530 Acta Orthopaedica 2013; 84 (6): 530–536
Monoblock all-polyethylene tibial components have a lower
risk of early revision than metal-backed modular components
A registry study of 27,657 primary total knee arthroplasties
Vivek Mohan1, Maria C S Inacio2, Robert S Namba1, Dhiren Sheth1, and Elizabeth W Paxton2
1Southern California Permanente Medical Group, Department of Orthopedic Surgery, Kaiser Permanente, Irvine; 2Department of Surgical Outcomes and
Analysis, Kaiser Permanente, San Diego, CA, USA.
Submitted 13-03-05. Accepted 13-07-11
Open Access - This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use,
distribution, and reproduction in any medium, provided the source is credited.
Background and purpose With younger patients seeking recon-
structions and the activity-based demands placed on the arthro-
plasty construct, consideration of the role that implant character-
istics play in arthroplasty longevity is warranted. We therefore
evaluated the risk of early revision for a monoblock all-polyeth-
ylene tibial component compared to a metal-backed modular
tibial construct with the same articular geometry in a sample of
total knee arthroplasties (TKAs). We evaluated risk of revision
in younger patients (< 65 years old) and in older patients (≥ 65
Method Fixed primary TKAs with implants from a single
manufacturer, performed between April 2001 and December
2010, were analyzed retrospectively. Patient characteristics, sur-
geon, hospital, procedure, and implant characteristics were com-
pared according to tibial component type (monoblock all-polyeth-
ylene vs. metal-backed modular). All-cause revisions and aseptic
revisions were evaluated. We used descriptive statistics and Cox
Results 27,657 TKAs were identified, 2,306 (8%) with mono-
block and 25,351 (92%) with modular components. In adjusted
models, the risk of early all-cause revision (hazard ratio (HR) =
0.5, 95% confidence interval (CI): 0.3–0.8) and aseptic revision
(HR = 0.6, CI: 0.3–1.2) was lower for the monoblock cohort than
for the modular cohort. In older patients, the early risk of all-cause
revision was 0.6 (CI: 0.4–1.0) for the monoblock cohort compared
to the modular cohort. In younger patients, the adjusted risk of
all-cause revision (HR = 0.3, CI: 0.1–0.7) and of aseptic revision
(HR = 0.3, CI: 0.1–0.7) were lower for the monoblock cohort than
for the modular cohort.
Interpretation Overall, monoblock tibial constructs had a
49% lower early risk of all-cause revision and a 41% lower risk
of aseptic revision than modular constructs. In younger patients
with monoblock components, the early risk of revision for any
cause was even lower.
Early total knee arthroplasty (TKA) designs used all-poly-
ethylene tibial components with good success rates (Rodri-
guez et al. 2001, Vessely et al. 2006, Nouta et al. 2012a).
With metallurgical enhancements and polyethylene modifi-
cations, metal-backed tibial modularity was introduced onto
the market. This enabled decisions regarding soft tissue
tensioning to be made even after cementation of the final
components. The added benefits of implant inventory reduc-
tion and improved polyethylene shelf-life, especially for
rarely used sizes, further increased the universal adoption
of tibial component modularity. However, micromotion of
the modular junction between the polyethylene liner and the
tibial base plate created a second interface for polyethyl-
ene wear debris production. A few studies have evaluated
the role of micromotion and backside wear in the eventual
development of osteolysis, and these have indicated that it
increases the risk of revision (Wasielewski et al. 1997, Parks
et al. 1998).
Clinically, the use of metal-backed TKAs has a high success
rate at long-term follow-up (Manley et al. 2008, Bae et al. 2012,
Nouta et al. 2012b), but backside wear continues to be an issue
that limits implant longevity. In contrast, as demonstrated by
radiostereometric analysis (RSA) techniques, the monoblock
all-polyethylene tibial component has less component migra-
tion and better tibial fixation than fixed, metal-backed designs
(Norgren et al. 2004, Hyldahl et al. 2005, Nouta et al. 2012b).
This would theoretically decrease or eliminate backside wear
and the possibility of eventual revision. Despite this factor and
its economic value, the all-polyethylene tibial component has
not been widely adopted.
Given the current demographic trends of younger patients
seeking reconstructions and the activity-based demands placed
on the arthroplasty construct (Kurtz et al. 2009), consideration
of the role that implant characteristics play in arthroplasty lon-
gevity is warranted. Using a community-based sample of pri-
Acta Orthopaedica 2013; 84 (6): 530–536 531
mary TKAs, we compared the early risk of revision (all-cause
and aseptic) of a monoblock all-polyethylene tibial compo-
nent with that for a metal-backed modular tibial construct of
the same articular geometry and from the same manufacturer,
while adjusting for potential confounders.
Study design, inclusion criteria, and data collection
We conducted a retrospective analysis of a prospectively
followed cohort of primary TKA patients using data from a
community-based Total Joint Replacement Registry (TJRR).
Patients with any diagnosis who underwent a primary fixed-
bearing TKA from a single implant manufacturer (Depuy)
were included in the sample. Cases were registered between
April 1, 2001 and December 30, 2010. Data collection, par-
ticipation, and coverage of the TJRR used to identify the cases
and data have been published elsewhere (Paxton et al. 2008,
2010a b, 2012). Briefly, the TJRR uses a hybrid data collec-
tion process (paper and electronic) to capture patient charac-
teristics together with implant and surgical information, and
also validated algorithms to capture the outcomes of interest.
Intraoperative information is collected by the surgeon at the
time of the procedure. The TJRR sample included in the study
covers 40 medical centers and 233 surgeons in 6 geographical
regions of the USA (Southern and Northern California, Colo-
rado, Hawaii, Northwest, and Mid-Atlantic). The voluntary
participation of the registry in 2010 was 95% (Paxton et al.
Exposure of interest
Implants were classified into metal-backed modular com-
ponents or monoblock all-polyethylene tibial components
using their catalog numbers and descriptions. Metal-backed
components included Depuy’s Press-Fit Condylar (PFC) and
Sigma designs. The monoblock all-polyethylene design inves-
tigated was Depuy’s PFC*Sigma. In these implants, the sur-
face geometry of the tibial liner includes concave portions to
accept the femoral bearing surface, which are neither flat nor
completely conforming. The surface geometries are identical
in the all-polyethylene versions and the modular versions. All
implants used during the TKA procedure are recorded using
the implant stickers that accompany each implant package.
These data are entered manually into the TJRR, and quarterly
quality control is conducted to check for data-entry errors and
Outcome of interest
The endpoint was all-cause revision and aseptic revision.
Revision was defined as any operation after the index TKA
where a component was replaced. Aseptic revisions were revi-
sions performed for any reason other than infection-related
causes. Reasons for revision were recorded by the surgeons
in the operative forms of the TJRR and confirmed with chart
review by a trained clinical research associate.
Several covariates were investigated as possible confound-
ers or effect modifiers of the association of tibial component
design and the outcomes of interest. Covariates investigated
included: patient characteristics (age, race, gender, BMI, dia-
betes status), primary TKA diagnosis (osteoarthritis, osteone-
crosis, posttraumatic arthritis, or rheumatoid arthritis), ASA
health status score, surgeons’ total joint arthroplasty fellow-
ship training status, surgeon and hospital average annual
volume, implant stability (cruciate retaining vs. posterior sta-
bilized), implant fixation (cemented, hybrid, or uncemented),
patellar resurfacing, operative time, and whether or not the
procedure was bilateral.
Frequencies, proportions, means, standard deviations (SDs),
medians and interquartile ranges (IQRs) were used to describe
the study sample. Comparisons between metal-backed modu-
lar and monoblock all-polyethylene tibial components used
chi-square tests for categorical variables and Student t-tests
for continuous variables. Crude revision rates and revision
rate per 100 years of observation were calculated for all-cause
revision and aseptic revision. Cox proportional hazard models
for multivariable survival data (adjusted for surgeon cluster-
ing using a sandwich covariance matrix estimator) were used
to assess hazard ratios (HRs) and 95% confidence intervals
(CIs) for the type of tibial component and risk of all-cause
and aseptic revision. Proportional hazard assumptions were
evaluated using graphs of survival function against survival
time. A model for the entire cohort was developed, as well as
models for a younger age group (< 65 years old) and an older
one (≥ 65 years old). Covariates were explored as confound-
ers of the associated of tibial component and risk of failure.
Variables not confounding the association between tibial com-
ponent type and risk of revision (changed estimates by > 20%)
were not included in the final models. The final models were
adjusted for age and sex. To account for missing values of
some variables, multiple imputations were performed to create
10 versions of the analytic data set and we then used Rubin’s
combining rules to calculate the final parameter estimates and
CIs from the 10 output sets (Rubin 1987). The imputation
model used included all covariates, event indicator, and Nel-
son-Aalen estimator of the cumulative baseline hazard at the
time of event or censoring for each case (Moons et al. 2006,
White and Royston 2009). The data were analyzed using SAS
software version 9.2 and α = 0.05 was used as the threshold
for statistical significance.
Internal Review Board (IRB # 5488) approval was obtained
before the study was started.
532 Acta Orthopaedica 2013; 84 (6): 530–536
27,657 fixed primary TKAs were included in the study sample.
The cohort had a higher percentage of women (63.7%), and the
mean age was 68.4 (SD 9.1) years. The majority of patients had
a diagnosis of osteoarthritis (97.2%) and were white (62.9%).
Of the total, 2,306 (8.3%) had monoblock all-polyethylene
tibial components and 25,351 (91.7%) had metal-backed
modular components. No monoblock pre-molded composite
tibial components were registered. During the study period,
5.4% (n = 1,501) of the cohort died and 8.2% (n = 2,264) were
followed for a median time of 1.8 (IQR: 0.8–3.3) years before
leaving the health plan and being categorized as lost to follow-
up. Patients who were lost to follow-up were younger than
those in the cohort (10% were < 55 years of age as compared
to 6% of those who remained in the study cohort) but no dif-
ference in sex distribution was observed. 12% (n = 272) of the
2,306 monoblock all-polyethylene tibial components and 8%
(n = 1,992) of the 25,351 metal-backed modular components
were implanted in patients who were lost to follow-up.
The cohort of patients who received monoblock all-poly-
ethylene tibial components had a higher mean age (71.8 vs.
68.1 years, p < 0.001), had a higher proportion of whites
(72.7% vs. 62%, p < 0.001), had lower mean BMI (30.1 vs.
31.6, p 0.001), and had a slightly different distribution of ASA
with metal-back modular components (87.4 vs. 92.7 min, p <
0.001) (Table 2).
The crude all-cause revision rate for the overall cohort
was 2.07%. The cohort was followed for a median time of
2.9 (IQR: 1.2–5.1) years. The rate was lower for monoblock
all-polyethylene components than for metal-backed modular
components (0.95% vs. 2.17%, p < 0.001). The incidence rate
of revision per 100 years of observation for all-cause revision
of the monoblock all-polyethylene components was 0.30, and
for the metal-backed modular components it was 0.65. For
aseptic revisions, this incidence rate was also lower for the
monoblock all-polyethylene cohort (0.18 vs. 0.35) (Table 3).
After adjusting for age and sex (no other variables were
found to be confounders), in the all-cause revision models,
the early risk of revision associated with using a monoblock
all-polyethylene tibial component was 0.51 (CI: 0.33–0.78)
times the risk of revision with a metal-backed modular com-
ponent. In the age-specific analysis, in younger patients (< 65
years old), the all-cause risk of revision for a monoblock
all-polyethylene component was 0.26 (CI: 0.10–0.72) times
lower than that for a metal-backed modular component. In the
older group (≥ 65 years old), the risk of early revision was
also lower with the monoblock all-polyethelene components
(HR = 0.59, CI: 0.35–0.99). In aseptic revision models, mono-
block all-polyethylene components were associated with a
Table 1. Study sample characteristics by component modularity, 2001–2010
Mean age, years
Age category, years
Mean BMI (mean, SD)
1 & 2
n (%) (%) (%) p-value
9.0 9.1 < 0.001
a Missing data: age (n = 3, 0.1%), sex (n = 1, 0.0%).
scores in comparison to the cohort of
patients with metal-backed modular
tibial components (Table 1).
There was a difference in the pro-
portion of cases operated by surgeons
with fellowship training (37.5%
in the monoblock all-polyethylene
cohort and 42.5% in the metal-backed
modular cohort; p < 0.001). A higher
proportion of monoblock all-poly-
ethylene components were used by
surgeons with high annual volumes
(70.4% vs. 60.6%, p < 0.001). All
arthroplasties using a monoblock
all-polyethylene component were
performed in hospitals with high
annual volumes (100% vs. 84.6%, p <
0.001). A higher proportion of mono-
block all-polyethylene components
than metal-backed modular compo-
nents were posterior stabilized con-
structs (77.8% vs. 64.8%, p < 0.001).
In addition, a higher proportion of
metal-backed cases than monob-
lock cases were performed without a
patellar resurface (2.4% vs. 1.2%, p
< 0.001). Cases with monoblock all-
polyethylene tibial components had
a shorter operative time than those
Acta Orthopaedica 2013; 84 (6): 530–536 533
lower risk of early revision than metal-backed modular com-
ponents only in the younger age-specific model (HR = 0.27,
CI: 0.11–0.65) (Table 4).
With 2.9 years of median follow-up, the fixed monoblock
all-polyethylene tibial component was identified as a superior
construct for TKA. For arthroplasties performed with a mono-
block all-polyethylene tibial component, the all-cause revision
rate per 100 years of observation was 0.30. During the same
period, for arthroplasties performed with a metal-backed mod-
ular tibial component, the crude revision rate per 100 years
of observation was 0.65. Based on adjusted models, the use
of a monoblock all-polyethylene component was associated
with a 50% lower risk of early revision. In younger patients
(< 65 years old), the lower risk of early revision associated
with monoblock all-polyethylene components was even more
pronounced, as it was approximately 74% lower than that for
metal-backed modular components.
Most studies have not found any statistically or clinically
significant difference in the risk of revision between arthro-
plasties performed with either a metal-backed tibial compo-
nent or a monoblock all-polyethylene tibial component (Gioe
et al. 2007b, Bettinson et al. 2009, Cheng et al. 2011, Voigt
and Mosier 2011). In a meta-analysis of 9 randomized con-
trolled trials published between 2000 and 2009 using 5 dif-
Table 2. Surgeon, hospital, procedure, and implant characteristics according to component modularity, 2001–2010
Surgeon, average annual volume
Hospital, average yearly volume
Patella not resurfaced
Operative time (min), (mean, SD)
n (%) (%) (%) p-value
37.5 < 0.001
40 40 13
26 24 2
CR: cruciate retaining; PS: posterior stabilized.
Table 3. Crude revision rate and revision rate/100 years of observation for all-cause revision and aseptic revision, for
the overall cohort and according to component modularity
Crude revision rate
Revision rate/100 y
534 Acta Orthopaedica 2013; 84 (6): 530–536
ferent implant systems, Cheng et al. (2011) reported similar
clinical results between the 2 groups in terms of knee scores,
quality of life, range of motion, radiographic implant align-
ment, and postoperative complications. In their analysis,
they did not find that metal-backed tibial components were
superior to the monoblock all-polyethylene tibial construct.
Another meta-analysis of 12 studies and 1,798 implants found
that the lower-cost monoblock all-polyethylene component
had clinical and functional results equivalent to the more
expensive fixed metal-backed modular component (Voigt and
Mosier 2011). Voigt and Mosier (2011) also reported that
there were no statistically significant differences in implant
longevity at 2, 10, and 15 years postoperatively. In a pro-
spective randomized controlled trial, Bettinson et al. (2009)
reported 10-year survivorships of 97.0% for 304 metal-backed
arthroplasties and 96.8% for 262 monoblock all-polyethylene
implants. Gioe et al. (2007b) conducted a randomized study
and reported a 10-year survivorship of 92% for 97 monob-
lock all-polyethylene tibial components with revision for any
reason and 100% for aseptic loosening. In their study, the sur-
vivorship of 70 metal-backed modular components was 89%
with revision for any reason and 94% for aseptic loosening. In
a case-series study, also by Gioe et al. (2007a), the reported
14-year survivorship of the all-polyethylene cases entered
in the institutional registry was 99% comparable with their
clinical trial results. In our cohort, however, we found a sta-
tistically and clinically significant difference between these 2
tibial constructs. There may be several reasons for the differ-
ences between our findings and those published in the litera-
ture. Firstly, the previously published studies may have been
underpowered. Secondly, from a patient and surgeon point of
view, our study sample was diverse, thus allowing us to mea-
sure the real-world performance of these constructs. Finally,
the follow-up periods of published studies varied from 2 to 10
years, whereas our median follow-up was 2.9 years.
Additional reports on the topic include annual reports from
national total joint arthroplasty registries (National Joint Reg-
istry for England and Wales 2010, AOA 2011). In the Eighth
Annual Report of the National Joint Registry for England and
Wales (NJREW), the revision rate at 7 years for all bicondylar
knees (n = 313,069) was 3.9% (National Joint Registry for
England and Wales 2010). In that same time period, 3.7% of the
43,708 monoblock tibias were revised. In the NJREW report,
no distinction was made between monoblock all-polyethylene
tibial components and pre-assembled metal-polyethylene
monoblock tibial components, but a slightly lower 7-year revi-
sion rate was seen for monoblock tibial constructs. In contrast,
the 2012 Report of the Australian Orthopaedic Association
National Joint Replacement Registry (AOANJRR) catego-
rized monoblock tibias into monoblock all-polyethylene com-
ponents and molded non-modular components (AOA 2011).
The incidence rate of revision per 100 years of observation
was highest with the monoblock all-polyethylene component
(0.74) and lowest with the molded non-modular component
(0.58). In comparison, the metal-backed modular revision rate
per 100 years of observation was 0.67. The cumulative revi-
sion rate at 10 years was 5.1% for the meta-backed modular
components, 5.4% for the monoblock all-polyethylene com-
ponents, and 4.7% for the molded non-modular components.
The different lengths of follow-up in the different cohorts and
the implant selection criteria are 2 observations of importance,
and possibly explain the observed differences when compar-
ing our data to those from the AOANJRR cohort.
The limitations of our investigation were related to the study
design, the retrospective nature of the study, and follow-up.
However, the outcomes were prospectively ascertained and
adjudicated by a trained research coordinator to reduce the
possibility of informational bias. It is possible that selection
bias may have existed in our sample. In our analysis, we tried
to adjust for the variables we found to be confounders of the
association between implant choice and risk of revision. How-
ever, there may still be certain patient-specific or implant-
based characteristics that we have not been able to account
for that may have influenced surgical and clinical decision-
making. The definition of failure was surgical revision, which
does not account for patient function and satisfaction as a
measure of success of arthroplasty. Radiographically failing
arthroplasties were not accounted for in our study. However,
we have no reason to believe that there would be a differen-
tial rate of radiographically defined failures between the study
groups, and if there was under-ascertainment of failures, this
was probably non-differential.
The purposeful restriction of our analysis to a single implant
(PFC Sigma) manufactured by the same company (DePuy)
with the same articular geometry in both the metal-backed
modular component and the monoblock all-polyethylene
component allowed us to focus on the role of tibial modularity
in arthroplasty survival. This strategy increased the internal
validity of our measurements as they applied to Depuy com-
ponents, but our results may not be generalizable to monob-
lock all-polyethylene components from other manufacturers
with possibly different design features.
Our loss to follow-up, another limitation, was slightly higher
in the monoblock all-polyethylene group (12%) than in the
Table 4. Adjusted risk of all-cause and aseptic revision for all-poly-
ethylene monoblocks compared to metal-backed modular tibial
components. Cox proportional hazard models
HR 95% CI p-value
All-cause revision models
Patients ≥ 65 years old
Patients < 65 years old
Aseptic revision models
Patients ≥ 65 years old
Patients < 65 years old
Acta Orthopaedica 2013; 84 (6): 530–536 535
metal-backed modular group (8%). It is unlikely that such a
small difference between the cohorts would cause our estima-
tions to be biased to the degree presented here. In addition, the
“lost to follow-up” cohort contributed a median of 1.8 years
of observation before attrition, thereby contributing to a sub-
stantial amount of the follow-up period (in which they did not
have a revision). Those lost to follow-up were also younger
than the patients who were not lost to follow-up, supporting
the idea that if the effects we are seeing are even greater in a
younger population, they must not be overestimated since that
would be the group most likely to miss event ascertainment. In
addition, these are short-term follow-up risk estimations and
the results should be interpreted as such. Our conclusions are
fit for the follow-up time of our cohort, and we hope that once
longer follow-up is available in this study cohort, we will be
able to re-evaluate this patient sample.
The strengths of the present study include its large sample
size, the diversity of the patient and surgeon sample included,
and the internal validity of the TJRR used for the study. Our
sample size allowed us not only to evaluate the relationship
between implant modularity and outcomes of TKA, but also
to evaluate models stratified by age and adjust our analysis for
possible confounders of the relationship studied. In addition,
the diversity of the sample included in our study (inclusive
of non-Medicare aged groups and inclusive of various racial
groups) increases the generalizability of our findings to vari-
ous patient populations. Similarly, the number of surgeons
and medical centers that contributed to this study, with sev-
eral levels of skill, volume, and training, as well as different
medical center characteristics increase the external validity
of our findings, which we believe are applicable to a range
of surgeons and medical centers. Finally, the TJRR collects
prospective information on all the registered TKA cases and
uses validated algorithms to ascertain the events evaluated in
this study. It also adjudicates every outcome via chart review.
This mechanism of event ascertainment together with the
integrated linkage to healthcare systems and access to patient
activity increases the internal validity of the information we
are reporting, as information bias is probably minimized.
An analysis of primary TKAs registered in a community-
based TJRR showed that in monoblock all-polyethylene tibial
components, the risk of revision was approximately 49% (CI:
22–67%) lower in the 2.9 years of median follow-up of our
cohort than in patients with metal-backed modular compo-
nents. For younger patients (< 65 years old), the risk of revi-
sion was even lower for the monoblock all-polyethylene com-
ponent, where the hazard ratio was approximately 0.3 (CI:
0.1–0.7) times that for the metal-backed modular component.
Study concept and design: VM, MCSI, and RSN. Extraction of data and
preparation of raw data: MCSI. Statistical analysis: MCSI. Interpretation of
data: VM, MCSI, RSN, DS, and EWP. Drafting of the text: VM and MCSI.
Drafting of th tables: MCSI. Critical revision of the manuscript for important
intellectual content: VM, MCSI, RSN, DS, and EWP.
We thank all Kaiser Permanente orthopedic surgeons and the staff of the
Department of Surgical Outcomes and Analysis who have contributed to the
success of the National Total Joint Replacement Registry.
No competing interests declared.
AOA. Australian Orthopedic Association. Hip and Knee Arthroplasty.
National Joint Replacement Registry. 2010 Annual Report.) Adelaide, Uni-
versity of Adelaide 2011.
Bae D K, Song S J, Park M J, Eoh J H, Song J H, Park C H. Twenty-year sur-
vival analysis in total knee arthroplasty by a single surgeon. J Arthroplasty
2012; 27: 1297-304 e1.
Bettinson K A, Pinder I M, Moran C G, Weir D J, Lingard E A. All-polyeth-
ylene compared with metal-backed tibial components in total knee arthro-
plasty at ten years. A prospective, randomized controlled trial. J Bone Joint
Surg (Am) 2009; 91: 1587-94.
Cheng T, Zhang G, Zhang X. Metal-backed versus all-polyethylene tibial
components in primary total knee arthroplasty. Acta Orthop 2011; 82: 589-
Gioe T J, Sinner P, Mehle S, Ma W, Killeen K K. Excellent survival of all-
polyethylene tibial components in a community joint registry. Clin Orthop
2007a; (464): 88-92.
Gioe T J, Stroemer E S, Santos E R. All-polyethylene and metal-backed tibias
have similar outcomes at 10 years: a randomized level I (corrected) evi-
dence study. Clin Orthop 2007b; (455): 212-8.
Hyldahl H, Regner L, Carlsson L, Karrholm J, Weidenhielm L. All-polyethyl-
ene vs. metal-backed tibial component in total knee arthroplasty-a random-
ized RSA study comparing early fixation of horizontally and completely
cemented tibial components: part 2. Completely cemented components:
MB not superior to AP components. Acta Orthop 2005; 76: 778-84.
Kurtz S M, Lau E, Ong K, Zhao K, Kelly M, Bozic K J. Future young patient
demand for primary and revision joint replacement: national projections
from 2010 to 2030. Clin Orthop 2009; (467): 2606-12.
Manley M, Ong K, Lau E,& Kurtz S M. Effect of volume on total hip arthro-
plasty revision rates in the United States Medicare population. J Bone Joint
Surg (Am) 2008; 90: 2446-51.
Moons K G, Donders R A, Stijnen T, Harrell F E, Jr. Using the outcome for
imputation of missing predictor values was preferred. J Clin Epidemiol
2006; 59: 1092-101.
National Joint Registry for England and Wales. 7th Annual Report 2010.
Norgren B, Dalen T, Nilsson K G. All-poly tibial component better than
metal-backed: a randomized RSA study. Knee 2004; 11: 189-96.
Nouta K A, Pijls B G, Nelissen R G. All-polyethylene tibial components in
TKA in rheumatoid arthritis: a 25-year follow-up study. Int Orthop 2012a;
Nouta K A, Verra W C, Pijls B G, Schoones J W, Nelissen R G. All-polyethyl-
ene tibial components are equal to metal-backed components: Systematic
review and meta-regression. Clin Orthop 2012b; (470): 3549-59.
Parks N L, Engh G A, Topoleski L D, Emperado J. The Coventry Award.
Modular tibial insert micromotion. A concern with contemporary knee
implants. Clin Orthop 1998; : 10-5.
Paxton E, Inacio M, Slipchenko T, Fithian D. The Kaiser Permanente National
Total Joint Replacement Registry. Perm J Summer 2008; 12: 12-6.
Paxton E, Namba R, Maletis G, Khatod M, Yue E, Davies M, et al. A prospec-
tive study of 80,000 total joint and 5,000 anterior cruciate ligament recon-
struction procedures in a community-based registry in the United States. J
Bone Joint Surg (Am) (Suppl 2) 2010a; 92: 117-32.
536 Acta Orthopaedica 2013; 84 (6): 530–536
Paxton E W, Inacio M C, Khatod M, Yue E J, Namba R S. Kaiser Permanente
National Total Joint Replacement Registry: aligning operations with infor-
mation technology. Clin Orthop 2010b; (468): 2646-63.
Paxton E W, Inacio M C, Kiley M L. The Kaiser Permanente Implant Regis-
tries: Effect on patient safety, quality improvement, cost effectiveness, and
research opportunities. Perm J 2012; 16 (2): 33-40.
Rodriguez J A, Bhende H, Ranawat C S. Total condylar knee replacement: a
20-year followup study. Clin Orthop 2001; :10-7.
Rubin D B. Multiple imputation for nonresponse in surveys, New York, Wiley.
Vessely M B, Whaley A L, Harmsen W S, Schleck C D, Berry D J. The Chi-
tranjan Ranawat Award: Long-term survivorship and failure modes of 1000
cemented condylar total knee arthroplasties. Clin Orthop 2006; (452):
Voigt J, Mosier M. Cemented all-polyethylene and metal-backed polyethyl-
ene tibial components used for primary total knee arthroplasty: a system-
atic review of the literature and meta-analysis of randomized controlled
trials involving 1798 primary total knee implants. J Bone Joint Surg (Am)
2011; 93: 1790-8.
Wasielewski R C, Parks N, Williams I, Surprenant H, Collier J P, Engh G.
Tibial insert undersurface as a contributing source of polyethylene wear
debris. Clin Orthop 1997; : 53-9.
White I R, Royston P. Imputing missing covariate values for the Cox model.
Stat Med 2009; 28: 1982-98.