Metal-on-metal hip resurfacing prostheses are a relatively recent intervention for relieving the symptoms of common musculoskeletal diseases such as osteoarthritis. While some short term clinical studies have offered positive results, in a minority of cases there is a recognised issue of femoral fracture, which commonly occurs in the first few months following the operation. This problem has been explained by a surgeon's learning curve and notching of the femur but, to date, studies of explanted early fracture components have been limited.

Tribological analysis was carried out on fourteen retrieved femoral components of which twelve were revised after femoral fracture and two for avascular necrosis (AVN). Eight samples were Durom (Zimmer, Indiana, USA) devices and six were Articular Surface Replacements (ASR, DePuy, Leeds, United Kingdom). One AVN retrieval was a Durom, the other an ASR. The mean time to fracture was 3.4 months. The AVNs were retrieved after 16 months (Durom) and 38 months (ASR).

Volumetric wear rates were determined using a Mitutoyo Legex 322 co-ordinate measuring machine (scanning accuracy within 1 micron) and a bespoke computer program. The method was validated against gravimetric calculations for volumetric wear using a sample femoral head that was artificially worn in vitro. At 5mm3, 10mm3, and 15mm3 of material removal, the method was accurate to within 0.5mm3. Surface roughness data was collected using a Zygo NewView500 interferometer (resolution 1nm).

Mean wear rates of 17.74mm3/year were measured from the fracture components. Wear rates for the AVN retrievals were 0.43mm3/year and 3.45mm3/year. Mean roughness values of the fracture retrievals (PV = 0.754nm, RMS = 0.027nm) were similar to the AVNs (PV = 0.621nm, RMS = 0.030nm), though the AVNs had been in vivo for significantly longer.

Theoretical lubrication calculations were carried out which found that in both AVN retrievals and in seven of the twelve cases of femoral fracture the roughening was sufficient to change the lubrication regime from fluid film to mixed. Three of these surfaces were bordering on the boundary lubrication regime. The results show that even before the femoral fracture, wear rates and roughness values were high and the implants were performing poorly.

Introduction. The identification of biological markers associated to implant failure in THA (total hip arthroplasty) patients remains a challenge in orthopedic surgery. In this search, previous studies have been mainly focused on typical mediators associated to bone metabolism and inflammation. Our group has evaluated changes in serum levels of insulin-like growth factor binding protein-1 (IGFBP-1), a protein which is not directly related to bone homeostasis, in patients undergoing THA. Method. We assessed IGFBP-1 levels in serum obtained from 131 patients (58% female, 42 % male; age: 68 ± 13 years) who underwent THA in the Orthopedic Surgery and Traumatology Department of our institution. In this cohort, 57% of patients had metal on polyethylene (MoP) as hip-bearing surface combination, 17 % had ceramic on ceramic (CoC) and 26% of them did not have any prosthesis. A test based on an enzyme-linked immunosorbent assay (ELISA) was used to determine IGFBP-1 levels in serum obtained from these patients. Result. Our results showed a significant increase in IGFBP- 1 levels in MoP group as compared to CoC and control groups, in which no differences in quantified levels were detected. Further analysis revealed no significant differences in IGFBP-1 between cemented and non-cemented MoP bearings. We performed a ROC curve to evaluate the accuracy of serum IGFBP-1 in discriminating MoP from the rest of patients (area under the curve: 0.7; 95% confidence interval: 0,6-0.8; p<0.05) and established a cut-off value of 10.2 ng/ml, according to the Youden´s Index. Logistic regression analysis showed that patients with MoP bearing surfaces had a higher risk of increased IGFBP- 1 levels in serum (p<0.05, Odds Ratio: 6.7, 95% Confidence Interval 3.1 to 14.8). Conclusion. IGFBP- 1 levels are significantly elevated in THA patients with MoP bearing surfaces, suggesting that this protein might be a reliable biomarker for the outcome of patients implanted with MoP

An increase in metal ion levels is seen after implantation of all MoM hip prosthesis due to release from the surface directly, more so during articulation and corrosion of the bearing surfaces. The bearing surfaces in MoM prosthesis consist of cobalt, chromium and molybdenum. Several case-reports of cobalt toxicity due to a MoM prosthesis have been published in the last decade. Cobalt intoxication may lead to a variety of symptoms: neuro-ocular toxicity (tinnitus, vertigo, deafness, blindness, convulsions, headaches and peripheral neuropathy), cardiotoxicity and thyroid toxicity. Nausea, anorexia and unexplained weight loss have been described. Systemic effects from metal ions even with well functioning implants or with ion concentrations lower than those associated with known adverse effects may exist and warrant investigation. The aim of this study is to investigate self-reported systemic complaints in association with cobalt ion concentrations in patients with any type of MoM hip prosthesis. A cohort study was conducted. Patients with both unilateral and bilateral, resurfacing and large head metal on metal total hip arthroplasties were included for the current study. Blood metal ion concentrations (cobalt and chromium) were measured by inductively coupled plasma mass spectrometry (ICP-MS). Based on the known cobalt toxicity symptoms of case-reports and toxicology reports a new non-validated questionnaire was developed. questions were subdivided in general questions/symptoms, vestibular symptoms, neurological symptoms, emotional health and cardio- and thyroid toxicity symptoms. Independent samples T test, Fishers Exact Test and Pearsons (R) correlation were used. Analysis was performed on two groups; a low cobalt ion concentration group and a high cobalt ion concentration group A total of 62 patients, 36 (58%) men and 26 (42%) women, were included with a mean age at surgery of 60.8 ± 9.3 years (41.6 – 78.1) and a mean follow up of 6.3 ± 1.4years (3.7 – 9.6). In these patients a total of 71 prosthesis were implanted: 53 unilateral and 9 bilateral. Of these, 44 were resurfacing and 27 large head metal on metal (LHMoM) total hip arthroplasties. Mean cobalt and chromium ion concentrations were 104 ± 141 nmol/L (9 – 833) and 95 ± 130nmol/L (6 – 592), respectively. Based on the different thresholds (120 – 170 or 220 nmol/L) the low cobalt ion concentration group consisted of 44 (71%), 51 (82%) or 55 (89%) subjects respectively. No differences were found in general characteristics, independently of the threshold. The composite score of vestibular symptoms (vision, hearing, tinnitus, dizziness) was significantly higher (p < .050) in all high cobalt ion concentrations groups, independent of the threshold value This study aimed to detect a trend in self-reported systemic complaints in patients with metal-on-metal hip arthroplasty due to raised cobalt ion concentrations. Vestibular symptoms were more common in high cobalt ion concentration groups independent of the three threshold levels tested. The upper limit of acceptable cobalt ion concentrations remains uncertain. With regards to proactively inquired, self-reported symptoms the threshold where effects may be present could be lower than values currently applied in clinical follow-up. It is unknown what exposure to elevated metal ion concentrations for a longer period of time causes with aging subjects. Further research with a larger cohort and a more standardized questionnaire is necessary to detect previously undiscovered or under-reported effects

It has been speculated that high wear at the head-stem taper may contribute to the high failure rates reported for stemmed large head metal-on-metal (LH-MOM) hips. In this study of 53 retrieved LH-MOM hip replacements, we sought to determine the relative contributions of the bearing and taper surfaces to the total wear volume. Prior to revision, we recorded the relevant clinical variables, including whole blood cobalt and chromium levels. Volumetric wear of the bearing surfaces was measured using a coordinate measuring machine and of the taper surfaces using a roundness measuring machine. The mean taper wear volume was lower than the combined bearing surface wear volume (p = 0.015). On average the taper contributed 32.9% of the total wear volume, and in only 28% cases was the taper wear volume greater than the bearing surface wear volume. Despite contributing less to the total material loss than the bearing surfaces, the head-stem taper junction remains an important source of implant-derived wear debris. Furthermore, material loss at the taper is likely to involve corrosion and it is possible that the material released may be more biologically active than that from the bearing surface

Introduction and Objective. Ceramic on Ceramic bearings in Total Hip Arthroplasty (THA) afford a low friction coefficient, low wear rates and extreme hardness. Significant complications include hip squeak, ceramic fracture and poor polyethylene performance in revision procedures due to imbedding of abrasive microscopic ceramic fragments. We report on the results of this bearing at a minimum of 10 years. Materials and Methods. A single-centre retrospective review of 449 THAs was performed. Primary outcome measures included aseptic revision and all-cause revision rates at a minimum of 10 years post operatively. Evaluation of functionality was performed with WOMAC and SF-36 scores which were performed pre-operatively and at intervals of 6 months, one year, 2 years, 5 years and 10 years post operatively. Results. There was a 6.2% (n=28) all-cause and 5.3% (n=24) aseptic revision rate for ceramic on ceramic total hip arthroplasty at minimum of 10 years with a mean time to revision 4.8 years (range 2 months − 11.6 years). Notably, there were 2 revisions for ceramic head fracture, one for ceramic liner fracture, 3 for aseptic loosening and 3 revisions for squeaking. Pain of unknown origin was the most common reason for revision. There was an improvement in postoperative WOMAC scores from a mean of 59.8 (range 15–95) pre-operatively to a mean of 15.6 (range 0–78) at 10 years. Conclusions. This study showed good functional outcomes but high revision rates for CoC THA at a minimum of 10 years. The role for CoC bearings in THA has been called into question in recent years and may continue to decline in popularity, even in younger patients. Further large scale studies are important to assess the long-term outcomes of this bearing surface

Introduction and Objective. Despite pure alumina have shown excellent long-term results in patients undergoing total hip arthroplasty (THA), alumina matrix composites (AMCs) composed of alumina and zirconium oxide are more commonly used. There are no comparative studies between these two different ceramics. We performed a retrospective case-control study to compare results and associated complications between AMC from two manufacturers and those with pure alumina from another manufacturer. Materials and Methods. 480 uncemented THAs with ceramic on ceramic (CoC) bearing surfaces (288 men and 192 women; mean age of 54.1 ± 12.4 years), were implanted from 2010 to 2015. Group 1: 281 THAs with pure alumina; Group 2A: 142 with AMC bearing in a trabecular titanium cup. Group 2B: 57 hips with AMC bearing with a porous-coated cup. Results. The mean follow-up was 7.3 years. There was one late infection in group 1, eight dislocations, three in group 1 (1.1%), three in group 2A (2.1%), all with a 36 mm femoral head, and two in group 2C (3.5%). Liner malseating was found in one hip in group 1, and in five hips in group 2C, of these, there were four liner fractures (7.0%). Four cups were revised for iliopsoas impingement (three in group 1 and one in group 2B). Two cups were revised for aseptic loosening, one in group 1 and one in group 2A, and four revised femoral stems in group 2A, three for subsidence and another for postoperative periprosthetic B. 2. fracture. The mean preoperative Harris Hip Score was 48.6 ± 3.3 in the whole series and 93.9 ± 7.2 at the end of follow-up. The survival rate of revision for any cause was 98.2% (95% Confidence Interval: 96.6–99.8) at ten years for group 1, 95.8% (95% CI: 92.1–99.5) for group 2A, and 91.1% (95% CI: 83.7–98.5) for group 2B (log-rank 0.030). Conclusions. Outcome of uncemented CoC THA in young patients was satisfactory at mid-term in all three groups. However, liner fractures were frequent in group 2B. All dislocated hips in group 2A had a 36 mm femoral head diameter, and revision due to any cause was less frequent in group 1. Pure alumina CoC THA can be used as a benchmark for comparison with newer CoC THAs

Introduction. Prosthetic joint infection (PJI) is an uncommon but serious complication of hip replacement. A recent systematic review of patient risk factors for PJI identified male gender, smoking status, increasing BMI, steroid use, previous joint surgery and comorbidities of diabetes, rheumatoid arthritis and depression as risk factors for developing PJI. Limitations of the current literature include the short term follow up of most published studies. We investigated the role of patient, surgical and healthcare factors on the risk of revision of a primary hip replacement for PJI at different time-points in the post-operative follow-up. It is important that those risk factors are identified so that patients can be appropriately counselled according to their individual risk profile prior to surgery and modifiable factors can be addressed to reduce the risk of PJI at an individual and healthcare system level. Materials and Methods. Primary hip replacements and subsequent revision procedures performed for PJI from 2003–2014 were identified from the National Joint Registry (NJR). Patient (age, gender, ASA grade, BMI), perioperative (surgical indication, type of anaesthesia, thromboprophylaxis regime, surgical approach, hip replacement and bearing surface and use of femoral or acetabular bone graft) and healthcare system characteristics (surgeon grade, surgical volume) were linked with data from Hospital Episode Statistics to obtain information on specific ethnicity and comorbidities (derived from the Charlson index). Multilevel piecewise exponential non-proportional hazards models were used to estimate their effects at different post-operative periods (0–3 months, 3–6 months, 6–12 months, 12–24 and >24 months post-operation). Results. The index hip replacements consisted of 623,253 primaries with 2,705 subsequently revised for PJI, 14% within 3 months, 8% between 3–6 months, 14% between 6–12 months, 22% between 1–2 years and 42% ≥2 years after the index procedure. Risk factors for revision of PJI included male gender, high BMI, high ASA grade and younger age. Their effects were period-specific. Patients with chronic pulmonary disease, diabetes or dementia had high early risk of revision for PJI, as did patients operated for a fractured neck of femur (<3 months). Metal-on-metal bearings (>12 months) and lateral surgical approach (≥3 months) also influenced the mid- and long-term revision risk for PJI. No or modest associations were found with the operating surgeon grade, surgical volume and hospital surgical volume. Conclusion. The effects of patient, perioperative and healthcare system risk factors for PJI after primary hip replacement are time-dependent. Modifiable risk factors such as the type of surgical approach and bearing surface have also been found

Background. As the number of ceramic THR bearings used worldwide is increasing, the number of implants that experience off-normal working conditions, e.g. edge loading, third bodies in the joint, soft tissues laxity, dislocation/subluxation of the joint, increases too. Under all such conditions the bearing surfaces can be damaged, leading eventually to a limitation of the expected performances of the implant. Methods. We characterised the damage resistance of different bearing surfaces (alumina matrix composite BIOLOXdelta, alpha-alumina BIOLOXforte, zirconia 3Y-TZP, oxidized zirconium alloy Zr-2.5Nb, CoCr-alloy) by scratch tests performed following the European standard EN 1071–3:2005. Also the scratch hardness of same materials has been assessed. Results. The Lc1 value (i.e., the load for the onset of a scratch) measured for BIOLOXdelta is about fivefold the one measured for the oxidized zirconium alloy (OXZr) surface and about tenfold the Lc1 measured for the CoCr alloy. The height of ridges along the scratch edges due to plastic flow in the composite ceramic BIOLOXdelta are only 21% in height than in CoCr, and only a small fraction (0.04%) of the height of ridges measured on OXZr surfaces. The scratch hardness of the metal samples tested (CoCr, OXZr) results one order of magnitude lower than the ones of ceramics. This behavior is not influenced by of the presence of the coating on OXZr surface. Conclusions. The transformation toughened ceramics tested (BIOLOXdelta, 3Y-TZP) are the materials that exhibit the higher resistance to scratching. Ridges at scratch edges are lower in ceramics than in coated or uncoated metals. The result show the superior scratch resistance behavior of toughened ceramics for THR wear couples with respect to coated or bare alloys. Level of Evidence. Level 1

Metal on metal hip replacements have been one of worst failures in recent years in terms of orthopaedic implants. Some of these devices have had catastrophic failure rates, with reports of 48% failure at 6 years. The failure of these devices has led to considerable suffering, pain and reduction in quality of life; consequently, they have given rise to high costs and multi-million-dollar legal cases. This talk will describe the history of the current metal on metal failure and discuss some of the reasons why might have occurred. It will also consider the reasons that wear debris arising from the trunnion is worse in terms of biological activity then that arising from the bearing surfaces

Material loss at the head-stem taper junction may contribute to the high early failure rates of stemmed large head metal-on-metal (LH-MOM) hip replacements. We sought to quantify both wear and corrosion and by doing so determine the main mechanism of material loss at the taper. This was a retrospective study of 78 patients having undergone revision of a LH-MOM hip replacement. All relevant clinical data was recorded. Corrosion was assessed using light microscopy and scanning electron microscopy, and graded according to a well-published classification system. We then measured the volumetric wear of the bearing and taper surfaces. Evidence of at least mild taper corrosion was seen in 90% cases, with 46% severely corroded. SEM confirmed the presence of corrosion debris, pits and fretting damage. However, volumetric wear of the taper surfaces was significantly lower than that of the bearing surfaces (p = 0.015). Our study supports corrosion as the predominant mechanism of material loss at the taper junction of LH-MOM hip replacements. Although the volume of material loss is low, the ionic products may be more biologically active compared to the particulate debris arising from the bearing surfaces

We have undertaken a series of clinical trials over the last 20 years to look at different bearing surface combinations in young adults. We continue to follow these patients well beyond the planned duration of the trials and new information is constantly becoming available. The first trial compared ceramic-on-ceramic with ceramic-on-standard-polyethylene. These patients have now been followed for 20 years with significant wear in the polyethylene group but virtually identical revision rates. The second trial ceramic-on-ceramic, cobalt-chrome-on-standard-polyethylene and cobalt-chrome-on-cross-linked-polyethylene. In this group the ceramic-on-ceramic patients have the lowest revision rate; the ceramic-on-polyethylene group demonstrates a lower wear rate than cobalt-chrome-on-polyethylene. The third trial looks at cobalt-chrome versus zirconium on either cross-linked polyethylene or conventional polyethylene. At 10 years there remains no evidence of improved performance from the zirconium surface as compared to cobalt-chrome. The cross-linked polyethylene group is clearly outperforming the conventional polyethylene in terms of wear rate but at 10 years the revision rates remain the same in all groups. Cross liked polyethylene appears to be the major determining factor in prosthetic longevity and appears to be more important than the counter face material

Finite element analysis was used to examine the initial stability after hip resurfacing and the effect of the procedure on the contact mechanics at the articulating surfaces. Models were created with the components positioned anatomically and loaded physiologically through major muscle forces. Total micromovement of less than 10 μm was predicted for the press-fit acetabular components models, much below the 50 μm limit required to encourage osseointegration. Relatively high compressive acetabular and contact stresses were observed in these models. The press-fit procedure showed a moderate influence on the contact mechanics at the bearing surfaces, but produced marked deformation of the acetabular components. No edge contact was predicted for the acetabular components studied. It is concluded that the frictional compressive stresses generated by the 1 mm to 2 mm interference-fit acetabular components, together with the minimal micromovement, would provide adequate stability for the implant, at least in the immediate post-operative situation

Summary Statement. The chemistry, amount, morphology, and size distribution of wear debris from silicon nitride coatings generated in the bearing surface can potentially reduce the negative biological response and increase the longevity compared to conventional materials in joint replacements. Introduction. Total hip implants have a high success rate at 15 years of implantation, but few survive over 25 years. At present, revisions are mostly due to aseptic loosening, believed to mainly be caused by the biological response to wear debris generated in the joint bearing. For the polymer liners the size of the wear debris determines the biological response, while for metal bearing surfaces a limitation is the metal ion release. When ceramics are used, the wear debris is in general small and mechanical factors may be the main cause for failure. A more recent, experimental alternative is to let the well-known metallic substrate serve as the soft, tough bulk, and additionally apply a hard and smooth ceramic coating. In this way a lower wear rate and reduced metal ion release could be obtained. Furthermore, the chosen composition, silicon nitride (SixNy), contains no detrimental ions, and silicon nitride debris has been shown to slowly dissolve in aqueous medium. Altogether, it can potentially increase the longevity of the implant. However, the debris from SixNy coatings has not yet been characterised. In this study, a wear model test was performed to generate wear debris from SixNy coatings. The debris was characterised using scanning electron microscopy (SEM) and energy-dispersive X-ray spectroscopy (EDS) in combination with computational calculations. Methods. Silicon nitride coatings deposited on flat cobalt chromium alloy (ASTM F75) were worn in a reciprocating ball on disc setup in a 25% serum solution at 37°C against an alumina ball with a load of 1.5 N. Wear debris was separated using serum digestion with hydrochloric acid (ISO 17853:2011) and examined in SEM in combination with EDS. As reference polyethylene (PE) was used to verify that relevant particles sizes were achieved. The SEM images were processed using a modified MATLAB-script originating from Cervera Gontard et al. [1], identifying the particles and calculating their size. Results. Particles generated from SixNy coatings (n=62) a size distribution D50 [D10-D90] of 0.29 µm [0.16–0.69] and were round to oval in shape. The PE particles (n=70) had a size distribution of 0.29 µm [0.13–1.3], shaped similar to the SixNy particles or with a more elongated shape. Discussion and conclusions. PE wear debris has been reported to lie in the size range of nm up to several μm in vivo, with a large proportion within the critical size for macrophage activation (0.2 to 0.8 μm). The model test reports relevant sizes and shape of PE debris, confirming the validity of the method. Particles generated from the SixNy coatings showed a smaller size range than PE, however most particles were within the critical size range for biological activation. In conclusion, this model test could be used to generate what we believe are relevant sizes and shapes of PE and SixNy wear debris and to learn more at an early stage of prediction of wear debris. Further dissolution studies as well as studies on the in vitro and in vivo cell response to these types of particles will be performed. The authors thank the Swedish Foundation for Strategic Research (SSF) through MS2E and FP7 NMP project LifeLongJoints for financial support, as well as Linköping University for the coating facilities and expertise

Background. Prosthetic implants used in primary total hip replacements have a range of bearing surface combinations (metal-on-polyethylene, ceramic-on-polyethylene, ceramic-on-ceramic, metal-on-metal); head sizes (small <36mm, large 36mm+); and fixation techniques (cemented, uncemented, hybrid, reverse hybrid), which influence prosthesis survival, patient quality of life, and healthcare costs. This study compared the lifetime cost-effectiveness of implants to determine the optimal choice for patients of different age and gender profiles. Methods. In an economic decision Markov model, the probability that patients required one or more revision surgeries was estimated from analyses of UK and Swedish hip joint registries, for males and females aged <55, 55–64, 65–74, 75–84, and 85+ years. Implant and healthcare costs were estimated from hospital procurement prices, national tariffs, and the literature. Quality-adjusted life years were calculated using utility estimates, taken from Patient-Reported Outcome Measures data for hip procedures in the UK. Results. Optimal choices varied between traditionally used cemented metal-on-polyethylene and cemented ceramic-on-polyethylene implants. Small head cemented ceramic-on-polyethylene implants were optimal for males and females aged under 65. The optimal choice for adults aged 65 and older was small head cemented metal-on-polyethylene implants. Conclusions. The older the patient, the higher the probability that small head cemented metal-on-polyethylene implants are optimal. Small head cemented ceramic-on-polyethelyne implants are optimal for adults aged under 65. Our findings can influence NICE guidance, clinical practice, and commissioning of services. Funding. NIHR Research for Patient Benefit programme PB-PG-0613-31032

Highly cross-linked polyethylene (HXLPE) is now a common used bearing surface in total hip arthroplasty. Current studies report superior wear rates with the use of HXLPE in total hip arthroplasty. However, there are few studies to support its long term use. The aim of this study is to measure the long term wear of HXLPE and evaluate patient satisfaction at more than 10 years follow up. 44 total hip arthroplasties were performed through a direct lateral approach by a single surgeon. All patients received the same uncemented acetabular component, mean liner thickness was 6.91mm (SD= 0.68). 16 of the femur components were cemented. Outcomes analysed include wear rates, osteolysis, revision rates, SF12 and Oxford hip scores. Wear rate was calculated using computer software (Polyware®) using edge detection software. Mean age at surgery was 58.9 years (SD= 11.67). The mean follow up was 11.3 years (SD= 1.19). There was no evidence of osteolysis and none had undergone revision surgery. Mean two dimensional wear was 0.38mm (SD= 0.25) and mean wear rate per year was 0.03mm (SD= 0.02, range 0.009 to 0.078). Oxford hip score at last follow up indicated satisfactory joint function (mean= 42 SD= 6.2). Our results support the use of highly cross-linked polyethylene in primary total hip replacements. The absence of osteolysis and need for revision surgery over a mean of 11.3 years is very encouraging

Background and aim. Recent proposals have been introduced to modify stem design and/or femoral fixation in total hip replacement (THR). New designs need to consider previous design features and their results. The aim of this study has been to evaluate the clinical and radiological results of six different designs of tapered uncemented stems implanted in our Institution. Methods. 1918 uncemented hips were prospectively assessed from 1999 to 2011 (minimum follow-up of five years for the unrevised hips). All hips had a 28 or 32 mm femoral head and metal-on-polyethylene or alumina-on-alumina bearing surface. Six uncemented femoral designs that shared a femoral tapered stem incorporating a coating surface were included in the study. The different design features included the type of coating, metaphyseal filling, and sectional shape. Results. Intra-operative proximal femoral crack was 6.7% in one of the designs (p=0.01), univariate analysis showing a lower risk of crack in the other designs. The position of the stem was neutral in 80% of the cases for all designs. Femoral canal filing was related to the stem design (p<0.001 at the three levels) and to the femoral level assessed (subset alpha=0.005). Twelve stems were revised for aseptic loosening (6 from two different designs). The survival rate for femoral aseptic loosening at 15 years was 96.6% (95% CI 93.8 to 99.4) for one of these two designs ad 97.4% (95% CI95.5 to 99.6) for the other. Regression analysis showed that stem design was the only factor related to aseptic loosening when adjusted for femoral canal filling (at the three levels) stem position (neutral or not) and femoral type (cylindrical or not). Conclusion. Tapered uncemented stems consistently provide excellent bone fixation. New designs need to avoid changing successful features and concentrate on the less successful aspects