and achieved correction in patients who underwent computer navigation-assisted UKA was 0.9 ± 1.1°, while for conventional UKA it was 2.8 ± 1.4° (p < 0.001). Furthermore, a significantly lower proportion of conventional UKA patients had their postoperative lower limb alignment within 2° of the pre-operative plan compared to patients who underwent computer navigation-assisted UKA (60% vs. 97%, p < 0.05).

4. Discussion

The main findings of the current study are as follows: (1) Robotic-assisted UKA was found to result in a significantly lower likelihood of all-cause complications compared to conventional UKA; (2) no significant differences were found in operative time, all-cause reoperations, and postoperative tibiofemoral alignment among the three treatment groups, though the use of robotic or computer navigation-assisted trended towards increased operative times; and (3) the majority of data not amenable to meta-analysis suggested that robotic-assisted and computer navigation-assisted UKA conferred an advantage in achieving target component positioning when compared to conventional UKA methods.

The mean difference in operative time as quantified through the network meta-analysis was 16.0 min longer when comparing robotic-assisted UKA to conventional UKA and 4.4 min longer when comparing computer navigation-assisted UKA to conventional UKA. However, these mean differences were not statistically significant. These findings are in opposition with previous literature which have suggested that the use of robotic or computer navigation for UKA consistently results in longer operative times than the conventional UKA techniques.^11,30,31 As the current study synthesized evidence from only level one RCTs, the discordance in results may be a function of confounding factors inherent in lower quality studies. The finding in the current study is clinically important, as it suggests that robotic-assisted UKA may not significantly increase operative time and therefore increased costs associated with longer operating room time requirements may be avoided. However, it is notable that when operative times were pooled for all RCTs and studies were not compared individually in the meta-analysis, there was notably a 40-min increase in operative time with robotic-assisted UKA compared to conventional UKA. This finding is more is consistent with previous literature and likely represents a clinically significant increase, though this comparison does not consider the variability of operative times among all included RCTs and the heterogeneity in studies as they were performed by different surgeons and settings on diverse patient cohorts. This discrepancy suggests that in some cases, the use of P-scores to rank treatments may be misleading, and we encourage readers to interpret these results with caution. Though operative times as a function of the P-score and mean differences from the meta-analysis were not significantly different, this suggests that variation in length of operative times were due to individual studies. For example, RA UKA operative time may have been largely weighted by one included study from 2006 where the authors reported a mean operative time of over 100 min.²⁶ It is possible that data from this study was published during a learning curve for the authors. As more studies continue to accumulate on this topic in conjunction with improved technology and greater surgeon experience, such discrepancies may dissipate, and the mean operative time for RA UKA in more contemporary series may approach times much shorter 104 min.

The likelihood of experiencing a postoperative complication with the follow up period (average: 15.6 months) was 71% lower when performing robotic-assisted UKA in comparison to conventional UKA. This finding is in accordance with a recent meta-analysis of 11 Level I-III studies comparing clinical outcomes between robotic-assisted and conventional UKA between 30 days and 5-years of follow-up, which found that robotic-assisted UKA was associated with a significantly reduced complication rate (relative risk: 0.62, 95% confidence interval: 0.45–0.85; p = 0.0041).³¹ Interestingly, none of the studies reported an early postoperative fracture at the pin insertion site in computer assisted or robotic UKA. Although the learning curve associated with utilizing robotic-assisted UKA must first be overcome, more inexperienced arthroplasty surgeons may benefit from adopting the use of such technology and by potentially avoiding early complications associated with conventional UKA that may ultimately lead to TKA conversion. In addition, the fixed and per-unit cost of robotic-assisted UKA should be considered in this context, as previous literature has demonstrated that robotic-assisted UKA is cost-effective compared to conventional UKA only when annual case volume exceeds 94 cases, while it was not cost-effective for low- and medium-volume centers.³² Therefore, low-volume arthroplasty surgeons may benefit from the use of this technology to gain the benefits of a potentially lower rate of early complications. However, it is worth noting that the majority of complications in the current study were likely not related to the choice of UKA technology, as 50% and 40.7% of the complications in the robotic-assisted and conventional cohorts respectively were related to wound dehiscence/leakage, and the next most frequent complications were surgical site infections. Future, longer-term studies with large sample sizes are warranted to further investigate the relationships between the use of computer assisted and robotic technologies and longer-term complications such as aseptic loosening, wear, and progression of osteoarthritis, as the low incidence of events identified by the current meta-analysis cannot provide information regarding these scenarios.

Interestingly, there were no differences the rate of reoperations across the three UKA groups. One liner dislocation of a mobile bearing was observed in the conventional UKA group which ultimately required a revision procedure, while another patient in the conventional group required a revision to TKA for aseptic loosening, one required a revision to TKA for infection, one required a revision to TKA for persistent pain, and two patients required polyethylene exchanges. Although the incidence of liner dislocations and aseptic loosening are low, none of these events were observed in the robotic-assisted studies. Though the current study may have been underpowered to detect a difference in the primary end-point of all-cause reoperations, the observation that no early reoperations occurred in either the robotic and computer-navigation UKA groups, while six occurred in the conventional UKA group, still warrants discussion. Indeed, a recent case-control study of 80 medial and lateral UKA demonstrated that 86% of revisions in the control group occurring in association with component malposition or limb malalignment, compared with none in the robotic-assisted group.³³ Furthermore, the results of this study corroborate survivorship data from previous literature that has reported comparable revision and reoperation rates between robotic or computer navigation-assisted and conventional UKA.^21,22,29,31 Ultimately, the short-term follow-up of the included studies presents a limitation in assessing all-cause revision rates, and it is also possible that the current meta-analysis was underpowered to detect a difference in reoperations, making it potentially inappropriate to draw conclusions given the low incidence of adverse events among the three treatment groups.

Despite no significant differences in postoperative tibiofemoral alignment among the three UKA groups, many studies indicated that a greater proportion of femoral and tibial components were placed within a planned target range when using robotic or computer navigation-assisted UKA when compared with conventional UKA.^20,23,28 Interestingly, results from the meta-analysis suggested relative overcorrection with the use of computer navigation-assisted systems (−0.09 and −1.58° compared to conventional and robotic-assisted UKA, respectively), though these differences were small and not statistically significant. Unfortunately, the heterogeneity with respect to positioning of tibial and femoral components, in addition to the multitude of different computer navigation systems, limits the ability to draw meaningful conclusions from these findings. Interestingly, a recent study by Sekiguchi et al. suggested that preferred tibial component alignment is between neutral and 2° varus in the coronal plane, while varus >4° or valgus alignment caused excessive translation, could be related to feelings of instability, and have potentially negative effects on clinical outcomes and implant durability.³⁴ Although not statistically significant, the pooled tibiofemoral angle of robotic-assisted UKA was 1.5° varus, while both conventional and computer navigation-assisted UKA were in slight valgus. Of note, the pooled 0.01-degree tibiofemoral angle of conventional UKA falls within the standard error of measurement error for this variable previously reported in the literature (±0.72–1.21°),³⁵ and therefore could be considered neutral.

4.1. Limitations

Several limitations should be considered in the context of the current study results. First, due to variable reporting and outcome heterogeneity, we were unable to perform a network meta-analysis of patient reported outcome measures and other measures of component positioning that may have clinical implications. Secondly, publication bias may have influenced study results, which is an inherent limitation of all systematic reviews. Third, a wide range of optimal tibiofemoral angle targets and degrees of correction were used across the studies, which limits the ability to draw conclusions for results of the analysis concerning the tibiofemoral alignment angle. Fourth, a multitude of different systems were used in the computer navigation-assisted studies, and therefore conclusions cannot be drawn regarding any one particular system based on the results from this study. Fifth, studies reporting on surgical site infections did not define their diagnostic criteria used for this outcome. Sixth, the ability to draw conclusions regarding long-term outcomes including complications and revisions are limited given the short-term follow-up in all of the included studies.

5. Conclusions

RA UKA results in fewer complications than conventional UKA with a clinically, but not statistically significant, increase in operative time. CN UKA confers similar operative times, rates of adverse events, and achievement of target tibiofemoral alignment to both RA and conventional UKA.

Author contributions

KNK wrote the manuscript and conducted all statistical analyses. DF collected and provided all data. All authors reviewed the final manuscript.

Appendix A. Supplementary data

The following is the Supplementary data to this article: