1 Introduction & aim of this study

Knee osteoarthritis (OA) is a widespread and debilitating condition that impairs mobility and quality of life globally.¹ It is a major cause of chronic disability, resulting in pain, stiffness, and difficulty with daily activities.² While various treatments exist, surgical interventions, especially knee arthroplasty, are often considered the final option for advanced cases.³ The specific surgical method chosen can significantly influence patient outcomes, functional recovery, and satisfaction.⁴

Kinematically Aligned Total Knee Arthroplasty (KA TKA) and Medial Uni-compartmental Knee Arthroplasty (UKA) are effective but differ in scope and application. KA TKA replaces the entire knee to restore natural alignment and function,⁵ whereas Medial UKA preserves healthy compartments, treating only the damaged medial section.⁶ This aligns with the increasing body of evidence supporting the benefits of KA TKA, particularly in comparison with MA TKA; however, the authors contend that these advantages position KA TKA as a surgical option whose outcomes more closely approximate those of medial UKA than those achieved with MA TKA. This paper aims to compare these approaches by analysing clinical outcomes, patient experiences, and long-term data to support tailored surgical decisions.

2 Methodology

3 Results

3.1

3.1 Screening and eligibility

Initial database searches identified 1,146 potential studies. We have conducted screenings of titles and abstracts to assess their relevance based on predefined inclusion and exclusion criteria. Duplicate studies (n = 756) and studies with irrelevant abstracts (n = 286) were removed, leaving 104 studies for further review. Full-text versions of these studies were obtained and thoroughly reviewed. Any further duplicates or studies that did not meet the inclusion/exclusion criteria were excluded at this stage. After this detailed screening, 47 studies remained. Unfortunately, we were unable to obtain full reports for 20 of these, leading to their exclusion. Of the remaining 27 studies, a further 9 were excluded because they were not true RCTs, and 1 was excluded as only the abstract was available. This resulted in a final selection of 17 RCTs for inclusion in our systematic review (Fig. 1). The characteristic of each included study is entailed in Supplementary file 1. Importantly, no automation tools were used during the selection process, ensuring that all decisions were made manually by the reviewers.

Close

Fig. 1 PRISMA flow chart.

Export to PPT

3.3

3.3 Oxford Knee Score

The mean OKS was reported across twelve studies with 1554 study participants comparing TKR-KA, UKR-M, and UKR-R at follow-up intervals ranging from 1 to 156 months^7–18 (Fig. 2). KA TKA and medial UKA demonstrate similar OKS values over time, with some variation among studies. Robotic UKA, reported in a single study, shows a comparable OKS to the other interventions at the corresponding time point (Fig. 2A). The indirect comparison between manual UKA and KA TKA yielded a non-significant weighted mean difference (WMD) of 0.84, with a 95 % confidence interval (CI) ranging from −1.68 to 3.37. Although the point estimate (0.84) suggests a potential marginal benefit of medial UKA over KA TKR, the wide confidence interval indicates that this difference lacks statistical significance (Fig. 2B). As a result, the perceived advantage of medial UKA cannot be definitively established, and the true effect may be neutral or even negative.

Close

Fig. 2 (A) OKS from individual studies at different time points after two strategies (KA-TKA and medial UKA) (B) Network Forest plot displaying the weighted mean difference (WMD) for OKS comparisons between KA TKA and medial UKA.

Export to PPT

Similarly, the extended indirect comparison of Robotic medial UKA versus KA TKA produced a WMD of −0.16, with a 95 % CI ranging from −4.01 to 3.69.

3.4

3.4 Knee Society Score (KSS)

Mean KSS values were reported across eleven studies for KA-TKR and medial UKA with 1412 study participants at follow-up durations ranging from 1.5 to 108 months^{7,8,12,14–16,18–22} (Fig. 3). The outcomes include total KSS and its subcomponents: KSS Function, KSS Objective, and KSS Pain. KA-TKA showed higher mean KSS values across most follow-up intervals and score components compared to UKA, with particularly noticeable differences at longer follow-up durations such as 60 and 108 months (Fig. 3). Indirect estimates comparing medial UKA to KA TKA showed an SMD of 0.012 (95 % CI: −0.362 to 0.386), indicating no significant difference between these two procedures (Fig. 3B). Additionally, the high level of heterogeneity observed in the data further complicates the interpretation of these findings, as it indicates variability in the studies included in the analysis.

Close

Fig. 3 (A) Mean KSS (total, function, objective, pain) for KA TKA and UKA across follow-up durations (1.5–108 months). (B) Network Forest plot displaying the standardized mean difference (SMD) for KSS comparisons between KA TKA and medial UKA.

Export to PPT

3.5

3.5 Forgotten Joint Score (FJS)

Mean FJS values were reported across nine studies for KA-TKR and UKR with 903 study participants at follow-up durations ranging from 1.5 to 156 months^{8,9,12,15–18,22,23} (Fig. 4) The results demonstrate consistently higher mean FJS values for medial UKA compared to KA TKA at most follow-up intervals. The difference was particularly evident at earlier follow-up periods (1.5–24 months), while FJS values for KA TKA approached those of medial UKA at longer follow-up durations such as 60 and 156 months (Fig. 4A) In the indirect comparisons, UKA versus KA TKA demonstrated a weighted mean difference (WMD) of −0.19 (95 % CI: −7.84 to 7.47), indicating no statistically significant difference between these two interventions. Similarly, the extended indirect comparison between Robotic UKA and KA-TKA showed a WMD of −4.94 (95 % CI: −18.91 to 9.03), indicating no significant difference (Fig. 4B).

Close

Fig. 4 (A) Mean FJS values for KA TKA and medial UKA across studies at follow-up periods from 1.5 to 156 months. (B) Network Forest plot displaying the weighted mean difference (WMD) for FJS indirect comparisons between KA TKR and medial UKA.

Export to PPT

3.6

3.6 KOOS

Mean KOOS values were reported across four studies for medial UKA and KA TKA with 393 study participants at follow-up durations ranging from 1.5 to 156 months^14,16,17,22 (Fig. 5). The outcomes included total KOOS and its subscales: KOOS Function (Daily Living and Sports), KOOS Jr, KOOS Pain, KOOS Quality of Life, and KOOS Symptoms. Medial UKA demonstrated higher mean KOOS scores across most follow-up intervals and score components compared to KA TKA. However, available data for KA-TKA were limited, with only a single study reporting KOOS Jr scores at 156 months (Fig. 5A). The indirect estimate comparing medial UKA with KA TKA demonstrated SMD for this comparison is 0.456, with a 95 % confidence interval ranging from −0.174 to 1.086. This positive SMD suggests that UKA may have a slight advantage over KA TKA (Fig. 5B). However, the confidence interval includes zero, indicating that the result is not statistically significant. Therefore, based on this indirect estimate, there is uncertainty regarding whether medial UKA truly outperforms KA TKA. This highlights the need for further research to draw more definitive conclusions.

Close

Fig. 5 (A) Mean KOOS values for medial UKA and KA TKA across studies at follow-up periods from 1.5 to 156 months. (B) Network Forest plot displaying the effect sizes (SMD) for KOOS comparisons between KA TKA and medial UKA.

Export to PPT

4 Sensitivity analysis

At the Less Than 1-Year Follow-Up: In the short-term follow-up period, the OKS reveals that medial UKA significantly outperforms KA TKA (WMD: 5.26; 95 % CI: 1.12 to 9.39). This result suggests that patients undergoing UKR-M experience better outcomes within the first year. For the Knee Society Score (KSS), the indirect comparison between UKR and TKR KA does not show a significant difference at this time point (SMD: 0.067; 95 % CI: −0.310 to 0.445) (Table 1).

At the 1–2 Year Follow-Up: By the 1–2-year mark, the differences between the interventions begin to diminish. The indirect comparison of the OKS between medial UKA and KA TKA indicates no significant difference (WMD: 0.07; 95 % CI: −2.07 to 2.21), suggesting comparable outcomes between the two procedures. For the KSS, a significant advantage emerges in favour of medial UKA over KA TKA (SMD: 0.432; 95 % CI: 0.118 to 0.746), highlighting the potential long-term benefit of medial UKA in these studies (Table 2).

Regarding the FJS, the analysis indicates that KA TKA may provide better patient-reported outcomes than medial UKA, with patients experiencing less joint awareness and a more "natural" feeling in the operated joint (WMD: −7.14; 95 % CI: −12.92 to −1.35). This difference suggests that KA TKA may result in a higher level of patient satisfaction regarding joint function and comfort. Furthermore, when comparing robotically assisted UKA to KA TKA in terms of FJS, the results indicate that KA TKA may still offer a slight advantage, although the difference is less pronounced than in the medial UKA comparison (WMD: −3.84; 95 % CI: −19.26 to 11.59). The wide confidence interval for this comparison reflects substantial variability in the data, meaning that while TKR KA may show a trend towards better outcomes, the evidence is less conclusive, and the true effect may vary considerably.

At the More Than 2-Year Follow-Up: In the long-term follow-up period, the advantages of one procedure over the other become less distinct. The comparison of OKS between medial UKA and KA TKA shows no statistically significant difference (WMD: −0.74; 95 % CI: −4.75 to 3.27), indicating that patients undergoing either procedure experience comparable outcomes in terms of knee function and pain relief over the long term (Table 2).

For the KSS, medial UKA continues to suggest a potential advantage over KA TKA, but the long-term indirect estimates show considerable uncertainty. The weighted mean difference (WMD: 3.116; 95 % CI: −2.240 to 8.473) highlights a positive trend in favour of medial UKA, yet the confidence interval spans both positive and negative values, suggesting that this advantage may not be statistically significant. This uncertainty indicates that while medial UKA may provide improved knee function according to the KSS, the evidence remains inconclusive over the long term.

Regarding the FJS, medial UKA maintains a slight lead over KA TKA in terms of joint awareness and comfort (WMD: 3.116; 95 % CI: −2.240 to 8.473), but again, this difference is not statistically significant due to the wide confidence interval. The variability in the data indicates that some patients may feel less joint awareness with medial UKA, but the true effect remains uncertain and could vary between individuals.

5 Heterogeneity and consistency

Heterogeneity was assessed across the comparisons, revealing high levels of variability, as indicated by I² values for different outcomes (Table 2). These values reflect considerable heterogeneity, suggesting that there were substantial differences among the included studies in terms of the effect sizes reported. The DOI plot (Fig. 6) displayed no symmetry, suggesting evidence of bias, and the funnel plot (Fig. 7) further corroborated this observation, especially for the direct estimates. Such variability can impact the precision of the overall estimates and should be interpreted with caution.

Close

Fig. 6 DOI plots assessing publication bias across (A) OKS, (B) FJS, (C) KSS, and (D) KOOS. Each plot displays the distribution of study effects for comparisons between KA TKA, MA-TKA, and Medial UKA, Asymmetry in the plots may indicate small-study effects or publication bias.

Export to PPT

Close

Fig. 7 Funnel plots assessing publication bias across (A) OKS, (B) FJS, (C) KSS, and (D) KOOS. Each plot displays the distribution of study effects for comparisons between KA TKA, MA TKA, and Medial UKA. Asymmetry in the plots indicate small-study effects or publication bias.

Export to PPT

In addition to DOI plot, the funnel plots assessing publication bias across (Fig. 6A) OKS (Fig. 6B), FJS (Fig. 6C), KSS, and (Fig. 6D) KOOS have confirmed what DOI plots have demonstrated. Each plot displays the distribution of study effects for comparisons between KA TKA, MA TKA, and UKA. Asymmetry in the plots indicate small-study effects or publication bias.

The consistency H value of 1.000 indicates perfect alignment between the direct and indirect comparisons included in the analysis. The agreement between direct and indirect evidence lends support to the overall conclusions suggesting that the indirect estimates can be confidently used to inform clinical decisions and support the broader conclusions drawn from the analysis.

6 Discussion

Total Knee Arthroplasty and medial UKA represent two fundamentally different philosophies in the surgical management of knee osteoarthritis. Existing meta-analyses have predominantly compared UKA to MA TKA, leaving a critical gap: Does KA TKA's kinematic approach narrow or eliminate the perceived benefits of medial UKA? This network meta-analysis (NMA) directly addresses this question by synthesizing indirect evidence from 17 RCTs (1,554 patients) to compare KA TKA and medial UKA across four key outcomes (OKS, KSS, FJS, KOOS). By leveraging the IVhet model and node-splitting to reconcile heterogeneity and consistency, this study provides the most comprehensive evaluation to date of whether KA TKA's theoretical advantages translate into measurable differences versus UKA, a question with immediate implications for surgical decision-making, patient counselling, and healthcare resource allocation.

Knee arthroplasty remains a gold-standard treatment for end-stage osteoarthritis. Historically, medial UKA has been perceived to offer superior functional outcomes compared to MA TKA, but the evolution of KA TKA appears to have narrowed this performance gap.

The analysis revealed several important findings regarding specific outcome measures. For the Oxford Knee Score, a minimal non-significant trend favoured UKR (WMD 0.84, 95 % CI: −1.68 to 3.37), while the Knee Society Score showed virtually no difference between groups (SMD 0.012, 95 % CI: −0.362 to 0.386). Similarly, KOOS results demonstrated no meaningful advantage for either procedure, whether considering conventional UKR (WMD -0.19, 95 % CI: −7.84 to 7.47) or robotic-assisted techniques (WMD -4.94, 95 % CI: −18.91 to 9.03). The Forgotten Joint Score showed a modest but non-significant trend toward better outcomes with UKR (SMD 0.456, 95 % CI: −0.174 to 1.086), though this marginal difference is substantially smaller than what has been historically reported between medial UKA and MA-TKA. These consistent findings across multiple validated PROMs strengthen the conclusion that KA TKA has successfully narrowed the functional outcome gap that previously existed between medial UKA and traditional MA TKA.

The clinical implications of these findings are significant for surgical decision-making. The comparable outcomes suggest that KA-TKR now provides an alternative that approaches the functional benefits traditionally associated with medial UKA, while maintaining the broader applicability of TKA. This narrowing of the performance gap between medial UKA and TKR through kinematic alignment techniques gives surgeons an important middle ground, particularly for patients who may be borderline candidates for medial UKA due to more advanced but not yet pan-compartmental disease. Medial UKA retains its advantages for ideal candidates with isolated medial compartment disease, particularly in terms of bone preservation and potentially faster recovery. However, the emergence of KA TKA as a viable alternative means surgeons no longer need to choose between the limited indications of medial UKA and the potentially inferior functional outcomes of traditional MA TKA. The wide confidence intervals observed across several outcomes do highlight that individual patient factors remain crucial in determining optimal surgical approach.

Several limitations must be considered when interpreting our results. The indirect comparison methodology inherently carries greater risk of confounding than direct randomized trials, as it relies on linking studies through common comparator groups. However, unfortunately there are no direct comparison between the two procedures hence indirect assessment have been selected. Significant heterogeneity existed across included studies regarding surgical techniques, implant designs, rehabilitation protocols, and follow-up durations. While the analysis suggests KA TKA has narrowed the gap with medial UKA, the included studies may not fully account for the learning curve associated with kinematic alignment techniques. The analysis focused on short-to medium-term outcomes, leaving open questions about whether the narrowed performance gap between KA TKA and medial UKA persists in long-term follow-up, particularly regarding implant survivorship. Additionally, the exclusive reliance on PROMs, while patient-centred, means subtle functional differences might not be captured.

Future research should address these limitations through well-designed randomized controlled trials directly comparing KA TKA and medial UKA with standardized techniques and longer follow-up periods.

7 Conclusion

This network meta-analysis found no statistically significant or clinically meaningful differences in patient-reported outcomes between kinematically aligned total knee replacement and medial uni-compartmental knee replacement across multiple validated assessment tools, including widely used instruments such as OKS and KOOS, however, this does not imply equivalence given indirect comparisons, high heterogeneity, and publication bias. Outcomes observed in KA TKA knees have closed the historic gap between MA TKA and medial UKR knees. Furthermore, these findings suggest that the functional and symptomatic advantages historically associated with medial UKR may no longer be as pronounced in the context of modern KA-TKR techniques, which aim to more closely replicate native knee kinematics.

Ethical statement

This work was a systematic review; there was no direct patient or public involvement in the design, conduct, or reporting of this research.

Patient and public involvement statement

This work was a systematic review; there was no direct patient or public involvement in the design, conduct, or reporting of this research.

Credit author statement

Mohammed Elmajee: Conceptualization, methodology, supervision, writing – original draft.

Chathura Munasinghe: Literature search, data curation, formal analysis, writing – review & editing.

Harriet Elson: Literature search, Data extraction, Writing – review & editing.

Nandita Nair: Literature search, Data extraction, Writing – review & editing.

Azka Syed: Data analysis, Visualization, and data pooling.

Anastasios Nikolaides: supervision, validation, writing review & editing.

All authors have read and approved the final manuscript.

Funding statement

The authors declare that no funding was received to support this study.