Introduction

It was reported that an estimated 45,000 primary unicompartmental knee arthroplasties (UKA) are performed every year in the USA for treating end-stage unicompartmental knee osteoarthritis [1, 2]. Despite numerous advances in prosthesis design, instrumentation, and postoperative rehabilitation, only 70–86% patients are satisfied with the functional outcome while revision rates have been reported between 10% and 20% [3,4,5,6]. It is known that component positioning and soft tissue balance are crucial for successful UKA [7]. Malalignment in the coronal, sagittal and axial direction can lead to eccentric loading, wear, instability, aseptic loosening of the components, and patellofemoral problems [8,9,10]. Consequently, robotic-assisted orthopedic surgery was implemented to improve preoperative planning, accuracy of implant positioning, and precision in bone cuts during UKA.

The first robotic-assisted orthopedic surgery was introduced clinically in 1992, in order to reduce complications, to improve patient satisfaction, and to improve total joint arthroplasty by maintaining the surgeon’s ability to reproduce prosthesis alignment and therefore better restore normal kinematics. A robotic system for total knee arthroplasty (TKA) was first reported in 1993, but it took several more years until a robotic system for UKA became available and used clinically [11, 12]. At present, approximately 15–20% of UKA operations are performed with robotic assistance [13].

Hansen et al. [14] performed a retrospective study in a matched sample of patients on robotic-assisted UKA versus conventional UKA to assess differences between procedures. The results showed no clinical and radiographic difference in outcomes; however, a recent, prospective, randomized controlled study conducted by Bell et al. [15] concluded that robotic-assisted UKA leads to observably improved accuracy of prosthesis positioning (sagittal, coronal, and axial planes) compared to conventional UKA. The complication rates in the previous studies varied greatly, which makes it difficult to estimate the safety outcomes of the two surgical techniques. The purpose of the current study was to conduct a systematic review and meta-analysis to evaluate the efficiency and safety outcomes of robotic-assisted UKA and conventional UKA for treatment of end-stage unicompartmental knee osteoarthritis, with the aim of providing more information to aid in clinical decision making.

Material and methods

This meta-analysis was carried out in accordance with the recommendations of the Cochrane Collaboration and the Quality of Reporting of Meta-analyses (QUORUM) guidelines.

Search strategy

Of the investigators two (JF and YNW) independently searched the PubMed, MEDLINE, OVID, Web of Science, Cochrane Library (Cochrane Central Register of Controlled Trials, Cochrane database of systematic reviews, Cochrane Methodology Register) and EMBASE databases to retrieve all relevant studies published before 1 April 2017. The search strategy was based on combinations of Medical Subject Headings (MeSH) and the following keywords: unicompartmental knee or unicondylar knee or partial knee or UK, and arthroplasty or replacement, and robotics or remote operation. The “related articles” function was used to expand the literature search.

Inclusion and exclusion criteria

Studies were included in our meta-analysis if they met the following criteria: (1) randomized controlled trial (RCT) or quasi-RCT and prospective cohort trial (PCT), (2) study objects with end-stage unicompartmental knee osteoarthritis, (3) undergoing robotic-assisted UKA or conventional UKA and (4) full-text manuscripts were available. With respect to quasi-RCTs, any trial that made mention of a quasi-randomized method of patient allocation, such as by medical record number, admission date, or date alteration was considered. On condition that there was no mention of randomized method, the corresponding authors of these studies were contacted in e‑mail for more information.

Exclusion criteria were as following: (1) non-English, (2) review, letter or comment, (3) duplicate publication and (4) the results of comparison were not reported or the data could not be extracted from the published results.

Data extraction

Two investigators (JF and YNW) manually checked out the entire literature lists of these selected studies in order to find out all potentially relevant studies that were not retrieved during the previous database searches. The extracted data included: the study characteristics (study date, study sample in intervention/control, study design, intervention/control), primary and secondary outcomes data and length of follow-up. According to the outcome type, we either extracted means and standard deviations or rates (as number of events out of total number). If standard errors were given in some studies, they were converted to standard deviations by multiplying the standard error by the square root of the number.

Primary and secondary outcomes

The primary outcome for this meta-analysis was implantation accuracy of component positioning, including implantation errors of tibial and femoral components and tibiofemoral component safe zone. The implantation accuracy of component positioning was evaluated in the literature by postoperative CT or X‑ray by comparing the target positioning values in the preoperative plan with the actual values achieved postoperatively. Accuracy was therefore calculated by the degree of deviation from the preoperative planned target values rather than by the absolute values of the component position. The component implant position was collected in seven degrees of freedom (three degrees of translation, in the medial-lateral, anteroposterior and proximal-distal directions, three degrees of rotation, in the varus-valgus, flexion-extension, and internal-external rotation directions and tibial posterior slope). Tibiofemoral component safe zone was determined as angles of tibiofemoral component implantation within 2° of the target position.

The secondary outcomes were to compare operation time and adverse events between robotic-assisted UKA and conventional UKA. Adverse events were divided into non-serious adverse events and serious adverse events. Non-serious adverse events were defined such as lower deep venous thrombosis, extremities swelling and continued medial knee pain. Serious adverse events were defined such as superficial or deep infection, tibial or femoral components loosening or failure, and other systemic complications (e.g., myocardial infarction, gastrointestinal ulcers bleeding). These adverse events were chosen since they were considered to be important to patient outcomes and most commonly reported by the included studies.

Assessment of methodological quality and risk of bias

The methodological quality for each included study in this current meta-analysis was evaluated using the Cochrane Collaboration’s tool for assessing risk of bias. Questions in this tool related to selection (random sequence generation and allocation concealment), performance (blinding of participants and personnel), detection (blinding of outcome assessment), attrition (incomplete outcome data), reporting (selective reporting) and other biases. The symbol of green, yellow and red represented low risk of bias, unclear risk of bias and high risk of bias, respectively. Of the reviewers two (JF and YNW) independently assessed the methodological quality of each included article. The authors of the included RCTs or quasi-RCTs were also contacted to provide further information regarding the methodology if necessary. An unweighted κ value was calculated to quantify the agreement between the two reviewers: 0.40 ≤ κ ≤ 0.59 reflected fair agreement, 0.60 ≤ κ ≤ 0.74 was considered as good agreement and κ ≥ 0.75 represented excellent agreement.

Statistical analysis

The statistical analysis was performed completely in the review manager from the Cochrane Collaboration (Version 5.3.5). The effect sizes of continuous and dichotomous variable were calculated by models of weighted mean difference and relative risk (RR), respectively, along with 95% confidence intervals (CI). Statistical heterogeneity among studies was evaluated using χ2 and I2 tests. A p-value from the χ2 test more than 0.05 and I2 test less than 50% were considered to be substantially lower heterogeneity, a fixed-effects model was employed; otherwise, the random-effects model was used. A p-value less than 0.05 was commonly considered statistically significant.

Results

Study inclusion and bias risk assessment

A total of 162 articles were identified based on search strategies for eligibility. After initial title and abstract review, 41 articles were assessed for full text eligibility, from which 34 were excluded for lack of controls and incomparable or incomplete data. Thus, a total of three RCTs [13, 15, 19], three quasi-RCTs [14, 16, 18] and one PCT [17] were included in this review (Fig. 1). The characteristics of the included studies are shown in Table 1, with a total of 242 robotic-assisted UKAs and 330 conventional UKAs. Fig. 2 shows the methodological quality and bias risk summary of all studies.

Fig. 1
figure 1

PRISMA flow diagram of search strategy and results

Full size image
Table 1 Characteristics of included studies
Full size table
Fig. 2
figure 2

Risk of bias summary of studies included

Full size image

Implantation accuracy of tibial component

Robotic-assisted UKA showed a significantly higher implantation accuracy of the tibial component compared to conventional UKA, resulting from lower implantation errors (P < 0.05) for all seven tibial component parameters (varus-valgus, flexion-extension, internal-external rotation, medial-lateral, anteroposterior, proximal-distal, and posterior slope directions). The greatest mean difference (MD) between the two techniques was identified in the tibial flexion-extension direction and was −3.02° (95% CI: −4.01, −2.02; P < 0.00001; Fig. 3).

Fig. 3
figure 3

Comparison of robotic-assisted and conventional UKA in implantation accuracy of tibial component. Varus-valgus (a), flexion-extension (b), internal-external rotation (c), medial-lateral (d), anteroposterior (e), proximal-distal (f), and posterior slope (g). UKA unicompartmental knee arthroplasty

Full size image

Implantation accuracy of femoral component

Fig. 4 shows lower implantation errors (P < 0.05) for all six femoral component parameters (varus-valgus, flexion-extension, internal-external rotation, medial-lateral, anteroposterior, and proximal-distal directions), i.e., robotic-assisted UKA showed a significantly higher implantation accuracy of femoral component compared to conventional UKA. The greatest MD between the two techniques was identified in the femoral component flexion-extension direction and was −2.84° (95% CI: −4.10, −1.58; P < 0.0001).

Fig. 4
figure 4

Comparison of robotic-assisted and conventional UKA in implantation accuracy of femoral component. Varus-valgus (a), flexion-extension (b), internal-external rotation (c), medial-lateral (d), anteroposterior (e), and proximal-distal (f). UKA unicompartmental knee arthroplasty

Full size image

Tibiofemoral component safe zone

The proportion of patients with component implantation within 2°of target position was significantly greater in robotic-assisted UKA (RR: 1.73; 95% CI: 1.12, 2.69; P = 0.01; Fig. 5).

Fig. 5
figure 5

Comparison of robotic-assisted and conventional UKA in 2°tibiofemoral component safe zone (a), operation time (b) and adverse events (c). UKA unicompartmental knee arthroplasty

Full size image

Operation time

Operation time was assessed by a total of four studies [12, 14, 18, 19]. Operation time of robotic-assisted and conventional UKA was 104 ± 16.6 min and 88 ± 18.3 min, respectively. Robotic-assisted UKA was found to significantly prolong the time of surgery by 15.69 min in comparison to conventional UKA (MD: 15.69, 95% CI: 11.92, 19.46; P < 0.00001; Fig. 5).

Adverse events

Across the studies, there were several non-serious and serious adverse events that were reported [12, 14, 17, 18], such as tibial or femoral components loosening or failure, superficial or deep infection, lower deep venous thrombosis; however, the rate of adverse events between robotic-assisted and conventional UKA was of no statistical significance (RR: 1.56; 95% CI: 0.77, 3.19; P = 0.22; Fig. 5).

Discussion

The fundamental purpose of the current meta-analysis was to compare the efficiency and safety outcomes of robotic-assisted and conventional UKA in treating end-stage unicompartmental knee osteoarthritis, including implantation accuracy of femoral and tibial components, operation time and adverse events. The main finding of this meta-analysis is that the use of robotic-assisted systems in UKA results in a more accurate prosthesis positioning compared to the conventional implantation technique and with no increase of adverse events; however, these advantages were accompanied of a prolongation in the operation time of approximate 16 min.

Several studies have measured the accuracy of component placement with the use of robotic-assisted systems on postoperative imaging data and were found to improve accurate prosthesis position and postoperative ligament balancing, even by means of minimally invasive incisions [20,21,22,23,24]. Compared to conventional UKA, the result of our meta-analysis also showed extensive reduction in error of component positioning and variability with use of robotic-assisted systems.

When evaluating an advanced and novel technique, there is undoubtedly concern that there will be increased operation time and procedure-related complications with this technique; however, our meta-analysis demonstrated that this longer operation time did not lead to a different rate between the two groups in terms of adverse events. If the additional time of 16 min spent during robotic-assisted UKA surgery could lead to a more accurate positioning of the component implantation and similar complications, this is acceptable in our opinion.

At present, there is no study reporting differences in survival rates or in postoperative clinical scores at a long-term follow-up after implantation of robotic-assisted and conventional UKA. Cobb et al. [12] measured the Western Ontario and McMaster Universities Osteoarthritis (WOMAC) score and the American Knee Society (AKS) score preoperatively, 6 and 18 weeks after surgery. The increased AKS score was significantly greater in robotic-assisted conventional UKA group but there was no significant difference in the changes of WOMAC scores. Hansen et al. [14] evaluated average postoperative range of motion (ROM) on the day of operation was statistically greater in the robotic-assisted UKA group, but there was no statistical difference in ROM on postoperative day 1 or day 2. Instead, ROM at 2 weeks postoperation was greater in the conventional UKA group. Therefore, more prospective follow-up studies are needed to compare the association with accuracy and clinical outcomes between robotic-assisted and conventional UKA.

In consideration of cost-effectiveness, Moschetti et al. [25] performed a Markov decision analysis to evaluate the costs, outcomes, and incremental cost-effectiveness of robotic-assisted UKA in patients with end-stage unicompartmental knee osteoarthritis. They found Robotic-assisted UKA was cost-effective compared with conventional UKA when the annual case volume exceeds 94 cases per year and it is not cost-effective at low-volume or medium-volume arthroplasty centers. Another study by Sinha [26] reported the limitations of increased cost of the robotic-assisted system. Future studies need to incorporate registration time to correctly portray the effect of utilizing robotic-assisted systems when analyzing cost-effectiveness. This advanced and novel technique offers opportunities to improve the quality of medical care delivered to patients and will be an indispensable part of patient care in the future, but its cost-effectiveness must be validated.

A limitation of this study is that some studies included in this meta-analysis were not RCTs (three RCTs, three quasi-RCTs and one PCT); however, we think that it is important to evaluate all the studies performed until now to know whether further investigations in this direction are worthwhile. On account of the evidence provided by this study, the authors suggest that robotic-assisted UKA should be accepted as providing more accurate component positioning; however, further well-designed RCTs or prospective cohort studies need to be performed to confirm these results.

Conclusion

Robotic-assisted UKA is a relatively new technique, but contradictory results about its efficiency have been reported in the literature. Based on this current meta-analysis, it could be shown that the use of robotic-assisted system in UKA is able to reduce the implantation errors and not to increase adverse events; however, there are few reports about clinical outcome and long-term follow-up results and it remains unknown whether the more accurate component positioning brings about a better clinical effect or a better long-term survival of the implants.