Unicompartmental knee arthroplasty (UKA) has currently evolved into the treatment of choice for localized arthritis of the knee in the middle-aged and elderly patients. However, the bearing modularity continues to incite debates since both bearings yield similar outcomes in spite of fundamentally different design concepts. Non-conforming articulating surfaces and restricted bearing motion inherent to fixed bearings predispose them to repetitive edge loading and peripheral wear1). Highly conforming mobile bearings were designed to overcome these limitations; thus, it offers better restoration of natural kinematics of the knee and reduction of contact stress and wear234). However, the superiority of mobile bearings to their predecessors is questionable, as concurrent advances in polyethylene processing, implant design and surgical technique enabled fixed bearings to offer competent results with a relatively forgiving technique5678).

The first meta-analysis that attempted to resolve this debate was constrained by paucity of studies that could support a robust statistical analysis9). A recently reported meta-analysis employed thorough statistical analysis and reached a conclusion that shows no essential differences in revision rates between mobile and fixed bearings10). Both studies910) faced the same limitations-paucity of randomized control trials (RCT) and variations in study objectives. For comparative analysis of data based on various sample sizes and follow-up periods, we calculated reoperation rates per hundred component years in this study.

Data pertaining to measures of survivorship and frequency and timing of failures was extracted and analyzed. We thought that the crux of the debate between fixed and mobile bearing UKA lies in propensity and timing of complications. Therefore, we focused on comparative analysis and discussion of complications.

This review followed the guidelines of Preferred Reporting Items for Systematic reviews and Meta Analyses (PRISMA)11). A search for articles was conducted on PubMed, Cochrane library and EMBASE for records available till June 2014. We used keywords such as "knee arthroplasty", "fixed", "mobile" and "bearing" and Boolean operators (AND, OR) for various combinations. All 723 studies comparing fixed and mobile medial UKAs were included, regardless of implant manufacturer, study period, study design, indication for surgery or language of reporting. A total of 169 unique titles were identified after elimination of duplicate results and screening for relevance. References in review papers and eligible studies were assessed manually. After stepwise exclusion of studies comparing in vivo behavior, retrieval analysis and cadaveric studies as shown in Fig. 1, only 10 comparative studies were found to be eligible. Since only three RCTs/ quasi-RCTs were found in our search, we included non-randomized comparative studies. After screening full text of the eligible studies, we formulated an analysis protocol that would ensure collection of all comparative data, including research methodology and level of evidence. One of the authors extracted data from each study regarding the number of patients, number of knees, study period, demographic data, implant, inclusion and exclusion criteria, indication for surgery, surgical technique, pre-operative scores, type of prosthesis, surgical approach, experience of surgeon and duration of follow-up.

There were no differences in the indications for the mobile type and fixed type UKAs. The outcome measures included reason and incidence of reoperation, timing of reoperation and survivorship. All data was verified by another author and discrepancies and disagreements were resolved through discussion.

The structure of study, sample size, follow-up period and outcome measures were not consistent in the literature. To compare the diverse data, we calculated the number of reoperations per 100 observed component years, as described by the Australian arthroplasty registry12). In this study, we combined different follow-up periods and number of knees reported in every study to calculate the observed component years at risk for each bearing design. Cumulative reoperations and cause-specific reoperations were then compared to the denominator using a simple formula: reoperations per hundred component years=number of reoperations for any reason×100total number of observed component years where the total number of observed component years=∑i=1nindividual years from implantation.

Interpretation of results is fairly straightforward, such that the value of 1 implies reoperation rate of 1% per year, or 10% at ten years. To evaluate whether heterogeneity of data confounded our results, we created subgroups within eligible studies and calculated the same values for the subgroups for comparison with the cumulative results. The number of knees reported may imply results of a high volume center, while follow-up of more than three years may identify complications dependent on time since surgery. Subgroups were created within eligible studies based on these criteria and results compared with the cumulative mean.

There are many existing studies showing favorable results of mobile and fixed bearing UKAs1024252627282930). However these studies have reported the results of a single implant instead of comparison of fixed and mobile bearing designs; therefore, we included only comparative studies. Although, the follow-up periods of 10 eligible studies were not sufficient to compare the long-term outcomes of UKA, the results of our review have confirmed that mobile bearings provide low rates of wear. In addition, the differences in the indication between the bearing types have not been established.

Durability and survivorship are perhaps the main reasons of preference for total knee arthroplasty over UKAs in eligible patients. Improvement in survivorship may transform UKA from an interim procedure to a preferred primary treatment indicated at an optimal stage of arthritis. Therefore, a comparative evaluation of complications of different bearing designs and their timings in UKA is at the heart of this discussion.

Aseptic loosening, progression of arthritis and wear of polyethylene insert are the main complications requiring reoperation following UKA. Our results suggest that first two complications are more common in the knees with mobile bearings than fixed bearings. While no knees with mobile bearings were re-operated for wear, the most common complication requiring reoperation for mobile bearing UKA was progression of arthritis. Mobile bearings were found to generate wear with a smaller particulate size (0.58 ?m) but greater granular debris (mean, 93% granular debris) as compared to Miller-Galante (5.23 ?m, 15%) and anatomic porous coated designs (1.17 ?m, 77%)31). Smaller wear particles of mobile bearing designs are biologically more active and have high propensity for osteolysis32). Bloom et al.33) compared the complexity of revision surgery after fixed and mobile bearing UKA. They reported that 46.7% of knees required tibial augments when the primary implant used was a mobile bearing design, compared to only 11.1% of knees that required augments when an implant with a fixed bearing design was revised. In the current review, aseptic loosening was the second highest complication requiring reoperation in mobile bearing UKAs. This was probably caused by osteolysis resulting from the wear of submicron particles32). Therefore, it is worthwhile to ponder whether aseptic loosening caused by wear of submicron particles that necessitates reoperation before the wear is grossly evident.

Considering that bearing dislocation is a complication unique to mobile bearings, understanding that observed rate of zero reoperations in fixed bearings is straightforward. However, observed rate of zero reoperations of mobile bearings for wear even in studies with follow-up exceeding 17 years is intriguing13). Further research in understanding wear behavior of bearings is warranted to establish the actual effectiveness of mobile bearings in reducing wear rates.

Progression of arthritis leading to reoperation was noted at a higher frequency in mobile bearing knees as compared to fixed bearing knees. Bearing dislocations are clearly a complication unique to mobile bearings. It is likely that the two causes are interrelated, such that prevention of one predisposes occurrence of the other. Smith et al9) theorized that a surgeon may tend to choose a tight knee and risk slight valgus overcorrection to avoid bearing dislocation. This can result in greater contact stresses and overt loading of the lateral compartment, accelerating the progression of arthritis. Conversely, any undercorrection of deformity may have more play in the medial compartment and predispose to bearing dislocation. Perfect ligament balancing and limb alignment is, therefore, mandatory to avoid both complications, making mobile bearing UKA an unforgiving and a technically challenging procedure. The low tolerance of mobile bearings to gap mismatch even in hands of high-volume surgeons reiterates the difficulty in widespread reproduction of their results7). Fixed bearing designs permit undercorrection and may offload the lateral compartment to decelerate the progression of arthritis in the aging knee.

Wear of polyethylene insert is the principal complication requiring reoperation of fixed bearings, occurring around 8.59 years after primary surgery58131415161718). Wear is a complication inherent to the design of fixed bearings, secondary to higher surface deformation and delamination in comparison to mobile bearings34). The mechanism of the polyethylene wear is different according to the type of bearing. In fixed bearings, wear of the polyethylene is caused by the fatigue and sheer stress-related mechanism. However, in mobile bearings, wear of the polyethylene is due to the abrasive and adhesive mechanism. Even though some in vitro studies based on knee stimulators report lower wear rates of fixed bearings in comparison to mobile bearings3536), our results suggest that higher wear rates are constantly seen in association with fixed bearing UKA, in spite of advancements in polyethylene processing and implant design.

Reoperations for persistent pain are higher for fixed bearings than mobile bearings. However, 50% of knees re-operated for persistent pain were all polyethylene UKAs of a single manufacturer reported in a single study that contributed less than 15% of the total study population18). The overall reoperation rates for persistent pain excluding the studies reporting all polyethylene UKAs are 0 for mobile bearings and 0.089 for fixed bearings. Authors of the study18) theorized that the all-polyethylene tibial components increased load transfer to the proximal tibia, resulting in persistent bone remodeling and unexplained medial knee pain.

We recognize several limitations in this review. Due to limited availability of published RCTs, nonrandomized comparative studies were included in the analysis.

Data used for review was observational data that relied primarily on reporting standards, which may vary with geography, surgeon's experience and study design. Although a subgroup of studies with level I evidence would be desirable, the cumulative sample size of these studies was found to be too small to be able to reflect findings that can be extrapolated to the study population.

Use of reoperation as a measure of failure restricts inclusion of patients who elected to forgo additional surgery. Sporadic causes for reoperation, such as arthroscopy for removal of cement segments or anterior cruciate ligament tear, could not be comparatively assessed.

Constant evolution of implant design may have affected the comparability of studies separated by decades. The confounding effect of these variables on results of our analysis cannot be controlled or ruled out. In spite of these factors, most studies reported equivalent outcomes of both bearing designs, and collation of their individual data on complications has allowed this review to draw cumulative results and inferences that may have overcome effect of an individual study. The method of calculation was simplistic and differences between values compared were small, but the pattern of these differences was fairly constant within subgroups. It reiterates the effect of heterogeneity of the data and overall equivalence of outcomes, similar to the meta-analyses previously published923). Nevertheless, the results should be interpreted with caution.

In conclusion, the overall incidence of complications is similar for fixed and mobile bearing UKAs. The discussion of complications presented above may assist surgeons in their choice of bearing design. Further research on progression of arthritis with undercorrection in UKA, biologic activity of wear debris and mechanisms of persistent pain after UKA is desirable for improvement of outcomes and survivorship of UKA.