The Journal of Arthroplasty xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

The Journal of Arthroplasty

j ourna l homepage: www.arth rop lasty journa l .o rg

Oxford Unicompartmental Knee Arthroplasty Versus Age and GenderMatched Total Knee Arthroplasty – Functional Outcome andSurvivorship Analysis

Jun W. Lim, BMSc, Gerard R. Cousins, MBChB BSc(MedSci) MRCS, Benedict A. Clift, BMSc FRCSOrtho,David Ridley, BSc MSc, Linda R. Johnston, RGN BN MScDepartment of Orthopaedics and Trauma, Ninewells Hospital, DD1 9SY Dundee, United Kingdom

a b s t r a c ta r t i c l e i n f o

The Conflict of Interest statement associated with thidx.doi.org/10.1016/j.arth.2014.03.043.

Conflict of Interest: No benefits in any form have befrom a commercial party related directly or indirectly to

Reprint requests: JunWei Lim, BMSc, Department of OrNinewells Hospital, Dundee, DD1 9SY.

http://dx.doi.org/10.1016/j.arth.2014.03.0430883-5403/© 2014 Elsevier Inc. All rights reserved.

Please cite this article as: Lim JW, et alArthroplasty – Functional Outcome and Su

Article history:Received 10 December 2013Accepted 31 March 2014Available online xxxx

Keywords:unicompartmental knee arthroplastytotal knee arthroplastyrevisionfunctional outcomesurvivorship

We compared the medium-term outcomes of age and gender matched patients with unicompartmental kneearthroplasty (UKA) and total knee arthroplasty (TKA). We retrospectively reviewed the pain, function andtotal knee society scores (KSS) for 602 UKAs and age and gender matched TKAs between 2001 and 2013.Function scores remained significantly better in UKAs from preoperative until 3 years follow up. The changeof function scores was not significantly different. TKAs performed better than UKAs for pain scores. Total KSSfor both groups were not significantly different in the study. Fewer medical complications were reported inUKA group. 6.30% of UKAs and 2.99% of TKAs were revised. The theoretical advantages of UKA were not borneout, other than in immediate postoperative complications.

s article can be found at http://

en received or will be receivedthe subject of this article.thopaedics and Trauma, Level 5,

, Oxford Unicompartmental Knee Arthroplarvivors..., J Arthroplasty (2014), http://dx.doi

© 2014 Elsevier Inc. All rights reserved.

The surgical treatment of unicompartmental osteoarthritis of theknee remains controversial. In the United Kingdom, unicompartmentalknee arthroplasty (UKA) has increased in popularity with The NationalJoint Registry for England andWales reporting thatUKAcomprises 8%ofall knee arthroplasties performed [1]. There have been impressivesurvivorship studies for the Oxford UKA (Biomet Ltd, Bridgend, UK),with a claimed implant survival rate of 95% at ten years [2,3].

UKA offers several potential advantages over total knee arthroplasty(TKA) including thepreservation of bone stock and aminimally invasivetechnique. In principle, the normal biomechanics of the knee arerecreated. Weale et al. reported no progression of osteoarthritis in thelateral compartment with medial UKA, which has been considereda significant potential risk [4]. Newman et al. conducted an industrysponsored prospective randomised controlled trials to establishwhether UKA outcomes are comparable with TKA. They reportedcomparable pain at 5 years but also significantly improved function atthat stage in the UKA group [5]. The same research group followed upthe same cohort of patients at 15 years and suggested that there was anincreased revision rate in the TKA group and improved function in theUKA group [6]. A randomised controlled trial – Total or Partial KneeArthroplasty Trial (TOPKAT) – has been designed to assess the clinical

and cost-effectiveness of UKAandTKA [7]. The results of this trial are notavailable yet, but it is hoped they may lead to evidence based guidancefor the treatment of unicompartmental knee osteoarthritis.

Contrary evidence regarding the role ofUKAexists in the literature. A2-year follow up patient-reported questionnaire-based study with aknee-specific Knee Injury and Osteoarthritis Outcome Score demon-strated better activities performancewith UKA, particularly with regardto the ability to bend the knee, possibly due to the preservation ofcruciate ligaments [8]. However, there was a little difference in pain orfunction between the TKA and UKA groups and there was a higher riskof revision with UKA [8]. A recent comparison of patient recordedoutcome measures (PROMS) from the National Joint Registry (NJR)further suggested that there was no difference in knee specific orgeneral health outcomes between TKA and UKA [9]. In addition, Lyonset al. demonstrated that although UKA is associated with better clinicaland functional outcomes, both preoperatively and postoperatively, thechangeof outcome scores is similar in both groups. The survivorship at 5and 10 years were significantly better in TKA group [10].

The conversion of failed UKA to TKA is believed to be safe, reliableand repeatable [11]. Multi-centre studies have suggested that therevision of UKA to TKA is not technically difficult and the results arecomparable to the results of primary TKA [12]. Willis-Owen et al.produced evidence demonstrating thatUKA is functionally superior andoffers a substantial cost saving over TKA, accounting for both primaryUKA and revision of failed UKA to TKA [13]. There is opposing evidencehowever, such as the study by Padgett et al. that suggested theconversion of a failed UKA carries a significant risk of poor outcomes[14]. The New Zealand National Joint Registry reviewed the rate of

sty Versus Age and Gender Matched Total Knee.org/10.1016/j.arth.2014.03.043

Table 1Patient Demographics.

UKA TKA

Age (years) 70 ± 10 70 ± 10Gender (F:M) 283 : 319 283 : 319Side (left:right) 319 : 283 291 : 311BMI (kg/m) 30.19 ± 5.03 30.33 ± 5.79Hospital stay (days) 4 ± 2 7 ± 4

2 J.W. Lim et al. / The Journal of Arthroplasty xxx (2014) xxx–xxx

revision of UKA and reported a relatively poor outcome of revision ofUKA to TKA [15]. Chou et al. further demonstrated that the clinicalresults of revised failed UKAs are inferior to primary TKA [16]. Thesestudies questioned the increasing use of UKA, suggesting that it does not

A

B

C

Fig. 1. A: Knee Society Score (Pain) over 5-year. B: Knee Society Score (Function) over 5-year. C: Knee Society Score (Total) over 5-year.

Please cite this article as: Lim JW, et al, Oxford UnicompartmentalArthroplasty – Functional Outcome and Survivors..., J Arthroplasty (201

confer a significant benefit and should not therefore be used as a more‘conservative’ procedure in the younger population [15,16].

An extensive literature review reveals little further evidence directlycomparing UKA to TKA. The relevant practical question is: do thereported improvements in function and recovery offset the possiblenegative aspects of reduced implant longevity and disappointingoutcomes with revision? This study aims to compare the medium-termoutcomesof age andgendermatchedpatients treatedwithUKAandTKA.We also further evaluated the outcomes of any subsequent revision.

Methods

We retrospectively reviewed all UKAs with an age and sex matchedgroups of primary TKAs entered into the Tayside Arthroplasty AuditGroup (TAAG) prospective database since 2001 [17]. The TKA groupwere selected from a possible 5146 cases where:

• Agewasbetween theminimumandmaximumages in theUKAgroup;• Date of operation was between the earliest and latest dates ofoperation in the UKA group;

• Surgerywas performed at the samehospitals as those in theUKAgroup.

An individual match for each UKA case was randomly selectedfrom a subset of TKA cases of the same gender and age. Where noneexisted, age matching was relaxed in increments of 1 year either wayuntil a match or matches was found. When more than one possiblematch existed, the TKA case was randomly chosen.

The functional outcomes for both UKA and TKA groups wereobtained (Appendix 1) [18]. The knee society score (KSS) for pain,function and total score were recorded preoperatively and postoper-atively for up to 10 years. For this study, the data was compared at 1, 3and 5 years. Subsequently, we identified all joint arthroplasties, bothUKAs and TKAs that required revisions. The functional outcomes wereanalysed and compared across groups.

Medical complications were recorded and compared across thegroups. Arrhythmias, ischaemic incidents and heart failure wereconsidered as cardiac complications. Anaemia was classified as ahaematology complication. Ileus, vomiting, bowel obstruction, jaun-dice, hiatus hernia and gastrointestinal bleeding were classified asgastrointestinal complications. Confusion, dizziness, cerebrovascularaccident and unresponsive episode were recorded as neurologicalcomplications. Urinary tract infection and retention were classified asurinary complications. Respiratory complication signified a lowerrespiratory tract infection. Any thrombotic or embolic incidents wereclassified as thromboembolism. Patients withmore than 1 complicationreported were placed into ‘N1 complications’ category. Woundcomplicationswere subdivided into superficial infection, deep infectionor “wound problem”, which included a leaking wound, delayed woundhealing or formation of haematoma.

The mean, standard deviation and range were used for descriptivepurposes. Statistical analysis was performed using Statistical Package forthe Social Sciences software (SPSS for Windows, Version 20.0). Data wastested for normal distribution using the Kolmogorov–Smirnov test. TheMann–Whitney U test was used to assess the statistical significancebetween UKA and TKA groups. The Kaplan–Meier test was used forsurvival analysis, up to 12 years; with 95% confidence intervals (CI)calculated. The comparison of revision-free survival was made using log-rank tests. P values less than 0.05were regarded as statistically significant.

Results

602 Oxford UKAs and 5146 primary TKAs were performed in NHSTayside between 2001 and 2013. A randomised matched group of 602unconstrainedprimary TKAswere collected from theTAAGdatabase. AllUKAs had a femoral component with a single fixation peg. The TKAimplants were: Kinemax (Stryker), Columbus (Braun), and LPS NexGen(Zimmer). Patient demographics are shown in Table 1. The average

Knee Arthroplasty Versus Age and Gender Matched Total Knee4), http://dx.doi.org/10.1016/j.arth.2014.03.043

Fig. 2. Medical complications for both groups.

3J.W. Lim et al. / The Journal of Arthroplasty xxx (2014) xxx–xxx

hospital stay for UKAwas 4 days compared to 7 days for the TKA group(P = 0.000). 11.7% (141 of 1204) of patientswere lost to follow-upoverthe 10 years. Of these, 52.5% (74 of 141) haddied fromunrelated events.There was a 12.0% (72 of 602) rate of loss from the database in the TKAcohort and 11.5% (69 of 602) rate of loss in the UKA group. This rate ofloss was not significantly different between the groups.

The comparison of KSS for pain, function and total are shownin Fig. 1 (A, B and C). The pre-operative KSS for pain and total scoreswere not significantly different between UKA and TKA (6.61 vs 6.05,P = 0.219 and 37.58 vs 36.43, P = 0.328 respectively) whereas thepreoperative function score was significantly better for the UKA group(55.65 vs 51.10, P = 0.000). There was a general trend for TKA toperform better than UKA for pain scores (Fig. 1A). At one year, the KSSfor pain was significantly better in the TKA group (41.08 vs 44.14,p = 0.009). However, it was not significantly different at 3 and5 years follow up (p = 0.314 and p = 0.064 respectively). Fig. 1Bdemonstrates that function in the UKA group was superior to TKApreoperatively. The KSS (function) remained significantly better withUKA until 3 years follow up. However, further analysis revealed nostatistically significant difference in the change of function scores over

Fig. 3. Kaplan Meier cumulative survival analysis of unicompartmental kneearthroplasty and total knee arthroplasty.

Please cite this article as: Lim JW, et al, Oxford UnicompartmentalArthroplasty – Functional Outcome and Survivors..., J Arthroplasty (201

time for both UKA and TKA (p = 0.487, p = 0.319, p = 0.453). Thetotal KSS for UKA and TKAwere not significantly different at any pointof the 5-year study (Fig. 1C).

The medical complications are shown in Fig. 2. In the majorityof complication categories, there were fewer complications in theUKA group. There was one intra-operative vascular injury where thepopliteal artery was damaged during UKA. There was one intra-operative neurological injury where the patient had a neuropraxia ofthe common peroneal nerve after TKA. All patients were givenchemical thromboprophylaxis postoperatively. However, there were5 thromboembolic events in each group. There were no immediatepostoperative or intraoperative deaths.

There were fewer wound complications in the UKA group. Inpatients who had undergone UKA, there were 1 deep infection, 7superficial infections and 5 wound problems. The TKA group had4 patients with deep infection, 10 patients with superficial infectionand 6 patients with wound problems. Despite the higher woundcomplication rate in TKA group, all superficial infections resolvedwithin 6 weeks in both groups. All the deep infections requiredwashout but none in TKA group required revision whereas 1 in UKAgroup required revision surgery. No revision surgery was reported forpatients with wound problems.

At 12 years, of the 602 UKAs recorded, 38 (6.30%) had requiredrevision (95% CI 10.47–11.13) whereas of the 602 TKAs recorded, only18 (2.99%) had required revision (95% CI 11.08–11.44). There was astatistical difference between UKA and TKA in the Kaplan-Meiersurvival analysis (P = 0.012) (Fig. 3). The reasons for revision surgeryare shown in Fig. 4. The main cause for revision surgery in theUKA group was aseptic loosening, followed by osteoarthritis ofthe contralateral compartment. Further analysis by the senior author(GC, BC) revealed that tibial component loosening accounted for mostof the aseptic loosening (Table 2). Further analysis was not possiblein 4 cases recorded as ‘aseptic loosening’ due to unavailability of pre-operative x-rays. Interestingly, more TKAs (4 TKAs vs 1 UKA) wererevised due to technical error such as rotational or other malalign-ment of the implants.

The outcome scores after revision at 1-year follow-up (Table 3)suggested that UKAs have a worse function score, but a better totalKSS after the revision surgery. The preoperative and postoperative pain,function and total KSS were not statistically different. Furthermore, thelow number of cases (n = 11 for TKA) suggested that the statisticalanalysis should be interpreted with caution.

Knee Arthroplasty Versus Age and Gender Matched Total Knee4), http://dx.doi.org/10.1016/j.arth.2014.03.043

Fig. 4. Reasons of revision for unicompartmental knee arthroplasty.

Table 3

4 J.W. Lim et al. / The Journal of Arthroplasty xxx (2014) xxx–xxx

Discussion

To our knowledge, this is the first paper to compare the outcomeof age and gender matched UKA and TKA groups. We found nosignificant difference in pain or total KSS for both UKA and TKAgroups. The function scores were significantly different preoperativelybetween the groups, and this continued throughout the period ofanalysis, presumably due to more widespread arthritis being presentin the knees of the TKAgroup. There is inevitably a preoperative selectionbias. Bremner-Smith et al. have suggested that the function score was ingeneral more likely to be influenced by demographic variables andtended to fall with advancing age in an elderly population [19], a factoraccounted for by age-matching the two groups in this study. Whenchange in function over time was analysed there was no differencebetween the groups. This result is contrary to reports of improvedfunction with UKA [6].

Our data suggested that there was peak improvement in all pain,function and total KSS at 1 year with a gradual decline in functionscore after 1 year, irrespective of which implant was used. On the otherhand, thepainand total KSS remained relatively constantpostoperatively.Although not statistically significant, the pain score tended to improvemore and remain better in the TKA group.

TKA was associated with more medical complications than the UKAgroup, especially gastrointestinal, respiratory and ‘N1’ complications.Furthermore, TKAwas associatedwith a higher rate ofwoundproblems,althoughnoneof theseTKAs required revision surgery. Interestingly, theUKA group was associated with a higher rate of cardiac complications.The incidence of thromboembolism was similar in both groups. Thisfinding is contrary to existing literature, which suggests that TKA carriesa higher risk of thromboembolic events than UKA [20,21]. The hospitalstay recorded in this study was before the widespread introduction ofenhanced recovery protocols. Therefore the difference may now be lessmarked and of less relevance. The lower rate of medical complicationsin UKA may have implications when deciding on which procedure to

Table 2Breakdown of Aseptic Loosening Group.

UKA TKA

Both components 3 1Tibial component 6 2Femoral component 3 1Without preoperative revision X-ray 4 0

Please cite this article as: Lim JW, et al, Oxford UnicompartmentalArthroplasty – Functional Outcome and Survivors..., J Arthroplasty (201

recommend in a physiologically frail individual. With respect torevision, our data demonstrated that UKA did not necessarily preventthe progression of osteoarthritis in other compartments as it was arelatively common reason for revision [4].

The UKA group had a two-fold increase in the rate of revisioncompared to the age and gender matched TKA group. The reasons forUKA revision are consistent with the literature [22]. It has beensuggested that the higher incidence of revision of UKA is often due to agenerally lower threshold for revision inUKAandpoorpatient selection.We found no difference in preoperative revision scores in both groups,suggesting that the revision thresholds were the same. Furthermore,the most common reason for revision was aseptic loosening whichwas unlikely to be a direct result of poor patient selection. We believethat the high aseptic loosening rates could be secondary to technicaldifficulty. As a non-specialist institute, our UKA revision rate isconsiderably higher than the originating centre, which is consistentwith the literature [16]. This may prompt further evaluation of the roleof UKA in non-specialist centre. The study is however a ‘pragmatic’ trialwhich reflects actual clinical practice by a mixed group of surgeons inthe National Health Service. All the surgeons had undergone compre-hensive specific training in the use of the UKA implant, as well as thegeneric United Kingdom orthopaedic training.

Current literature suggests that if revision is required for UKA,a good outcome can be achieved by conversion to TKA [11]. Thenumbers were too small in this study to provide meaningful statistics.However, the trend was for UKAs that required revision to have worsefunction scores than the TKAs that required revision. This finding isconsistent with previous studies that have shown poorer outcomesthan expected for UKAs that required revision [15].

The authors acknowledge that the current study has certainlimitations. Firstly, due to the decreasing numbers of patients as time

Revision Score.

Time(Years)

KneeSocietyScore

UKA TKAPValueN Mean Range N Mean Range

Pre-op Pain 32 4.06 0–30 11 4.55 0–10 0.483Function 32 49.84 0–100 10 57.00 30–90 0.475Total 29 38.79 0–73 10 34.36 18–55 0.308

1 Pain 25 30.60 0–50 8 30.63 10–50 0.853Function 25 63.20 0–100 8 65.00 0–100 0.757Total 25 73.84 32–100 8 63.00 9–98 0.665

Knee Arthroplasty Versus Age and Gender Matched Total Knee4), http://dx.doi.org/10.1016/j.arth.2014.03.043

5J.W. Lim et al. / The Journal of Arthroplasty xxx (2014) xxx–xxx

progressed, partly due to the loss of patients and despite the cohortsize, a relatively small number who have reached the relevant stage,long-term deductions about implant survival need to be interpretedwith caution. Despite that, the performance for UKAs was still inferiorto TKAs in the Kaplan Meier cumulative survival analysis at 5 years.Secondly, due to the low number of TKAs that required revision in thecurrent study, no conclusion can be made regarding the comparativeoutcomes of revisions. Furthermore, TAAG does not record preoperativeco-morbidity and we are unable to comment about the postoperativecomplications profile in each group. We therefore made the assump-tions that both cohorts are in similar fitness level with the matchingof patients.

In conclusion, our study suggests that UKA is associatedwith fewerpostoperative complications, however TKA provides better initialpain relief and is less likely to require revision. The revision rate forUKA was twice as much as TKA. The theoretical advantages of UKA arenot borne out by the findings in this study other than immediatepostoperative complications.

Acknowledgments

The authors would like to thank the Tayside Arthroplasty AuditGroup team who have contributed data to this study. The authors aregrateful to Dr. Weijie Wang for statistical support and Mr. Ian Christiefor image illustrations.

Appendix 1. Knee Society Score

Category Knee Society Score

Excellent 80-100Good 70-79Fair 60-69

References

1. National Joint Registry for England & Wales: 9th Annual Report 2012. http://www.njrcentre.org.uk/NjrCentre/Portals/0/Documents/England/Reports/9th_annual_report/NJR%209th%20Annual%20Report%202012.pdf. [date last accessed 8th June2013].

2. Murray DW, Goodfellow JW, O’Connor JJ. The Oxford medial unicompartmentalarthroplasty: a ten-year survival study. J Bone Joint Surg Br 1998;80-B(6):983.

Poor b60

Please cite this article as: Lim JW, et al, Oxford UnicompartmentalArthroplasty – Functional Outcome and Survivors..., J Arthroplasty (201

3. Svärd UCG, Price AJ. Oxford medial unicompartmental knee arthroplasty: a survivalanalysis of an independent series. J Bone Joint Surg Br 2001;83-B(2):191.

4. Weale AE, Murray DW, Crawford R, et al. Does arthritis progress in the retainedcompartments after ‘Oxford’ medial unicompartmental arthroplasty? A clinicaland radiological study with a minimum ten-year follow-up. J Bone Joint Surg Br1999;81-B(5):783.

5. Newman JH, Ackroyd CE, Shah NA. Unicompartmental or total knee replacement?Five-year results of a prospective, randomised trial of 102 osteoarthritic knees withunicompartmental arthritis. J Bone Joint Surg Br 1998;80-B(5):862.

6. Newman J, Pydisetty RV, Ackroyd C. Unicompartmental or total knee replacement:the 15-year results of a prospective randomised controlled trial. J Bone Joint Surg Br2009;91-B(1):52.

7. Beard D, Price A, Cook J, et al. Total or Partial Knee Arthroplasty Trial – TOPKAT:study protocol for a randomised controlled trial. Trials 2013;14:292.

8. Låstad Lygre SH, Espehaug B, Havelin LI. Pain and function in patients after primaryunicompartmental and total knee arthroplasty. J Bone Joint Surg 2010;92(18):2890.

9. Baker PN, Petheram T, Jameson SS, et al. Comparison of patient-reported outcomemeasures following total and unicondylar knee replacement. J Bone Joint Surg Br2012;94-B(7):919.

10. Lyons MC, MacDonald SJ, Somerville LE, et al. Unicompartmental versus total kneearthroplasty database analysis: is there a winner? Clin Orthop Relat Res 2012;470(1):84.

11. Johnson S, Jones P, Newman JH. The survivorship and results of total kneereplacements converted from unicompartmental knee replacements. Knee2007;14(2):154.

12. Saldanha KAN, Keys GW, Svard UCG, et al. Revision of Oxford medial unicompart-mental knee arthroplasty to total knee arthroplasty – results of a multicentre study.Knee 2007;14(4):275.

13. Willis-Owen CA, Brust K, Alsop H, et al. Unicondylar knee arthroplasty in the UKNational Health Service: an analysis of candidacy, outcome and cost efficacy. Knee2009;16(6):473.

14. Padgett DE SS, Insall JN. Revision total knee arthroplasty for failed unicompart-mental replacement. J Bone Joint Surg 1991;73(2):186.

15. Pearse AJ, Hooper GJ, Rothwell A, et al. Survival and functional outcome afterrevision of a unicompartmental to a total knee replacement: the New Zealandnational joint registry. J Bone Joint Surg Br 2010;92-B(4):508.

16. Chou DT, Swamy GN, Lewis JR, et al. Revision of failed unicompartmental kneereplacement to total knee replacement. Knee 2012;19(4):356.

17. Tayside Arthroplasty Audit Group (TAAG). http://www.dundee.ac.uk/orthopaedics/audit/taag.htm . [date last accessed 09 June 2013].

18. Insall JN, Dorr LD, Scott RD, et al. Rationale of the Knee Society clinical ratingsystem. Clin Orthop Relat Res 1989;248:13.

19. Bremner-Smith AT, Ewings P, Weale AE. Knee scores in a ‘normal’ elderlypopulation. Knee 2004;11(4):279.

20. Willis-Owen CA, Sarraf KM, Martin AE, et al. Are current thrombo-embolicprophylaxis guidelines applicable to unicompartmental knee replacement? J BoneJoint Surg Br 2011;93-B(12):1617.

21. Furnes O, Espehaug B, Lie SA, et al. Failure mechanisms after unicompartmentaland tricompartmental primary knee replacement with cement. J Bone Joint Surg2007;89(3):519.

22. Dudley T, Gioe T, Sinner P, et al. Registry outcomes of unicompartmental kneearthroplasty revisions. Clin Orthop Relat Res 2008;466(7):1666.

Knee Arthroplasty Versus Age and Gender Matched Total Knee4), http://dx.doi.org/10.1016/j.arth.2014.03.043