The revision rate of the UKA is much higher in national registers than in most published series. Although some authors (Labek et al, 2011) have implied that this is due to biased or even fraudulent reporting of the cohort studies by the designer surgeons, the main reason is likely to be due to surgical experience. In the national registers, most surgeons are found to be doing very small numbers whereas, in published series, surgeons tend to do large numbers. The data from the NJR would suggest that about half the surgeons doing knee replacement do some unicompartmental replacement. For those doing unicompartmental replacement, the most common number implanted per year is one and the second most common number is two (Fig. 10.6) (Liddle, 2014). The average number is five.
Figure 10.6 Surgeon numbers of UKA per year in NJR.
When the number of UKA performed per surgeon per year were compared to the revision rate, it was found, not surprisingly, that the surgeons doing small numbers have a very high revision rate (Fig. 10.7). The surgeons doing one or two UKA per year have a 4% failure rate per year which would equate to about a 60% survival at 10 years. The revision rate dramatically decreases with increasing numbers. Surgeons doing about 10 UKA per year have a revision rate of 2% per year whereas surgeons doing about 30 per year or more have a revision rate of 1% per year (Fig. 10.7). The data would suggest that surgeons should do at least 10 per year and ideally at least 30 per year. When surgeons perform at least 30 per year, the re-operation rate is the same with UKA and TKA out to eight years.
Figure 10.7 Relationship between revision rate and the number of UKA done per surgeon per year in NJR (Liddle, 2014).
Over the years we, and others, have encouraged surgeons to increase their numbers of UKA but these efforts have had little effect. This is not surprising because surgeons cannot easily increase the size of their practice. However, there is an alternative way of increasing the number of UKA: increasing the proportion of their knee replacements that are unicompartmental. In order to do this, they would have to broaden their indications for UKA.
Figure 10.8 shows the relationship between the revision rate and the percentage of a surgeon’s knee replacements that are UKA, which we have called usage. The shape of the graph is not what would be expected. In 1989, Kozinn and Scott defined the ideal indications for UKA and implied that surgeons who extended the indications would have a higher revision rate (Kozinn & Scott, 1989). Subsequently Stern et al. (1993), and Ritter et al. (2004), found that, based on Kozinn and Scott criteria, about 5% of patients would be considered ideal for unicompartmental replacement. It would therefore be expected that the lowest revision rate would be achieved when surgeons did 5% per year and that the revision rate would increase with higher usage. This is far from what actually happens: the revision rate for surgeons with a 5% usage is very high, about 3% per year, which equates to a 70% survival at 10 years. As the usage increases, the revision rate dramatically decreases until 20% usage. Thereafter, for the OUKA, with increasing usage there is a slow but steady decrease in the revision rate with the optimal usage being around 50%.
Figure 10.8 Annual Revision Rate for OUKA plotted against the proportion of a surgeon’s practice that is unicompartmental (based on Liddle et al., 2013).
With usage higher than about 65%, the revision rate increases although the data at these high usage rates is unreliable because of the small numbers of surgeons doing these rates and because these surgeons tend to have atypical referral practices.
There are two main reasons why the revision rate decreases with increased usage up to 50%. Firstly with increased usage, surgeons do increased numbers and their results improve. Secondly, their indications change. Surgeons doing small numbers tend only to use UKA when there is early disease and the rest of the knee is pristine. In these circumstances, particularly with partial thickness cartilage loss, the revision rate is very high (Niinimaki et al, 2011; Pandit et al, 2011). To achieve good results, surgeons not only need to do reasonable numbers but also to use the appropriate indications. The evidence would suggest that, if surgeons have a low usage, they should either stop doing UKA or change their indications so they are doing at least 20%. We would consider usage of 20% or more to be acceptable. For optimal results, surgeons should increase their usage to about 50%. If the recommended indications for the OUKA are adhered to, then about 50% of a surgeon’s knee replacements would be UKA.
Liddle et al.(2014) in their propensity matched cohorts of UKA and TKA, assessed the effect of usage. Overall, in the unmatched cohort, the revision rate of UKA was 3.2 times higher than TKA but, when the patients were matched, the hazard ratio for revision became 2.12 (95% CI 1.99 – 2.26) that of TKA. However, in the subset of patients in whom the surgeons were considered to have optimal usage rates (40 – 60%), the revision rate of UKA was only 1.4 times higher than TKA. Revision is in some ways a biased outcome measure and reoperation rate is probably a fairer measure with which to compare UKA and TKA. Out to eight years for unicompartmental replacements done by surgeons with 40 – 60% usage, there was no significant difference in reoperation rate between matched UKA and TKA (Fig 10.9) (Liddle, 2014).