Statistics

Descriptive data are presented as unadjusted means with standard deviations (SD) or as proportions (%) and between-group differences as means with 95% confidence intervals in parentheses (95% CI) and p values. The univariate analyses were conducted using Welch’s test and Likelihood Ratio Test (or if violated Fisher’s exact test) respectively. The paired sample t-test was used for the intra-group MCID analysis. The adjusted data are presented as regression coefficients (β) and odds ratios (OR) respectively, with 95% confidence intervals in a manner similar to that above. The stepwise multivariate analyses were conducted using multiple linear or logistic regression, depending on whether the data was numerical or categorical. To enable comparisons of summarized scores and/or dimensions from different scores, with different scales and skewness, numerical variables were also converted into standardized values (z-scores).

A p value less than 0.05 was considered statistically significant. Furthermore, the statistical considerations are based on unequal sample sizes. Although both groups were considered large enough to run parametric tests, each group’s variables were still confirmed visually regarding normality by Q–Q probability plots. Data analyses were conducted using SPSS^® Statistics, Version 25 (IBM^®, Armonk, New York, USA).

Ethical approval

This observational study (case-control designed) was carried out in compliance with the 7^th version (2013) of the Helsinki Declaration and according to the STROBE Statement (www.strobe-statement.org). The study was approved by the Swedish Ethical Review Authority (2020–02001) as of June 17, 2020 and by the regional Ethical Review Board of Region Skåne (173–20) as of Sep 29, 2020.

Adverse events

Yet, uniform register data show higher revision rates for UKA than TKA [1–4]. Furthermore, two [1] to four [4] times higher revision rates (to TKA) were reported if UKA was index surgery compared with primary TKA. The same is true of corresponding inferior patient-reported outcomes [25]. With that said, superior patient-reported outcomes have been shown in UKA-TKA compared with TKA-TKA revision [16], probably caused by a lower risk of needing augments, stems, or more constrained revision implants in the former scenario [1, 25]. Therefore, UKA as an index operation may well be favored over TKA if “revision implant”, or even worse, were to be endpoint variables [1]. This may especially concern SONK as registry data from Australia has shown 26% higher 10-year revision rates for osteonecrosis compared with osteoarthritis in TKA [3]. As was also demonstrated in this study, revision only counts for one of many adverse events that would be defined as severe complications [1, 9, 10]. Although complications should foremost be valued on a personal level, they are indisputably also very much linked to the overall costs. UKA has been reported more cost-effective than TKA in registry-based studies [26–29]. Mid-term results (five years) from the largest RCT so far are in line with these results [17]. The initial cost savings seem to persist for life [29], where surgeon-usage appears to have significant impact on the overall cost-effectiveness of the UKA procedure [27]. Correspondingly in this study, it was not until surgeon-volume was included in the regression model that the relative risk of complications significantly decreased in the UKA group (Table 2B).

PROMs

Most studies have not been able to show any between-group differences in PROMs when comparing TKA and UKA [11, 14–19]. One explanation could be, somewhat surprisingly, that out of different osteoarthritic disease patterns AMOA is indicated to achieve the best improvements when exclusively analyzing TKA procedures [30]. Other plausible explanations to the many non-significant findings in the literature could be that traditional PROMs are either affected by too many confounding factors in their general health-weighted output designs or related to their moderate ability to differentiate well-functioning patients. To a certain extent, both effects may have been present in this study as it also used rather general health-weighted scores.

A recent matched study of 135 TKAs and 135 UKAs reported both significantly improved pain and function, activity level, and satisfaction scores at 1-year follow-up for the UKA using the new Knee Society Score (KSS), the University of California Los Angeles (UCLA) activity-level score, and a Likert satisfaction scale respectively [31]. The Total Knee Questionnaire (TKQ) and the Forgotten Joint Score (FJS-12) have both demonstrated low ceiling effects in patient satisfaction measures and found UKA to be superior in this aspect [32, 33]. In a systematic review comparing TKA and UKA regarding knee awareness (FJS-12) five studies (930 patients) were eligible for a Forest plot meta-analysis two years after surgery and reported a mean difference of 7.6 (95% CI 3.7–11.6), favoring the UKA procedure [34].

Volume

Arguably, UKA constitutes one of two main surgical options for AMOA and SONK, but what percentage should be recommended? In the literature a span of around five [35] to fifty [36] percent is reported. Recent findings have reported significantly lower revision rates if both a yearly surgeon-volume of more than 12 cases [7] and surgeon-usage of more than 20% [8] are reached, then even suggested to reach similar revision rates as for TKA [8], arguments not contradicted in this study (Table 2B). (During the data collection for this article, according to the stipulated radiological consensus criteria [21], a minimum of 27.4% of all knee arthroplasties performed within Region Skåne in 2016 were determined to be suitable for a UKA (as only the TKA procedures reported “medial” were evaluated).) Newer findings on age, chondrocalcinosis, weight and patellofemoral disease [21, 36] have suggested a broadening of the original indications [35]. Accordingly, around 10% of the UKA cases in high-volume surgeons appear to be on the lateral side with similar performance as on the medial side [37]. Furthermore, a meta-analysis comprising five cohort studies found no difference in functional outcome or revision rates when comparing UKA with and without patellofemoral joint arthritis [38]. Beard et al. found no relationship between anterior knee pain (AKP) and degeneration, nor between preoperative AKP and postoperative function score in UKA [39]. There is, however, consensus that traumatic ACL injuries, high tibia osteotomy (HTO) and inflammatory arthritis should be considered non-recommendable for UKA surgery [21].

Age

Although two studies have reported younger patients’ performance as inferior after TKA than after UKA [40, 41], these findings should not be translated into the somewhat persistent view that the younger population is more suitable for a UKA than the elderly population. According to the literature this is far from correct. In an effort to save bone stock, the younger population may even be better off with a HTO [42], whereas two recent meta-analyses have shown UKA to be especially favorable in the elderly population [43, 44]. As was also shown in this study, irrespective of age (Table 2B), there is solid evidence that UKA is associated with a substantially lower risk of adverse events than TKA [9]. Patients older than 75 years were reported to have lower rates of postoperative transfusions, greater postoperative range of motion, and higher levels of activity at the time of discharge if operated on with a UKA compared with a TKA [15], which speaks further in favor of the UKA procedure in the elderly population. Another study showed that patients older than 75 years had no between-group difference in revision rate at five years [15]. Data from the Swedish Knee Arthroplasty Registry, in analogy found a two-times better 10-year UKA survival in those above 75 years of age compared to those below 65 years of age [1]. UKA is also reported to be more cost-effective than TKA as the initial health improvements appear to be maintained for the elderly patients, something that is not as evident for the younger patients depending on the lifetime risk of revision [29]. A study of the Norwegian Joint Registry concluded UKA to be a more cost-effective strategy than TKA for the elderly low-demand population as long as the annual probability of revision was below four percent [28].

Strengths

To the author’s knowledge, this is the largest radiograph-based comparative study ever undertaken that addresses adverse events and patient-reported outcomes between TKA and UKA, while also adjusting for patient characteristics and perioperative settings. The study is authentic in that it comprises several surgeons at five centers, yet with a minimum of confounding factors as it was conducted during a short period of time within one county council in the south of Sweden, i.e., demographics, patient characteristics, and perioperative routines were expected to be very similar. Nevertheless, the fact that it was adjusted for known confounding patient characteristics such as age, ASA class, BMI, Charnley class, and gender makes the data robust (Tables 1 and 2). An important strength of this study is the fact that only one single individual evaluated all the radiographs, rolling out any kind of inter-observer bias. The response rates were considered high, especially among the high-volume cases (TKA 82–100% and UKA 90–100%, depending on variable). Finally, the study is unique in that it provides evidence of the importance of adjusting for surgeon-volume (which could shed some light on the so often diametrical differences in results when comparing cohort with registry-based studies).

Weaknesses

First, the study was not randomized, but given the retrospective approach, it was possible to conduct a large multi-center comparative study with similar perioperative routines, demographics, and adjustments for potential confounders. One other limitation would be the strict radiograph-based inclusion/exclusion criteria [21] used, without evaluating any medical notes regarding the choice of implant prior to surgery based on center and/or surgeon preferences. Furthermore, although the consensus criteria include both evaluation of the cartilage and indirectly the viability of the ACL [21], the suitability for a UKA should finally be confirmed during surgery, which was obviously not the case in this study. With that said, conventional lateral x-ray alone is proved 93% sensitive and 96% specific in determining UKA suitability for medial degeneration [45], and has also shown to be more reliable than both MRI and clinical examination for assessing ACL viability in OA [46]. Finally, the fact that the approximately 60% higher chance of achieving a better pain and function response, according to OMERACT-OARSI responder criteria, did not reach statistical significance in the UKA group indicates a type II error.

A personal reflection

For doctor and patient alike a risk/reward discussion is often meaningful but requires data to be presented in a relevant manner, of which the set of OMERACT-OARSI responder criteria is one such attempt. A clear aim of the presentation in this study was to summarize the results in order to make them useful in such a doctor/patient scenario, both in relative and absolute terms. The author argues that caution should be taken not to exaggerate non-significant summarized overall PROM results, as most traditional scores are either weighted towards or incorporate evaluation of general health, with individual dimensions often overlapping, and thereby risk blurring a specific area of interest. (This fact was deliberately demonstrated in this study by a summarized mean UKA z-score of 0.00 (ns) in the high-volume procedures using the included PROM structures (while both respecting the differences in scale and skewness but assumed equal weight) and at the same time, somewhat contradictory, a trend of both an absolute higher rate of any knee specific “response” (5%) and “high response” (4%), as well as both a significant absolute lower rate of any complications (10%) and severe complications (5%) compared with TKA.)