We are using a non image based navigation system on a routine basis for unicompartmental knee replacement (UKR). We prospectively studied 60 patients who underwent navigated minimally invasive UKR for primary medial osteoarthritis at our hospital between October 2005 and October 2006. We established a navigated control group of 60 patients who underwent conventional implantation of a UKA at our hospital between April 2004 and September 2005. There were 42 male and 78 female patients with a mean age of 65 years (range, 44–87 years). There were no differences in all preoperative parameters between the two groups.

The accuracy of implant positioning was determined using predischarge standard anteroposterior and lateral radiographs. The following angles were measured: femorotibial angle, coronal and sagittal orientation of the femoral component, coronal and sagittal orientation of the tibial component. When the measured angle was in the expected range, one point was given. The accuracy was defined as the sum of the points given for each angle, with a maximum of five points (all items fulfilled) and a minimum of 0 point (no item fulfilled). Our primary criterion was the radiographic accuracy index on the postoperative radiograph evaluation. All other items were studied as secondary criteria.

The mean accuracy index was similar in the two groups: 4.1 ± 0.8 in the study group and 4.2 ± 1.2 in the control group. 36 patients (60%) in the control group and 37 patients (62%) in the study group had the maximum accuracy index of five points. All measured angles were similar in the two groups. There were no differences between the percentages of patients in the two groups achieving the desired implant positions. Mean operating time was similar in the two groups. There were no intraoperative complications in either group. The groups had similar major postoperative complication rates during hospital stay (3% for both).

The used navigation system is based on an anatomic and kinematic analysis of the knee joint during the implantation. The modification of the existing software for minimal invasive approach has been successful. It enhances the quality of implantation of the prosthetic components and avoids the inconvenient of a smaller incision with potential less optimal visualization of the intra-articular reference points. However, all centers observed a significant learning curve of the procedure, with a significant additional operative time during the first implantations. The postoperative rehabilitation was actually easier and faster, despite the additional percutaneous fixation of the navigation device. This system has the potential to allow the combination of the high accuracy of a navigation system and the low invasiveness of a small skin incision and joint opening.

Introduction: Total knee arthroplasty in obese patients remains a challenge to most surgeons. Surgical complication rates as well as perioperative morbidity are higher than total knee arthroplasty in the nonobese. The purpose of this paper is to review our experience with total knee arthroplasty in superobese patients (BMI> 50).

Methods: From 1998–2005, 84 patients underwent 148 knee arthroplasties. Sixty-four patients underwent simultaneous bilateral total knee arthroplasties and 20 patients underwent unilateral knee arthroplasties. They were compared with similar group of nonobese patients who underwent knee arthroplasties during the same time period. All patients received combined regional and general anesthesia.

Results: Mean follow-up was 3.8 years (2–7). Knee society scores improved by 36 points in the superobese (pre-op 47 to 83 post-op) and by 45 points in the non-obese (pre-op 47 to 93 post-op) (p< .05). There was a greater incidence of complications in the superobese group, namely superficial wound infections and deep vein thrombosis. There was late loosening in three tibial components and instability in two patients that required revision in the superobese group. No reoperations in the nonobese group.

Conclusion: Although total knee arthroplasty may be safely performed in the superobese, it may be complicated by infection, loosening, instability, and lower knee scores.

Revision total knee replacement (TKR) is a challenging procedure, especially because most of the standard bony and ligamentous landmarks used during primary TKR are lost due to the index implantation. However, as for primary TKR, restoration of the joint line, adequate limb axis correction and ligamentous stability are considered critical for the short- and long- term outcome of revision TKR. There is no available data about the range of tolerable leg alignment after revision TKR. However, it is logical to assume that the same range than after primary TKR might be accepted, that is ± 3° off the neutral alignment. One might also assume that the conventional instruments, which rely on visual or anatomical alignments or intra- or extra-medullary rods, are associated with significant higher variation of the leg axis correction, especially in cases with significant bone loss which prevents to control the exact location of the usual, relevant landmarks. Navigation system might address this issue.

We used an image-free system (ORTHOPILOT TM, AESCULAP, FRG) for routine implantation of primary TKR. The standard software was used for revision TKR. Registration of anatomic and cinematic data was performed with the index implant left in place. The components were then removed. New bone cuts as necessary were performed under the control of the navigation system. The size of the implants and their thickness was chosen after simulation of the residual laxities, and ligament balance was adapted to the simulation results. The system did not allow navigation for intra-medullary stem extensions and any bone filling which may have been required. This technique was used for 54 patients. The accuracy of implantation was assessed by measuring following angles on the post-operative long-leg radiographs: mechanical femoro-tibial angle (normal = 0°, varus deformation was described with a positive angle); coronal orientation of the femoral component in comparison to the mechanical femoral axis (normal = 90°, varus deformation was described with an angle < 90°); coronal orientation of the tibial component in comparison to the mechanical tibial axis (normal = 90°, varus deformation was described with an angle < 90°); sagittal orientation of the tibial component in comparison to the proximal posterior tibial cortex (normal = 90°, flexion deformation was described with angle < 90°).

Individual analysis was performed as follows: one point was given for each fulfilled item, giving a maximal accuracy note of 4 points. Prosthesis implantation was considered as satisfactory when the accuracy note was 4 (all fulfilled items). The rate of globally satisfactory implanted prostheses and the rate of prostheses implanted within the desired range for each criterion were recorded.

Limb alignment was restored in 88%. The coronal orientation of the femoral component was acceptable in 92% of the cases. The coronal orientation of the tibial component was acceptable in 89% of the cases. The sagittal orientation of the tibial component was acceptable in 87% of the cases. Overall, 78% of the implants were oriented satisfactorily for the four criteria.

The navigation system enables reaching the implantation objectives for implant position and ligament balance in the large majority of cases, with a rate similar to that obtained for primary TKA. The navigation system is a useful aid for these often difficult operations, where the visual information is often misleading.

Navigation systems are able to measure very accurately the movement of bones, and consequently the knee laxity, which is a movement of the tibia under the femur. These systems might help measuring the knee laxity during the implantation of a total (TKR) or a unicompartmental (UKR) knee replacement.

20 patients operated on for TKR (13 cases) or UKR (7 cases) because of primary varus osteoarthritis have been analyzed. Pre-operative examination involved varus and valgus stress X-rays at 0 and 90° of knee flexion. The intra-operative medial and lateral laxity was measured with the navigation system at the beginning of the procedure and after prosthetic implantation. Varus and valgus stress X-rays were repeated after 6 weeks. X-ray and navigated measurements before and after knee replacement were compared with a paired Wilcoxon test at a 0.05 level of significance.

The mean pre-operative medial laxity in extension was 2.3° (SD 2.3°). The mean pre-operative lateral laxity in extension was 5.6° (SD 5.1°). The mean pre-operative medial laxity in flexion was 2.2° (SD 1.9°). The mean pre-operative lateral laxity in flexion was 6.7° (SD 6.0°). The mean intra-operative medial laxity in extension at the beginning of the procedure was 3.6° (SD 1.7°). The mean intra-operative lateral laxity in extension at the beginning of the procedure was 3.0° (SD 1.3°). The mean intra-operative medial laxity in flexion at the beginning of the procedure was 1.9° (SD 2.6°). The mean intra-operative lateral laxity in flexion at the beginning of the procedure was 3.5° (SD 2.7°). The mean intra-operative medial laxity in extension after implantation was 2.1° (SD 0.9°). The mean intra-operative lateral laxity in extension after implantation was 1.9° (SD 1.1°). The mean intra-operative medial laxity in flexion after implantation was 1.9° (SD 2.5°). The mean intra-operative lateral laxity in flexion after implantation was 3.0° (SD 2.8°). The mean post-operative medial laxity in extension was 2.4° (SD 1.1°). The mean post-operative lateral laxity in extension was 2.0° (SD 1.7°). The mean post-operative medial laxity in flexion was 4.4° (SD 3.3°). The mean post-operative lateral laxity in flexion was 4.7° (SD 3.2°).

There was a significant difference between navigated and radiographic measurements for the pre-operative medial laxity in extension (mean = 1.4° – p = 0.005), the pre-operative lateral laxity in extension (mean = 2.6° – p = 0.01), the pre-operative lateral laxity in flexion (mean = 3.3° – p = 0.005). There was no significant difference between navigated and radiographic measurements for the pre-operative medial laxity in flexion (mean = 0.3° – p = 0.63). There was a significant difference between navigated and radiographic measurements for the postoperative medial laxity in flexion (mean = 2.5° – p = 0.004). There was no significant difference between navigated and radiographic measurements for the postoperative medial laxity in extension (mean = 0.3° – p = 0.30), the post-operative lateral laxity in extension (mean = 0.2° – p = 0.76), the post-operative lateral laxity in flexion (mean = 1.7° – p = 0.06). These differences were less than 2 degrees in most of the cases, and then considered as clinically irrelevant.

The navigation system used allowed measuring the medial and lateral laxity before and after TKR. This measurement was significantly different from the radiographic measurement by stress X-rays for pre-operative laxity, but not statistically different from the radiographic measurement by stress X-rays for post-operative laxity. The differences were mostly considered as clinically irrelevant. The navigated measurement of the knee laxity can be considered as accurate. The navigated measurement is valuable information for balancing the knee during TKR. The reproducibility of this balancing might be improved due to a more objective assessment.

Revision TKR is a challenging procedure, especially because most of the standard bony and ligamentous landmarks are lost due to the primary implantation. However, as for primary TKR, restoration of the joint line, adequate limb axis correction and ligamentous stability are considered critical for the short- and long-term outcome of revision TKR. There is no available data about the range of tolerable leg alignment after revision TKR. However, it is logical to assume that the same range than after primary TKR might be accepted, that is ± 3° off the neutral alignment. One might also assume that the conventional instruments, which rely on visual or anatomical alignments or intra- or extramedullary rods, are associated with significant higher variation of the leg axis correction.

We used an image-free system (ORTHOPILOT TM, AESCULAP, FRG) for routine implantation of primary TKA. The standard software was used for revision TKA. Registration of anatomic and kinematic data was performed with the index implant left in place. The components were then removed. New bone cuts as necessary were performed under the control of the navigation system. The size of the implants and their thickness was chosen after simulation of the residual laxities, and ligament balance was adapted to the simulation results. The system did not allow navigation for centromedullary stem extension and any bone filling which may have been required. This technique was used for 54 patients. The accuracy of implantation was assessed by measuring the limb alignment and orientation of the implants on the post-operative radiographs.

Limb alignment was restored in 88%. The coronal orientation of the femoral component was acceptable in 92% of the cases. The coronal orientation of the tibial component was acceptable in 89% of the cases. The sagittal orientation of the tibial component was acceptable in 87% of the cases. Overall, 78% of the implants were oriented satisfactorily for the five criteria.

The navigation system enables reaching the implantation objectives for implant position and ligament balance in the large majority of cases, with a rate similar to that obtained for primary TKA. The navigation system is a useful aid for these often difficult operations, where the visual information is often misleading. The navigation system used enables facilitated revision TKA.

Anterior cruciate ligament (ACL) reconstruction allows overall good results, but there is still a significant rate of failure. It is well accepted that the main reason for ACL reconstruction failure is a misplacement of tibial or femoral tunnels. Conventional techniques rely mainly on surgical skill for intra-operative tunnel placement. It has been demonstrated that, even by experienced surgeons, there was a significant variation in the accuracy of tunnel placement with conventional techniques. Navigation systems might enhance the accuracy of ACL replacement.

10 cadaver knees with intact soft-tissue and without any intra-articular abnormalities were studied. We used a non image based navigation system (OrthoPilot ®, Aesculap, Tuttlingen, FRG). Localizers were fixed on bicortical screws on the distal femur and on the proximal tibia. Both kinematic and anatomic registration of the knee joint were performed by moving the knee joint in flexion-extension and palpating relevant intra- and extra-articular landmarks with a navigated stylus. The most anterior, posterior, medial and lateral point of both tibial and femoral attachment of the ACL were marked with metallic pins. The navigated stylus was positioned on these points, and the system recorded its position in comparison to the bone contours. Subsequently, we performed conventional plain AP and lateral X-rays and a CT-scan, and measured the position of the pins in comparison to the bone contours. Finally, all measurements were made again with a caliper after disarticulating the knee joint. We calculated the center of the footprint as the mid-point between the four pins of both tibial and femoral attachment for each measurement technique. All measurements were expressed as percentages of the bone size to compensate for the different sizes.

There were no significant difference in the paired measurements of the location of the ACL footprints on both femur and tibia between anatomic, radiographic, CT-scan and navigated measurements. There was a significant correlation between the paired measurements of the location of the ACL footprints on both femur and tibia with either measurement techniques.

Anatomic measurement is the gold standard experimental technique for the positioning of the ACL foot-print, and CT-scan measurement is currently the gold standard technique in clinical situation. According to this reference, the position of ACL attachments on the tibia and on the femur can be accurately defined by the navigation system. Intra-operative measurement of the location of the bone tunnels during ACL replacement with this navigation system should be accurate as well.

Unicompartmental knee replacement (UKR) is accepted as a valuable treatment for isolated medial knee osteoarthritis. Minimal invasive implantation might be associated with an earlier hospital discharge and a faster rehabilitation. However these techniques might decrease the accuracy of implantation, and it seems logical to combine minimal invasive techniques with navigation systems to address this issue.

The authors are using a non image based navigation system (ORTHOPILOT ^™, AESCULAP, FRG) on a routine basis for UKR. We prospectively studied 60 patients who underwent navigated minimally invasive UKR for primary medial osteoarthritis at our hospital between October 2005 and October 2006. We established a navigated control group of 60 patients who underwent conventional implantation of a UKA at our hospital between April 2004 and September 2005. There were 42 male and 78 female patients with a mean age of 65 years (range, 44–87 years). There were no differences in all preoperative parameters between the two groups.

The accuracy of implant positioning was determined using predischarge standard anteroposterior and lateral radiographs. The following angles were measured: femorotibial angle, coronal and sagittal orientation of the femoral component, coronal and sagittal orientation of the tibial component. When the measured angle was in the expected range, one point was given. The accuracy was defined as the sum of the points given for each angle, with a maximum of five points (all items fulfilled) and a minimum of 0 point (no item fulfilled). Our primary criterion was the radiographic accuracy index on the postoperative radiograph evaluation. All other items were studied as secondary criteria.

The mean accuracy index was similar in the two groups: 4.1 ± 0.8 in the study group and 4.2 ± 1.2 in the control group. 36 patients (60%) in the control group and 37 patients (62%) in the study group had the maximum accuracy index of five points. All measured angles were similar in the two groups. There were no differences between the percentages of patients in the two groups achieving the desired implant positions. Mean operating time was similar in the two groups. There were no intraoperative complications in either group. The groups had similar major postoperative complication rates during hospital stay (3% for both).

The used navigation system is based on an anatomic and kinematic analysis of the knee joint during the implantation. The modification of the existing software for minimal invasive approach has been successful. It enhances the quality of implantation of the prosthetic components and avoids the inconvenient of a smaller incision with potential less optimal visualization of the intra-articular reference points. However, all centers observed a significant learning curve of the procedure, with a significant additional operative time during the first implantations. The postoperative rehabilitation was actually easier and faster, despite the additional percutaneous fixation of the navigation device. This system has the potential to allow the combination of the high accuracy of a navigation system and the low invasiveness of a small skin incision and joint opening.

Accuracy of implantation is an accepted prognostic factor for the long term survival of total knee replacement (TKR). The use of navigation demonstrated a significant higher accuracy of implant orientation in comparison to conventional methods. However, these systems are often thought to be technically demanding, to increase operating time and to involve a long learning curve. We performed a prospective, multicenter study to compare the accuracy of implantation of a TKR measured on post-operative X-rays in experienced and less experienced centers.

All centers used the same navigation system (Ortho-Pilot ®, Asculap, Tuttlingen, FRG): 4 had already a significant experience with it (group A – 182 cases), 9 centers were considered as beginners with less than 10 cases performed prior to the study (group B – 221 cases). Accuracy of implantation was measured on post-operative antero-posterior and lateral long leg X-rays with five items: mechanical femoro-tibial angle, coronal orientation of the femoral component, sagittal orientation of the femoral component, coronal orientation of the tibial component, sagittal orientation of the tibial component.

When the measured angle was in the expected range, one point was given. The accuracy note was defined as the sum of all points given for each patient, with a maximum of 5 points (all items fulfilled) and a minimum of 0 point (no item fulfilled). The mean accuracy note was compared in the two groups by a Student t-test at a 0.05 level of significance. Power of the study was 0.80.

There were no significant differences in pre-operative parameters between the two groups, except for the clinical KSS. The mean operative time was significantly longer in group B than in group A (110 minutes vs 90 minutes, p=0.01). However this difference occurred mainly during the first twenty cases in the beginner centres where we observed a clear tendency to achieve the same operative time as the experienced centres at the end of the study. The mean accuracy note was 4.3 ± 0.8 (range, 1 to 5) in the control group and 4.3 ± 0.9 (range, 1 to 5) in the study group (p > 0.05). The power of the study to detect a 0.25 point difference in the post-operative accuracy note was retrospectively calculated to be 0.80. There were no significant differences between the two groups for all individual radiographic items.

This study is, to our knowledge, the first one which investigates the learning curve of navigated TKR The used navigation system allowed a very accurate implantation of a TKR in both experienced and less experienced centers. There was no detectable learning curve with respect to accuracy of TKR implantation, clinical outcome and complication rate. The duration of the learning curve when considering the operating time was 30 cases.

Purpose of the study: Navigation systems have proven efficacy for the implantation of unicompartmental knee prostheses. Minimally invasive methods, which limit access to non-operated compartments, might compromise system accuracy.

Material and methods: A standard navigation software was used for kinematic acquisition of the lower limb and to acquire anatomic landmarks for both femorotibial compartments. A modified version of the navigation software designed for minimally invasive surgery replaed palpation of the anatomic landmarks of the non-operated compartment by a computation method based on other data. Three groups of patients were analyzed. Group 1 included 64 patients who underwent minimally invasive surgery for implantation of a medial unicompartmental prosthesis. Group B included 60 patients selected randomly among 140 cases of medial unicompartmental prosthesis patients treated with the standard navigation technique. Group C included 30 patients selected randomly among 180 patients who underwent total knee arthroplasty with the standard navigation system. The quality of the implantation was assessed on the postoperative ap and lateral views by comparing five criteria describing the desired prosthetic alignment. The number of criteria describing correct alignment was noted for each patient, thus yielding a quality score from 0 to 5. ANOVA was used to compare the mean scores of the three groups using Boneffini-Dunn correction at the 5% risk level.

Results: The mean quality score was 3.5±1.2 for group A, 4.5±0.8 for group B and 4.2±1.0 for grup C (p< 0.001). Ther was no significant difference between groups B and C (p=0.24). The quality score was significantly lower in group A (A versus B: p=0.015; A versus C: p< 0.001).

Discussion: The minimally invasive approach is proposed to enable more rapid functional recovery after implantation of a unicompartmental knee prosthesis. The long-term outcome however depends on the quality of the implantation. The quality of the implantation with a minimally invasive method should thus be equivalent to that achieved with the standard method. Conventional minimally invasive methods are more difficult. Navigation could be expected to overcome this difficulty without sacrificing implantation quality. However, the version used here did no enable an implantation equal to the quality achieved with the standard navigation system.

Conclusion: The standard navigation system for the conventional access remains the gold standard for implantation quality. Changes resulting from a less invasive approach should be validated before routine use.

Purpose of the study: Navigation systems have proven their capacity to improve the quuality of total knee arthroplasty (TKA) implantation. The navigation system coud also be used to record knee kinematics intraoperatively.

Material and methods: Twenty TKA implantations were studied. The series included six males and 14 females, mean age 71 years (range 63–78 years). All underwent surgery for overall osteoarthritis. A TKA with a mobile plateau was implanted with preservation of the posterior cruciate ligament. The OrthoPilot® imageless navigation system (Aesculap, Tuttlingen, German) was used. The software was modified to enable recording the relative movement of the femur in relation to the tibia during flexion-extension movements. Infrared locators were fixed on the lower part of the femur and the proximal part of the tibia. After kinematic and anatomic acquisition of conventional navigation data, the kinematic recordings were made during passive flexion-extension before performing any procedures on the bones. The system recorded femur rotation in relation to the tibia in the frontal plane (varus-valgus), in the sagittal plane (flexion-extension), and in the horizontal plane (internal-external rotation) as well as anteroposterior translation of the femur on the tibia. The prosthesis was implanted using the conventional navigation technique. After implantation, the same kinematic recordings were repeated. Each measurement was taken in duplicate to study reproducibility in the same patient. Pre- and postoperative kinematic recordings in the same patient were compared to obtain objective evidence of changes induced by prosthesis implantation. The pre- and postoperative results were compared with those reported to date in the literature.

Results: The recorded kinematic curves, both before and after TKA implantation, were coherent with generally accepted values, particularly for rotation and antero-posterior translation. Paradoxical kinematic recordings were noted after implantation. There was no significant difference between the two recordings in the same patient.

Discussion: The software enables a reliable study of knee kinematics before and after TKA implantation. This could be useful to test new prosthetic solutions, but also to choose for a given patient, the best kinematic compromise. It would be interesting to compare these results with data on in vivo kinematic recordings made in the same patients.

Conclusion: Intraoperative kinematic analysis is a research tool at the present time, but could be useful to improve the quality of TKA implantations.

Purpose of the study: Navigation systems have proven efficacy for implantation of total knee arthroplasty (TKA). Navigations have been accused of being complex, requiring a long learning curve. We compared the results obtained with the same navigation system in centers with experienced operators and centers with new operators.

Material and methods: Thirteen European centers participated in this prospective consecutive study. Inclusion criteria was indication for a TKA using a gliding prostheis with preservation of the posterior cruciate ligament. Four experienced cents(group A) with a mean experience of four years, and nine new centers (group B) with no prior experience participated in the study. The study concerned 403 TKA (182 in group A and 221 in group B). The main indications were primarily lateralized osteoarthritis. The navigation system was an imageless system based on intaoperative kinematic anatomic and kinematic analysis. A mobile plateau prosthesis was inserted. The following items were compared between the two groups: overall operative time and its variation over time, postoperative HKA, orientation of the femoral and tibial components in the ap and lateral views, complications and revisions.

Results: No significant difference was observed between the two groups for the preoperative items so comparison between the groups was licit. Correction of the frontal mechanical axis was satisfactory in 90% of patients in group A and 88% in group B (p> 0.05). There was no difference between the groups in quality of implantation for each prosthetic element on the ap and lateral views. There was no difference for rate of complications or reoperations. Longer operative time in group B disappeared after 15 implantations.

Discussion: The results from centers using navigation systems for prosthetic implantations shows that the performance in centers starting use is the same as in experienced centers. The only difference is an operative time slightly longer for the first 15 cases.

Introduction: The accuracy of implantation is an accepted prognostic factor for the long term survival of a unicompartmental knee prosthesis (UKP). Minimal invasive technique is recommended for faster post-operative recovery. We developed an adaptation of a non image based system for either conventional or minimal invasive UKP implantation. We hypothesized that the used non image based navigation system will allow to place a UKP in the same position for both conventional and minimal-invasive approach.

Methods: 20 patients were operated on with this experimental minimal invasive navigated technique (group A) and compared to a group of 20 cases operated with the conventional navigated technique (group B), matched to the study group according to age, gender andseverity of the coronal deformation. Coronal mechanical femorotibial angle and coronal and sagittal orientation of the components were measured on post-operative antero-posterior and lateral long leg X-rays. The rate of satisfactory implanted prostheses was compared in both groups with a Chi-square test with a 0.05 limit of significance.

There was no significant difference in the pre-operative data between both groups. The post-operative coronal group A and 17 cases in group B. The prosthesis was optimally implanted in 17 cases in group A and 18 cases in group B. No difference was statistically significant.

Discussion-Conclusion: The used navigation system allowed a very precise implantation of a UKP for both conventional and minimal invasive navigated technique.

Purpose: Implantation quality is an important prognostic factor for long-term outcome of unicompartmental knee prostheses. Minimally invasive techniques allow more rapid rehabilitation but at the price of potentially diminished implantation quality. Navigation systems have been developed to overcome this problem.

Material and methods: We analysed a preliminary series of 20 patients (group A) whose unicompartmental medial femorotibial prosthesis (Search(r), Aesculap, Tuttlingen, Germany) was implanted with the CT-free Orthopilot(r) system (Aesculap, Tuttlingen, Germany). This system uses intra-operative kinematic and anatomic analysis to define the mechanical axes of the femur and tibia in space. The femoral and tibial cut lines are aligned on these axes. This series was compared with a retrospective historical series (group B) of 60 knees with the same prostheses implanted with the same navigation system but with a conventional approach requiring patellofemoral subluxation. Implantation quality was measured using the following angles: AP mechanical femorotibial angle, orientation of the tibial and femoral prostheses (AP and lateral), vertical level of the prosthetic joint space in relation to the preserved joint space.

Results: The AP mechanical femorotibial angle was in the desired range in 16 knees in group A (80%) and in 48 in group B (80%). The femoral component exhibited optimal position in 18 knees in group A (90%) and in 54 in group B (90%). The tibial component exhibited optimal position in 17 knees in group A (85%) and in 53 in group B (88%). Thirteen prostheses in group A (65%) and 37 in group B (62%) were implanted optimally using the studied criteria. The length of the incision varied from 7 to 10 cm in group A. There was no significant difference.

Discussion: This navigation system allows very precise implantation of the medial unicompartmental knee prosthesis, both with the conventional technique and the minimally invasive technique. Use of the minimally invasive technique does not decrease the radiographic quality of the implantation in comparison with the conventional navigation technique. This technique could become the gold standard for implantation of unicompartmental knee prostheses.

Purpose: We reviewed retrospectively 349 cases of infected total hip arthroplasty treated by prosthesis replacement. The surgical strategy, 127 single-stage procedures and 222 two-stage procedures, was determined by the surgeon on a case by case basis.

Material and methods: At least one positive sample during the clinical history was required for inclusion in the series. Results of all bacteriological samples collected pre- and intra-operatively were noted. Samples were considered reliable if obtained from a deep site (puncture, biopsy, intraoperative specimen) and non-reliable if obtained from any other site. We studied the agreement between preoperative and intraoperative samples, taking the intraoperative samples as the reference, in order to determine the effect of complete preoperative knowledge of the causal germ on the outcome of infection treatment at last follow-up.

Results: For single-stage replacement procedures, preoperative samples were reliable in 74 cases (58%) and non reliable in seven (6%); they were sterile or absent in 46 cases (36%). Intra-operative samples were positive in 103 cases (81%). Agreement between the preoperative and intraoperative samples was observed in 48 cases (38%). The rate of success was not different if the surgeon had or did not have reliable knowledge of the causal germ(s) preoperatively: successful treatment in 66 cases (89%) with knowledge and successful in 46 cases (87%) without knowledge. For two-stage procedures, preoperative samples were reliable in 155 cases (70%) and non-reliable in 15 (7%); they were sterile or absent in 52 cases (23%). Intraoperative samples were positive in 178 cases (80%). Agreement between preoperative and intraoperative samples was observed in 107 cases (48%). The rate of success was not different if the surgeon had or did not have reliable knowledge of the causal germ(s) preoperatively: successful treatment in 133 cases (86%) with knowledge and successful treatment in 56 cases (84%) without knowledge.

Conclusion: Reliable preoperative knowledge of the causal germ(s) did not affect the rate of success for single-stage or two-stage total hip arthroplasty replacement procedures. These findings do not corroborate the notion that it is absolutely necessary to recognise the germ(s) causing the infection before undertaking a single-stage replacement procedure.

Background: The aim of introduction of navigation in knee arthroplasty was to further contribute to precision of endoprosthetic alignment.

Methods and material: A multicentre comparative study was conducted including 821 patients. The SEARCH knee system was used throughout the series. 555 TKA’s were implanted with the use of a navigation system (OrthoPilot) and 266 cases were operated using manual instrumentation. Alignment was radiographically evaluated at the three months follow-up with respect to mechanical axis and femoral and tibial axes using one-leg stance x-rays and standardized lateral radiographs.

Results: The summarized results of the series are shown in the table below. The chi-square test was applied for the statistical analysis.

Conclusions: Endoprosthetic alignment using the navigation system was superior to manual implantation technique on the average with respect to all parameters. Results were more consistent on the tibial side. The navigation system proved to be reliable. The overall results justify the further use and development of navigation tools in knee arthroplasty.

Purpose: The design of the contact surfaces of total knee prostheses is a recognised factor affecting polyethylene wear and thus prosthesis survival. Flat-on-flat prostheses have a limited surface area of contact and are thought to favour polyethylene wear. They are not currently recommended for implantation. Nevertheless, several series have reported similar survival with other more congruent prostheses. We studied a series followed for eight years.

Material and methods: We implanted 223 flat-on-flat design total knee prostheses between 1992 and 1996 (Search®, Aesculap, Chaumont). All patients were followed prospectively and seen at regular intervals for physical examination and x-rays. We noted any intervention for implant revision and recorded time to any such procedures as well as the underlying cause. Kaplan-Meier survival curves were plotted taking revision for any cause other than infection as the endpoint.

Results: Ninety-four percent of the patients were reexamined or questioned by phone for this study conducted during 2001. Six percent of the patients were lost to follow-up after a mean 24 months. Seventy-four percent of the prostheses were still in situ at the time of this study at a mean 78 months follow-up. Ten percent of the patients died with their initial implant in place at a mean 50 months. Ten percent of the patients underwent revision surgery at a mean 37 months, half of them for infection and one quarter for a mechanical cause. The overall rate of revision at eight years was 11%; The rate of revision, infection excluded, at eight years was 6%.

Discussion: Survival of this prosthesis in non-infected patients is similar to that of other more congruent implants. This study confirms earlier clinical findings.The undesirable effect of the linear contact surfaces is proven in the laboratory but must not be considered to be automatically transferable to the clinical level. Polyethylene wear is a multifactorial phenomenon which cannot be reduced to a simple question of prosthesis design.

Purpose: According to the Ottawa rules, x-rays are not needed after knee trauma unless one or more of the following clinical criteria are present: age over 55 years, pain at palpation of the head of the fibula, pain at palpation of the anterior aspect of the patella, impossible knee flexion beyond 90°, inability to walk four steps immediately after trauma and at the emergency consultation. We conducted a prospective study in a consecutive series of patients to check the validity of this rule in daily practice.

Materials and methods: From December 2001 to January 2002, we included all patients consulting in an emergency situation for recent trauma involving only the knee joint. We excluded patients aged less than ten years, wounds without trauma, trauma more than two days before consultation, and patients with a history of trauma involving the same knee. An emergency physical exam was performed in all cases with identification of the study criteria. Standard x-rays (AP and lateral view in the supine position) were obtained for all patients. The patients and the x-rays were seen later by a senior orthopaedic surgeon and a senior radiologist who noted the presence of fracture requiring specific therapeutic management. The sensitivity, specificity and positive and negative predictive values of the Ottawa rule were determined for search for fracture.

Results: One hundred thirty-eight patients met the inclusion criteria during the study period. The sensitivity and negative predictive value of the Ottawa rule were 100%; the specificity was 36%, and the positive predictive value was 25%. Nineteen fractures (14%) requiring specific therapeutic management were identified: all patients had at least one positive sign. Seventy-six patients (55%) without fracture had at least one positive sign. Forty-three patients (31%) without fracture did not have any positive sign. The x-rays were not contributive for these patients.

Discussion and conclusion: This study demonstrated the validity of the Ottawa rule in the clinical setting of our practice. With widespread use of this rule, approximately one-third of the x-rays performed for recent trauma involving the knee alone could be avoided.

Purpose: Knowledge of the radiological axes in the normal lower limb is important for correction and reconstruction surgery. Classically, the femorotibial mechanical axis presents a zero angle on the anteroposterior view, with 3° femoral valgus being compensated by an equivalent tibial varus. Reference data have however been established with questionable methodology because they have been obtained with small selected samples.

Material and methods: We obtained teleradiograms of the lower limbs in 100 healthy volunteers free of any disease of the lower limbs and selected randomly among patients undergoing surgery for trauma or degenerative lesions of the upper limb. The following angles were measured by the same senior surgeon: mechanical femorotibial angle, orientation of the femoral condylar complex in relation to the mechanical axis of the femur, angle between the mechanial axis and the anatomic axis of the femur, orientation of the tibial plateaux in relation to the mechanical axis of the tibia.

Results: Sixty-nine men and 31 woman, mean age 39 years (range 17 – 62 years) participated in this study. The mean mechanical femorotibial angle was 179° (SD 3°, median 179°, range 168°–185°). The mean orientation of the femoral condylar complex in relation to the femoral mechanical axis was 91° (SD 2°, median 91°, range 86°–98°); 17 subjects had the classical value of 93°. The mean angle between the mechanical and anatomic axis of the femur was 6° (SD 1°, median 6°, range 3°–9°); 29 subjects had the classical value of 7°. The mean orientation of the tibial plateaux in relation to the mechanical axis of the tibia was 88° (SD 2°, median 88°, range 82°–84°); 14 subjects had the classical value of 87°.

Discussion and conclusion: The values considered to be normal in the literature only included 15–20% of the subjects in this study. Although there could be a theoretical selection bias in this series, it can be assumed that there is a wide dispersion of “normal” values around the means. The pertinence of this dispersion in clinical practice remains to be established. The question of individualising reconstruction or prosthetic procedures is raised.

Purpose: The purpose of this study was to conduct a multicentric comparison of total knee arthroplasty using the conventional technique versus digitalized navigation.

Material and methods: A prospective comparative study was conducted in five centres in 821 patients using the same implant (Search®, Aesculap, Chaumont): 555 procedures with the Orthopilot® navigation system (Aesculap, tutligen, group 1) and 266 conventional procedures, group 2). Radiographic results were analysed by an independent investigator who examined telemetric images obtained three months after surgery.

Results: The mechanical femorotibial axis was within desired limits (3° frontal deformation) in 88.6% of the knees in group 1 and in 72.2% of the knees in group 2 (p< 0.001). The rate of unacceptable implantations (> 5° deviation) was 2.5% in group 1 and 9.8% in group 2).

Frontal orientation of the femoral component was satisfactory in 89.4% of the knees in group 1 and in 77.1% in group 2. Sagittal orientation of the femoral component was satisfactory in 75.5% of the knees in group 1 and in 70.7% of the knees in group 2. Frontal orientation of the tibial component was satisfactory in 91.9% of the knees in group 1 and in 83.5% of the knees in group 2. The sagittal orientation of the tibial piece was satisfactory in 81.3% of the knees in group 1 and in 69.9% of the knees in group 2. Optimal implantation, considering all criteria studied, was achieved in 275 patients (49.5%) in group 1 and in 82 patients (30.8%) in group 2 (p< à.001). Ther was no difference in results between centres.

Discussion: Computer-assisted navigation facilitated prosthesis implantation with the desired orientation in comparison with manual instrumentation. The number of unacceptable implantations was significantly lower. After a short learning curve, the reliability of this system has proven very satisfactory, facilitating its use since this study.

INTRODUCTION

Nowadays, longevity of total knee arthroplasties is very acceptable. Survivorship analyses demonstrate a success in a range of 80% to more than 95% over a period of more than ten years (1–4). However, long-term results largely depend, amongst other factors, on restoration of physiological alignment of the lower limb (5–11). Jeffery et al. (12) reported a three percent loosening rate over eight years when knees were correctly aligned whereas insufficient alignment lead to prosthetic loosening in 24 percent. Rand and Coventry (13) found a 90 percent survivorship rate at ten years when the mechanical axis was aligned in a range from nought to four degrees of valgus. Valgus position of more than four degrees or varus alignment resulted in only 71 percent and 73 percent of survivorship respectively.

Recently, computer aided instrumentation systems (14,15) became available and preliminary results of small series were reported (16–17).

The purpose of this study was to assess the accuracy of computer integrated instrumentation for knee alignment.

MATERIAL AND METHOD

The OrthoPilot^® represents a computer controlled image supported alignment system. A 3-D Optotrak™ camera localizes infra-red diodes fixed to rigid bodies within the surgical field. Thereby a spatial coordinate reference system is provided. The localizer is linked to a UNIX work station which performs the operative protocol using a graphical interface and a foot pedal. The rigid bodies are fixed to the bones by bicortical screws. An intraoperative kinematic analysis and various additional landmarks lead to definition of the centres of hip, ankle and knee joint and sizing of endoprosthetic components. With the use of LED-equipped alignment instruments the femoral and tibial resection planes are determined.

The OrthoPilot^® navigation system is not dependant on CT data and no additional preoperative planning is therefore necessary.

A prospective comparative multicentre study in five institutions, four in Germany and one in France, was carried out. 821 patients with primary tricompartimental knee arthroplasty using the SEARCH LC knee (B|Braun AESCULAP) were included in the study. The OrthoPilot^® Navigation system was used in 555 cases and 266 knees were implanted with the use of conventional instrumentation. At the three months follow-up alignment was assessed using standardized one leg stance radiographs with regard to the mechanical axis and the femoral and tibial angels in the coronal plane. For the lateral femoral and tibial angels standard lateral x-rays were used. Prosthetic alignment was verified by an independent observer.

RESULTS

The radiographically assessed results were subdivided into three groups. An error of ± one degree in the radiographical measurements and small deviations caused by the play of surgical instruments have to be considered. With respect to the femoral and tibial angels in the ap and lateral view the group of very good clinical results was, therefore, defined in the range between ninty degrees and ± two degrees. Deviations of three and four degrees from the optimum were classified as being clinically acceptable. Aberrations of more than four degrees were classified as outliers. When measuring the mechanical axis deviations from fully precise femoral and tibial angels may add up. For this reason zero degrees ± three degrees were rated as a very good result, deviations of four to five degrees were considered to be acceptable and alignment beyond five degrees from the optimum was classified as an unsatisfactory result.

Mechanical axis:

35. 2% of the navigated cases were aligned at exactly zero degrees. This was achieved in only 24. 4% of the manual cases. 88. 6% of cases using navigation and 72. 2% in the manual group showed zero degrees and varus or valgus angles of up to three degrees. 8. 9% and 18. 1% of cases respectively showed deviations of four or five degrees of valgus or varus alignment representing an acceptable clinical result. In only 2. 5% of the navigation group aberrations of more than five degrees occurred. The rate of dissatisfying results was 9. 8% in the manual group.

Femoral axis (coronal plane):

In the navigation group 48. 1% of cases showed an alignment at exactly 90 degrees which was the case in only 33. 5% of the control group. Altogether, in 89. 4% of the navigated cases a very good result was observed. In the conventionally instrumented cases only 77. 1% very good results were found. There were 1. 6% outliers beyond the limits of four degrees in the navigation group in comparison to 4. 9% amongst the control cases.

Femoral axis (sagittal plane):

Very good results with up to two degrees of deviation from a ninety degree position were obtained in 75. 5% of navigated cases and 70. 7% of manual cases. 37. 3% and 34. 6% respectively showed an ideal alignment of exactly ninety degrees. Unsatisfactory results were observed in 9. 5% of the navigated cases and 9. 4% of the manual cases.

Tibial axis (coronal plane):

58. 7% of the computer assisted and 40. 6% of the reference cases were exactly aligned at rectangles. All in all, in 91. 9% navigated and only 83. 5% manual cases a very good result was obtained. Only 1. 1% outliers had to be observed in the navigation group whereas 3. 4% unsatisfactory results were registered with manual technique.

Tibial axis (sagittal plane):

44. 3% of the navigated cases and only 26. 7% of cases in the control group were aligned perpendicular to the dorsal tibial cortex, thus showing no posterior slope. Altogether, 81. 3% could be classified as very good clinical results in the computer assisted group. The corresponding rate of the manual group was 69. 9%. Equivalent values of 8. 6% in the navigation group and 8. 3% in the reference group were registered beyond the limits of four degrees deviation.

The additional operation time for the use of the navigation system is calculated between eight and ten minutes after having passed through the learning curve.

CONCLUSIONS

Knee navigation facilitates proper alignment of endoprosthetic components and with the use of the Ortho-Pilot^® system results are clearly more favourable in comparison to conventional instrumentation technique. In addition, the data obtained from literature demonstrate that the use of this navigation system contributes to reducing outliers in number. With the learning curve the OrthoPilot^® alignment system proved to gain in reliability.

Deviations from perfect alignment are still difficult to be classified into surgical or technical deficiencies.

Many technical and software improvements which were introduced in the meantime will, in addition, contribute to reliability and time saving.

Comparative studies with different navigation systems are not yet available. They might allow an even more profound insight into the possibilities and advantages or disadvantages of computer assisted knee alignment.

LITERATURE

(1) Knutson K, Lindstrand A, Lidgren L. Survival of knee arthroplasties, a nation-wide multicenter investigation of 8000 cases. J Bone Joint Surg. 1986; 68B: 795-803

(2) Scuderi GR, Insall JN, Windsor RE, Moran MC. Survivorship of cemented knee replacement. J Bone Joint Surg. 1989; 798-409

(3) Nafei A, Kristensen O, Knudson HM, Hvid I, Jensen J. Survivorship analysis of cemented total condylar knee arthoplasty. J Arthoplasty 11, 1996;07-10

(4) Ranawat CS, Flynn WF, Saddler S, Hansraj KH, Maynhard MJ. Long-term results of total condylar knee arthroplasty. A 15-years survivorship study. Clin Orthop 1993; 286:94-102

(5) Lotke PA, Ecker ML. Influence of positioning of prosthesis in total knee replacement. J Bone Joint Surg 1977;59-A:77-79

(6) Hood RW, Vanni M, Insall JN. The correction of knee alignment in 225 consecutive total condylar knee replacements. Clin Orthop 1981;160:94-105

(7) Bargren JH, Blaha JD, Freeman MAR. Alignment in total knee arthroplasty. Clin Orthop 1983;173:178-183.

(8) Hvid I, Nielsen S. Total condylar knee arthroplasty. Acta Orthop Scand 1984;55:160-165

(9) Tew M, Waugh W. Tibial-femoral alignment and the results of knee replacement. J Bone Joint Surg 1985;67-B:551-556

(10) Jonsson B, Astrom J. Alignment and long-term clinical results of a semi-constrained knee prosthesis. Clin Orthop 1988;226:124-128

(11) Ritter MA, Faris PM, Keating EM, Meding JB. Postoperative alignment of total knee replacement its effect on survival. Clin Orthop 1994;299:153-156

(12) Jeffery RS, Morris RW, Denham RA. Coronal alignment after total knee replacement. J Bone Joint Surg 1991;73-B:709-714

(13) Rand JA, Coventry MB. Ten-year evaluation of geometric total knee arthroplasty. 1988;232:168-173

(14) Leitner F, Picard F, Minfelde R, Schulz HJ, Clinquin P, Saragaglia D. Computer assisted knee surgical total replacement. In: CVRMed-MRCAS. Troccaz J, Grimson E, Mösges R (Eds). 1997; 630-638, Springer

(15) Delp SL, Stulberg SD, Davies BL, Picard F, Leitner F. Computer assisted knee replacement. Clin Orthop 1998; 354:49-56

(16) Picard F, Saragaglia D, Montbarbon E, Chaussard C, Leitner F, Raoult O. Computer assisted knee arthroplasty - preliminary clinical results with the Ortho-Pilot System. Abstract, 4th International CAOS Symposium, Davos, Switzerland, 1999

(17) Miehlke RK, Clemens U, Jens J-H, Kershally S. Navigation in der Knieendoprothetik - vorläufige klinische Erfahrungen und prospektiv vergleichende Studie gegenüber konventioneller Implantationstechnik, Z Orthop 2001; 139: 109-116

Purpose: Treatment of infection in patients with an unstable bone is based on removal of implants, bone resection, reconstruction, and external fixation. We report a retrospective series of 11 patients who developed post-traumatic osteitis of the tibia on an unstable bone who were treated by removal of all implants, cleaning, antibiotics, and internal fixation using a centromedullary locked nail.

Material and methods: The series included seven men and four women, mean age 32.4 years (16–56). Initially, there were two closed fractures and nine open fractures (Gustilo II: 4, IIIA: 1; IIIB: 4) treated by external fixation in six cases, centromedullary locked nailing in four and plate fixation in one. Bacteriology results were available for all deep surgical samples. The initial implants were removed in all cases, followed by debridement sparing soft tissue, and reaming of the bone. Adapted antibiotics were prolonged for three months. Refixation using a centromedullary locked nail was performed at the first revision time in two cases and later after cleaning in nine (mean delay 28 days, range 2–53 days). Two cases required a flap for cover.

Results: There were two failures: one due to recurrent infection with a different germ, the other due to necrosis of a latissimus dorsi flap followed by amputation. There were nine successes with bone healing in all cases (first intention in eight and after complementary bone graft in one) and no recurrent infection at the current mean follow-up of 2.6 years.

Discussion: These eleven cases have a common feature of no extensive bone necrosis or major bone defect. Bone resection was basically related to reaming with a minimalistic approach for soft tissue debridement. Reliable bacteriological examinations, effective antibiotic therapy, and prolonged and rapid skin cover are essential elements for success.

Conclusion: This experience is limited but does demonstrate that locked centromedullary nailing can be successful for the treatment of long bone infections on unstable bones, considering that this could be the ideal fixation method.

Purpose: Centromedullary nailing with reaming is a recognised treatment for open leg fractures with a well-measured risk of postoperative infection. The development nailing procedures without reaming might reduce this risk.

Material and methods: We performed a Medline search using the following key words: nailing, tibia, open fracture, infection. To be retained for analysis, articles had to evaluate infectious risk of nailing with or without reaming, in clinical trials or experimental studies, with precise diagnostic criteria. Clinical articles retained were classed in three categories by decreasing value of their methodology: prospective randomised comparative studies, case-control studies, comparative observation studies, simple observation studies. Only comparative experimental and prospective comparative randomised studies were considered to be pertinent.

Results: Five articles met the predefined quality inclusion criteria and were retained for analysis: three experimental studies and two clinical trials. The experimental studies by Melcher (1995 and 1996) demonstrated a significant increase in infection rate and bacterial counts after nailing with reaming; there were two confounding factors however, steel or titanium nail and full or hollow nail which also had a significant effect on the rate of infection. The experimental work by Curtis (1995) did not find any difference in incidence and severity of infection between nailings with and without reaming. The two prospective comparative randomised clinical trials by Keating (1007) and Finkemeier (2000) included a total of 132 cases. the risk of infection was 8% after nailing with reaming and 7% after nailing without reaming (NS). The relative risk of infection after nailing with reaming was 1.02-fol greater than that without reaming (NS).

Discussion, conclusion: There is experimental evidence that would tend to prove that the risk of infection is lower after nailing with reaming, but it is insufficient to explain the mechanism of this lower rate. Inversely, although the clinical observation series tend to confirm these results, the two methodologically valid prospective comparative randomised studies did not find any difference. To date, there is no objective evidence ruling out the usefulness of nailing with reaming because of higher infection risk in open leg fractures.

Purpose: The quality of implantation of single-compartment knee prostheses is a recognised prognostic factor. Acceptable reproducibility can be achieved with traditional instrumentations, although the rate of error can be significant. Computer-assisted implantation might improve results. Most of the currently proposed techniques require supplementary preoperative imaging or implantation of metallic material for guidance. The Orthopilot® system is a purely peroperative system and could thus provide better cost-effectiveness.

Material and methods: We implanted 30 single-compartment knee prostheses using the Orthopilot® computerised system (Aesculap, Chaumont, Group A) and compared the radiographic quality of the implant on telemetric AP and lateral views with those from a control group of 30 single-compartment prostheses implanted with a traditional instrumentation with a femoral centromedullary aiming device (group B). All patients underwent surgery for primary degeneration and were operated on by the same surgeon using the same implant (Search®, Aesculap, Chaumont). The control group was selected among a consecutive series of 250 implants to match the study group for age, gender, importance of the degeneration and frontal femorotibial mechanical angle.

Results: The mechanical femorotibial angle was within desired limits (177±3°) in 26 patients in group A and in 20 patients in group B. Frontal orientation of the femoral component was within desired limits (90±2°) in 27 patients in group A and in 19 in group B (p< 0.05). Frontal orientation of the tibial piece was within desired limits (90±2°) in 27 patients in group A and in 19 patients in group B (p < 0.02). The original level of the joint line was reconstructed with a 2 mm margin in 30 patients in group A and in 24 patients in group B (p < 0.05). Eighteen patients in group A and four patients in group B had optimal implantation for all criteria studied (p < 0.001). There were no system-related complications.

Discussion, conclusion: Computer-assisted implantation is more reliable and more reproducible than traditional instrumentation for the implantation of a single-compartment knee prosthesis. Follow-up results with these prostheses may be better. Systematic preoperative imaging, or preoperative implantation of metallic guide pins is not necessary with this system. The system appears to offer a better cost-effectiveness.

Purpose: The biepicondylar axis of the femur is considered by many authors as a reliable reference axis for flexion-extension of the knee and to establish desirable orientation of the femoral component of a total knee arthroplasty. We studied the reproducibility of axis measurments made using an automatic digital acquisition system (OrthoPilot®, Aesculap, Chaumont, France). The system localises anatomic points in space from information obtained with a palpation probe carrying an infrared diode.

Material and methods: A consecutive series of 20 total knee arthroplasties (Search®, Aesculap, Chaumont, France) implanted by two senior surgeons on the same surgical team were studied. The mechanical axis of the femur was calculated prior to the study using kinematic acquisition of the position of the centres of rotation of the hip and the knee. The frontal reference plane was then defined from the most posterior point on the femoral condyles palpated with the probe as the plane containing the mechanical axis of the femur and parallel to the posterior bicondylar line. The apex of the two femoral epicondyles was obtained by direct palpation with the probe. A second plane passing through the apex of the epicondyles and parallel to the mechanical axis of the femur was thus defined. Three acquisitions were made for the same patient by each of the two surgeons without changing the posterior bicondylar reference plane. The angle between the frontal plane of reference and the biepicondylar plane was calculated directly by the software for each acquisition. The variability of the three measurements taken by each operator and between the two operators was studied with the Wilcoxon test for paired series and with Spearman’s coefficient of correlation.

Results: Mean intraobserver variability for the orientation of the biepicondylar axis was 4° for the two operators, with a maximum of 11° for the first operator and 9° for the second, the directions being random. The mean interobserver variability for this orientation was 4° with a maximum of 14°, again at random. All differences were statistically significant.

Discussion, conclusion: Measurements of the biepicondylar axis exhibit high intra- and interobserver variability, probably due to the anatomic conditions; the apex of the epicondyles is a blunt surface difficult to identify with precision. Use of this axis to determine the rotation of the femoral component of a total knee arthroplasty is thus an element of wide variability with measurement inaccuracy of a mean ± 5° but with a maximum that can reach 10°. The question remains to determine whether this uncertainty is tolerable or whether more precision is required.