Unicompartmental knee arthroplasty (UKA) is an attractive option for isolated medial compartmental osteoarthritis with good long-term results [1–3]. A substantial proportion of patients undergoing knee arthroplasty are suitable for UKA, which would result in a functionally superior outcome with function similar to the native knee at a reduced cost to the health service . However the use of UKA has been declining  in recent years and this may be due to technically challenging surgery and difficulties in the accurate placement of the implants, which is key to a successful clinical outcome.
Computer-assisted surgical navigation has the potential to improve the accuracy of implant positioning, however its effect on clinical outcome is still debatable. The relatively recent introduction of computer navigation means that long-term studies are not available yet. However, short- to mid-term studies in TKAs [22, 23] and a short-term study in UKA  found no statistical difference between navigated and non-navigated knees.
This study did not demonstrate a significant difference in the longer term survivorship and clinical outcomes of navigated and non-navigated UKAs. A larger proportion of well aligned knees had good or excellent clinical outcomes and a higher proportion of navigated knees were well aligned, though these trends were not statistically significant. The importance of accurate mechanical alignment in TKAs has been debated recently  and our poorer (although statistically not significant) survivorship results show that more accurate and reproducible implant positioning may not necessarily lead to a better survival.
Our previous study  showed that computer navigation facilitated a higher rate of knees to be in the desired zone for leg alignment. In the present study there is a tendency, but the difference is statistically not significant using the same statistical tests. We demonstrated minor changes in leg alignment over time in 9 knees and substantial changes in 4 knees. It is not clear how much these minor changes represent an actual deterioration and how much they represent an intra-observer error, as only measurements were available from the previous study.
The limitations of our study lie mainly in the small sample size and thus a loss of statistical power. The differences in survival between the two groups was statistically not significant (p = 0.0625), however with longer follow-up this may become significant in favour of the non-navigated group. The implants used in the two groups were different, however both were fully cemented, fixed bearing unicompartmental knees with a similar design rationale. We had good results with the Allegretto, but a change to the EIUS was necessary to enable us to use the navigation system in our hospital. The cohort in our study also represents the initial part of the senior surgeon's learning curve with computer navigation, which may have affected our results unfavourably . At the time of the change the EIUS was relatively new without long-term registry data. The latest National Joint Registry  reports higher revision rates for the EIUS (3.3 vs. 1.8 revisions per 100 obs. years) which may be a factor in our survival analysis. Since our navigated cohort followed on our non-navigated group, ranges of follow-up do not overlap. Therefore outcome measures are obtained on average 2 years apart and any difference in the groups may be attributed to a natural disease progression.
Although there is evidence that increased operating times can result in higher infection rates , it is our impression that the time spent on setting up the computer referencing does not significantly add to the overall operating time and may even be offset by the time taken to place jigs and perform bone resections.