Comparing Outcomes of Medial Parapatellar and Subvastus Approaches in Total Knee Arthroplasty

The Journal of Arthroplasty Vol. 27 No. 3 2012

Comparing Outcomes of Medial Parapatellar and Subvastus Approaches in Total Knee Arthroplasty A Randomized Controlled Trial Michael G. Bourke, BPhty,*y Gwendolen A. Jull, PhD, MPhty, FACP,y Peter J. Buttrum, BPhty,* Prudence L. FitzPatrick, MBBS, FRACS,z Philip A. Dalton, BPhty (Hons), MMedSc, MBBS (Hons), FRACS CIME,z and Trevor G. Russell, PhD, BPhtyy

Abstract: The medial parapatellar (MP) approach in total knee arthroplasty is more common, but the subvastus (SV) approach is less insulting to the quadriceps. Whether the SV approach affords better outcomes was investigated using 90 participants with knee osteoarthritis, randomized to receive either SV or MP approaches and followed for 18 months. The primary outcome was the American Knee Society Score (AKSS); secondary outcomes included pain, knee range, quadriceps lag, Oxford Knee Score, 3-m timed “Up and Go” test, days to straight leg raise, surgeon perceived difficulty, operation duration, and length of stay. Analysis (n = 76) revealed no significant difference in AKSS (P = .076) or other outcomes, except the following: AKSS Functional scores at 12 and 18 months, favoring the MP (P = .032 and P = .028 respectively); surgeon's perceived difficulty, favoring the MP (P = .001); and days to straight leg raise, favoring the SV (P = .044). This study found that the SV approach offers no clinical benefit over the MP approach. Keywords: arthroplasty, knee, medial parapatellar, subvastus. Crown Copyright © 2012 Published by Elsevier Inc. All rights reserved.

Primary and revision total knee arthroplasties (TKAs) performed in Australia have risen by 42.6% since 2003 and, in 2009, totaled 40 675 [1]. With health economists predicting continual increases, it is important to optimize surgical procedures to ensure efficiency and to produce the best possible outcomes. The medial parapatellar (MP) approach is arguably the most common approach in TKA, despite disrupting the quadriceps muscle up to 50 to 60 mm proximally from the superomedial border of the patella. The subvastus (SV) approach is less widely used but aims to preserve the quadriceps muscle by bluntly dissecting deep to the From the *Department of Physiotherapy, QEII Jubilee Hospital, Queensland Health, Brisbane, Queensland, Australia; yDivision of Physiotherapy, School of Health and Rehabilitation Sciences, The University of Queensland, Brisbane, Queensland, Australia; and zDepartment of Orthopaedic Surgery, QEII Jubilee Hospital, Queensland Health, Brisbane, Queensland, Australia. Submitted March 8, 2011; accepted June 2, 2011. Supplementary material available at www.arthroplastyjournal.org. The Conflict of Interest statement associated with this article can be found at doi:10.1016/j.arth.2011.06.005. Reprint requests: Michael G. Bourke, BPhty, Department of Physiotherapy, QEII Jubilee Hospital, Locked Bag 2, Archerfield, Qld, Australia 4108. Crown Copyright © 2012 Published by Elsevier Inc. All rights reserved. 0883-5403/2703-0003$36.00/0 doi:10.1016/j.arth.2011.06.005

vastus medialis muscle [2]. Preservation of the quadriceps during TKA is argued to result in improved quadriceps function postoperatively [3-7], less pain [3,4,7], and shorter length of hospital stay [3]. Other proposed advantages of the SV approach include a reduction in analgesic requirements [3] and preservation of patellar vascularity [8-10]. Despite the proposed benefits of the SV approach, a number of surgical technical difficulties have been reported; for example, it is more difficult to visualize the surgical field and evert the patella [10]. Factors such as obesity, muscle bulk, and contractures around the knee are also known to make the SV approach technically difficult [2,11-13]. A recent systematic review highlighted the paucity of high-quality studies providing evidence for the efficacy of the SV compared with the MP approach [14]. Studies to date have generally lacked the methodological rigor required to derive definitive conclusions about the outcomes of SV or MP approaches. Methodological issues included poor randomization processes where allocation was decided by the senior surgeon [5,6] and poor complication reporting [3,4,7], which affected the generalizability of results and contributed to the trials' relatively low-quality methodology component scores [15-17]. The highest quality trial in the review reported

347

348 The Journal of Arthroplasty Vol. 27 No. 3 March 2012 appropriate randomization procedures, but their followup extended only 3 months postoperatively [3]. In one clinical trial published subsequent to this review [18], the SV was found to be superior to the MP approach on the outcomes of range of motion (ROM) and American Knee Society Scores (AKSS) at 1 week and Western Ontario and McMaster Osteoarthritis Index (WOMAC), Short-Form 36 (SF-36), and European Quality of Life (EuroQol) scores at 1 year. This trial did not investigate immediate inpatient outcomes in the first few days postoperatively, nor did it follow up beyond 12 months. Therefore, further trials are required to investigate if there are any definitive benefits of the SV compared with the MP approach in the immediate postoperative phase and in the long term. To this end, a single-center, prospective, randomized, controlled trial was conducted in a metropolitan hospital to further compare the SV and MP approaches. Outcomes were assessed preoperatively, intraoperatively for some outcomes, on days 1 to 3 postsurgery, and at discharge as well as at week 6, month 6, month 12, and month 18 postsurgery. It was hypothesized (i) that participants receiving the SV approach would experience better early outcomes than those receiving the MP approach and (ii) that outcomes would converge by 18 months after surgery.

Materials and Methods The study was conducted in a hospital in Brisbane, Queensland, Australia, from May 2006 to November 2009. Institutional ethics committee approval was granted for this trial, and written informed consent was obtained from each participant. The trial was registered on the Australian and New Zealand Clinical Trials Registry (ACTRN12606000376549), and there were no external sources of funding. Participants Eligible participants were recruited by the participating orthopedic surgeons from patients attending their outpatient clinics and who were subsequently scheduled to undergo a TKA at the hospital. To be included, the participants were required to be 18 years or older, diagnosed with knee osteoarthritis and scheduled to undergo primary unilateral TKA, and appropriate for either the MP or SV approach. Participants were also required to possess normal mentation, be able to attend supervised outpatient physiotherapy rehabilitation sessions over a period of 6 weeks, and be able to provide signed informed consent. Persons excluded were those with comorbidities preventing participation in rehabilitation (eg, severe obstructive pulmonary disease, hemiplegia after stroke); undergoing revision or bilateral TKA, having knee stiffness with less than 70° of flexion or flexion contracture of greater than 20°, or having undergone previous high tibial osteotomy or major arthrotomy on the operative knee (indicating inability to

perform SV approach as originally described) [2]. Also excluded were persons who were unable to preoperatively mobilize full weight-bearing in a bipedal manner with or without a walking aid or who were unlikely to be able to follow the TKA clinical pathway. In addition, participants were excluded intraoperatively if they required an intervention outside the standardized surgical protocol, for example, requiring a femoral nerve block or a lateral surgical release. This enabled a better comparison of successfully completed SV and MP procedures. Randomization and Blinding Randomization was undertaken using a computergenerated randomization sequence that was created with stratification to surgeon and a 1:1 allocation using a block size of 4. Block randomization was used to ensure an equal number of participants in each group as the trial progressed, and the stratification ensured that each surgeon had a similar number of participants in each surgical group. The allocation sequence was generated by a research team member (T.R.) who was not involved in participant contact or data collection. The allocations were placed in sequentially numbered, sealed envelopes, and the envelope was drawn out by a theater nurse. Group allocation was revealed to the orthopedic surgeon in the operating suite immediately before surgery. The assessors, physiotherapists, and nurses providing the postoperative care and participants were blind to the surgical procedure received by the patient. The operating surgeon was unblinded. The surgical approach and procedure were described in the original operation report, which was then placed in a sealed envelope and not opened until the conclusion of the trial. To maintain blinding of the physiotherapists and nurses, a second printed temporary operation report with the surgical approach blackened on both sides was placed in the medical record after surgery. Procedure All participants who had agreed to enter the study attended a preoperative education clinic conducted approximately 4 weeks before surgery. A preoperative assessment of all baseline measures was undertaken, and participants were randomized at this point. All participating surgeons, regardless of experience with either operative approach, completed 5 supervised sessions with the senior consultant surgeon (J.D.) to standardize the SV approach, which is less commonly performed than the MP approach. All 6 surgeons performed both procedures to enhance the generalizability of the results. Surgical Procedure and Postoperative Care Surgeons used their preferred prosthesis from the choice of either the Smith & Nephew Genesis II (Brisbane, Australia) or LCS Depuy mobile-bearing total knee prosthesis (Brisbane, Australia). A tourniquet was fitted

The Medial Parapatellar Versus Subvastus Approach in TKA  Bourke et al

but only inflated for cementing of the prostheses. An identical midline skin incision was used in both procedures to ensure blinding of the other researchers in the postoperative period. For the SV approach, the knee was flexed for the skin incision and incision of the inferior aspect of the capsule. The vastus medialis was then dissected with the knee flexed or extended. The patella was not everted, but rather subluxed laterally. The femur and tibia were prepared, and the prosthesis was inserted. For the MP approach, the knee was flexed, and an MP incision extending 6 to 7 cm above the proximal pole of the patella was performed. The incision was extended inferiorly to the medial aspect of the tibial tubercle. The patella was everted for the duration of the surgery if required. As for the SV approach, the femur and tibia were prepared, and the prosthesis was inserted. For both the MP and SV approaches, wound drains exited laterally avoiding vastus lateralis where possible and were removed on the first day after surgery. Incisions were closed with the knee in flexion of approximately 90°. Postoperative pain relief for all participants was administered via intravenous patient-controlled analgesia for the first 48 hours and then subsequently by oral analgesia. Postoperative nursing care (including use of thromboembolic deterrent stockings and removal of surgical drain) and length of stay criteria were standardized for both groups. The rehabilitation of all patients was standardized according to the hospital's clinical pathway for TKA and physiotherapy guidelines for rehabilitation. After discharge, all patients attended at least 2 postoperative physiotherapy rehabilitation sessions before their 6-week review, as is conventional practice at the hospital. Outcomes The primary outcome measure was the AKSS (x/200 points) [19]. For better discrimination, its component scores, the AKSS Objective (x/100 points) and AKSS Functional (x/100 points) were also considered. Secondary outcome measures included the Oxford Knee Score (OKS; 12-60/60 points) [20], 3-m timed “Up and Go” test (TUG; in seconds) [21], knee flexion and extension ROM (in degrees), quadriceps lag on straight leg raise (SLR; in degrees), days to SLR, pain (Numerical Assessment Scale 1-10), and knee girth (in millimeters). These outcomes were measured at time points 4 weeks preoperatively (baseline), postoperatively on each of days 1 to 3, on discharge, week 6, month 6, month 12, and month 18. Intraoperative secondary outcome measures were knee flexion and extension range of movement, operation duration (in minutes), surgeon's perceived level of difficulty with the operative approach (Numerical Assessment Scale 1-10), and tourniquet duration (in minutes). Length of postoperative stay in the orthopedic ward (in days) was also recorded. All data were collected on an electronic personal digital assistant by physiotherapists and surgeons who were

349

trained in its use, with the exception of the self-rated OKS, which was collected as a paper-based questionnaire. Flexion, extension, and quadriceps lags were measured using software on the personal digital assistant, which was adapted from telerehabilitation research [22]. This software has demonstrated criterion validity (limits of agreement = −1.66° to 1.76° for the measurement of knee joint excursion from extension to flexion) and excellent intrarater (Intraclass Correlation Coeficient [ICC](2,1) = 0.97 to N0.99) and interrater (ICC(2,1) = 0.93 to N0.99) reliability. Sample Size A sample size calculation was conducted using the AKSS data from the first 22 participants from the MP group (AKSS: 56.75 ± 14.19) and 21 participants from the SV group (AKSS: 51.82 ± 14.04). To detect a 5% improvement in the AKSS (10 points) with a 2-sided 5% significance level and a power of 80%, a sample size of 32 participants per group was necessary (total n = 64). Sufficient recruitment (final n = 90) was undertaken to allow for a dropout rate of at least 10% during the 18month follow-up period, as well as any loss to the trial due to surgical variations from protocol.

Statistical Methods The data retained for analysis were of the participants who underwent the MP or SV surgical procedure as per study protocol. All data were inspected for normality and transformed as required before analysis. Linear mixed models (LMMs) were performed for analysis of continuous variables (AKSS, AKSS Objective, AKSS Functional, OKS, TUG, flexion, extension, quadriceps lag, pain, and girth), using surgical approach (SV or MP) and outcome assessment points (and their interaction effect) as fixed effects. The LMM was favored over Student t tests because of its ability to correctly handle correlation errors observed with repeated measures, its ability to support the random effects and hierarchical effects of the variables, and ability to account for baseline differences [23]. Analysis of secondary outcomes (days to SLR, operation time, tourniquet duration, surgeon's rated difficulty with the procedure, length of stay) was performed using 1way analysis of variance to determine differences between the 2 treatment groups. Independent-samples Mann-Whitney U tests were used to analyze baseline demographics. A significance level of P b .05 was chosen for all analyses.

Results Ninety participants were accepted into the trial between 11 May 2006 and 18 October 2007. The follow-up of all participants ended on 12 November 2009. Fig. 1 illustrates participant recruitment and the flow of participants throughout the study. Of the original 90 participants, 76 received the intervention and were

350 The Journal of Arthroplasty Vol. 27 No. 3 March 2012 Assessed for eligibility (n = 158)

Excluded (n = 68) • Declined to participate (n = 51) • Other (n = 17)

Randomized (n = 90)

Allocated to MP group (n = 45)

No operation within study period (n = 4)

Allocated to SV group (n = 45)

No operation within study period (n = 5)

Received MP approach (n = 41)

Intra-operative complications contravening standardized surgical protocol e.g. required lateral release (n = 1)

Received SV approach (n = 40)

Intra-operative complications contravening standardized surgical protocol e.g. required lateral release (n = 4)

Included in analysis (n = 40)

Included in analysis (n = 36)

Lost to follow-up (n = 2) • Lost after 6 months (n = 1) • Lost after 12 months (n = 1)

Lost to follow-up (n = 1) • Lost after 6 weeks (n = 1).

Fig. 1. A diagram of participant flow through the trial.

retained in the study for analysis. Nine participants, although randomized, did not receive surgery within the study period because they subsequently declined the procedure or experienced medical complications subsequent to recruitment (MP group, 4; SV group, 5). Five participants had unplanned interventions intraoperatively. In the MP group, 1 participant received a unicompartmental knee arthroplasty. In the SV group, 4 participants variously received the following: simultaneous bilateral TKA, femoral nerve block, femoral nerve block and lateral release, and postoperative bracing and restricted weight-bearing due to intraoperative medial collateral ligament avulsion. Because these interventions/complications contravened the standardized surgical protocol, these participants were excluded from further statistical analysis. Therefore, the intention-totreat analysis was performed using the data from 76 participants. Within the cohort of 76, 3 participants were

lost to follow-up, 1 from the MP group at 6 months and another at 12 months and 1 from SV at 6 weeks. The data of these participants were included in the analysis for the time points for which it was available. Baseline analysis revealed no significant betweengroup differences for gender (MP: female 26, male 14; SV: female 19, male 17: P = .282) or age (MP: 67.7 ± 6.5 years; SV: 68.1 ± 8.2 years: P = .83). There were no significant differences between groups in baseline functional or clinical characteristics (Table 1). The values for the primary and secondary outcomes for the MP and SV groups at each follow-up time point along with results of LMM analysis for normally distributed outcomes are presented in Table 2 (available online at www.arthroplastyjournal.org). Compared with baseline values (Table 1), all outcomes progressively improved from 6 weeks postoperatively onward in both the MP and SV groups. However, there were no statistically

The Medial Parapatellar Versus Subvastus Approach in TKA  Bourke et al

351

Table 1. Baseline Comparisons of Physical and Functional Outcome Measures in the MP and SV Groups MP (n = 40) Outcome Primary outcome AKSS (x/200) AKSS Objective (x/100) AKSS Functional (x/100) Secondary outcome Pain (x/10) Extension (deg) Flexion (deg) Quadriceps lag (deg) Girth (mm) OKS (12-60/60) TUG, 3-m course (s)

n

Mean ± SD

40 40 40

100.5 ± 30.6 55.7 ± 13.7 44.6 ± 22.2

40 40 40 40 38 30 40

4.9 ± 6.5 ± 117.6 ± 2.7 ± 425 ± 38.6 ± 13.7 ±

SV (n = 36) Range

2.0 5.0 12.0 4.1 38 8.6 13.5

n

Mean ±SD

Range

35-173 34-93 0-90

36 36 36

102.3 ± 32.6 54.6 ± 15.5 44.7 ± 23.1

36-179 24-90 0-90

1-8 −5.7 to 20.0 88.5-140.0 −5.6 to 13.8 344-510 23-53 8.0-29.3

36 36 36 36 33 26 34

4.9 ± 2.1 7.6 ± 5.7 119.0 ± 12.5 2.4 ± 3.5 425 ± 40 38.1 ± 8.3 15.9 ± 7.2

1-7 0.0-20.0 88.5-140.0 −3.1 to 10.0 360-535 25-52 7.7-34.6

There was no significant between-group differences for any baseline measure.

significant differences overall in the primary outcomes between the groups based on LMM analysis (against fixed effects of group and outcome assessment point). Post hoc analysis revealed significant differences at 3 specific time points. On day 1 postoperatively, the SV group demonstrated a significantly better AKSS Objective score (P = .029). However, the MP group demonstrated significantly better AKSS Functional score at 12 months (P = .032) and 18 months (P = .028). There was also a trend toward greater improvement in the MP group on AKSS (P = .076) at 18 months. Analysis of secondary outcomes revealed 2 significant differences only. There was a significant difference in days to SLR (MP: 2.76 ± 1.89 vs SV: 1.92 ± 1.59, P = .044), indicating an earlier return of quadriceps function in the SV group. There was a difference in surgeon's perceived difficulty between surgical approaches (MP: 3.32 ± 1.89 vs SV: 5.38 ± 2.33, P = .001), indicating that surgeons considered the SV approach to be more technically difficult. There were no differences in other secondary outcomes, including operation duration, tourniquet duration, and length of hospital stay (Table 3), but there was a trend toward less pain in the MP (MP: 1.66/ 10 vs SV: 2.63/10, P = .097) group at 6 weeks. Within the 18-month follow-up period, there were 6 knee-related adverse effects. In the SV group, there were 2 complications; aseptic loosening requiring revision and stiffness requiring manipulation under anesthetic. In the

MP group, the 4 complications were aseptic loosening requiring revision, deep vein thrombosis, stiffness requiring manipulation under anesthetic, and wound breakdown requiring intravenous antibiotics.

Discussion The proposed clinical benefits of the SV approach such as less pain, earlier quadriceps function, and the potential to reduce length of hospital stay and costs make this approach an attractive option. Although the SV group was significantly better on day 1 AKSS Objective scores, at all other time points there was either no difference between the approaches, or the difference or trend favored the MP group. Therefore, the trial hypotheses of better early outcomes with the SV approach were rejected, and the convergence of SV and MP group outcomes at 18 months was conditionally accepted as AKSS Functional scores were better in the MP group by 12 months postoperatively. The results of this trial showed that both groups made marked improvements in the primary outcomes measured over the course of the study when compared with baseline values. We are not aware of any studies that have investigated the minimum clinically important difference for the AKSS; therefore, it is difficult to rate the improvement in this study on these terms. Although the LMM did not demonstrate any difference for AKSS Functional scores on the factor of group

Table 3. Results of Analysis for Interval Outcome Measurements MP (n = 40) Outcome Days to SLR Surgeon perceived difficulty (x/10) Operation duration (min) Tourniquet duration (min) Length of stay (d)

n 37 28 40 40 39

Mean ± SD 2.8 ± 3.3 ± 91.6 ± 28.8 ± 4.5 ±

1.9 1.9 (n = 28) 20.1 (n = 40) 14.4 (n = 29) 1.2 (n = 39)

SV (n = 36) n

Mean ± SD

36 26 36 36 35 *

1.9 ± 1.6 5.4 ± 2.3 (n = 26) 97.9 ± 18.1 (n = 36) 34.1 ± 20.71 (n = 31) 4.7 ± 1.2 (n = 35)

Bold-italic font indicates significance, P b .05. * One participant from MP received inpatient rehabilitation (length of stay, 37.5 days). One participant from SV received inpatient rehabilitation (length of stay, 14.5 days). These statistical outliers were removed from analysis.

352 The Journal of Arthroplasty Vol. 27 No. 3 March 2012 by outcome assessment point, post hoc analysis revealed a significant difference favoring the MP approach at 12 and 18 months. Inspection of the raw data revealed a generalized improvement in the MP group, which cannot be attributed to a small number of cases. Given that no significant difference at these time points was found with any other variable in the study, we are left to attribute this improvement in the AKSS Functional score to the MP surgical approach. This trial found no statistically significant differences in pain scores but did reveal a trend favoring the MP group at 6 weeks postoperatively, similar to Bridgman et al [18]. Our results for pain do not agree with earlier studies that found less pain associated with the SV approach [4,6,7,18]. For example, a trial conducted by Cameron [7] found less pain in the SV group at day 1 postoperatively but did not state whether the patient or assessor was blinded to the approach, which may have affected the outcome. The result of earlier SLR in the SV group concurs with preexisting literature describing earlier quadriceps function with the SV approach [3,4,7]. Despite this difference, earlier SLR did not result in significantly better outcomes on the AKSS or other outcomes (eg, quadriceps lag on SLR, TUG), or a reduction in hospital length of stay for the SV group in this trial. Some of the outcomes on which previous study conclusions were based (ie, Cybex isokinetic dynamometer system [5] and LIDO isokinetic dynamometer [6]) were not performed in this trial. Our study found no significant differences in ROM overall, or at any isolated time points, between the MP and SV groups. This is in contrast to earlier studies that found improved ROM with the SV approach. Of these, 2 studies found differences favoring the SV group at or earlier than 4 weeks. One used outcomes that we did not (days to full passive extension, days to passive 90°) [4], and the other had unusually low flexion range in both groups (MP 55° vs SV 78° at 1 week) [3]. The third study found small but significant differences in ROM at 6 and 12 months [7]. A more recent randomized controlled trial by Bridgman et al [18] with 231 participants found that the SV group had greater ROM at 1 week (7° greater increase in SV group compared with baseline), although the scores for both groups were also relatively low (MP: 57.2° vs SV: 61.2°) compared with our results (MP: 84.2° vs SV: 84.0°, at discharge). The population demographics of Bridgman et al [18] were similar to the present study; however, the larger sample and use of a standardized prosthesis may have contributed to their findings. The significant difference in surgeon's perceived difficulty between surgical approaches is consistent with Bridgman et al [18] where “ease of exposure” scores were significantly worse in the SV group. A surgeon's perceived difficulty with the SV approach is a major reason given for undertaking the more common

MP approach. Although the SV approach was reported as more difficult in our trial, this did not translate to a higher complication rate; in fact, there were more complications in the MP group (n = 4) than in the SV group (n = 2). Our more experienced surgeons rated the SV approach less difficult than did the inexperienced surgeons. Notably, despite the difference in difficulty scores, neither the operation duration nor tourniquet duration differed significantly between groups. The OKS is a widely used measure in TKA research, with evidence supporting its reliability, construct validity, and freedom from bias [20,24]. We did not find any significant differences between the MP and SV groups, although we feel that it is probably less suited as an inpatient questionnaire and less sensitive at these time points. This relates to the 7 measures pertaining to function, 5 of these had usually not been attempted by participants before hospital discharge. We are only aware of one other study investigating surgical approach in TKA that used the OKS at such an early time postoperatively, but this study did not investigate SV outcomes [25]. Learning from the limitation of previous trials, great care was exercised in this study to maximize its methodological rigor. This included prospective randomization, blinding of participants and assessors, power analysis to determine sample size, and an intention-totreat analysis. It is possible that some of our findings are in contrast to existing literature because of the enhanced methodology. The quality of methodology component score described by Bourke et al [14] was self-determined to be 11 of 12 for this trial, whereas the scores for existing literature ranged between 4 of 12 and 7 of 12 [14]. Although an 18-month follow-up is arguably short for an arthroplasty study, the primary intention of this trial was to investigate early outcomes because it was anticipated that this is where the differences would occur between the groups. We attempted to mitigate the influence of the type of prosthesis with our stratification process, ensuring that each surgeon had a similar number of participants in each group. A systematic review comparing the efficacy of the SV and MP approaches recommended that blood loss be collected alongside the outcomes used in this study [14]. An attempt to collect these data was made; however, this was not feasible because of trial resources. Future trials should include the collection of blood loss data and should also record the volume and type of pain medication.

Conclusions This study aimed to investigate the physical and functional outcomes of the SV compared with the MP approach for TKA. We found no evidence on our primary outcome (AKSS) to support our original hypothesis that the SV approach affords better early outcomes when compared with the MP approach. In

The Medial Parapatellar Versus Subvastus Approach in TKA  Bourke et al

addition, it was evident from our study that surgeons found the SV a more technically difficult surgical approach and that the AKSS Functional scores favored the MP group by 12 months postoperatively.

Acknowledgments We thank Dr. John Dodsworth, Department of Orthopaedic Surgery, QEII Jubilee Hospital, Queensland Health, Brisbane, Queensland, Australia, for the surgical procedure standardization and Matthew Lumchee, Department of Physiotherapy, QEII Jubilee Hospital, Queensland Health, Brisbane, Queensland, Australia, for the data management and manuscript formatting.

References 1. Australian Orthopaedic Association National Joint Replacement Registry. Annual report. Adelaide: AOA; 2010. 2. Hofmann A, Plaster R, Murdock L. Subvastus (Southern) approach for primary total knee arthroplasty. Clin Orthop 1991;70. 3. Roysam G, Oakley M. Subvastus approach for total knee arthroplasty: a prospective, randomized, and observerblinded trial. J Arthroplasty 2001;16:454. 4. Weinhardt C, Barisic M, Bergmann E, et al. Early results of subvastus versus medial parapatellar approach in primary total knee arthroplasty. Arch Orthop Trauma Surg 2004; 124:401. 5. Cila E, Guzel V, Ozalay M, et al. Subvastus versus medial parapatellar approach in total knee arthroplasty. Arch Orthop Trauma Surg 2002;122:65. 6. Faure B, Benjamin J, Lindsey B, et al. Comparison of the subvastus and paramedian surgical approaches in bilateral knee arthroplasty. J Arthroplasty 1993;8:511. 7. Cameron S. Subvastus versus medial parapatellar arthrotomy in total knee arthroplasty. J Appl Res 2001;1(suppl 2):95. 8. Holtby R. Osteonecrosis and resorption of the patella after total knee replacement: a case report. Clin Orthop 1996;155. 9. Kayler D. Surgical interruption of patella blood supply by total knee arthroplasty. Clin Orthop 1988;221. 10. Matsueda M, Gustilo R. Subvastus and medial parapatellar approaches in total knee arthroplasty. Clin Orthop 2000;161. 11. Cushner F. The subvastus approach to the knee. J Knee Surg 2003;16(suppl 1):52.

353

12. Knezevich S. Comparison of subvastus quadriceps-sparing and standard quadriceps-splitting approaches in total and unicompartmental knee arthroplasty. Orthop Trans 1992; 16:615. 13. Sporer S. The minimally invasive subvastus approach for primary total knee arthroplasty. J Knee Surg 2006; 19(suppl 1):58. 14. Bourke M, Buttrum P, Dalton P, et al. Systematic review of medial parapatellar and subvastus approaches in total knee arthroplasty. J Arthroplasty 2010;25:728. 15. Higgins JPT, Green S. (eds). Cochrane handbook for systematic reviews of interventions. Main text. Version 5.0.2. [updated September 2009]. The Cochrane Collaboration; 2009. http://www.mrc-bsu.cam.ac.uk/cochrane/ handbook502/. 16. Jadad A. Assessing the quality of reports of randomized and controlled trials: is blinding necessary. Control Clin Trials 1996;17:1. 17. Khan R. Surgical approaches in total knee arthroplasty (protocol). Cochrane Database Syst Rev 2005;CD005329, doi:10.1002/14651858.CD005329. 18. Bridgman S, Walley G, MacKenzie G, et al. Sub-vastus approach is more effective than a medial parapatellar approach in primary total knee arthroplasty: a randomized controlled trial. Knee 2009;16:216. 19. Insall J, Dorr L, Scott R, et al. Rationale of the knee society clinical rating system. Clin Orthop 1989;248:13. 20. Dawson J, Fitzpatrick R, Murray D, et al. Questionnaire on the perceptions of patients about total knee replacement. J Bone Joint Surg Br 1998;80:63. 21. Podsiadlo D, Richardson S. The timed “Up & Go”: a test of basic functional mobility for frail elderly persons. J Am Geriatr Soc 1991;39:142. 22. Russell T. Goniometry via the internet. Aust J Physiother 2007;52:136. 23. Garson DG. Linear mixed models: Heirarchical, linear, random effects, multilevel, random coefficients, and repeated measures models. In: Garson DG, editor. Statnotes: Topics in Multivariate Analysis. North Carolina: Raleigh; 2011. url: http://faculty.chass.ncsu.edu/garson/ PA765/multilevel.htm. 24. Conaghan P, Emerton M, Tennant A. Internal construct validity of the Oxford Knee Scale: evidence from Rasch measurement. Arthritis Rheum 2007;57:1363. 25. Karachalios T, Giotikas D, Roidis N, et al. Total knee replacement performed with either a mini-midvastus or a standard approach: a prospective randomised clinical and radiological trial. J Bone Joint Surg Br 2008;90-B:584.

Table 2. The Physical and Functional Outcome Measures in MP and SV Groups at All Follow-Up Points Including Results of LMM Analysis for Normally Distributed Outcomes, Presented for Time by Allocation Intervention Effect

AKSS/200 Intraoperative Day 1 Day 2 Day 3 Discharge Week 6 Month 6 Month 12 Month 18 LMM

MP SV MP SV MP SV MP SV MP SV MP SV MP SV MP SV MP SV

57.5 ± 13.1 57.8 ± 15.9 53.0 ± 13.3 53.1 ± 16.9 125.9 ± 29.2 127.7 ± 37.9 151.9 ± 28.3 140.9 ± 34.1 162.7 ± 23.0 153.1 ± 29.7 167.3 ± 27.5 153.1 ± 36.6 P1.111;473 = .355

AKSS Objective/100

50.2 ± 16.4 59.2 ± 15.0 56.7 ± 15.4 56.6 ± 13.7 56.9 ± 13.0 57.8 ± 15.9 50.1 ± 10.8 48.4 ± 12.1 70.2 ± 15.7 69.2 ± 13.0 80.4 ± 12.6 72.8 ± 15.4 80.9 ± 8.8 82.2 ± 9.9 84.2 ± 14.7 81.1 ± 12.8 P1.729;605 = .089

Bold-italic font indicates significance, P b .05. * All participants scored 0/100 at this time point.

Secondary Outcomes AKSS Funtional/100

Pain/10

0.5 ± 2.3 0.0 ± 0.0 * 2.9 ± 5.3 4.7 ± 8.4 55.7 ± 21.0 58.5 ± 28.0 71.4 ± 20.6 68.1 ± 24.3 81.9 ± 17.9 70.9 ± 25.7 83.1 ± 18.2 72.0 ± 27.1 P1.773;481 = .103

3.8 ± 1.8 3.1 ± 1.8 3.0 ± 1.9 3.2 ± 1.5 3.0 ± 1.4 3.1 ± 1.7 2.9 ± 1.8 2.4 ± 1.6 1.7 ± 1.3 2.6 ± 2.2 1.4 ± 1.0 1.7 ± 1.0 1.4 ± 0.9 1.6 ± 1.0 1.8 ± 1.8 2.0 ± 1.9 P1.366;619 = .208

Knee Extension (deg) 1.2 ± 3.0 0.8 ± 4.0 5.3 ± 4.2 4.6 ± 5.2 4.7 ± 4.3 6.3 ± 4.1 5.6 ± 4.0 7.4 ± 4.8 6.1 ± 3.5 7.5 ± 4.2 6.8 ± 4.6 7.6 ± 4.9 3.7 ± 5.0 3.7 ± 5.2 3.0 ± 4.5 3.2 ± 4.2 2.7 ± 4.2 2.5 ± 3.8 P0.769;700 = .930

Knee flexion (deg) 123.8 ± 10.7 122.5 ± 8.3 81.27 ± 10.1 83.94 ± 11.69 81.3 ± 10.1 83.9 ± 11.7 78.7 ± 6.9 79.5 ± 10.4 84.2 ± 9.1 84.0 ± 10.5 91.1 ± 9.7 90.8 ± 10.4 108.4 ± 12.9 109.5 ± 11.8 116.6 ± 13.2 114.3 ± 9.3 119.6 ± 11.8 119.1 ± 9.3 P0.410;695 = .958

Quadriceps lag (deg)

Girth (mm)

OKS

10.2 ± 4.2 11.1 ± 6.6 12.0 ± 6.2 14.2 ± 6.4 14.2 ± 5.8 12.8 ± 5.1 11.1 ± 5.8 8.7 ± 5.3 4.3 ± 3.9 4.2 ± 4.3 3.1 ± 4.3 2.7 ± 2.3 1.9 ± 3.0 2.1 ± 2.8 2.0 ± 2.5 1.3 ± 2.3 P1.122;515 = .346

443.4 ± 43.8 441.5 ± 37.6 465.6 ± 41.9 462.5 ± 37.5 463.0 ± 48.4 461.6 ± 42.0 463.1 ± 45.0 463.9 ± 37.7 438.1 ± 42.6 442.0 ± 39.4 427.3 ± 40.8 428.2 ± 43.7 422.4 ± 41.0 427.3 ± 38.5 427.5 ± 40.8 425.8 ± 40.0 P0.075;621 = 1.000

43.8 ± 9.3 39.1 ± 9.4 38.0 ± 8.5 35.2 ± 8.8 26.7 ± 7.0 27.5 ± 8.9 21.2 ± 8.8 24.7 ± 9.4 19.5 ± 6.9 22.1 ± 8.9 18.8 ± 8.9 21.0 ± 8.3 P1.934;426 = .074

TUG (transformed)

0.36 ± 0.02 0.04 ± 0.01 0.10 ± 0.03 0.10 ± 0.03 0.11 ± 0.03 0.10 ± 0.03 0.13 ± 0.02 0.10 ± 0.03 0.11 ± 0.02 0.10 ± 0.03 p1.354;411 = .241

The Medial Parapatellar Versus Subvastus Approach in TKA  Bourke et al

Primary Outcome

353.e1