2.1 Eligibility criteria

All prospective randomized controlled trials (pRCTs) comparing outcomes for primary total knee arthroplasty (TKA) performed using conventional and navigated (computer-assisted) alignment techniques were eligible for analysis. Radiographic outcomes relating to postoperative coronal, sagittal and axial tibial and femoral component alignment were collected. The hip-knee-ankle (HKA) mechanical axis measurement was also collected. Clinical outcomes included all-cause revision and aseptic revision. Functional outcomes were analyzed where reported. All studies were required to be prospective randomised controlled trials. This study was compliant with PRISMA guidelines.

2.2 Search strategy

On March 7th, 2023, a number of electronic databases were searched using the following MeSH terms: ‘TKA’, ‘Total Knee Arthroplasty’, ‘TKR’, ‘Total Knee Replacement’, ‘Navigation’, ‘Navigated’, ‘Conventional’, ‘RCT’ and ‘Randomized Controlled Trial’ in various combinations to maximize the number of studies returned for review. The locations searched included the Cochrane Library, PubMed, the EU clinical trials register, the International Clinical Trials Registry Platform (World Health Organisation) and ClinicalTrials.gov. Results were analyzed on two separate occasions. This was to ensure the accuracy of data retrieval. The flow diagram in Fig. 1 illustrates this selection process. Full article review was performed by two authors (GS and MM) with all contention being resolved through consensus. Only those studies meeting the above mentioned criteria were included. After comprehensive review, there were 17 pRCTs included for review in the meta-analysis.

Close

Fig. 1 Flow diagram.

Export to PPT

2.3 Data extraction

An electronic data extraction form was used to record extracted data. Extracted information included year, journal, author, country of origin, total navigated TKAs, total conventional TKAs, navigation device used, gender, age, all-cause revision, aseptic revision, minimum follow-up and loss to follow-up for both groups. Functional outcomes were collected where they were recorded for each study.

There was a cumulative total of 2,201 TKAs included. There were 1,063 TKAs in the conventional group and 1,138 in the navigated group. The pertinent characteristics of each study are listed in Table 1^8–24. In addition to this, in the conventional group there were 362 males and 776 females. In the Navigation group there were 340 males and 856 females. The conventional group mean age range was 65–75.5 years. The navigation group mean age range was 65.2–76 years. Mean BMI was recorded in 12 studies reporting a range of 26.2–34.4 in the conventional group and 25.4–31.2 in the navigation group.

2.4 Statistical analysis

Descriptive statistics were used for demographic data. All studies except for Nam et al. and Minoda et al. used a cut-off value of 3°from neutral to define whether a component was malaligned in the coronal, sagittal or axial plane. Nam et al. and Minoda et al. used a 2°cutoff from neutral as their definition. For this reason, two major radiographic analyses were performed - The first analyzed all studies including Nam et al. and Minoda et al. The second analysis was exclusively comprised of studies taking 3° as the definition for malalignment. The purpose of this analytic approach was to ensure that the varying definition of a “normally aligned” implant did not have an impact on the validity of the results reported in this study.

A random-effects meta-analysis of all available cases was performed. This allowed for any missing data. The method used for the first stage model was the Taylor series approximation. The method used for the second stage model was random-effects meta-analysis. Statistical significance was determined by a p-value of less than 0.05. Inter-study heterogeneity was assessed for using the chi-squared test and the I² statistic. Heterogeneity-induced variation was expressed as a percentage. With a p-value of greater than 0.05, we concluded that heterogeneity had no significant impact on the results. Functional outcome scores were expressed as interval variables. Given the wide variety of reported functional outcome scores, a histogram was used to illustrate the comparative results across the studies reporting on functional outcomes. Postoperative conventional and navigated results were reported. If a number of time points were reported on for each score, the 1-year postoperative results were illustrated on the histogram. All studies reported on absolute postoperative score figures except for Ali et al. who reported on the improvement in functional scores without listing the absolute score. For this reason, Ali et al. was not listed in the histogram illustrating comparative functional scores but their results are mentioned in the results section. Statistical software [Stata/IC 13.1 for Mac (64-bit Intel)] was used for all statistical analyses.

2.5 Bias

Small study effects were analyzed using a funnel plot. This eliminated the effect of publication bias. Egger's test for small-study effects was used to assess the funnel plot for asymmetry that was statistically significant. Again, a p-value of less than 0.05 was deemed to be significant.

3.1 Radiographic outcome- Large group analysis (alignment within 2°or 3° of neutral)

iCoronal Alignment

Normal coronal tibial component mechanical alignment was achieved significantly more often with navigation (p < 0.001) (Fig. 2). Normal coronal femoral component mechanical alignment was achieved significantly more often with navigation also (p = 0.001) (Fig. 3). Funnel plot review demonstrated no evidence of publication bias (Fig. 4). There was no statistically significant asymmetry in the funnel plot when using Egger's test for small-study effects (p = 0.103). There was no evidence of publication bias affecting the current study validity.iiSagittal Alignment

Close

Fig. 2 Coronal alignment of the Tibial Component (Large Group Analysis).

Export to PPT

Close

Fig. 3 Coronal alignment of the Femoral Component (Large Group Analysis).

Export to PPT

Close

Fig. 4 Funnel Plot with pseudo-95% confidence intervals.

Export to PPT

Normal sagittal tibial component mechanical alignment was achieved significantly more often with navigation (p < 0.010) (Fig. 5). There was no significant difference in the sagittal alignment of the femoral component between the conventional and navigated groups (p = 0.617) (Fig. 6).iiiHip-Knee-Ankle mechanical axis

Close

Fig. 5 Sagittal alignment of the Tibial Component (Large Group Analysis).

Export to PPT

Close

Fig. 6 Sagittal alignment of the Femoral Component (Large Group Analysis).

Export to PPT

The lower limb HKA mechanical axis was restored to normal significantly more often in the navigated compared to the conventional group (p < 0.001) (Fig. 7).ivRotational alignment

Close

Fig. 7 Mechanical (HKA) Axis alignment of the lower limb (Large Group Analysis).

Export to PPT

There were two studies reporting on the rotational alignment of tibial components and three studies reporting on the rotational alignment of femoral components. There was no significant difference noted in the rotational alignment of tibial components (p = 0.583) or femoral components (p = 0.532). Forest plots were not generated given the low number of studies reporting on these outcomes.

3.2 Radiographic outcomes – Subgroup analysis (alignment within 3° of neutral only)

iCoronal Alignment (3°sub-group)

When exclusively analyzing studies that considered 3° as the normal limit for alignment, normal coronal tibial component mechanical alignment was achieved significantly more often with navigation (p < 0.001) (Fig. 8). Normal coronal femoral component mechanical alignment was achieved significantly more often with navigation also (p = 0.001) (Fig. 9).iiSagittal Alignment (3°sub-group)

Close

Fig. 8 Coronal alignment of the Tibial Component (Sub-Group Analysis).

Export to PPT

Close

Fig. 9 Coronal alignment of the Femoral Component (Sub-Group Analysis).

Export to PPT

There was no difference in the studies included in the large group analysis for the sagittal tibial or femoral alignment and so no further analysis was performed here.iiiHip-Knee-Ankle mechanical axis (3°sub-group)

The mechanical axis of the lower limb (HKA axis) was restored to normal significantly more often in the navigated compared to the conventional group (p < 0.001) (Fig. 10).ivRotational alignment (3°sub-group)

Close

Fig. 10 Mechanical (HKA) Axis alignment of the lower limb (Sub-Group Analysis).

Export to PPT

There was no difference in the studies included in the large group analysis and so no further analysis was performed here.

3.3 Clinical outcomes

iOperative time

There was a clinically important reduction in the mean operative time in the conventional group relative to the navigated group. The mean conventional operative time was 87 min (σ = 16.6, 95% CI 76.4–98.8) compared to 97.6 min (σ = 16.9, 95% CI 86.2–109.1) in the navigated group (p = 0.17).iiAll-cause revision

Within the follow-up times reported in Table 1, there was a 0.66% (n = 7) all-cause revision rate in the conventional group. There was a 0.62% (n = 7) all-cause revision rate in the navigated group. There was no significant difference in the all-cause revision rate between groups (p = 0.649) (Fig. 11).iiiAseptic revision

Close

Fig. 11 All-cause revision.

Export to PPT

Within the follow-up times reported in Table 1, there was a 0.38% (n = 4) aseptic revision rate in the conventional group. There was a 0.26% (n = 3) aseptic revision rate in the navigated group. There was no significant difference in the aseptic revision rate between groups (p = 0.79) (Fig. 12).

Close

Fig. 12 Aseptic revision.

Export to PPT

3.4 Functional outcomes

Six studies reports on functional outcomes as illustrated in Fig. 13. All outcomes reported are postoperative outcomes comparing both groups. Ali et al. reported on the change in Knee injury and Osteoarthritis Outcome Score (KOOS) and Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) score between both groups. Absolute figures were not reported and so the Ali et al. results were not reported in Fig. 13. Regarding the KOOS score, there was a 43 point improvement in the conventional group (conventional improvement – symptoms 36, pain 47, ADLs 43, Sports 40, QoL 49) and a 41 point improvement in the navigation group (navigation improvement – symptoms 38, pain 43, ADLs 40, Sports 42, QoL 44). Regarding the WOMAC score, there was a 41 point improvement in the conventional group (conventional improvement – pain 9, stiffness 3, function 29) and a 39 point improvement in the navigation group (navigation improvement – pain 8, stiffness 3, function 28).

Close

Fig. 13 Functional Outcomes* *VAS: visual analogue scale, EQ-5D: EuroQol 5D, KOOS: Knee injury and Osteoarthritis Outcome Score, QoL: Quality of Life, S&R: Sports and Recreation, ADL: Activity of Daily Living, ROM: Range of Motion, KSS: Knee Society Score, OKS: Oxford Knee Score, HSS: Hospital for Special Surgery Score.

Export to PPT

4.1 Radiographic alignment

The current study demonstrates the superiority of computer-assisted navigated alignment methods when compared with conventional techniques for primary TKA, for restoration of desired alignment. The coronal alignment of the tibial component, femoral component and lower limb (as demonstrated by the HKA mechanical axis) was shown to be significantly more accurate in the N-TKA group. Sagittal alignment was reported less often. Five studies did report on the tibial component sagittal alignment and reported a significant improvement in the N-TKA group over the C-TKA group^{9,13,19,20,24}. One radiographic parameter that was not deemed to be significantly better in the N-TKA group was sagittal alignment of the femoral component. It should be noted that only 4 studies reported on this outcome and so a type II statistical error should be considered here.

There was no significant difference in the rotational alignment of the tibial and femoral components were not significantly different between groups. It should again be noted that only two studies reported on tibial rotation while only three studies reported on femoral component rotation after TKA. This stands to reason since the rotational alignment of components is more likely to be accurate with conventional methods due to the ability of the surgeon to directly visualise the femoral epicondylar axis and the tibial tubercle which serve as the commonest anatomical reference points for component rotational positioning. Meijer et al. report on the differences in femoral and tibial component rotation between a N-TKA and C-TKA cohort concluding that no appreciable difference was demonstrated and that further evidence was required to investigate this topic. The ability of a surgeon to appreciate rotational alignment intraoperatively is in contrast to their ability to accurately assess coronal and sagittal alignment of conventionally-inserted components. This is because an appreciation of the hip-to-ankle alignment is required which is not readily available in the conventional setting.

4.2 Clinical outcomes

Table 1 demonstrates the minimum follow-up time for each study. There were only two studies with minimum follow-up of 1 year. Mullaji et al. report minimum follow-up of 1 year while Ali et al. report a minimum follow-up of 38 months^8,18. The mean follow-up time was reported in five studies. Ali et al. was the only study to report a mean follow-up time in excess of 1 year (51 months)⁸. Given the relatively short clinical follow-up times, no meaningful interpretation can be made from the all-cause and aseptic revision rates between both groups. Within the limitation of a short-term follow-up, no significant difference in the rates of all-cause and aseptic revision were noted. Outside of the scope of Level I evidence, there is no consensus in the published literature regarding which technique provides superior implant survivorship. Baumbach et al. report on 10-year survivorship outcomes for 113 navigated and 104 conventional TKAs²⁵. They reported a 98% survival rate for the navigated group and 87% survivorship in the conventional group. This was statistically significant. Song et al. published results of a 9-year follow-up study analyzing navigated and conventional survival rates²⁶. The navigated group had a better survival rate (95.6% v 88.4%) compared to the conventional group. These results were not statistically significant however. A recent study by Lee et al. reported on 10-year survivorship in 152 C-TKAs and 62 N-TKAs²⁷. The cumulative survival rate in the C-TKA group was 96.1% compared to 96.8% in the N-TKA group (p = 0.962). While there are suggestions that long term survivorship may be superior in N-TKA, there is no evidence suggesting that C-TKA is associated with superior survivorship when compared to N-TKA. It appears that there is no disadvantage to N-TKA for long-term survivorship but there may be a disadvantage with C-TKA at long term follow-up based on published evidence.

4.3 Functional outcomes

Functional outcomes are reported in five studies. As demonstrated in Fig. 13, a wide range of scoring systems are used to capture this data. Gothesen et al. report on the widest range of outcomes, with 10 scores reported in total²⁴. All scores were better in the N-TKA group except for the VAS score. Similarly, Minoda et al. report on 5 outcomes consisting of the Knee Society Score (KSS) and the EQ-5D with all 5 scores being better in the N-TKA group. Interestingly, Narkbunnam et al. and Tsuda et al. report a slight improvement in the C-TKA scores over the N-TKA scores. In 2019, Budhiparama et al. reported on the functional outcome scores in both N-TKA and C-TKA²⁷. Three studies were included and there were no differences in functional outcome scores. Similar to this, in 2020, Lee et al. reported on functional outcomes in 7 RCTs comparing N-TKA and C-TKA using KSS, WOMAC, pain and range of motion after surgery²⁸. The conclusion from that study was that there is no difference in the functional outcomes reported between the N-TKA and C-TKA groups. The current study supports the hypothesis that there is no significant functional difference between C-TKA and N-TKA.

4.4 Operative time

Up to this point, C-TKA has demonstrated no advantage over N-TKA however there is one clear clinically significant advantage to C-TKA. There was a clinically important reduction in the mean operative time in the conventional group relative to the navigated group. The mean conventional operative time was 87 min (σ = 16.6, 95% CI 76.4–98.8) compared to 97.6 min (σ = 16.9, 95% CI 86.2–109.1) in the navigated group (p = 0.17). Although not statistically significant, a 10 min reduction in operating time is clinically relevant. It is especially relevant in the context of a high volume center where a reduction in multiple 10 min periods per case per day may allow for an additional TKA to be performed within a defined time frame or single operating room session.

The evidence in the current study supports the superiority of N-TKA in achieving normal mechanical alignment. The benefits in relation to clinical survivorship are less obvious and a limited follow-up time in the current study restricts conclusions that may be drawn here, although there is substantial long-term (non-Level I) evidence in the literature supporting the claim that survivorship is better in N-TKA. Functional improvement is not apparently better with N-TKA in either the current study or in other studies reported in the literature. As technology advances, surgeons will need to select a technique that optimizes all variables discussed above.

4.5 Cost-effectiveness

A final consideration is cost-effectiveness. In a recent simulation based analysis, Hickey et al. found that surgical navigation would be the most cost-effective in the widest number of clinical scenarios. Patient-specific implants were thought to possibly be at least cost-neutral and robotic surgery is only likely to be cost effective in high volume institutions with patients who are at a higher risk of revision at their baseline²⁹. Given the significant increase in the cost of R-TKA relative to N-TKA, coupled with the superior ability of N-TKA to achieve normal mechanical alignment when compared to C-TKA, it is reasonable to suggest that N-TKA may be the preferred technique for executing accurate TKA procedures, inserting implants that will remain durable and optimizing cost-effectiveness in primary TKA into the future.

4.6 Limitations

There was no comprehensive reporting for all outcomes across all included studies. Specifically sagittal and rotational alignment results were not reported in the majority of studies. Clinical follow-up times were short in general and so limited conclusions may be drawn from the survival rates of TKA implants at such short-term follow-up. The definition of normal alignment varied between authors, with the majority agreeing that alignment within 3°of neutral was normal. Two studies (Nam et al. and Minoda et al.) did use a definition of 2°within neutral as their definition. For this reason we performed two analyses, one including all studies and the other including studies exclusively with the 3° definition for alignment. This allowed for the varying definitions of normal component alignment in TKA.