We report the five-year outcome of a randomised controlled trial which used radiostereometric analysis (RSA) to assess the influence of surface oxidised zirconium (OxZr, Oxinium) on polyethylene wear in vivo.

A total of 120 patients, 85 women and 35 men with a mean age of 70 years (59 to 80) who were scheduled for primary cemented total hip arthroplasty were randomly allocated to four study groups. Patients were blinded to their group assignment and received either a conventional polyethylene (CPE) or a highly cross-linked (HXL) acetabular component of identical design. On the femoral side patients received a 28 mm head made of either cobalt-chromium (CoCr) or OxZr.

The proximal head penetration (wear) was measured with repeated RSA examinations over five years. Clinical outcome was measured using the Harris hip score.

There was no difference in polyethylene wear between the two head materials when used with either of the two types of acetabular component (p = 0.3 to 0.6). When comparing the two types of polyethylene there was a significant difference in favour of HXLPE, regardless of the head material used (p < 0.001).

In conclusion, we found no advantage of OxZr over CoCr in terms of polyethylene wear after five years of follow-up. Our findings do not support laboratory results which have shown a reduced rate of wear with OxZr. They do however add to the evidence on the better resistance to wear of HXLPE over CPE.

Cite this article: Bone Joint J 2015;97-B:1463–9.

We performed a randomised controlled trial comparing computer-assisted surgery (CAS) with conventional surgery (CONV) in total knee replacement (TKR). Between 2009 and 2011 a total of 192 patients with a mean age of 68 years (55 to 85) with osteoarthritis or arthritic disease of the knee were recruited from four Norwegian hospitals. At three months follow-up, functional results were marginally better for the CAS group. Mean differences (MD) in favour of CAS were found for the Knee Society function score (MD: 5.9, 95% confidence interval (CI) 0.3 to 11.4, p = 0.039), the Knee Injury and Osteoarthritis Outcome Score (KOOS) subscales for ‘pain’ (MD: 7.7, 95% CI 1.7 to 13.6, p = 0.012), ‘sports’ (MD: 13.5, 95% CI 5.6 to 21.4, p = 0.001) and ‘quality of life’ (MD: 7.2, 95% CI 0.1 to 14.3, p = 0.046). At one-year follow-up, differences favouring CAS were found for KOOS ‘sports’ (MD: 11.0, 95% CI 3.0 to 19.0, p = 0.007) and KOOS ‘symptoms’ (MD: 6.7, 95% CI 0.5 to 13.0, p = 0.035). The use of CAS resulted in fewer outliers in frontal alignment (> 3° malalignment), both for the entire TKR (37.9% vs 17.9%, p = 0.042) and for the tibial component separately (28.4% vs 6.3%, p = 0.002). Tibial slope was better achieved with CAS (58.9% vs 26.3%, p < 0.001). Operation time was 20 minutes longer with CAS. In conclusion, functional results were, statistically, marginally in favour of CAS. Also, CAS was more predictable than CONV for mechanical alignment and positioning of the prosthesis. However, the long-term outcomes must be further investigated.

Cite this article: Bone Joint J 2014; 96-B:609–18.

We evaluated the rates of survival and cause of revision of seven different brands of cemented primary total knee replacement (TKR) in the Norwegian Arthroplasty Register during the years 1994 to 2009. Revision for any cause, including resurfacing of the patella, was the primary endpoint. Specific causes of revision were secondary outcomes.

Three posterior cruciate-retaining (PCR) fixed modular-bearing TKRs, two fixed non-modular bearing PCR TKRs and two mobile-bearing posterior cruciate-sacrificing TKRs were investigated in a total of 17 782 primary TKRs. The median follow-up for the implants ranged from 1.8 to 6.9 years. Kaplan-Meier 10-year survival ranged from 89.5% to 95.3%. Cox’s relative risk (RR) was calculated relative to the fixed modular-bearing Profix knee (the most frequently used TKR in Norway), and ranged from 1.1 to 2.6. The risk of revision for aseptic tibial loosening was higher in the mobile-bearing LCS Classic (RR 6.8 (95% confidence interval (CI) 3.8 to 12.1)), the LCS Complete (RR 7.7 (95% CI 4.1 to 14.4)), the fixed modular-bearing Duracon (RR 4.5 (95% CI 1.8 to 11.1)) and the fixed non-modular bearing AGC Universal TKR (RR 2.5 (95% CI 1.3 to 5.1)), compared with the Profix. These implants (except AGC Universal) also had an increased risk of revision for femoral loosening (RR 2.3 (95% CI 1.1 to 4.8), RR 3.7 (95% CI 1.6 to 8.9), and RR 3.4 (95% CI 1.1 to 11.0), respectively). These results suggest that aseptic loosening is related to design in TKR.

Cite this article: Bone Joint J 2013;95-B:636–42.

Background

Improving quality and efficiency is a priority in health policy. Several studies have shown a correlation between high hospital volume and improved results of surgery. In Norway, orthopaedic surgeons operate a relatively low number of total knee replacements compared with other countries. The number of total knee replacements has, however, increased significantly over the past 10 years.

Some studies have also shown an association between surgeon volume and outcomes after total knee arthroplasty (TKA).

Background

Improving positioning and alignment by the use of computer assisted surgery (CAS) might improve longevity and function in total knee replacements. This study evaluates the short term results of computer navigated knee replacements based on data from a national register.

Introduction

The treatment of trochanteric and subtrochanteric fractures remains controversial, and new implants are constantly being developed trying to improve outcome and minimize the number of complications in these fractures.

In Norway the Sliding Hip Screw(SHS), with or without a Trochanteric Stabilizing Plate (TSP), is still the most commonly used implant, but worldwide nailing of these fractures is increasing. This trend, however, has not been supported by documentation of better clinical results compared to the SHS in well designed studies. Therefore, in the present study we compared the recently launched Trigen Intertan nail (Smith and Nephew) with the SHS in the treatment of trochanteric and subtrochanteric fractures.

In this prospective study we studied the effect of the inclination angle of the acetabular component on polyethylene wear and component migration in cemented acetabular sockets using radiostereometric analysis.

A total of 120 patients received either a cemented Reflection All-Poly ultra-high-molecular-weight polyethylene or a cemented Reflection All-Poly highly cross-linked polyethylene acetabular component, combined with either cobalt–chrome or Oxinium femoral heads. Femoral head penetration and migration of the acetabular component were assessed with repeated radiostereometric analysis for two years. The inclination angle was measured on a standard post-operative anteroposterior pelvic radiograph. Linear regression analysis was used to determine the relationship between the inclination angle and femoral head penetration and migration of the acetabular component.

We found no relationship between the inclination angle and penetration of the femoral head at two years’ follow-up (p = 0.9). Similarly, our data failed to reveal any statistically significant correlation between inclination angle and migration of these cemented acetabular components (p = 0.07 to p = 0.9).

Objectives: The major objective of the present study was to investigate the risk of revision of infection after primary total knee replacements (TKR) in patients with rheumatoid arthritis (RA) during a 13-year period. We wished to compare RA patients with OA patients in order to detect differences in the risk of revision for infection, and to compare changes in the risk for the two patient groups over time. Furthermore we studied the time from primary implantation to revision for infection in the two groups.

Patients and Methods: From January 1994 to June 2008, 2482 primary TKRs in patients with RA and 25189 in OA patients were identified in the Norwegian Arthroplasty Register. Kaplan-Meier survival curves, with revision for infection as the endpoint, were constructed. Cox regression analyses were performed to calculate relative risk (RR) of revision for infection according to diagnosis, age, gender, year of surgery (from 1994 through 2000 and from 2001 to June 2008) and time of revision related to the time of primary TKR. All relative risks were adjusted for the other variables.

Results: The 5-year revision rate for infection of TKR was 1.1% in RA patients and 0.5% in OA patients. Rheumatoid arthritis patients had a 1.7 (95%CI 1.2–2.6) times higher risk of revision for infection compared to the control group (OA patients). The patients who had TKR surgery in the later period had a decreased risk of revision for infection (RR 0.72, 95%CI 0.53–0.98) compared to the time period 1994–2000. Furthermore, the risk of revision for late infection in RA patients, increased compared to OA from 4 years after the index operation.

Conclusion: Patients with RA undergoing total knee replacement surgery were at a higher risk of revision for prosthetic joint infection and had a higher risk of late infections leading to revision, than patients with OA. These findings emphasizes the importance of preoperative management and optimizing preventive strategies, especially in this patient group.

Background and Purpose: The purpose of the present study was to assess the risk for revision due to deep infection for primary uncemented total hip arthroplasties (THAs) reported to the Norwegian Arthroplasty Register (NAR) over the period 1987–2007.

Methods: All primary uncemented THAs reported to NAR from the period 1987–2007 were studied. Adjusted Cox regression analyses with first revision due to deep infection as the end-point were performed. Changes in the revision rate as a function of year of operation were investigated, as was impact of risk factors (gender, age, type of diagnosis, duration of surgery, operation room ventilation and systemic antibiotic prophylaxis) on risk for revision due to deep infection.

Results: 14,348 primary uncemented THAs met the inclusion criteria. 97 THAs had been revised due to deep infection (5-year survival 99.56). Risk for revision due to deep infection increased through the period studied. Compared to the uncemented THAs implanted 1987–1992, the risk for revision due to infection was 1.2 times higher (95%CI 0.6–2.4, p=0.6) for those implanted 1993–1997, 1.4 times (95%CI 0.7–2.9, p=0.3) for 1998–2002, and 5.3 times (95%CI 2.6–10.7, p=< 0.001) for 2003–2007. The increase in risk for revision due to infection for primary uncemented THAs was most pronounced after the year 2000. No risk factor registered had any statistically significant impact on risk for revision due to infection in this study.

Interpretation: The results of this study indicate an increase in incidence of deep infection after uncemented THAs during the period 1987–2007. Concomitant changes in confounding factors, however, complicate the interpretation of these results.

Background: The treatment of trochanteric and subtrochanteric fractures is still controversial. In Norway the most commonly used implant for these fractures is the Sliding Hip Screw (SHS), with or without a trochanteric support plate. The Intertan nail (Smith & Nephew) has been launched as a nail with improved biomechanical properties for the treatment of these fractures, but so far it has not been shown that the clinical results are superior to the traditional Sliding Hip Screw.

We wanted to investigate any differences in pain and function between the new Intertan nail and the Sliding Hip Screw in the early postoperative phase.

Materials and Methods: 665 patients older than 60 years with a trochanteric or subtrochanteric fracture were randomized to either a SHS (CHS/DHS) or an Intertan nail in 5 hospitals. For practical reasons only 315 patients (47%) were evaluated at day 5 postoperatively (163 Intertan and 152 SHS), and these patients were used for our analysis. Pain was measured using a Visual Analog Scale (VAS), and early functional mobility by the “Timed Up and Go”- test (TUG-test). T-tests and chi-square tests were used to examine differences between the groups.

Results: The average pain at rest was similar for the 2 groups (VAS 21). Pain at mobilization, however, differed, where patients operated with the Intertan nail had less pain than those operated with the SHS (VAS 47 vs. 53, p = 0.02). The difference between the implants was most pronounced for the simple two-part fractures (AO Type A1). More patients treated with the nail than with the SHS performed the TUG-test at day 5 (85/163 vs. 63/152, p = 0.06), but there was no statistically significant difference regarding the average speed the TUG-test was performed with (71 vs. 66sec, p = 0.36). The implant type did not influence the length of hospital stay.

Discussion/Conclusion: Regarding early postoperative pain and function, there seems to be similar or better results for trochanteric and subtrochanteric fractures treated with the Intertan nail compared to the SHS. The difference in measured pain level was statistically significant, but may not be clinically significant (a difference of VAS 6). We could not detect any significant differences in terms of early functional mobility between the two implants.

In our opinion it still remains to show good long-term results and acceptable complication rates before the new Intertan nail is widely taken into use. Due to the additional costs for the Intertan nail also economic aspects should be considered when choosing the implant and operative method for these fractures.

Background and purpose: Development of minimal invasive operation techniques has given unicompartemental knee arthroplasty (UKA) renewed interest. Indications for use of UKA are however debated, and short-term advantages of UKA over total knee arthroplasty (TKA) should be weighed against the higher risk of reoperation. More knowledge on long term results of pain and function after knee arthroplasties is therefore needed and was the purpose of this study.

Methods: Patient-reported pain and function were collected at least two years after the operations in postal questionnaires from 1643 osteoarthritis patients reported to the Norwegian Arthroplasty Register with intact primary TKA (n=1271) or UKA (n=372). The questionnaire contained instruments for calculation of the knee specific Knee Osteoarthritis Outcome Score (KOOS), and for quality of life (EQ-5D, post- and pre-operative). 5 subscales from KOOS were used as outcome. To incorporate an outcome for anterior knee pain an additional subscale based on 7 questions from KOOS, clinically accepted to be related to such discomfort, was calculated. Pain and satisfaction from a visual analogue scale (VAS) were also used as outcomes together with improvement in EQ-5D index score. The outcomes were measured on a scale from 0 (worst) to 100 (best) units with an argued minimal perceptible clinical difference of 8–10 units. Group differences were analysed with multiple linear regression, adjusted for confounding by age, gender, Charnley category, time since operation and preoperative EQ-5D index score.

Results: UKA performed better than TKA for the KOOS subscales Activity in Daily Living (difference in mean outcome score =−3.4,p=0.02) and Sport and Recreation (difference =−4.4,p=0.02) and for Anterior Pain (difference=−4.5,p< 0.01). The difference was also significant for the outcome Pain(VAS) but now in favour of TKA (difference=3.3,p=0.02). Motivated by the discrepancy in the results of the pain related outcomes, Anterior Pain (UKA best), Pain(VAS) (TKA best) and Pain(KOOS) (no difference) further investigation of the questions (0=best to 4=worst) used for calculation of Pain(KOOS) and Anterior Pain were performed. Patients that had undergone UKA had more often pain from the knee (difference=0.26,p< 0.01), while they had less pain when they were bending the knee fully (difference=−0.37,p< 0.01) and less problems when squatting (difference=−0.25,p< 0.01).

Interpretation: Estimated differences did not reach the level of minimal perceptible clinical difference. There are however indication of differences in the way the two treatment groups experience knee related discomfort. Even though UKA offers a lower level of pain and less problems in activities involving bending of the knee, these patients seem to experience pain from the knee more often.

Background: Few studies have compared long-term survival for different types of cemented primary total hip arthroplasties, and prostheses are still used without adequate knowledge of their endurance.

Patients and Methods:We compared the 10 most used prosthesis brands cemented with Palacos or Simplex in primary THAs reported to the Norwegian Arthroplasty Register in 1987–2007, totaling 62305 operations. Survival analyses with revision as endpoint (any cause or aseptic loosening) were performed with Kaplan-Meier and multiple Cox regression analyses. Risk estimates were established for different time intervals after the primary operation (0–5 years, 6–10 years, > 10 years).

Results: At 20 years, the Kaplan-Meier estimated revision per cent with any revision as endpoint was 15 % (95 % CI: 14–17), and with revision due to aseptic loosening as endpoint, 11 % (9.8–13). The adjusted revision percentage for aseptic loosening at 10 years could be established for 7 of the prostheses and varied from 0.4 % (0.0–0.8) for the Lubinus SP (I, II) to 6.6 % (4.1–9.0) for the Reflection all poly/Spectron-EF (cup/stem) combination. Only Charnley, Exeter, Titan and Spectron/ITH could be compared beyond 10 years. While long-term results were similar for these stems, Exeter (RRcup = 1.7, p = 0.001) and Spectron (RRcup = 2.4, p = 0.001) cups had higher revision rates due to aseptic loosening than Charnley cups. Comparing Charnley with prostheses with shorter follow-up, we observed an increased revision risk for aseptic loosening in the 6–10 year time interval also for Reflection all poly/Spectron-EF (RRcup = 5.5, p< 0.001; RRstem = 2.4, p< 0.001), Elite/Titan (RRcup = 7.5, p< 0.001; RRstem = 5.4, p< 0.001) and for the cup in the Reflection all poly/ITH combination (RRcup = 2.1, p = 0.03). Only the Lubinus SP had statistically significant better results than the Charnley prosthesis (RRcup = 0.2, p = 0.09; RRstem = 0.1, p = 0.01). Since several of the prostheses were introduced in recent years, analyses were also performed on operations from 1998 and onwards. Except for Lubinus SP that now had results similar to that of Charnley, differences in survivorship as compared with Charnley were enhanced. This was mainly due to a marked improvement in results for the Charnley prosthesis.

Conclusion: We observed in the Norwegian Arthroplasty Register clinically important differences among cemented prosthesis brands and identified inferior results for previously undocumented prostheses. Overall results at 20 years were, however, satisfactory according to international standards.

Background: This study was done to compare the early failure of primary cemented unicompartmental knee arthroplasties (UKA) with that of total knee arthroplasties (TKA).

Methods: The Kaplan-Meier survial-method and the Cox multiple regression model were used to compare the failure rates of the primary cemented UKAs (n=1410) and the primary cemented TKAs (patellar resurfaced) (n=2818) that were reported to the Norwegian Arthroplasty Register between 1st January 1994 and 1st April 2003.

Results: 8 years survival for UKAs was 85.2 % (95% CI: 81.5–88.9) compared to 93.0 % (91.5–94.5) for TKAs, relative revision risk (RR) 1.8 (1.4–2.4), p< 0.001. The increased revision risk in UKAs was seen in all age categories. Among the UKAs the 8 years survival showed no statistically significant difference for MOD III, Genesis uni and Oxford II. However, Duracon uni knees had, statistically significantly higher rates of revision, although the numbers of prostheses were low. Two UKAs were introduced recently and the follow up was short. After 3 years the Miller Galante uni had 82.8 % (75.6–90.0) survival compared to 93.8 (91.0–96.6) for the Oxford III knee, p< 0.002. The higher failure rates of the Miller Galante and Duracon knees were mainly due to more loosening of the tibial components. UKAs had an increased risk of revision due to pain, aseptic loosening of the tibial and femoral components and periprosthetic fractures compared to TKAs. The UKAs had a lower risk of infection compared to TKAs.

Conclusions: This prospective study has shown that the prostheses survival of cemented UKAs was not as good as for cemented TKAs. There were differences between the UKAs, but the best UKA had results inferior to the average of the TKAs.

Aims: To assess the influence of hip disease on the risk of revision, we studied different disease groups among 53 698 primary total hip replacements (THRs) reported to the NAR between 1987 and 1999. Methods: the revision rate in the 8 most common hip diseases were compared by kaplan-meier survival analyses and cox multiple-regression. To eliminate the influence of prosthesis type a subgroup of 16217 charnley prostheses were analysed. Results: we found statistically significant differences in prosthesis survival among the hip diseases, but after adjustment for prosthesis type most of the differences disappeared. In patients ≤60 years, 59% of the prostheses were uncemented and 33% could be defined as inferior uncemented prostheses. In the charnley subgroup only complications after fracture of the femoral neck had an increased risk for revision compared to primary osteoarthritis (rr 1.5, p=0.005). 10 years survival for cemented charnley prostheses with osteoarthritis was 92.0% for patients ≤60 years and 93.5% for patients > 60 years. Conclusions: after adjustment the results for all disease groups were good. The results of thrs in disease-groups where patients are operated on at a young age were less good because these patients had often been given inferior uncemented prostheses.

Aims: The mid- and long-term results of uncemented cups are uncertain. The aim of this study was to assess their results and to compare them with the most common cemented cup, the Charnley. Material and methods: In the study we included only patients under the age of 60 that had been operated with one of the 10 most common uncemented cup brands or the Charnley cup. Only brands that had been in use for at least 6 years were included. There were only minor differences among the brands concerning age, gender and diagnosis. Survival percentages were calculated with the Kaplan-Meier method. Results: With all cup revisions (change of cup or polyethylene liner) as end-point, the overall 5 year survival probability was 97% (95% CI: 96.3 97.3) which decreased to 86% (95% CI: 84.4 Ð 87.5) at 10 years. The 10 years survival varied from 74% for the Atoll cup to 88% for the Optiþx and the Harris-Galante cup, whereas for the cemented Charnley cup the 10 years survival probability was 94%. For the uncemented cups the increase in revision risk after 6 years was mainly due to wear and osteolysis. For the HA-coated cups also aseptic loosening increased after 6 years of follow-up. Conclusion: Uncemented cups in young patients had good short-term results. At 10 years of follow-up for the investigated uncemented cup brands, the results were inferior to the Charnley cup.

Aims: We wanted to investigate the association between risk factors recorded prospectively before primary hip replacement, and the risk for later revision hip surgery. Methods: During the years 1977–83 The National Health Screening Service in Norway conducted an investigation of risk factors for cardiovascular disease. 56,818 persons born 1925–42 were invited, and 92% participated. We matched these screening data with data from the Norwegian Arthroplasty Register concerning primary and revision hip arthroplasty. Results: We identified 504 men and 834 women who had received a primary total hip replacement after the screening. Of these 75 and 94 were revised during follow-up. Mean age at screening was 49 years; mean age at primary hip replacement was 62 years. Mean age at censoring was 68 years. Men vs women had a relative risk of 1.9 of undergoing hip revision during follow-up (95% CI 1.3–2.8). For each years increase in age at primary hip arthroplasty, the risk of revision surgery during follow-up decreased with 14% for men and 17% for women. Men who at screening had the highest level of physical activity during leisure had 5.5 times the risk of later revision, relative to those with the lowest level of physical activity (95% CI 1.0–31.9). Conclusions: Men have a higher risk for revision hip surgery. There is less risk of revision the older the patient is at primary hip arthroplasty. Men with intense physical activity at middle age are at increased risk of undergoing revision hip surgery before they are 70 years old.

Purpose: The outcome of primary total hip arthroplasty (THA) after a previous paediatric hip disease was studied in data from the Norwegian Arthroplasty Register (NAR).

Materials and Methods: 72,301 primary THAs were reported to the NAR for the period 1987 – February 2002. Of these, 5,459 (7.6%) were performed because of sequela after developmental dysplasia of hip (DDH), 737 (1.0%) because of DDH with dislocation, 961 (1.3%) because of Perthes’/ slipped femoral capital epiphysis (SFCE) and 50,369 (70%) because of primary osteoarthritis (OA). Prosthesis survival was calculated by the Kaplan-Meier method and relative risks for revision in a Cox model with adjustments for age, gender, type of systemic antibiotic, operation time, type of operating theatre and brand of prosthesis.

Results: Without any adjustments the THAs for all three groups of paediatric hip diseases had 1.4 – 2.0 times increased risk for revision compared to that of OA (p< 0.001). Due to huge differences in the studied groups, a more homogenous subset of the data had to be analysed. In this subset, only THAs with well documented prostheses, high-viscosity cements and antibiotic prophylaxis both systemically and in the cement were included (16,874 THAs). In this homogenous subset, no differences in the survivals could be detected for DDH without dislocation and for Perthes’/SFCE compared to OA. For DDH with dislocation the revision risk with all reasons for revisions as endpoint in the analyses was increased 3.3 times compared to OA (p< 0.001), 2.7 times with aseptic loosening as endpoint (p< 0.01) and 10 times with infection as endpoint (p< 0.001).

Conclusions: If well-documented THAs are used after paediatric hip diseases the results are just as good as after osteoarthritis, except for DDH with dislocation where increased revision risk is found.

We have compared the survival of two hydroxyapatite (HA)-coated cups, 1208 Atoll hemispheric and 2641 Tropic threaded, with cemented Charnley all-polyethylene cups (16 021) using the Cox regression model.

The Tropic cup used in combination with an alumina ceramic femoral head, had good results, similar to those of the Charnley cup. When used in combination with a stainless-steel head, however, the risk of revision beyond four years was increased 3.4 times for the Tropic cup compared with the Charnley cup (p < 0.001). Over the same period, the Atoll cup had an increased risk of revision of 3.8 times when used with the alumina heads (p < 0.001) and an increased risk of 6.1 times when used with stainless-steel heads (p < 0.001). Revision because of wear and osteolysis was more common with both types of HA-coated cup than with the Charnley cup. The rate of revision of the Atoll cup because of aseptic loosening was also increased.

We found that HA-coated cups did not perform better than the Charnley cup.

Using data from the Norwegian Arthroplasty Register, we have the assessed survival of 17 323 primary Charnley hip prostheses in patients with osteoarthritis based upon the type of cement used for the fixation of the implant.

Overall, 9.2% had been revised after follow-up for ten years; 71% of the failures involved aseptic loosening of the femoral component. We observed significantly increased rates of failure for prostheses inserted with CMW1 and CMW3 cements. Using implants fixed with gentamicin-containing Palacos cement as the reference, the adjusted Cox regression failure rate ratios were 1.1 (95% CI 0.9 to 1.4) for implants cemented with plain Palacos, 1.1 (95% CI 0.7 to 1.6) for Simplex, 2.1 (95% 1.5 to 2.9) for gentamicin-containing CMW1, 2.0 (95% CI 1.6 to 2.4) for plain CMW1 and 3.0 (95% CI 2.3 to 3.9) for implants fixed with CMW3 cement. The adjusted failure rate at ten years varied from 5.9% for implants fixed with gentamicin-containing Palacos to 17% for those fixed with CMW3.