Abstract

Abstract

We have previously reported on the improved all-cause revision and improved revision for instability risk in lipped liner THAs using the NJR dataset. These findings corroborate studies from the Australian (AOANJRR) and New Zealand (NZOA) joint registries. The optimal orientation of the lip in THAs utilising a lipped liner remains unclear to many surgeons. The aim of this study was to identify impingement-free optimal liner orientations whilst considering femoral stem version, cup inclination and cup version.

A cementless THA kinematic model was developed using a 20 degree XLPE liner. Physiological ROM and provocative dislocation manoeuvre analyses were performed. A total of 9 cup positions were analysed (inclination 30–40–50 degrees, anteversion 5-15-25 degrees) and combined with 3 stem positions (anteversion 0-15-30 degrees) and 5 lip orientations (right hip 11 to 7 o'clock).

Some lip orientation/component position combinations lead to impingement within the physiological ROM range. Using a lipped liner increases the femoral head travel distance prior to dislocation when impingement occurs in the plane of the lip. In THAs with a cup inclination of 30 and 40 degrees, inferior lip orientations (7–8 o'clock for a right hip) performed best. Superior lip orientation performed best with a cup inclination of 50 degrees. Femoral stem version has a significant effect on the range of movement prior to impingement and hence the preferred lip orientation.

The optimal orientation of the lip in lipped liner THA is dependent on the position of both the acetabular and femoral components. In the common component orientation combination of stem anteversion 15, cup inclination 40 and cup anteversion 15, the optimal lip orientation was postero-inferiorly (8 o'clock for a right hip). Preventing impingement during physiological ROM is possible with appropriate lip liner orientation.

Our previous work presented at BHS revealed a reduced risk of revision for all reasons in THAs using lipped (asymmetric) liners. Some audience members felt that this finding may be due to unaccounted confounders and the hip surgery community remains sceptical.

A fully adjusted Cox model was built after exploratory Kaplan-Meier analyses. The following surgical approaches were included in the analysis: Posterior, Hardinge/anterolateral, Other. The variables included in the final Cox model included: Gender, liner asymmetry, age, head composition, stem fixation method, head diameter, indication for implantation and surgical approach. An additional analysis of the 3 most commonly used polyethylene liners with both a flat and asymmetric version was performed.

In the fully adjusted Cox model, the use of a flat liner was associated with increased risk of revision for instability (HR: 1.79, 95%CI: 1.52–2.10) and increased risk of revision for all reasons (HR 1.195, 95%CI: 1.104 – 1.293) when compared to THAs utilising flat liners. This finding was upheld in the product specific analysis.

When utilising flat liners, the Hardinge/anterolateral approach was associated with a reduced risk of revision for instability compared to the posterior approach (HR 0.56, 95%CI: 0.446 to 0.706). When an asymmetric liner was used, there was no significant difference in the risk of revision for instability between the Hardinge/anterolateral approach and the posterior approach (HR 0.838, 95%CI: 0.633 – 1.110).

The use of polyethylene-based bearings gives the surgeon the option to use asymmetric liners. The reduced risk of revision in THAs with asymmetric liners was seen in the analysis of the complete dataset and in the product specific analysis. Our results demonstrate that the posterior approach, when used with asymmetric XLPE liners, is not associated with a higher risk of revision for instability as historical data suggested.

Computer aided Total Hip Arthroplasty (THA) surgery is known to improve implantation precision, but clinical trials have failed to demonstrate an improvement in survivorship or patient reported outcome measures (PROMs). Our aim was to compare the risk of revision, PROMs and satisfaction rates between computer guided and THA implanted without computer guidance.

We used the National Joint Registry dataset and linked PROMs data. Our sample included THAs implanted for osteoarthritis using cementless acetabular components from a single manufacturer (cementless and hybrid). An additional analysis was performed limiting the sample size to THAs using cementless stems (fully cementless). The primary endpoint was revision (of any component) for any reason. Kaplan Meier survivorship analysis and an adjusted Cox Proportional Hazards model were used.

41683 non computer guided, and 871 (2%) computer guided cases were included in our cementless and hybrid analysis. 943 revisions were recorded in the non-guided and 7 in the computer guided group (adjusted Log-rank test, p= 0.028). Cumulative revision rate at 10 years was 3.88% (95%CI: 3.59 – 4.18) and 1.06% (95%CI: 0.45 – 2.76) respectively. Cox Proportional Hazards adjusted HR: 0.45 (95%CI: 0.21 – 0.96, p=0.038). In the fully cementless group, cumulative revision rate at 10 years was 3.99% (95%CI: 3.62 – 4.38) and 1.20% (95%CI: 0.52 – 3.12) respectively. Cox Proportional Hazards adjusted HR: 0.47 (95%CI: 0.22 – 1.01, p=0.053). There was no statistically significant difference in the 6-month Oxford Hip Score, EQ-5D, EQ-VAS and success rates. Patient Satisfaction (single-item satisfaction outcome measure) was improved in the computer guided group but this finding was limited by a reduced number of responses.

In this single manufacturer acetabular component analysis, the use of computer guided surgery was associated with a significant reduction in the early risk of revision. Causality cannot be inferred in view of the observational nature of the study, and further database and prospective studies are recommended to validate these findings.

The use of routine sampling for histological analysis during revision hip replacement has been standard practice in our unit for many years. It is used to identify the presence of inflammatory processes that may represent peri-prosthetic infection.

This study follows up on a smaller study in the same unit in 2019 where an initial 152 cases were scrutinised. In this follow up study we examined 1,361 consecutive patients over a 16-year period whom had undergone revision hip replacement in a tertiary orthopaedic centre for any reason excluding primary bone tumour or malignant metastasis. All patients had tissue sampling for histopathological analysis performed by consultant histopathologists with a specialist interest in musculoskeletal pathology. The presence of bacteria in greater than 50% of samples sent for microbiological analysis in each patient was used as the gold standard diagnostic comparator for infection. This was then compared with the histology report for each patient.

After excluding 219 patients with incomplete data and 1 sample rejection, 1,141 cases were examined. Microbiology confirmed infection in 132 cases (prevalence of infection 11.04%) and histopathology analysis suggested infection in 171 cases. Only 64 cases with confirmed infection in more than 50% of microbiology samples had concurrent diagnosis of infection on histological analysis (5.60% of total; PPV 51.20%). Furthermore, microbiology analysis confirmed infection in 62 cases where histological analysis failed to identify infection (5.43% of total; False negative rate 49.21%). Overall, histopathology analysis was seen to have a good specificity of 93.99% but poor sensitivity of 50.79%.

We believe that this is the largest series in the literature and is somewhat unique in that all histology analysis was performed by consultant histopathologists with specialist interest in musculoskeletal pathology. Based on the costs incurred by this additional investigation our experience does not support routine sampling for histological analysis in revision hip arthroplasty. This is a substantial paradigm shift from current practice among revision arthroplasty surgeons in the United Kingdom but would equate to a substantial cost saving.

Abstract

In light of recent regulatory initiatives, medical devices now require additional clinical evidence to prove their safety and efficacy. At the same time, patients' own assessment of their devices' function and performance has gained in importance. The collection of these data allows for a more comprehensive picture of clinical outcomes and complications following total knee arthroplasty (TKA).

These trends have led researchers to search for new methods of acquiring, interpreting and disseminating patient-reported outcome measurements (PROMs). The current study assesses the feasibility of a digital platform for collecting PROMs that was recently adapted for TKA patients. It sought to determine patient engagement, survey completion rates, and satisfaction with this platform.

Eighty-two patients (mean age, 63.7 years, 59% females) scheduled for TKA were enrolled from one US and six UK sites between January 12, 2018 and April 30, 2018. Patients were supplied with a mobile application (app) that collects a variety of PROMs, including four domains based on the Patient-Reported Outcome Information System (PROMIS™): physical function, depression, pain interference and pain behavior. The platform electronically administers questionnaires using computer-adaptive tests (CATs), which reduce the burden on patients by tailoring follow-up questions to account for their previous answers. Satisfaction with the app was assessed in subset of patients who evaluated its ease-of-use (n=45), likelihood that they would recommend it to family/friends (n=35), and whether they successfully used the information it provided during their recovery (n=31). These scores were taken on a 1 to 10 (worst to best) scale.

Patients demonstrated regular engagement with the platform, with 73% using the app at least once a week. Weekly engagement remained high throughout the seven-week post-operative period (Figure 1). There was a 69% completion rate of all PROMIS™ CAT surveys during the study. The four PROMIS™ CAT domains had similar survey completion rates (Figure 2). The subset of patients queried regarding their satisfaction with the app gave it favorable mean scores for ease-of-use (8.8), likelihood to recommend to a family member or friend (8.1), and their success at using its information to improve their recovery (7.4).

Initial results support this digital platform's potential for successfully and efficiently collecting large volumes of PROMs. Patients reported high levels of engagement and satisfaction.

For any figures or tables, please contact authors directly.

Aim

To assess the effect of different polyethylene modifications on Total Hip Replacement survival.

Aim

To assess the effect of the bearing surface and head size on the survival of total hip replacements with modern bearing surface combinations.

The use of routine sampling for histological analysis during revision hip replacement has been standard practice in our unit for many years. It is used to assess for the presence of inflammatory processes that may represent peri-prosthetic infection.

Our study examines 152 consecutive patients who underwent revision hip replacement in our centre for all reasons, excluding malignant neoplasm or metastasis. We reviewed the cases from a prospectively collated database, comparing microbiology results with histology results. Both microscopic and macroscopic analysis by specialist musculoskeletal histopathologist was included in our study.

We found 17 (11.2%) patients had cultured bacteria from intra-operative samples. Eight patients (5.3%) had histological findings interpreted as infection. Only one patient who had macroscopic and microscopic histology findings suggestive of infection also had culture results that identified a pathogen. Furthermore, the macroscopic analyses by the histopathologist suggested infection in nine patients. Only one patient with positive culture in greater than 2 samples had histological features of infection.

Of the 4 patients who were found to have 3 or more samples where an organism was identified only one had histological features of infection. This represents 25% sensitivity when using histology to analyse samples for infection. Of the 8 patients who had both macroscopic and microscopic features of infection only 1 patients cultured bacteria in more than 3 samples (PPV 12.5%).

Our experience does not support the routine sampling for histology in revision hip replacement. We suggest it is only beneficial in cases where infection is suspected or where a multi-procedure, staged revision is performed and the surgeon is planning return to theatre for the final stage. This is a substantial paradigm shift from the current practice among revision arthroplasty surgeons in the United Kingdom but will equate to a substantial cost saving.

Polymethylmethacetate (PMMA) is a bone cement used in over 725,000 primary hip arthroplasties in 2018. Cement integrity is affected by external factors, including temperature, mixing technique and moisture uptake, which can influence cement microstructure. Changes in the cement microstructure may ultimately threaten the survivorship of the implant.

The introduction of enhanced recovery and various local anaesthetic infiltration techniques have been adopted in an attempt to facilitate early mobilisation and reduce length of stay. Our study aims to investigate if the mechanical properties of PMMA are altered with exposure to Ropivacaine LA.

Cements were cured in three separate states (air, serum and serum with LA) and the mechanical properties tested at 24 hours and 28 days. Using Refobacin bone cement provided by ZimmerBIOMET, cylindrical molds (12×6mm) were constructed with a split-mold. The LA used was 2mg/ml Ropivacaine hydrochloride solution. Using pilot data, this study was powered to 80% and a sample size of 10 per group (n=60) was calculated.

Cement samples were subjected to compressive loading using a universal testing apparatus (Zwick/Roell). Yield-strength and modulus values were extracted from the respective stress versus strain curves. Significant differences were determined by one-way anova for each time point, and Bonferroni post-hoc testing to determine significance between actual groups.

At 24-hours there were no significant differences in strength or modulus between groups. At 28-day strength and modulus increased in all groups. Compared to the air group, both serum and LA groups show a significant decrease in compressive strength. The modulus for the LA group is significantly less stiff compared to the air group.

The results suggest that the initial exposure to LA has a significant impact on the physical properties of the PMMA. We propose increased awareness of the potential effects this may have on the longevity and survivorship of cemented implants.

Surgical site infections (SSIs) are associated with significant consequences in orthopaedic surgery, where their presence can lead to ultimate revision of the implant. Furthermore, infections and impaired wound healing can prolong length of hospital stay following orthopaedic surgery, which can place additional financial burdens on healthcare systems. The current analysis was conducted to determine whether the use of the PICO single-use negative pressure wound therapy (sNPWT) system after orthopaedic surgery reduced the incidence of SSIs and length of hospital stay compared with using conventional dressings. A systematic literature review (SLR) was performed using the PubMed, Embase and Cochrane Library databases. English-language studies comparing PICO sNPWT to conventional dressings published from 2011 to August 2018 with ≥10 patients in each treatment arm were included. Reference lists of included studies were searched for further relevant studies. Meta-analyses were performed using a fixed effect (I² < 50%) or random effects model (I² ≥ 50%). The SLR identified 6,197 studies, of which 5 relevant studies (607 patients) were included. The odds of an SSI were reduced by 57% (odds ratio [OR]: 0.43; 95% confidence interval [CI]: 0.21–0.86; p = 0.02) and there was consistency between studies (I² = 0%). Three studies reported on length of hospital stay. The mean difference between patient groups indicated that PICO sNPWT was associated with a 1-day reduction in hospital stay (mean difference [MD]: −0.99; 95% CI: −1.32 to −0.65; p < 0.00001) and there was again consistency between studies (I² = 0%). These results suggest that the use of PICO sNPWT system after closed surgical incisions can reduce the incidence of SSIs and shorten the duration of hospital stay when used in orthopaedic patient populations.

Optimum component orientation in hip arthroplasty is vital in an effort to avoid dislocation and excessive wear. Computer navigation in hip arthroplasty surgery has the potential to improve accuracy in component placement. However, it has been slow to gain widespread acceptance. One of the major concerns surgeons have is the difficulty in registering pelvic landmarks.

We used a retrospective series of 200 pelvic CT scans to validate a new methodology to construct the anterior pelvic plane, using anatomical landmarks that are easily palpated with the patient positioned and draped in the lateral decubitus position. Analysis of the scans was also made in an effort to stimulate the inaccuracies of obtaining the anterior pelvic plane through soft tissue.

When comparing the new registration methodology to the anterior pelvic plane, the error in acetabular component inclination was 0.69° (SD 2.96) and anteversion was 1.17° (SD 3.53). This compares favourably to the error in acetabular component inclination of −0.92° (SD 0.26) and anteversion of −5.24° (SD 2.09) when the anterior pelvic plane is registered through soft tissue. The data also shows that using the new registration method in more than 99.6% of cases the acetabular placement is within the safe zone as described by Lewinnek.

This study appears to show that through the identification of anatomical constants we are able to construct the anterior pelvic plane from anatomical landmarks that are easily palpable in the lateral decubitus position during hip arthroplasty. These landmarks also appear to be more accurate in obese patients than registering the anterior pelvic plane.

Implant alignment in knee arthroplasty has been identified as critical factor for a successful outcome. Human error during the registration process for imageless computer navigation knee arthroplasty directly affects component alignment. This cadaveric study aims to define the error in the registration of the landmarks and the resulting error in component alignment.

Five fresh frozen cadaveric limbs including the hemipelvis were used for the study. Five surgeons performed the registration process via a medial parapatellar approach five times. In order to identify the gold standard point, the soft tissues were stripped and the registration was repeated by the senior author. Errors are presented as mm or degrees from the gold standard registration.

The error range in the registration of the femoral centre in the coronal plane was 6.5mm laterally to 5.0mm medially (mean: −0.1, SD: 2.7). This resulted in a mechanical axis error of 5.2 degrees valgus to 2.9 degrees varus (mean: 0.1, SD: 1.1). In the sagittal plane this error was between −1.8 degrees (extension) and 2.7 degrees (flexion). The error in the calculation of the tibial mechanical axis ranged from −1.0 (valgus) to 2.3 (varus) degrees in the coronal plane and −3.2 degrees of extension to 1.3 degrees of flexion. Finally the error in calculating the transepicondylar axis was −11.2 to 6.3 degrees of internal rotation (mean: −3.2, SD: 3.9).

The error in the registration process of the anatomical landmarks can result in significant malalignment of the components. The error range for the mechanical axis of the femur alone can exceed the 3 degree margin that has been previously been associated with implant longevity. The technique during the registration process is of paramount importance for image free computer navigation. Future research should be directed towards simplifying this process and minimizing the effect of human error.

Mal-positioning of the acetabular component is associated with increased dislocation rate, increased wear and component impingement. Navigation provides real time feedback to the surgeon and allows the accurate position of implants. Compared to conventional techniques of total hip replacement; use of the imageless navigation system has shown to improve accuracy of implant positioning.

When impacting uncemented acetabular components under navigation, there is often a deviation from the planned abduction and anteversion measurement due to deflection of the implant in the reamed cavity. Although there exists the ability to navigate the reaming of the acetabular cavity; this is not widely performed. The ability to ream the acetabular cavity in the exact orientation of the planned acetabular component may provide some theoretical advantage on the final acetabular position. The purpose of this study was to compare the effect of navigated Vs free hand acetabulum reaming on achieving the planned orientation of acetabular component.

In a retrospective study we reviewed two groups of patients who underwent computer navigated placement of the acetabular component with reference to the anterior pelvic plane. We used an imageless computer navigation system for all cases (Brainlab, Munich). All procedures were performed by single surgeon (ETD) through a standard posterior approach. The patients were divided into two groups depending on the availability of the navigated reamer. In the first group (n = 57), acetabulum reaming was done under navigation and in the second group (n = 37) a non-navigated reamer was used. The acetabular cavity was reamed “line to line” or under reamed by 1 or 2mm. Intra-operative acetabular abduction and anteversion angles were planned using navigation at the discretion of the surgeon. Results of planned acetabular abduction and anteversion angles were compared with intra-operative verification using the navigation system.

In the navigated reamer group, the mean error from the planned to verified abduction angle was 1.7 degrees (SD 2.1 degrees) and in the non-navigated reamer group the mean error was 2 degrees (SD 2.6 degrees). In the navigated reamer group, the mean error from the planned to verified anteversion angle was 0.5 degrees (SD 2.8), and in the non-navigated reamer group the mean error was 0.1 degrees (SD 1.6). There was no statistically significant difference in the mean error between the navigated and non-navigated reaming groups for abduction angle (p = 0.54) or anteversion angle (p = 0.24). There was no statistical difference between the mean acetabular component size in the navigated (mean 53mm) and non-navigated (53mm) reamer groups (p = 0.8). There was no statistical difference in the mean difference in reamer size and the acetabular component size in the navigated (0.8mm) and non-navigated reamer groups (0.8mm, p = 0.52).

This study appears to show that performing reaming of the acetabular cavity under navigation does not improve the final orientation of the acetabular component when compared to using conventional non-navigated reamers. However, this study only considered the abduction and anteversion orientation of the component. The move to a range of movement or kinematic orientation of the acetabular component in hip arthroplasty requires control over the off-set of the acetabular component which may be more easily achieved when the reaming is performed under navigation. This study used a conventional posterior approach rather than a minimal incision technique, where the use of navigated reaming may also provide some theoretical advantage when visibility is limited. Further study is required in these two areas.

There appears to be a slightly higher standard deviation for the anteversion measurement in the navigated reamer group when compared to the non navigated reamer group, although this is not significant. It is difficult to account for this as it appears to be opposite of what one would predict. One explanation for this may come in the difference in the angled geometry of the navigated reamer when compared to the straight non navigated reamer. The angled reamer can be more difficult to control forming a cavity in the correct orientation but with the possibility for the cavity to not been perfectly hemispherical.

When using navigation to insert the acetabular component in a planned abduction and anteversion position during hip arthroplasty through a standard incision, navigating the reaming of the acetabular component does not appear to provide any advantage over the use of conventional non-navigated reamers in the final acetabular orientation.

Aims

Leg length inequality following total hip replacement remains common. In an effort to reduce this occurrence, surgeons undertake pre-operative templating and use various forms of intra-operative measurements, including computer navigation. This study aims to delineate which measurement technique is most appropriate for measuring leg length inequality from a pelvic radiograph.

We aimed to identify whether patients in lower socioeconomic groups had worse function prior to total knee arthroplasty and to establish whether these patients had worse post-operative outcome following total knee arthroplasty.

Data were obtained from the Kinemax outcome study, a prospective observational study of 974 patients undergoing primary total knee arthroplasty for osteoarthritis. The study was undertaken in thirteen centres, four in the United States, six in the United Kingdom, two in Australia and one in Canada. Pre-operative data were collected within six weeks of surgery and patients were followed for two years post-operatively. Pre-operative details of the patient's demographics, socioeconomic status (education and income), height, weight and co-morbid conditions were obtained. The WOMAC and SF-36 scores were also obtained. Multivariate regression was utilised to analyse the association between socioeconomic status and the patient's pre-operative scores and post-operative outcome. During the analysis, we were able to control for variables that have previously been shown to effect pre-operative scores and post-operative outcome.

Patients with a lower income had a significantly worse pre-operative WOMAC pain (p=0.021) and function score (p=0.039) than those with higher incomes. However, income did not have a significant impact on outcome except for WOMAC Pain at 12-months (p=0.014). At all the other post-operative assessment times, there was no correlation between income and WOMAC Pain and WOMAC Function. Level of education did not correlate with pre-operative scores or with outcome at any time during follow-up. This study demonstrates that across all four countries, patients with lower incomes appear to have a greater need for total knee arthroplasty. However, level of income and educational status did not appear to affect the final outcome following total knee arthroplasty. Patients with lower incomes appear able to compensate for their worse pre-operative score and obtain similar outcomes post-operatively.

In perfroming hip resurfacing arthroplasty, concern has been expressed as to the proximity of the femoral neurovascular bundle during the anterior capsulotomy and the risk of damage during this maneuver. We therefore aimed to identify the proximity of the femoral nerve, artery and vein during an anterior capsulotomy done during a hip resurfacing procedure using the posterior approach.

A standard posterior approach was performed in 5 fresh frozen cadavic limbs. An anterior incision was then used to measure the distance of the femoral neurovascular structures to the anterior capsule. Measurements from the most posterior aspect of the vessels and nerves to the most anterior aspect of the anterior capsule were taken prior to hip dislocation. The femoral head was then dislocated, and measurements were made with the hip in both flexion and extension. In a separate group of eleven patients that underwent routine MR imaging of the hip, measurements were taken to assess the proximity of the anterior joint capsule to the femoral neurovascular bundle, by a specialist musculoskeletal radiologist who had no prior knowledge of the results obtained during the cadaveric dissection

All 5 cadaveric limbs were utilised. 3 were male and 2 were female. The average age was 72.4 years (range 56–84). The patients whom underwent routine MR imaging incorporated 6 males and 5 females with a mean age of 43.7 years (age range 18–64 years). There was no significant difference between the mean distances to the nerve (p=0.21), artery (p=0.21) or vein (p=0.65) between the MR and cadaveric groups. Prior to dislocation the femoral artery and vein were closest to the anterior capsule (mean distance of 21mm) and the femoral nerve was the furthest away (mean distance 25mm). Following dislocation there was a significant increase (25mm to 31mm) in mean distance to the femoral nerve when the superior capsule was cut with the hip in a flexed position (p=0.01) and to the femoral artery in flexion (increase mean distance from 21mm to 35mm) (p< 0.0001) and in extension(increase mean distance from 21mm to 31mm) (p=0.005). When the inferior capsule was cut, there was a significant increase (25mm to 31mm) in mean distance to the femoral nerve and femoral artery when the hip was dislocated and the capsule cut with the hip in flexion (increase mean distance from 21mm to 27mm) (p=0.019) and in extension(increase mean distance from 21mm to 28mm) (p=0.015).

This study suggests that the neurovascular structures are relatively well protected during an anterior capsulotomy performed during hip resurfacing. The procedure may be safer if the capsulotomy is performed with the hip dislocated and the hip in a flexed position while cutting the antero-superior aspect and in an extended position while cutting the antero-inferior aspect.

The purpose of this study was to assess the accuracy of clinical assessment compared to imageless computer navigation in determining the amount of fixed flexion during knee arthroplasty.

In fourteen cadaver knees, a medial para-patella approach was performed and the navigation anatomy registration process performed. The knees were held in various degrees of flexion with two crossed pins. The degree of flexion was first recorded on the computer and then on lateral radiographs. The cadaver knees were draped as for a knee arthroplasty and nine examiners (three arthroplasty surgeons, three fellows, and three residents) were asked to clinically assess the amount of fixed flexion. Three examiners repeated the process one week later.

The mean error from the radiograph in the navigation group was 2.18 degrees (95%CI 2.18+/−0.917) compared to 5.57 degrees (CI 5.57+/− 0.715) in the observer group. The navigation was more consistent with a range of error of only 5.5 degrees (standard deviation 1.59). The observers had a range of error of 18.5 degrees (S.D. = 4.06). When analysing the observers’ error with respect to flexion (+) and extension (−), they tended to under-estimate the amount of knee flexion (median error=−4) whereas the navigation was more evenly distributed (median error=0). The highest correlation was found between navigation and the radiograph r=0.96. The highest observer correlation with the radiograph was a consultant surgeon (r=0.91) and the worst was from a resident (r=0.74). The intra-class correlation coefficient was 0.88 for the three surgeons who repeated the measurements; their mean error was 3.5 degrees with a range of fifteen degrees.

The use of computer navigation appears to be more accurate in assessing the degree of knee flexion, with a reduced range of error when compared to clinical assessment. It is therefore less likely to leave the patient with residual fixed flexion after knee arthroplasty.